**** BEGIN LOGGING AT Fri Nov 04 02:59:57 2011 Nov 04 03:25:49 r28746 broke my 11na radio on dir825 Nov 04 03:26:05 iw list shows every channel "(disabled)" Nov 04 03:26:12 the 2.4 radio worked fine Nov 04 03:28:26 nbd: ping Nov 04 05:46:34 build #84 of xburst is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/xburst/builds/84 Nov 04 09:06:37 build #76 of etrax is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/etrax/builds/76 Nov 04 10:02:38 jow_laptop, you know last night after i updated latest trunk on my dlink 825 and the ip6 firewall prevented my dhcp6c to get the prefix, its happening again on latest today trunk on this 1043n Nov 04 10:03:25 I already said you most likely need a rule Nov 04 10:04:29 yes, but i think i borked something, is the ip6 firewall copying ipv4 firewall, right? Nov 04 10:05:06 mostly Nov 04 10:19:51 ok jow, uhm.. is it possible to add a dependency in openwrt trunk build like if i select wide-dhcp6c to automagically add a firewall rule like: http://pastebin.com/M6cUvTLR Nov 04 10:19:59 would be awesome Nov 04 10:20:53 no Nov 04 10:21:18 did you even test an accept rule yet? Nov 04 10:21:42 yes, it works Nov 04 10:22:47 i just allowed 546 in INPUT so my wan dhcp6c be able to get the PD Nov 04 10:23:18 udp? Nov 04 10:24:32 yup, udp only for ipv6 so ISP's dhcp server can communicate with my dhcp6c Nov 04 10:44:18 jow_laptop, something's happening: http://pastebin.com/RwLcJsgr Nov 04 10:45:09 the oops is useless without symbols Nov 04 10:45:47 but could be an oom Nov 04 10:45:57 9k free Nov 04 10:48:06 right, 1k now Nov 04 10:49:29 jow_laptop, could it be something from last hostapd updates? more output: http://pastebin.com/45eMELaM Nov 04 10:49:46 this line is interesting: WARNING: at /home/dani/openwrt/tp-link_wr1043n/trunk/build_dir/linux-ar71xx_generic/compat-wireless-2011-11-01/net/mac80211/tx.c:693 0x81f9d55c() Nov 04 10:50:24 again, those oopses are useless without symbols Nov 04 10:51:03 you want me to rebuild the firmware with all debugging or..? Nov 04 10:51:26 no, I just don't want to see oopses without symbols Nov 04 10:51:44 if you tell me what to do.. i will :) Nov 04 10:52:08 there is a global option to build the kernel with symbol table ifnormation Nov 04 10:52:24 got it Nov 04 11:07:54 shut down uhttpd until i reflash with debug and everything seems ok at dmesg Nov 04 11:20:00 my dlink 825 is absolutely fine in this matter, its 28738 Nov 04 11:20:32 the wr1043 is at 28746 Nov 04 11:31:40 jow_laptop, ok http://pastebin.com/FYgX16Dk Nov 04 11:32:33 any other info i can get? Nov 04 11:35:20 this is the system log: http://pastebin.com/nUtcVxY9 Nov 04 11:43:41 seems to trigger only when i do heavy stuff via wireless Nov 04 11:44:06 like if i scp the openwrt firmware to /tmp again it ooms or something Nov 04 11:49:37 nbd hello, i updated the nfs-kernel-server package and i found out that the malloc.h patch isnt needed anymore - at least for my system - do you remember why you added it? was malloc.h missing in older versions of uclibc? Nov 04 11:49:42 its was https://dev.openwrt.org/browser/packages/net/nfs-kernel-server/patches/003-no_malloc_h.patch?rev=25519 Nov 04 11:50:07 the package compiles fine with and without it Nov 04 11:51:05 <_trine> jow_laptop, can I install lucu-app-pbx on my dockstar because I dont see it when I do opkg list Nov 04 11:58:55 tripolar * r28747 /packages/net/nfs-kernel-server/patches/101-no_malloc_h.patch: [packages]: nfs-kernel-server - 101-no_malloc_h.patch isn't needed anymore Nov 04 12:09:19 on my dlink 825 that works okay hostapd is at 0.8.x , on this wr1043 which crashed when i scp stuff like a firmware its at 2.0-devel ... Nov 04 12:11:48 tripolar: if i remember correctly, it was needed for compiling on non-linux systems Nov 04 12:17:01 dape_at_work: still no symbols Nov 04 12:17:15 did compiled the kernel with.. Nov 04 12:17:53 now it wont let me take a peek at all, it seem