**** BEGIN LOGGING AT Tue Jun 16 02:59:59 2015 Jun 16 05:04:58 rmilecki r45996 branches/chaos_calmer/package/kernel/linux/modules/usb.mk * kernel: use bcma-hcd.ko on bcm53xx for ohci Jun 16 05:30:54 hm, wonder why babel (duplicate struct ethhdr) and procps (PATH_MAX missing) do not compile me with the latest trunk anymore; worked fine with Thursday's Jun 16 05:31:11 giess I will do bit more clean rebuild Jun 16 05:46:12 idli: if it matters, this happened about 6-ish hours ago: r45995 trunk/ rules.mk toolchain/Config.in * toolchain: switch to musl by default, except for mips64 Jun 16 05:46:22 nods, I suspect it's that Jun 16 05:46:32 my clean rebuild had same issues, now running make -k to get list of things that broke Jun 16 05:50:30 seems to be only those two packages in my case; guess I will go out of musl until they get fixed Jun 16 05:50:50 IIRC musl uses non-system headers and at least Babel compilation failure looks like it has both musl and Linux include, with conflicting definitions = boom Jun 16 05:51:13 you could post a ticket with logs; it's more likely to be fixed that way Jun 16 05:56:41 fair enough - added minimal https://dev.openwrt.org/ticket/19874 Jun 16 05:57:30 I wonder why C library devs are so keen to embed Linux kernel includes - it is recipe for disaster as those structs do change over time Jun 16 05:58:59 ah, GPLv2 license poisoning is the reason Jun 16 06:06:21 yep, HEAD^ worked fine - guess no musl for me yet Jun 16 06:20:06 google autocomplete search sucks Jun 16 06:20:30 "babeld musl" -> muslims rebuild tower of babel Jun 16 06:20:50 lmao Jun 16 08:15:42 build #31 of mpc83xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc83xx/builds/31 Jun 16 08:29:23 rmilecki r45997 trunk/target/linux/ (8 files) * bcm53xx: use USB patches sent upstream by Hauke Jun 16 08:32:21 rmilecki r45998 branches/ (8 files) * bcm53xx: use USB patches sent upstream by Hauke Jun 16 08:38:57 build #31 of ep93xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/31 Jun 16 09:55:02 idli: because e.g. signal, syscall and error numbers differ from arch to arch, and anything exposed to userspace/libc is considered stable ABI and isn't supposed to ever change except through new things Jun 16 10:51:22 nbd r45999 trunk/package/system/opkg/Makefile * opkg: fix duplicate check_signature line for smime variant (#19863) Jun 16 10:51:42 nbd r46000 branches/chaos_calmer/package/system/opkg/Makefile * opkg: fix duplicate check_signature line for smime variant (#19863) Jun 16 11:55:29 nbd r46001 trunk/toolchain/ (5 files in 2 dirs) * musl: update to version 1.1.10+git Jun 16 12:03:55 build #31 of mpc52xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/mpc52xx/builds/31 Jun 16 12:58:23 are we really goign to just wholesale start patching packages? are we sending any of tehse patches upstream at all? Jun 16 13:38:42 nbd r46002 branches/ chaos_calmer/package/base-files/files/etc/opkg/keys/de98a2dd1d0f8a07 chaos_calmer/package/base-files/files/etc/opkg/keys/af22f7a88858c8e9 * base-files: replace snapshot key with 15.05 release key Jun 16 14:25:13 build #33 of lantiq.xrx200 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/lantiq.xrx200/builds/33 Jun 16 15:56:49 nbd: what is responsible for the permissions of /dev/shm? they are 0755 when they should be 1777 Jun 16 17:28:04 cyrus r46003 trunk/config/Config-build.in * hardening: enable format security checking by default Jun 16 17:28:09 cyrus r46004 trunk/config/Config-build.in * hardening: enable RELRO by default Jun 16 17:28:17 cyrus r46005 trunk/package/libs/openssl/Makefile * openssl: disable parallel builds (spurious linking break) Jun 16 18:20:47 What populates /dev, anyway? Jun 16 18:44:48 procd started process Jun 16 