**** BEGIN LOGGING AT Wed May 28 02:59:59 2014 May 28 03:37:01 <5EXAA8SU8> build #71 of pxa is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/pxa/builds/71 May 28 07:05:42 <5EXAA8SU8> build #524 of ar71xx is complete: Failure [failed shell_15] Build details are at http://buildbot.openwrt.org:8010/builders/ar71xx/builds/524 May 28 07:21:00 <5EXAA8SU8> build #72 of adm8668 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/adm8668/builds/72 May 28 07:29:03 <5EXAA8SU8> build #471 of gemini is complete: Failure [failed shell] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/471 May 28 09:46:30 nbd r40867 trunk/package/libs/uclient/Makefile May 28 09:46:30 uclient: update to the latest version, fixes an issue with http authentication handling May 28 11:32:56 hi all! May 28 11:34:42 nbd who is supposed to populate/use device->settings I am accessing it to get the macaddress but they are all 0 while i see macvlandevice use another config structure to store macaddr May 28 11:37:15 should i call device_init_settings from vlanad_config_init to get it populated ? May 28 11:39:02 mmm it seems i shouldn't at lest in vlanad_config_init May 28 11:39:50 i already call it in vlanad_reload May 28 11:39:54 so it should work May 28 12:48:24 <5EXAA8SU8> build #73 of mpc83xx is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/mpc83xx/builds/73 May 28 16:34:42 ping nbd jow_laptop May 28 16:40:48 anyone can help about netifd development ? May 28 17:40:09 <5EXAA8SU8> build #578 of rb532 is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/578 May 28 17:41:15 <5EXAA8SU8> build #578 of ppc44x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/578 May 28 17:41:29 <5EXAA8SU8> build #547 of sibyte is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/547 May 28 17:44:06 <5EXAA8SU8> build #485 of iop32x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/485 May 28 18:06:23 g10h4ck: pong May 28 18:06:37 i have some questions May 28 18:06:39 first of all May 28 18:07:19 how to make netifd logging D messages like D(DEVICE, "Create new device '%s' (%s)\n", name, type->name); May 28 18:08:05 i have tried pasing -d 0xff to netifd as sugested by jow but i didn't see more debug messages... May 28 18:08:12 it doesn't print them to syslog, it currently outputs them to stderr when the bit is set in the mask provided via -d May 28 18:08:21 it would probably make sense to change this to syslog May 28 18:08:26 and make it configurable via uci May 28 18:08:30 other question May 28 18:08:50 if in a config device i set mtu i see different behaviour May 28 18:08:55 if the type is bridge May 28 18:09:10 the mtu is applied while it is not for macvlan and vlanad May 28 18:09:40 i was looking to the bridge code but didn't find where the mtu is setted so i thinked it was something implemented in device May 28 18:09:48 but i didn't understood how it work... May 28 18:10:50 might be a kernel restriction May 28 18:11:12 but why if i do it with ip link it works ? May 28 18:11:48 or for example specifing mtu in interface section it works May 28 18:11:52 the first thing to check is if the netifd code even tries to apply the mtu May 28 18:11:57 if it's tracked internally May 28 18:12:23 while looking at netifd code it seems something related to device more that to interface (and it make sense) May 28 18:12:53 macvlan specific code doesn't look for mtu (i have checked it) May 28 18:13:15 so if it is tracked somewhere i expect it is in device.c May 28 18:13:26 but didn't found some clue there :( May 28 18:14:15 device_init_settings should track it internally May 28 18:15:25 yeah it seems so May 28 18:16:18 to do some debug i checked if flag is setted in system_vlanad_add May 28 18:16:42 because i was ipotizing that vlanad specific code was in charge of this stuff May 28 18:17:10 but both mtu and macaddr flags are never 1 ever if setted in config May 28 18:17:37 another question i was thinking is May 28 18:17:56 if macaddr is already handled by device.c why it is handled too in macvlan.c ? May 28 18:18:34 i understand macaddr is something important in macvlan but it seems a redundancy to handle it in both (decive.c macvlan.c) May 28 18:22:31 also ifname and type should be already present in device->settings why are they specified another type a macvlan specific settings ? May 28 18:22:35 ping jow_laptop nbd May 28 18:23:26 i am misunderstunding something? May 28 18:23:30 ifname has a different type and is used differently May 28 18:23:54 and for type and macaddr ? May 28 18:23:58 as for type, i don't see that in macvlan.c May 28 18:24:09 sorry it is not there May 28 18:24:27 why macaddr ? May 28 18:24:46 checking May 28 18:25:47 i guess it would be possible to remove the attribute, but the mac address is still stored in macvlan_config for a reason May 28 18:26:14 for grouping all macvlan related config files in one struct May 28 18:26:51 ok May 28 18:27:47 so it is ok to access device->settings, and moreover vlanad code shold not take care of stuff like mtu and macaddr that are already handled by device.c right ? May 28 18:31:37 nbd the strange thing is that i never get the flag setted also if macaddr is present as option in device section https://gitorious.org/netifd/gsoc2014-netifd/source/af2dc2bc0300eb28c6eb0f94d69331f401ce67fa:system-linux.c#L825 May 28 18:31:48 the if at thet line is never true... May 28 18:33:36 and i believe the same is going for mtu and also for macvlan May 28 18:35:48 in macvlan the problem doesn't happen for macaddr because it have specific code doing it, but it happens for mtu May 28 18:44:05 so you should check if the appropriate fields in the tb_* arrays are set on loading the config May 28 18:44:33 and maybe