to reset the board on every scp attempt i made on it Nov 04 12:19:43 <_trine> jow_laptop, how do I obtain luci pbx because I dont see it in opkg update Nov 04 12:21:25 _trine: don't know, maybe its broken Nov 04 12:22:56 I could compile it successfully locally Nov 04 12:23:11 jow_laptop, do i need to install some kernel debugger to check those symbols? Nov 04 12:23:17 dape_at_work: no Nov 04 12:23:43 dape_at_work: simply enable "compile kernel with sybol table information" Nov 04 12:23:47 its even default in trunk afair Nov 04 12:23:57 you probably stripped it too much Nov 04 12:24:09 just recompiled it earlier with the option and reflashed.. Nov 04 12:24:29 wasn't applied Nov 04 12:24:31 let me recheck the menuconfig again Nov 04 12:24:39 did you run make target/linux/clean ? Nov 04 12:24:47 nop Nov 04 12:24:56 just enabled the option and did make Nov 04 12:24:57 well, do it Nov 04 12:25:13 aight, it will last forever on this 1.3 ghz celeron Nov 04 12:25:38 CONFIG_DEBUG_FS is needed? Nov 04 12:25:42 no Nov 04 12:25:52 <_trine> jow_laptop, should it be possible to see the luci pbx in opkg list update ? Nov 04 12:25:52 just that option and make clean and make again you say.. Nov 04 12:26:14 _trine: probably not, depends on how current the buildbots are Nov 04 12:26:33 <_trine> k Nov 04 12:27:30 damn, didnt have CONFIG_KERNEL_KALLSYMS on Nov 04 12:27:58 CPU 0 Unable to handle kernel paging request at virtual address ffffd4d0 <- this looks like a null pointer dereference Nov 04 12:28:23 nico * r28748 /packages/net/xinetd/ (Makefile files/xinetd.init): Nov 04 12:28:23 packages/xinetd: fix initscript (closes: #10342) Nov 04 12:28:23 Create a pid file by default when starting xinetd (resolves #10342). Nov 04 12:28:23 Signed-off-by: Stefan Agner Nov 04 12:29:15 _trine: well the xburst snapshots have luci-app-pbx so I suppose its okay Nov 04 12:29:22 maybe you didn't update your feeds Nov 04 12:29:55 <_trine> jow_laptop, what do I need to do to update the fedds Nov 04 12:29:58 <_trine> feeds Nov 04 12:30:30 <_trine> do you mean ./scripts/feeds update -a Nov 04 12:30:36 yes, of course Nov 04 12:30:45 <_trine> and install cos I did do that Nov 04 12:31:17 <_trine> but how would that affect opkg update Nov 04 12:31:45 opkg update depends on how current the snapshots are Nov 04 12:31:51 and they are not very current Nov 04 12:31:58 ar71xx is ~10 days old for example Nov 04 12:32:18 <_trine> ok Nov 04 12:32:49 xburst is from today and has luci-app-pbx Nov 04 12:33:36 <_trine> whats xburst never heard of that Nov 04 12:33:47 another target Nov 04 12:35:09 <_trine> I have removed freeswitch from my compile and now the kernel decompresses on my dockstar Nov 04 12:35:20 <_trine> but it has another problem now Nov 04 12:35:21 nico * r28749 /packages/libs/libv4l/Makefile: packages/libv4l: mark as 2.6.30-37 only, v4l1 support has been removed in 2.6.38 Nov 04 12:36:28 <_trine> it has stuck at this point on the serial connection Nov 04 12:36:31 <_trine> http://dpaste.com/647232/ Nov 04 12:40:33 <_trine> interestingly the dockstar is up and working but the serial line is jammed Nov 04 12:41:17 did you try to press enter? Nov 04 12:41:22 <_trine> yes Nov 04 12:41:26 <_trine> and ctl C Nov 04 12:41:36 ok Nov 04 12:47:12 <_trine> I rebooted the dockstar and again the serial line has jammed in the same place Nov 04 12:48:35 <_trine> I have never seen the serial line jam before and ssh works ok this is very odd Nov 04 12:50:12 <_trine> jow you may be pleased to know that luci is up as well Nov 04 12:50:32 "up" ? Nov 04 12:50:42 <_trine> up= working ok Nov 04 12:51:32 <_trine> I must have left something out of the compile which has affected the serial line Nov 04 12:53:42 jow * r28750 /trunk/target/linux/ (5 files in 5 dirs): add timeserver config to target specific system configs as well Nov 04 12:58:42 build #74 of ep93xx is complete: Failure [failed shell_10] Build details are