**** BEGIN LOGGING AT Thu Nov 28 02:59:58 2013 Nov 28 03:48:55 build #456 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/456 Nov 28 04:56:15 build #388 of sibyte is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/sibyte/builds/388 Nov 28 07:04:45 build #398 of avr32 is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/avr32/builds/398 Nov 28 07:52:04 build #18 of mxs is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mxs/builds/18 Nov 28 08:53:21 build #336 of iop32x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/iop32x/builds/336 Nov 28 09:28:47 build #432 of uml is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/uml/builds/432 Nov 28 10:17:19 build #19 of mvebu is complete: Failure [failed compile_4] Build details are at http://buildbot.openwrt.org:8010/builders/mvebu/builds/19 Nov 28 11:28:37 jow r38935 trunk/package/system/procd/files/procd.sh * procd: fix processing of datatype specification with spaces Nov 28 11:29:09 gracia Nov 28 11:30:41 jow r38936 trunk/package/utils/busybox/files/sysntpd Nov 28 11:30:41 busybox: do not abort sysntpd init if enable_server is unset, that variable just controls the server part for relaying NTP Nov 28 11:35:50 jow r38937 trunk/package/system/ubox/files/log.init Nov 28 11:35:50 Revert "ubox: the validation code has problems with the "(" char. use string for now" Nov 28 11:40:40 happy thanksgiving for those in the US, even though EU is more common I guess Nov 28 12:53:10 * russell-- just noticed ar71xx has broken remote syslog and ntpclient (as of r38899), going to try reving up Nov 28 12:53:29 on wzr600dhp, anyway Nov 28 13:12:35 Hello. Nov 28 13:13:46 I have created a package for openwrt- not my first package.. But game has changed since last time, when I used svn to checkout sources.. Now I used git.. And it seems that making a patch with git - is harder than with svn, and dev.openwrt.org doesn't give any guidance (doc for git missing).. Nov 28 13:14:01 So I was thinking that maybe someone could give me a hint..? Nov 28 13:15:18 Found few other sites that saying that my patch could be generated by running git diff - but this doesn't output anything.. Nov 28 13:15:19 more information please Nov 28 13:15:44 Okay, host is archlinux on a virtualbox.. Nov 28 13:16:37 russell--: just tell me what you want/need.. Nov 28 13:16:47 did you create a branch for your changes? Nov 28 13:17:53 russell--: eh, nope. I just checked out git.. Made changes to the tree and built image to see that it works out.. Nov 28 13:18:04 russell--: ran git diff - without any output and came for guidance.. Nov 28 13:18:13 did you commit anything? Nov 28 13:18:29 have you ever used git before? Nov 28 13:18:41 eh- no.. I don't think it's anonymously even allowed to.. Nov 28 13:19:22 yeah, but I use a nice frontend to it on my mac :) Nov 28 13:19:24 git status | curl -F 'sprunge=<-' http://sprunge.us Nov 28 13:20:22 okay.. Now we are talking.. So I add first my directory containing changes, and then all files on it, right? Don't happen to know if there's a switch like "-r" ? Nov 28 13:20:25 * russell-- guessing your diff is empty because you haven't added the files to your local tree Nov 28 13:20:55 Yeah, tryied git status and realized it from the error mssage := Nov 28 13:20:58 create a branch first Nov 28 13:21:11 okay, how is that done? Nov 28 13:21:23 git checkout -b my-packae Nov 28 13:21:34 my-package* Nov 28 13:21:50 I should add * to the end..? Nov 28 13:21:53 no Nov 28 13:21:56 ok.. Nov 28 13:22:09 switched to a new branch, ncm.. Nov 28 13:22:11 this way, you can switch back to master branch and pull Nov 28 13:22:21 you don't want to commit local changes to the master branch Nov 28 13:22:46 yeah I do- but guess nbd or someone wants to take a look first.. Nov 28 13:23:06 okay, then I add files with "git add -r " ? Nov 28 13:23:17 when you commit, it is just to your local tree Nov 28 13:23:38 ok.. Nov 28 13:23:43 