18:44:55 somethingplugd, perhaps Jun 16 18:45:06 (the code is within procd repo IIRC) Jun 16 18:46:10 Hmm i'll have to read that Jun 16 18:47:02 under plug/ in the procd git repo Jun 16 18:48:02 Github? Jun 16 18:54:23 build #31 of adm5120 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/31 Jun 16 19:14:40 goddamn it Jun 16 19:14:49 ping nbd Jun 16 19:16:52 mkdir("/dev/shm", 0755); in both plug/coldplug.c and initd/early.c need to be changed to 01777 Jun 16 19:18:05 and that also needs to be backported to CC Jun 16 19:18:21 background: https://forum.openwrt.org/viewtopic.php?id=57073 Jun 16 19:52:32 build #33 of cobalt is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/cobalt/builds/33 Jun 16 20:04:09 build #33 of cns21xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/cns21xx/builds/33 Jun 16 20:04:13 build #33 of ppc40x is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ppc40x/builds/33 Jun 16 20:04:21 build #33 of orion is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/33 Jun 16 20:04:46 build #33 of pxa is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/pxa/builds/33 Jun 16 20:06:49 build #32 of avr32 is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/32 Jun 16 21:36:16 hi, I have ubiquite edgemax, builded myself with kernel 3.10.49 , new BB. configuration is very simple - just IP address on lan side, eveyrhting else default. After some time, device set itself 'ethX down'. (can see im dmesg - [ 34.538878] eth1: Link down ) , but NIC is definitelly connected. mii-tool is showing it as OK, but I cannot fix it anyway ( from serial console). ifconfig up, ip link up - nothing is helping. Jun 16 21:36:16 eth is all the time in 'down' state. can someone suggest some reason and/or solution please ?? Jun 16 21:37:30 sounds like a kernel bug Jun 16 21:37:42 btw CC is now a release candidate Jun 16 21:38:04 BB is not new Jun 16 21:39:28 DonkeyHotei: I had to use BB, becasue I have some custom kernel driver, had to port it from 3.0.80. 3.10.49 from BB was (quite) OK. Im not sure how newer kernel from CC.. Jun 16 21:41:36 DonkeyHotei so you are suggesting that this definitely is not openwrt fault itself? I suspected some daemon is breaking this, I saw (new to me) netifd is doing something there. Jun 16 21:43:33 matteo r46006 trunk/package/network/config/ltq-vdsl-app/patches/101-musl.patch * ltq-vdsl-app: build fix for MUSL Jun 16 21:44:05 iirc the edgemax is an octeon target Jun 16 21:44:23 ys Jun 16 21:44:25 yes Jun 16 21:44:43 build #33 of x86.64 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/x86.64/builds/33 Jun 16 21:45:01 octeon ethernet is a kernel driver afaik Jun 16 21:45:20 yes Jun 16 21:46:49 i could be wrong but i don't believe userspace could fake the lack of a link Jun 16 21:52:13 Im not sure, but as I understand (I'v read a little yet) , about netifd , its using netlink - talking directly with kernel about networking stuff. I tought *maybe* something is forcing kernel to not allow device to be in 'up' state ? Jun 16 21:55:37 i haven't read the netifd code so i can't say for sure Jun 16 21:56:26 yeah, good suggestion, read the code ;) Jun 16 21:57:07 I just went here to IRC, hopping to get faster answers to my doubts ;) Jun 16 21:58:18 voyo, did you tried the bb or cc openwrt from download.openwrt.org yet? Jun 16 22:00:12 rundfreifunk: hm, good suggestion. Actually I tried "orginal" BB sometime ago, just to check if it is working at all with my device. but I didnt made longer stability tests. I went very quickly to re-build it from source myself.. Indeed - I will check orginal bb yet. Jun 16 22:00:31 also cc Jun 16 22:01:28 best would be to check with official cc-rc2.. then you could help us with finding current problems