**** BEGIN LOGGING AT Thu Sep 05 02:59:59 2013 Sep 05 06:17:48 Devastator: your patch gets deleted if you didn't compile it Sep 05 06:18:58 Devastator: run "make target/linux/{clean,refresh}" Sep 05 06:19:10 Devastator: after that build entire image Sep 05 07:00:39 Devastator: /me keeps this open in a browser tab: http://wiki.openwrt.org/doc/devel/patches Sep 05 09:45:03 nbd r37896 branches/attitude_adjustment/package/busybox/config/coreutils/Config.in * busybox: enable readlink by default, it is used by various scripts Sep 05 10:09:40 kaloz r37897 trunk/target/linux/generic/config-3.10 * Add missing symbol Sep 05 11:10:58 luka thanks! Sep 05 11:11:15 russell-- me too, but some things are still a mistery to me Sep 05 11:43:58 nothing like some verbosity: http://codepad.org/46ieWlEw Sep 05 11:44:08 now I have to figure why.. Sep 05 11:46:13 maybe it fails because I've patched latest kernel and ramips uses 3.10.4?! Sep 05 12:04:19 Devastator: your patches need to be compile time and run time tested, it's not enough if the look nice and we are able to apply them Sep 05 12:04:45 Devastator: and fyi ramips uses 3.10.10 Sep 05 12:26:23 luka my branch might be old compared to master, I will ask in #git how to put master and my branch in sync Sep 05 12:26:40 maybe then it will apply Sep 05 12:27:07 Greetings. Sep 05 12:27:21 cyrusff: Hello. Sep 05 12:29:17 cyrusff: I've been told to ask you about 6relayd: would it be possible to use it on plain linux or does it rely on something specific to openwrt? Sep 05 13:10:15 cousin_luigi: go ahead, there are no dependencies Sep 05 13:10:45 cousin_luigi: just follow the README and see 6relayd -h for options Sep 05 13:11:45 cousin_luigi: feel free to ask if anything is unclear Sep 05 13:49:17 cyrusff: Thanks. Sep 05 13:59:30 cyrusff: My network is made of three segments, I've been trying to replicate the same setup I have on IPv4 on IPv6: three /64 ULA subnets (with a few static leases) and a single public /64. I've been trying to subnet it in to three /68s, but I couldn't make it work. Someone suggested me NDP proxying as alternative, but I don't know enough of this to appreciate the difference. Do you think 6relayd could solve my problem either ( Sep 05 14:00:02 well you can do NDP-proxying Sep 05 14:00:10 but thats only a last resort Sep 05 14:00:18 if you can try to get a bigger public prefix Sep 05 14:00:35 *can't Sep 05 14:00:48 well you know what i mean Sep 05 14:00:49 my ISP won't give me more than a /64 for the time being and I'm not comfortable with third-party tunnels. Sep 05 14:00:54 ok Sep 05 14:01:01 cyrusff: do you think subnetting is a big no-no ? Sep 05 14:01:20 I know I'd lose SLAAC, but since I'd be already using DHCPv6... Sep 05 14:01:32 well it won't work very well Sep 05 14:01:39 how so? Sep 05 14:01:41 hi Sep 05 14:01:42 routing Sep 05 14:02:02 is there a standard way except the sys interface to access the leds on different routers? Sep 05 14:02:09 well Sep 05 14:02:11 luka I appreciate your help, but I give up, this patch seems to apply cleanly to latest kernel, but I can't for the life of me, make it apply to openwrt.. Sep 05 14:02:23 cyrusff: could you give me an example? Sep 05 14:03:19 the problem is that dhcpv6 cannot be used to distribute routing information or even a subnet mask Sep 05 14:04:08 so its hard to tell your clients which clients are on-link Sep 05 14:04:17 and can be reached without a router Sep 05 14:04:54 if you give out addresses via dhcpv6 they are always given out with a /128 prefix, that means the clients will send everything throught the router Sep 05 14:05:13 even if they are communicating with devices in the same network Sep 05 14:05:21 cyrusff: My idea was to use ULA addresses for clients to communicate with each other. Sep 05 14:05:33 ok Sep 05 14:05:37 not sure if that would make a difference in that regard Sep 05 14:05:53 well yes that would help Sep 05 14:06:37 however you would still have the problem that