git push would be adding your changes to the upstream, but you don't have that permission Nov 28 13:24:20 argh.. Nov 28 13:24:40 should I had used something else than add? like.. mkdir instead? Nov 28 13:25:24 git add Nov 28 13:25:30 ping florian Nov 28 13:26:31 then "git diff master" should show them, i think (not certain if you have to commit them first, don't think so) Nov 28 13:28:28 okay.. Files added.. Trying now.. Nov 28 13:30:28 * russell-- confirms remote syslog isn't working with r38937, also, ntpclient doesn't seem to run, anyway i get the wrong date/time on boot Nov 28 13:30:43 russell--: I am working on the ntp thing Nov 28 13:30:50 its a bit more complicated Nov 28 13:31:15 something has the port, /me presumes the busybox ntpd Nov 28 13:31:30 the problem is the validation code Nov 28 13:31:40 it doesn ot understand uci lists Nov 28 13:31:50 therfore it faisl to read the server names from uci Nov 28 13:32:07 if you change the list server ... to option server 'srv1 srv2 srv3' it will work for now Nov 28 13:32:59 and prior to my recent commit above it wrongly prevents starting if option_enable server is 0 Nov 28 13:33:24 okay, sorry it took- noticed I had some TABS on end of few lines + on some empty lines, so I wanted to remove them.. Nov 28 13:34:34 is there a switch that I can use to redirect output of giff diff master to a file so I can send it to trac ticket, or has submitting changed..? Nov 28 13:35:10 giff diff master=git diff master Nov 28 13:35:14 what's wrong with me?-) Nov 28 13:35:46 jake1981: if it's all in one patch you can just redirect it to a file Nov 28 13:36:05 git runs it with |more.. Nov 28 13:36:26 single feature=single patch.. Nov 28 13:36:33 russell--: fyi, having issues with sysntpd here on ar71xx Nov 28 13:36:37 there are also fancy things that git can do to create a series of emails if there is more than one patch Nov 28 13:37:19 all in one patch. It's a protocol support for ncm devices (3g dongles that create a netifd) Nov 28 13:37:41 git format-patch etc Nov 28 13:38:03 git am etc Nov 28 13:38:46 etc is not supported by git.. Nov 28 13:39:16 * russell-- has to dig out manpage/google for those, usually, so you can as well, if you want to try them out Nov 28 13:40:01 git diff master > /tmp/my-awesome-patch.patch Nov 28 13:40:16 oh crap.. Nov 28 13:40:23 ... is probably all you need Nov 28 13:40:34 my patch broke when I made those changes removing extra tabs.. Nov 28 13:40:43 Now it's showing only those changes :/ Nov 28 13:41:10 oops.. my bad.. Nov 28 13:41:17 forgot to add "master" to git diff.. Nov 28 13:41:41 jake1981: also please submit patches to the mailing list, not attach them to trac tickets Nov 28 13:42:16 KanjiMonster: Okay. Nov 28 13:42:33 jake1981: see https://dev.openwrt.org/wiki/SubmittingPatches Nov 28 13:42:35 But I'll add it to trac first.. Nov 28 13:42:42 no, don't Nov 28 13:42:44 KanjiMonster: that error I got for sysntpd: syntax error: unexpected "(" Nov 28 13:42:56 it's because I used "" in the ubus call Nov 28 13:43:02 there is a wiki page about submitting patches iirc https://dev.openwrt.org/wiki/SubmittingPatches Nov 28 13:43:35 stintel: that error is fixed with https://dev.openwrt.org/changeset/38935 Nov 28 13:43:46 stintel: but there are two more bugs preventign the start of sysntpd Nov 28 13:44:01 jake1981: no patch redundancy please ;) Nov 28 13:44:09 jow_laptop: I see Nov 28 13:44:22 (you can always just add a link to your submission on http://lists.openwrt.org/) Nov 28 13:45:16 or say in the patch description "Addresses #12345" Nov 28 13:45:23 is what I would do :) Nov 28 13:46:20 KanjiMonster: sorry, lingual problem. Redundancy? What does it mean? Nov 28 13:46:31 should i not use ntpclient anymore? historically, i used it to initialize the clock at boot time, under the theory that ntp didn't like big time changes or something Nov 28 13:46:41 jake1981: having something at more than