use blobmsg_format_string(config, 1) to dump the data that goes in there May 28 18:44:52 nbd i was suspecting the data disappears because of May 28 18:45:02 ah, i noticed one more thing May 28 18:45:02 sec May 28 18:45:24 https://gitorious.org/netifd/gsoc2014-netifd/source/af2dc2bc0300eb28c6eb0f94d69331f401ce67fa:config.c#L187 May 28 18:45:56 vlanad.c and macvlan.c are missing the reference to the generic device attributes May 28 18:46:01 in the struct uci_blob_param_list May 28 18:46:04 compare with bridge.c May 28 18:46:07 seems only device_type specific param are putted in the blob May 28 18:47:11 so you need to set next and n_next May 28 18:47:24 yeah May 28 18:47:34 i was thinking they should inerithed some way May 28 18:47:37 but i didn't get how May 28 18:47:39 .next = { &device_attr_list }, May 28 18:47:52 should a i fix this for macvlan too ? May 28 18:47:57 yes May 28 18:48:00 there is another problem i noticed May 28 18:48:09 what device name are stuff like May 28 18:48:10 you can submit that as a separate patch before submitting the vlan-ad work May 28 18:48:14 eth0.2 May 28 18:48:28 the compatibility vlan code is called too May 28 18:48:47 at moment i solved using eth0-2 in my config, but what should we do ti fix it really ? May 28 18:49:20 many thanks nbd i was going crazy May 28 18:49:47 btw. i think you should call the vlanad stuff something like vlandev May 28 18:49:50 create it via config device sections May 28 18:49:54 with arbitrary names May 28 18:49:57 specified in the cnofig May 28 18:50:03 and a field for specifying the vlan proto May 28 18:51:09 at moment May 28 18:51:34 the type attribute is used to specify 8021ad or 8021q seems good to you ? May 28 18:52:31 nbd specifing .next it is safe to access device->settings to get info about the device like ifname ? May 28 18:52:59 what the .next does is ensure that the config parser code will also emit the attributes present in the generic device attribute list May 28 18:55:01 so device->settings will be populated correctly May 28 18:55:46 yes May 28 18:55:58 the config parser code only propagates config values that are specified in the attribute list May 28 18:56:09 cool May 28 18:58:46 I don't suppose anyone's seen anything like this, but with a usb A female socket? (ie, like a phone charger brick, just din rail mounted) http://www.aliexpress.com/item/20W-24V-MINI-Din-Rail-Single-Output-Switching-power-supply/761270723.html May 28 18:59:33 <5EXAA8SU8> build #513 of au1000 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/au1000/builds/513 May 28 19:00:45 nbd what is the copyright policy ? May 28 19:01:06 keep the same license that is in the code already May 28 19:01:13 <5EXAA8SU8> build #618 of cobalt is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/618 May 28 19:01:16 i have participated il projects they add copyright line also for just for one line contribution May 28 19:01:28 is this the policy for netifd too ? May 28 19:01:49 i don't like adding copyright lines for small contributions May 28 19:01:51 the question is May 28 19:02:00 ok May 28 19:02:02 especially since contributions are tracked in the git log anyway May 28 19:06:06 <5EXAA8SU8> build #624 of orion is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/624 May 28 19:08:34 <5EXAA8SU8> build #588 of uml is complete: Failure [failed shell compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/588 May 28 19:46:29 <[florian]> :x May 28 19:46:50 nbd r40868 trunk/package/network/ (7 files in 6 dirs) * uqmi: Add support for QMI-based mobile broadband modems May 28 20:27:26 <5EXAA8SU8> build #473 of mcs814x is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mcs814x/builds/473 May 28 20:32:19 <5EXAA8SU8> build #558 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/558 May 28 21:05:16 i managed to get some output from netifd debug in particular i sa this line May 28 21:05:19 system_vlandev_add(850): Error adding vlandev 'eth1-12' over 'eth1': -6 May 28 21:08:58 the strange thinkg is that is happening only for the second vlan builded on top of an interface May 28 21:18:13 it seems [NLE_EXIST] = "Object exists" May 28 21:18:26 maybe i am not passing VID well... May 28 21:23:52 could be May 28 21:24:59 nbd did the patch resend arrived well ? May 28 21:29:44 nbd r40869 trunk/package/system/ubox/Makefile * ubox: update to latest version, fixes kmodloader issues with musl May 28 21:30:10 g10h4ck: yep, thaks May 28 21:30:20 :) May 28 21:30:29 applied May 28 21:35:00 nbd r40870 trunk/toolchain/gcc/patches/4.8-linaro/210-disable_libsanitizer_off_t_check.patch * gcc: disable libsanitzier off_t check to fix musl build May 28 21:36:16 :D May 28 21:45:09 luka r40871 trunk/target/linux/ (5 files in 4 dirs) * kirkwood: add support for Linksys EA4500 May 28 21:45:12 luka r40872 trunk/package/boot/uboot-envtools/ Makefile files/kirkwood * [package] uboot-envtools: add support for kirkwood ea4500 May 28 21:45:13 luka r40873 trunk/target/linux/ kirkwood/profiles/120-pogoplug.mk kirkwood/profiles/110-nas.mk * kirkwood: cosmetic: move POGOE02 profile to different file May 28 22:39:57 libtool is failing compiling May 28 22:40:04 ../libtool: line 1198: libtool: compile: cannot determine name of library object from `': command not found May 28 22:40:05 make[6]: *** [libsysfs_la-sysfs_utils.lo] Error 1 May 28 22:42:01 nbd r40874 trunk/toolchain/gcc/patches/4.8-linaro/860-uclibc_use_eh_frame.patch * gcc: define USE_PT_GNU_EH_FRAME for musl as well, reduces binary object size May 28 23:20:17 nbd r40875 trunk/toolchain/musl/patches/900-iconv_size_hack.patch * musl: add a hack to rip out excessive iconv bloat May 28 23:39:00 good night! **** ENDING LOGGING AT Thu May 29 03:00:00 2014