e.g. android-devices wouldn't get an ipv6 address as they do only support slaac Sep 05 14:07:28 android and google are the devil for me, so they're unlikely to join my network anytime soon:) Sep 05 14:07:35 lol Sep 05 14:07:48 haha, have fun searching the web then ;) anyway Sep 05 14:08:07 you have 3 options Sep 05 14:08:11 more or less Sep 05 14:08:40 anyway I've been writing scripts for the last 3 days to find a way to automate this and couldn't, so I wondered if I've been using the wrong tools. Sep 05 14:09:02 well you could bug your ISP, try to setup dhcpv6 with a smaller public Sep 05 14:09:18 which could be painful if your isps prefix changes regularly Sep 05 14:09:21 ipv6 is still experimental, so it's this way or the highway Sep 05 14:09:22 or try relaying Sep 05 14:10:01 how many routers do you have or how does your topology look? Sep 05 14:10:17 and where does the publc /64 come from? Sep 05 14:10:51 one router with three LAN ports and one WAN port connecting via ppp. Sep 05 14:10:58 ok Sep 05 14:11:57 and the /64 is on the wan-side or is it a delegated prefix? Sep 05 14:12:25 I've tried using isc-dhcp-server, but for some reason I could make it work only with linux clients. Dibbler on the other hand, works, but I can't seem to have clients assigned both an ULA and a public IP. Sep 05 14:12:29 cyrusff: delegated Sep 05 14:13:22 6relayd doesn't support pools >/64 yet so there is that :P Sep 05 14:13:30 :D Sep 05 14:13:50 you could use the dirty method for now and assign the same /64 to each of your 3 networks Sep 05 14:14:03 in a way Sep 05 14:14:04 and hope there won't be a collision? Sep 05 14:14:21 I know it's unlikely, but it feels dirty Sep 05 14:15:10 are you saying 6relayd wouldn't work with a /56? Sep 05 14:15:30 well i mean it doesn't work with 65-128 Sep 05 14:15:38 ah Sep 05 14:16:25 So that goes out of the window. Could it do NDP proxying à la ndppd? Sep 05 14:16:38 yes the problem is that it might not work Sep 05 14:16:48 you can try to set it up on the router Sep 05 14:17:16 but it would use the wan /64 not the delegated one and your ISP might simply just filter your traffic away Sep 05 14:17:56 :\ Sep 05 14:18:55 i mean you can try to set it up using the delegated prefix Sep 05 14:19:23 but you would end up assigning the same /64 to every interface and just proxy ndp between them Sep 05 14:19:44 What are the drawbacks of that? Sep 05 14:20:00 well you have the same /64 on every interface Sep 05 14:20:21 but collisions should be taken care of as ndp is proxied Sep 05 14:20:25 its not a clean network setup :) Sep 05 14:20:27 ok Sep 05 14:20:44 but then pushing a single /64 to a customer is not a clean ipv6 support either Sep 05 14:20:50 yeah Sep 05 14:21:04 I understand it works at L2: would that introduce security concerns? Sep 05 14:21:17 I mean, how much is it comparable to a bridge? Sep 05 14:21:29 well only certain icmp packages are bridged Sep 05 14:21:31 you can still filter based on interfaces Sep 05 14:21:39 its not bridged through below netfilter Sep 05 14:21:43 yes Sep 05 14:21:59 you can do arp-spoofing Sep 05 14:21:59 just classifying traffic based on ip ranges becomes impossible Sep 05 14:22:03 kind of Sep 05 14:22:13 since its the same range in every direction Sep 05 14:22:28 I see. Sep 05 14:22:32 I could live with that. Sep 05 14:22:58 if you have a malicious device in one subnet it could prevent all other devices from getting an ipv6-address Sep 05 14:23:09 by just responding to the duplicate address detection Sep 05 14:23:20 I see. Sep 05 14:23:33 thats what you would have to live with Sep 05 14:23:53 But speaking of DHCPv6 again, does the standard allow for multiple addresses per client? Sep 05 14:23:58 yes Sep 05 14:24:05 is there a way to apply a patch in patches-3.10 manually? I'm trying to understand why I have "Hunk #1 FAILED" Sep 05 14:24:33 cyrusff: Is it the server that pushes them or the client who has to ask for more? Sep 05 14:24:49 the server pushes