one place Nov 28 13:48:09 russell--: its not really needed Nov 28 13:48:22 busybox ntpd can also cope with huge time warps (1970->now) Nov 28 13:48:32 cool Nov 28 13:49:29 KanjiMonster: eh.. My patch doesn't repeat itself.. Was this what you mean, or do you mean re-inventing a wheel? Nov 28 13:50:21 jake1981: I mean submitting the patch to the mailing list *and* attaching it to a trac tickets - this is once too often, and it should be only on the mailing list. Nov 28 13:51:21 KanjiMonster: okay, which should I then do- I want to give this available for easy to add by anyone, I have a forum post about this.. And was thinking about adding a link to trac ticket from there.. Nov 28 13:52:25 jake1981: http://patchwork.openwrt.org/ collects all patches submitted to openwrt-devel, so gives you an easy way to link to it Nov 28 13:53:15 KanjiMonster: okay, I'll add it to there then if that's rather used than trac.. Nov 28 13:53:39 jake1981: read that wiki page i linked above Nov 28 13:54:32 helpfully titled "SubmittingPatches" ... it tells you how to submit patches, lol Nov 28 13:55:41 KanjiMonster: Jeez.. what a list.. Building this thing too a less time than all the signing necessary :) Nov 28 13:58:31 It's supposed to include the patch as text copy pasted to the message instead of as a patch? Nov 28 13:58:55 you mean, "instead of as an attachment" i think Nov 28 13:59:53 yes, that's what I meant.. Nov 28 14:00:06 I'm tired :) Nov 28 14:00:19 see number 6 in the wiki page Nov 28 14:01:22 That's what I was referring.. Nov 28 14:02:04 if you read it, you will have your answer, along with an explanation of why Nov 28 14:02:06 This patch contains a bunch of shell and comgt scripts.. Nothing to compile, sources don't exist anywhere.. So patch is a bit long.. Nov 28 14:02:40 It is 1396 lines. Nov 28 14:03:16 You sure you want it to be posted as is and not as a.. attachment instead.. Nov 28 14:03:30 someone is going to read it before they apply it Nov 28 14:04:23 if they didn't mean it, they'd change the wiki page ;-) Nov 28 14:04:53 Okay then. Where in the doc is the address where this is going to be sended? While trying to figure out I was able to fix another package that was a long time ago marked as @broken@.. Nov 28 14:05:50 (changes to cutter, so it now uses nf_conntrack instead of ip_conntrack) Nov 28 14:06:17 found it. Nov 28 14:06:23 section 5. Nov 28 14:11:24 patch size 41KB Nov 28 14:13:18 Here's a description what I made: https://forum.openwrt.org/viewtopic.php?pid=218745#p218745 Nov 28 14:19:45 huh, there.. Done.. Nov 28 14:20:30 http://patchwork.openwrt.org/patch/4474/ Nov 28 14:24:46 jake1981: what is NCM? and QMI? Nov 28 14:25:35 russell--: interfaces used by modern 3g/lte usb dongles for broadband, they use ppp internally and expose and network interface.. Nov 28 14:26:33 russell--: kernel support is available from package kmod-usb-net-ether-ncm but there was no support for it in the system. Now there is. QMI is similar.. NCM is used atleast by Huawei. Nov 28 14:26:51 russell--: But as I only had this one device- I wasn't able to provide support for more devices than this one. Nov 28 14:28:35 russell--: but I did decide system in a way, that it's easy to add support for other similar devices.. Nov 28 14:30:09 jake1981: be prepared to send a v2 ;) (I see some issues, like the luci stuff should be moved into its own package to the luci repository, not in the openwrt main one, as it's technically an external package. Also there should be no dependencies from trunk/package to external feeds) Nov 28 14:30:25 but before you do that, wait for comments on the ML (including mine) Nov 28 14:30:33 this was just a quick glance Nov 28 14:30:33 i guess i'm suggesting to spell network-control-model out or link to a reference (maybe http://en.wikipedia.org/wiki/Ethernet_over_USB) so people know what you are referring to. though maybe other people aren't as ignorant as