at http://buildbot.openwrt.org:8010/builders/ep93xx/builds/74 Nov 04 13:00:00 ok Nov 04 13:20:09 <_trine> jow where can I read about luci pbx and about what needs to be in asterisk.ctl Nov 04 13:24:49 forum Nov 04 13:25:54 <_trine> ok I will try to find it Nov 04 13:31:51 jow_laptop, is this useful? http://pastebin.com/ek3wddq7 Nov 04 13:33:48 http://permalink.gmane.org/gmane.linux.suse.kernel/2723 Nov 04 13:33:58 nbd * r28751 /packages/net/nfs-kernel-server/patches/101-no_malloc_h.patch: Revert "[packages]: nfs-kernel-server - 101-no_malloc_h.patch isn't needed anymore" Nov 04 13:34:07 tripolar_: the patch is needed Nov 04 13:34:50 didnt plugged anything, i just reflashed the router with kernel with debug, restarted and dmesged Nov 04 13:35:07 well it crashes somewhere Nov 04 13:35:24 http://pastebin.com/8TqAHp7B Nov 04 13:35:50 something seems to eat memory, 1k left almost all the time Nov 04 13:36:15 then try to track it down Nov 04 13:36:25 kill all services until the leak stops happening Nov 04 13:36:51 nbd but why? Nov 04 13:37:21 tripolar_: because some non-linux systems (e.g. mac os x) don't have malloc.h Nov 04 13:37:28 and the include statement isn't necessary Nov 04 13:38:57 nbd good to know Nov 04 14:06:37 meh, official trunk image for wr1043n works smooth Nov 04 14:07:23 i really think something happened last 24 hours, i didnt changed configs, just svn update, make reflashed.. Nov 04 14:07:36 even factory so i can be sure nothings left dirty Nov 04 14:14:44 maybe the mac80211 update caused it Nov 04 14:15:01 the trunk snapshot build doesn't have that yet Nov 04 14:15:27 * dape_at_work sighs Nov 04 14:15:33 though i don't see how that would cause the filesystem related crap Nov 04 14:15:48 after you built your image, did you install any packages from the default location, especially kernel related stuff? Nov 04 14:16:05 or did you include everything in your own build? Nov 04 14:16:33 didnt added anything else, didnt use opkg on the router, just svn updated make again and flashed the factory image Nov 04 14:16:51 ok Nov 04 14:17:09 nico * r28752 /packages/net/freeradius2/Makefile: Nov 04 14:17:09 packages/freeradius2: fix dependencies Nov 04 14:17:09 freeradius2: fix some dependencies Nov 04 14:17:09 Signed-off-by: Alexander Couzens Nov 04 14:17:20 i did it after the hostapd update, yesterdays the router was awesomely fine, i spend all day in the office to see it resetting on intensive stuff like.. scp-ing :) Nov 04 14:17:50 if you need any extra info, im here with the router on Nov 04 14:18:39 so if i understand the issues correctly there's two parts Nov 04 14:18:44 a crash with kernel oops Nov 04 14:18:46 and a memleak Nov 04 14:18:47 right? Nov 04 14:20:04 im not a programmer so excuse me please, its acting ok if i dont do anything intensive.. this morning when i loaded its luci in browser it hanged, i barely saw some oopses in dmesg before it crashed, Nov 04 14:20:32 ok, but you mentioned that memory usage keeps increasing Nov 04 14:20:34 until now , through all the day , it seems something eats the memory until almost every components ooms or smth Nov 04 14:20:44 ok Nov 04 14:21:35 so first we need to figure out if it's the kernel or some process Nov 04 14:22:14 you can check the kernel memory usage by running cat /proc/meminfo and looking at the "Slab:" line Nov 04 14:22:20 ok, im here, have 28601 on it Nov 04 14:22:35 you want me to flash latest trunk again? Nov 04 14:22:55 yes Nov 04 14:23:13 okay, still with debugging stuff in right? be right back then Nov 04 14:25:24 uhm, i have some custom /etc/config/ files, i think they are ok, its the pppoe connection prepared in /etc/config/network , wireless on in /etc/config/wireless , i dont think they have anything wrong in them, want me to post them somewhere to check them just in case? Nov 04 14:25:50 i mean they worked fine until today ofc. Nov 04 14:27:40 i don't think it's a config issue