them Sep 05 14:25:20 That's good news. Sep 05 14:25:43 but the client might ignore them if its badly designed Sep 05 14:26:13 expect difficulties when it comes to apple devices Sep 05 14:26:31 Do you happen to know if networkmanager belongs to the same category? Sep 05 14:27:04 no linux and windows are in general fine Sep 05 14:29:09 maybe it's dibbler then, or the patches in the kernel I'm using. Sep 05 14:29:38 the kernel has nothing to do with dhcpv6 Sep 05 14:29:56 you should isc dhcp client if in doubt Sep 05 14:29:57 true that Sep 05 14:30:53 I still wonder why isc-dhcp-server won't play well with windows (nor symbian) Sep 05 14:34:31 cyrusff: So, to sum things up, can I use 6relayd to handle ULA ranges (with user-defined leases), and NDP proxying the public /64 across my three interfaces? Sep 05 14:34:44 yes Sep 05 14:36:11 Are there other configuration samples to see? Sep 05 14:36:21 6relayd -Nr -Dserver -Rserver -m0 . eth1 eth2 eth3 Sep 05 14:36:29 that would be your command line Sep 05 14:36:32 out of interest; where is 6relayd drawing its ranges from on non-openwrt? Sep 05 14:36:47 simply read from the interface routes? Sep 05 14:36:49 same as on openwrt, it queries netlink Sep 05 14:36:58 and reads /proc/.../ipv6_routes Sep 05 14:37:09 ah Sep 05 14:37:10 or whatever that proc file was called Sep 05 14:37:29 i'll add a ubus adapter soon though Sep 05 14:38:21 cyrusff: oh thanks! Sep 05 14:38:26 cousin_luigi: not sure if with or without the dot Sep 05 14:38:33 cyrusff: but can I also set my own leases? Sep 05 14:38:34 before eth1 Sep 05 14:39:04 well in a way Sep 05 14:41:15 set -m1 and then use the -a 001122334455667788:1234 Sep 05 14:41:30 where the part before the : is the DUID and the part after is the address suffix Sep 05 14:41:50 e.g. if your prefix is 2001:db80::/64 then it would be 2001:db80::1234/64 in that case Sep 05 14:42:16 ok Sep 05 14:42:29 can I differentiate between ULA and non-ULA in that regard? Sep 05 14:42:34 no Sep 05 14:42:37 :'( Sep 05 14:43:42 ok, you've been most helpful, I will make some experiment right away Sep 05 14:43:45 bye! Sep 05 14:44:32 bye Sep 05 15:28:28 how do I checkout specific git branch within OpenWrt Makefile? Sep 05 15:31:52 does PKG_SOURCE_VERSION variable specify it? Sep 05 15:37:07 I really don't understand why this patch doesn't apply.. Sep 05 15:38:27 does anybody wants to take a look? Sep 05 15:46:43 build #340 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/340 Sep 05 15:48:43 build #340 of ppc44x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/340 Sep 05 16:05:53 Devastator: that makes me want to help you in the future too... Sep 05 16:40:32 build #353 of uml is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/353 Sep 05 16:44:18 Devastator: (assuming git) when i have a separate branch, i first make sure to commit everything to my local branch, checkout master, pull, then checkout my branch again, and say "git merge master" (which will prompt me for a commit message), then say "git rebase master". this is just local cargo cult understanding, though it seems to work for me most of the time, my local branch then becomes master + my local patches. Sep 05 16:46:21 e.g. i have been carrying around my local libtool-2.4.2 patches that way for months ;-) Sep 05 16:48:32 git checkout master ; git pull ; git checkout update-libtool ; git merge master ; git rebase master Sep 05 16:48:45 (where update-libtool is my local branch) Sep 05 17:06:56 build #372 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/372 Sep 05 17:26:01 russell-- thanks! that helps! Sep 05 17:36:19 bbl Sep 05 18:17:43 build #321 of sibyte is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/321 Sep 05 19:35:41 https://forum.openwrt.org/viewtopic.php?id=45820 any news on this? Sep 05 19:35:57 hauke r37898 trunk/package/kernel/mac80211/Makefile * ath10k: enable Atheros 802.11ac driver Sep 05 19:40:36 hauke r37899 trunk/package/kernel/linux/modules/usb.mk * kernel: kmod-usb-serial-keyspan: fix dependencies Sep 05 19:41:41 hauke r37900 trunk/package/kernel/linux/modules/fs.mk * kernel: kmod-fs-xfs: fix dependencies Sep 05 19:44:20 hauke r37901 trunk/package/kernel/mac80211/Makefile Sep 05 19:44:20 mac80211: do not download carl9170 firmware separately, but use the one from linux-firemware Sep 05 19:54:03 hauke r37902 trunk/package/kernel/linux/modules/firewire.mk * kernel: kmod-firewire: fix dependencies Sep 05 19:56:45 hauke r37903 trunk/package/kernel/linux/modules/other.mk * kernel: kmod-ssb, kmod-b44: fix load order Sep 05 19:58:25 hauke r37904 trunk/package/kernel/linux/modules/netdevices.mk * kernel: kmod-libphy, kmod-tg3: fix load order Sep 05 20:11:13 yep, I think I've found the problem Sep 05 20:18:10 hauke r37905 trunk/package/kernel/mac80211/patches/300-pending_work.patch * mac80211: kmod-rtl_pci: fix dependencies Sep 05 20:31:16 jogo r37906 trunk/target/linux/ generic/files/drivers/net/phy/b53/b53_mdio.c generic/files/drivers/net/phy/b53/b53_phy_fixup.c * kernel: b53: support phy ids for BCM5365 Sep 05 20:37:17 i am in need for some source routing :). As i see it can be done using the normal tools I would use on my server. But i'd like to configure it using luci Sep 05 20:37:33 is there any chance, that this might be possible? Sep 05 20:41:35 mkkl: (no idea about luci) it might help to describe your source routing scenario Sep 05 20:43:10 russell--: ^ Sep 05 20:44:00 KanjiMonster: ah, sweet, i was just noticing it hadn't been applied yet ... and now it has Sep 05 20:52:42 russell--, high sorry in meant uci Sep 05 20:53:26 i have two wireless lans one should be routed via vpn the other directly via the "standard" uplink Sep 05 20:58:16 blogic: have you made any progress on the vgv7519 yet? Sep 05 21:00:00 it only arrived today Sep 05 21:01:12 ok Sep 05 21:01:13 can you send me a full snapshot of what you have so far ? Sep 05 21:01:13 snapshot? as in code? Sep 05 21:01:13 the lantiq target is moving to 3.10, right? Sep 05 21:04:16 The_Lizard: yes Sep 05 21:06:17 I already mailed my brn_patches... Sep 05 21:06:36 well mail everything you have Sep 05 21:06:37 @23-8 Sep 05 21:06:38 unless that is all Sep 05 21:06:47 i thought you havd a nor flash atch aswell Sep 05 21:07:06 that is the configuration that I used in combination with the original bootloader Sep 05 21:07:12 mkkl: have you considered mwan3 or similar? https://forum.openwrt.org/viewtopic.php?id=39052 Sep 05 21:08:14 when you have the bootloader extracted, can I have a copy of it? to check if you have the same version? Sep 05 21:08:26 if so, I will return my patched edition if you like Sep 05 21:08:46 or if it is different, I will patch out the code checks for you Sep 05 21:10:19 russell--, no not yet, but might be just what I was looking for Sep 05 21:12:43 I have been trying to get JTAG to work on my other board, the chain is 'working' but it doesn't detect the device correctly, openocd tells me it is returning all zeroes Sep 05 21:14:32 cyrus r37907 trunk/package/network/ipv6/odhcp6c/Makefile * odhcp6c: more compatibility fixes Sep 05 21:14:36 anyone has any idea? Sep 05 21:14:57 the altera software reported unknown_jtag_id Sep 05 21:15:54 cyrus r37908 branches/attitude_adjustment/package/odhcp6c/Makefile * AA: odhcp6c: more compatibility fixes Sep 05 22:43:12 heil r37909 packages/net/haproxy/ (8 files in 2 dirs) Sep 05 22:43:13 package: haproxy Sep 05 22:43:13 - add backported patches from 1.5dev Sep 06 00:52:57 build #282 of ep93xx is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/282 Sep 06 01:13:37 build #331 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/331 Sep 06 02:22:22 luka you are probably not there anymore, but after some digging I've found the problem and now the patch is refreshed, so basically I "git commit", then "git format-patch -s --stdout master > ~/mypatch.patch" and "git send-email mypatch.patch", right? **** ENDING LOGGING AT Fri Sep 06 02:59:59 2013