Jun 16 22:02:33 rundfreifunk : heh, I need something stable, not more current problems ;) Jun 16 22:02:50 but ok, I'll check cc-rc2 too, just for curiosity Jun 16 22:04:09 build #32 of x86.kvm_guest is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/x86.kvm_guest/builds/32 Jun 16 22:10:18 build #32 of x86 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/x86/builds/32 Jun 16 23:28:51 DonkeyHotei , rundfreifunk - my issue is related to BB firmware. I tried 'orginal one ' BB, not modified by me. Its acting same, interfaces are unstable. I suspect its something maybe with switch (ports eth2-eth4 are actually switch).. not sure... Jun 16 23:29:38 i expected that which was why i suggested seeing if cc is different in that regard Jun 16 23:30:18 voyo, do you have edge-router lite or something different? Jun 16 23:30:30 he said edgemax Jun 16 23:30:39 rundfreifunk: edgeroute max (version with POE) Jun 16 23:31:15 HOWEVER - I just booted it in failsafe mode, and configured simple network manually - it works ! Jun 16 23:31:36 :-) Jun 16 23:32:02 feck, can't send break on serial console . ping .. :P Jun 16 23:32:57 in failsafe the interface is not in promiscuous mode required for bridging Jun 16 23:33:22 hm... good to know.. what else is different ? Jun 16 23:33:46 hm, not odhcpd started, and no netifd ... Jun 16 23:34:06 but lets try CC yet... Jun 16 23:35:37 hm... octeon-er or octeon-erlite ? Jun 16 23:36:02 er I guess... Jun 16 23:37:07 no, wrong platform. so erlite then... Jun 16 23:41:18 argh.. reboot command is not working. to be precisely - its not sync'ing Jun 16 23:52:39 sync,reboot, uci commit - all these commands are freezed . openwrt-15.05-rc2-octeon-erlite-squashfs-sysupgrade on EdgeRouter POE (octeon) Jun 17 00:02:43 i'm more interested in what *is* working Jun 17 00:05:47 lol. so bad ? I tought it is RC :-/ Jun 17 00:06:23 not a lot of people use it on octeon Jun 17 00:07:15 I think this is related to https://dev.openwrt.org/ticket/19838 Jun 17 00:10:28 I left it on 'uci commit' , then it crashed kernel after 120s with hung_task_timeout_secs, I got prompt back again, changes were _not_ saved, tried modified with vi , then 'find /' , then sync again - success (no crash), reboot - I got my changes. Jun 17 00:10:36 i don't see how it's related Jun 17 00:11:51 maybe. not savinfg changes - becasue it have problems with syncing data from fs cache to disk/flash. Jun 17 00:14:00 now seems it is working. (uci commit) . strange... seems it was related somehow to sysupgrade process ? it is doing something major yet on 1st boot after sysupgrade ? Jun 17 00:23:55 DonkeyHotei : I have to go, sleep... I can check other things on this device and CC-rc2 tomorrow if developers dont have octeon it for testing. Jun 17 00:25:36 for now as an issue I see this behaviour with syncing after 1st boot after sysupgrade , and unfortunatelly - as with BB - not all interfaces are working. these wich belongs to switch are not usable :( Jun 17 00:27:03 so no difference with regard to your original issue? Jun 17 00:27:15 correct Jun 17 00:27:17 ;-( Jun 17 00:28:02 question yet - I couldnt sysupgrade to 'er' version, have to use 'erlite' , was it correct ? Jun 17 00:28:40 then i recommend asking on the dev mailing list about the interface down issue Jun 17 00:35:36 tcpdump - segmentation fault, on all interfaces. Jun 17 00:35:53 swconfig list - empty, not detecting anything (not sure if it should ? ) Jun 17 00:36:16 the broken tcpdump on octeon is a compiler bug iirc Jun 17 00:36:16 fair enoug.. sleep time. tomorrow willl try to ask on mailing list... Jun 17 00:36:28 oh hi jow Jun 17 00:48:04 jow_laptop: who has git push access to procd? **** ENDING LOGGING AT Wed Jun 17 02:59:59 2015