Nov 04 14:28:20 ok, building Nov 04 14:31:24 build #72 of gemini is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/gemini/builds/72 Nov 04 14:38:39 juhosg * r28753 /trunk/target/linux/generic/files/drivers/net/phy/swconfig.c: kernel: swconfig: add a missing unlock in error path Nov 04 14:46:52 nbd, flashed, configured on pppoe for inet acces (ipv6 too) rebooted, logged via ssh and i see the first strange thing after 6 minutes: http://pastebin.com/sQ1TR038 Nov 04 14:47:54 weird Nov 04 14:47:59 doesn't look like anything wifi related Nov 04 14:49:31 nbd, logread - http://pastebin.com/5LJfWxBC Nov 04 14:49:51 do i see sysinit seg fault...? Nov 04 14:51:11 the segfault is probably a consequence of the kernel crash issue (triggered by radvd) Nov 04 14:52:44 hmm, didnt do any updates on my gentoo host last days, im trying to review what changed but i cant find anything relevant, i just updated from svn , the feeds and recompile again Nov 04 14:54:22 you could try doing a make dirclean Nov 04 14:55:53 alright, doing make again Nov 04 14:57:35 if this wont make any difference then could be hardware fault? like ram memory ? Nov 04 15:24:28 can Package/conffiles also be used for directories? and are those kept by sysupgrade then? Nov 04 15:46:33 hmm, started the aircooling in this room, wasnt too hot but.. also i had the wifi at 20dbm-100mw, maybe hot heated.. Nov 04 16:09:19 soma: yes and yes Nov 04 16:12:22 great **** ENDING LOGGING AT Fri Nov 04 16:14:09 2011 **** BEGIN LOGGING AT Fri Nov 04 16:18:15 2011 **** ENDING LOGGING AT Fri Nov 04 17:02:03 2011 **** BEGIN LOGGING AT Fri Nov 04 17:03:22 2011 **** ENDING LOGGING AT Fri Nov 04 17:15:12 2011 **** BEGIN LOGGING AT Fri Nov 04 17:24:14 2011 Nov 04 18:28:21 build #71 of rdc is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/rdc/builds/71 Nov 04 19:39:22 uhm, right so.. earlier in the office i tried to reflash the wr1043n but the goddam thing resetted few seconds after i ran mtd -r write bin firmware, is it possible i totally destroyed it? i can see leds flashing right after i plug it on power now.. (didnt finished the serial soldering yet..) Nov 04 20:46:49 uhm, trying to build a ramdisk image to rescue this bad burned 1043n and make stops like this: http://pastebin.com/AfP8VY0k Nov 04 20:56:14 nah, stuckd Nov 04 20:57:06 in /home/dani/openwrt/tp-link_wr1043n/trunk/build_dir/linux-ar71xx_generic/linux-2.6.39.4/usr/Makefile:58: on that line its just $(deps_initramfs): ; Nov 04 21:12:35 dape-home: question is what is in "$(deps_initramfs)" ... probably some colon injected from the environment Nov 04 22:12:40 nico * r28754 /packages/libs/re/ (Makefile patches/ patches/100-re.mk.patch): packages/re: update to 0.3.0 (thanks to Alfred E. Heggestad) Nov 04 22:12:41 nico * r28755 /packages/libs/rem/ (. Makefile): packages/rem: add (thanks to Alfred E. Heggestad) Nov 04 22:12:43 nico * r28756 /packages/net/baresip/ (5 files in 3 dirs): packages/baresip: update to 0.3.0 (thanks to Alfred E. Heggestad), make it modular, use start-stop-daemon Nov 04 22:12:48 nico * r28757 /packages/net/restund/ (Makefile files/restund.init patches/ patches/101-cross.patch): packages/restund: update to 0.3.0 (thanks to Alfred E. Heggestad), make it modular, use start-stop-daemon Nov 04 22:26:12 build #67 of octeon is complete: Failure [failed compile_6] Build details are at http://buildbot.openwrt.org:8010/builders/octeon/builds/67 Nov 04 22:29:03 jow_laptop i downloaded trunk again and recompiling everything to other location.. Nov 04 22:34:01 ok Nov 04 23:28:10 nico * r28758 /packages/net/openl2tp/ (3 files in 2 dirs): packages/openl2tp: enable config file support (thanks to Bogdan Roman), provide -full & -mini variants (closes: #10164) Nov 04 23:28:44 <3 {Nico} Nov 04 23:28:48 planned to do this myself Nov 04 23:38:39 <{Nico}> jow_laptop: oops, sorry Nov 04 23:39:20 no, I'm happy you did it Nov 04 23:40:17 <{Nico}> we still