i am. Nov 28 14:31:52 Oh crap- just when I tried to get back to my work with weston on rpi.. Nov 28 14:32:17 Should had just kept this with myself after all.. Nov 28 14:40:08 russell--: I guess everyone trying to get ncm work.. Know what ncm is, and this is pretty useless for one that doesn't need it. Nov 28 14:44:31 jake1981: if it is applied then everyone will see it, so it's useful to write the commit message for a broad audience, imho. Nov 28 14:52:22 yeah, well.. I think I made the biggest work here by building the system, unfortunately I have a very limited time for a hobby like this and my days off from work are over.. Not trying to be rude here, but.. The thing works and problems are on the description/placement of package in the source tree/etc.. I never thought it would go as-is.. Nov 28 14:57:19 Another patch: http://patchwork.openwrt.org/patch/4475/ Nov 28 14:58:13 * russell-- notices that syslogging was removed from procd in mid-feb ... where did it land? "remove log support, this is an external service now" (commit 916f95cb58604038695347ee41a430d8ca1f0556) Nov 28 14:58:32 sorry, mid november Nov 28 14:59:05 ? Nov 28 15:02:19 blogic: ? Nov 28 15:05:41 ah, found it: r38822 (it landed in ubox) Nov 28 17:30:50 hmmm something wrong with qos-scripts recently? enabled with downstream 150000 I get only ~105Mbps, while disabled I get ~142Mbps, advertised line speed is 150Mbps Nov 28 18:11:04 wigyori: ping Nov 28 18:11:46 I have many error on boot such as "ipv6: no symbol version for module_layout". Any clue why? Nov 28 18:42:47 Is there a guide of the init process? my neuron died with uloop :-$ Nov 28 18:44:04 no Nov 28 18:44:09 what do you need to know ? Nov 28 18:45:51 when mount_root is called Nov 28 18:47:06 then tomorrow will be the next round :-/ Nov 28 18:48:24 mount_root is called in preinit Nov 28 18:48:27 /lib/preinit Nov 28 18:48:31 or for initramfs images in /etc/init.d/boot Nov 28 18:49:08 perfect, thanks! Nov 28 18:49:44 nobody had this module_layout issue? Nov 28 18:59:44 build #328 of mcs814x is complete: Failure [failed compile_5] Build details are at http://buildbot.openwrt.org:8010/builders/mcs814x/builds/328 Nov 28 19:11:22 This is often explained by a missing Module.symver file when compiling the modules, Nov 28 19:11:40 but I cannot figure out why I have this with OpenWRT right now Nov 28 20:49:27 jow r38938 trunk/package/system/ubox/Makefile Nov 28 20:49:28 ubox: fix handling of uci lists, prevent shell code injection from uci values (#14548) Nov 28 20:49:41 stintel: ^ this should fix the sysntpd issue Nov 28 20:55:48 jow_laptop: cheers, I'll give it a try soon Nov 28 20:59:36 so with qos-scripts enabled I get little over 100Mbps downstream, softirqd goes up to 26%, qos-scripts disabled, ~142Mbps and softirqd only 3% Nov 28 22:28:04 ping florian Nov 28 22:28:51 build #330 of adm5120 is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/adm5120/builds/330 Nov 28 22:32:19 jow_laptop: /me just tried r38938 and date comes up 1970 Nov 28 22:34:50 using sysntpd rather than ntpclient ? Nov 28 22:35:03 yeah, i removed ntpclient Nov 28 22:35:05 (haven't tried it yet, messing around with my UDOO) Nov 28 22:35:25 where are ntp servers set? /me doesn't see them in /etc/config/system Nov 28 22:35:34 weird, should be there Nov 28 22:35:39 check /rom/etc/config/system Nov 28 22:36:02 i think i probably overrode the file, checking Nov 28 23:01:02 yes Nov 28 23:01:43 so, what's the reason that ntp seems to wait until after the jffs2 is formatted? Nov 28 23:02:01 on a buffalo wzr600dhp, that's over 3 minutes Nov 29 01:32:20 nbd: ping Nov 29 02:01:11 build #420 of rb532 is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/rb532/builds/420 Nov 29 02:08:13 build #420 of ppc44x is complete: Failure [failed shell_12] Build details are at http://buildbot.openwrt.org:8010/builders/ppc44x/builds/420 **** ENDING LOGGING AT Fri Nov 29 02:59:58 2013