need sample configuration files i guess Nov 04 23:44:06 I think someone will submit some eventually Nov 04 23:45:22 hello Nov 04 23:48:00 fast and simple query - Im patching (trying..) my external kernel with owrt changes, (lets say its 2.6.32). I see files in target/linux/generic-2.6/{files,patches-*} and in target/linux/ar71xx/{files,patches*} (ar71xx is my board). Nov 04 23:48:19 you need to apply both series Nov 04 23:48:23 generic first Nov 04 23:48:27 do I need to get all those files+patches, if yes in what order. or do I need something else to apply ? Nov 04 23:48:53 order is alphabetical Nov 04 23:49:11 I mean generic THEN ar71xx ? Nov 04 23:49:15 yes Nov 04 23:49:21 00:48 < jow_laptop> generic first Nov 04 23:49:23 ok Nov 04 23:49:52 oki, tnx a lot Nov 04 23:50:30 patches can be applied with -p1 Nov 04 23:50:38 ah you also need yet another thing Nov 04 23:50:54 ah no, you alrady took files into account Nov 04 23:51:06 those files must be simply copied into the kernel tree Nov 04 23:51:12 order does not matter there Nov 04 23:52:45 ok, I figured that out already Nov 04 23:53:18 btw, do you know where/which script is responsible for kernel building in make world ? Nov 04 23:55:29 include/kernel-build.mk, include/kernel-default.mk Nov 04 23:55:58 tnx Nov 05 00:05:46 damn Nov 05 00:06:15 fresh trunk, made the usual firmware, then checked ramdisk in menuconfig and bam: /home/dani/openwrt/wr1043nd/trunk/build_dir/linux-ar71xx_generic/linux-2.6.39.4/usr/Makefile:58: *** multiple target patterns. Stop. Nov 05 00:07:33 replace "$(deps_initramfs)" with "$(info $(deps_initramfs))" re-run make V=99 and paste the output Nov 05 00:10:48 http://pastebin.com/4FEu0aVn .. Nov 05 00:12:21 do I need to clean everything after I change kernel tree in menuconfig to 'external' ? I changed this, run make and I see it still have some references to old-orginal owrt kernel. what I should clean/delete ? Nov 05 00:13:17 dape-home: looks like the colons in the usb-modeswitch files are tripping it up Nov 05 00:13:36 let me uncheck the pack Nov 05 00:14:41 this usually does not happen because ramdisk images are rather minimal Nov 05 00:15:04 right, i compiled ipv6 and 3g usb stuff in Nov 05 00:20:16 uhm, is there a quick way to reset package selection in menuconfig to default? or just edit .config ? Nov 05 00:21:18 duhh, was right there in the menuconfig, down :) Nov 05 00:26:49 this is unbelieveable, still the same error for ramdisk creation Nov 05 00:29:51 ok, erased .config, just selected ar71xx and tplink wr1043nd profile and make still failes for ramdisk Nov 05 00:35:21 dape-home : try to clean tmp folder maybe yet ? Nov 05 00:36:02 oki Nov 05 00:37:51 no, same error Nov 05 00:43:57 ok, make clean and repeat then.. Nov 05 00:49:03 I think Im with similar problem like with you, my changes with kernel tree are not working. Im making copy of whole owrt folder, and will try to continue building after manualy remove things from build_directory, allowing 'make world' to rebuild it freshly. Nov 05 00:49:21 this should be faster than clean everything and rebuild from scratch... Nov 05 00:49:54 be aware that the buildroot is not relocatable Nov 05 00:51:26 in worst case scenario I will still have my not touched owrt-copy folder. Nov 05 00:52:30 make clean and make again seem to work, still compiling Nov 05 00:53:56 on my machine fresh make takes about 3-4 hours .. so I can recomend you to run it on 'screen' and go to sleep ;) Nov 05 00:54:59 nah, it takes like 10 min on this pentium-m like computer Nov 05 01:37:39 mazilo * r28759 /packages/net/freeswitch/ (Makefile patches/002-Makefile.am.patch): bumped to the latest FS git (54b4b088d911d4c80f224ed561d8e278229a75ff) Nov 05 01:54:40 nico * r28760 /packages/libs/apr/ (Makefile patches/101-fix_apr_time_now.patch): packages/apr: force LFS support, fix apr_time_now function (closes: #9287) **** ENDING LOGGING AT Sat Nov 05 02:59:57 2011