**** BEGIN LOGGING AT Fri Nov 08 02:59:58 2013 Nov 08 08:19:36 juhosg r38689 trunk/target/linux/ ar71xx/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_phy.c ar71xx/files/drivers/net/ethernet/atheros/ag71xx/ag71xx_main.c * ar71xx: ag71xx: fix a race involving netdev registration Nov 08 09:05:25 build #361 of kirkwood is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/kirkwood/builds/361 Nov 08 11:35:44 hi, this VLAN setup brings up no interfaces. did i make an obvious mistake, or is there a known bug right now?: http://paste.ubuntu.com/6381495/ Nov 08 11:59:10 towolf: are you sure the switch name is correct? Nov 08 12:00:14 DonkeyHotei, i think i messed up with the human readable uci labels Nov 08 12:00:43 are you using squashfs? Nov 08 12:01:05 DonkeyHotei, yeah Nov 08 12:01:35 look in /rom/etc/config/network to compare Nov 08 12:01:41 oh wait Nov 08 12:02:00 i forgot the file is dynamically generated Nov 08 12:02:29 rename /etc/config/network, reboot, compare the autogenerated file Nov 08 12:05:02 man, such a weird this this whole AR9380/AR9381 Nov 08 12:05:19 DonkeyHotei, i’ve got it sorted out. i reused labels, which is no good Nov 08 12:05:31 towolf: ok good Nov 08 12:07:00 AR9381 is supposedly only 802.11bgn but it shows up as 802.11abgn just like AR9380 Nov 08 12:07:22 blogic: are you around? i booted an initramfs image on a device not connected to anything, just left it running, and after a while http://pastebin.com/raw.php?i=QPWdLZBY Nov 08 12:07:34 you can even configure it as 802.11a and it shows up as enabled Nov 08 12:09:38 DonkeyHotei: was it a ramdisk? Nov 08 12:09:56 initramfs = ramdisk Nov 08 12:10:47 that same thing happens to me every time i leave my ramdisk running Nov 08 12:11:12 JyZyXEL: what target? Nov 08 12:11:17 lantiq Nov 08 12:11:37 but it does not happen when running from flash? Nov 08 12:11:39 figured its only an issue when running in ramdisk though, the memory just runs out Nov 08 12:11:49 i'm not so sure Nov 08 12:11:59 hence why i ask Nov 08 12:13:15 JyZyXEL: does it happen when running from flash? Nov 08 12:15:11 so far i haven't flashed anything so i have easy access to the OEM system Nov 08 12:15:25 same here Nov 08 12:15:38 so we don't know Nov 08 12:15:48 but blogic might know Nov 08 12:17:11 i think in my case the ramdisk reserves around 32MiB of the total 64MiB of memory Nov 08 12:18:10 juhosg r38690 trunk/package/base-files/ Makefile files/lib/functions.sh * base-files: mtd_get_mac_ascii bugfix Nov 08 12:18:14 root@OpenWrt:/# free Nov 08 12:18:14 total used free shared buffers Nov 08 12:18:14 Mem: 61228 26528 34700 0 0 Nov 08 12:18:14 -/+ buffers: 26528 34700 Nov 08 12:18:14 Swap: 0 0 0 Nov 08 12:20:59 Mem: 61168 32260 28908 0 0 Nov 08 12:21:04 tmpfs 29.9M 36.0K 29.8M 0% /tmp Nov 08 12:22:14 kinda of weird that at some point something eats up all the free memory Nov 08 12:32:21 root@OpenWrt:/# while true; do ps -w; free -m; sleep 60s; done Nov 08 12:32:32 ill leave it running :p Nov 08 12:39:21 kaloz r38691 trunk/ (6 files in 5 dirs) * [omap]: rename the omap4 target to omap Nov 08 12:46:20 kaloz r38692 trunk/ (9 files in 5 dirs) * [omap]: switch to 3.12, enable support for AM33xx/OMAP3 Nov 08 13:44:33 DonkeyHotei: it seems slowly the free memory is being used for something Nov 08 13:45:04 does ps give any clues? Nov 08 13:45:28 nope, no process seems to be hogging it Nov 08 13:46:24 perhaps /proc/meminfo would be better Nov 08 13:46:52 if it's not a process, it must be in the kernel Nov 08 13:48:10 ill loop cat /proc/meminfo for some time Nov 08 13:53:52 DonkeyHotei: we had a weird memleak on a ar71xx some time ago... it seemed to be fixed now Nov 08 13:54:12 DonkeyHotei: in our case on a tplink 4300 running as vlan switch/wifi ap Nov 08 13:55:04 i only know it was in kernelspace, since it didnt show any extra mem use which i could see in userland Nov 08 13:56:35 roh: do you know where in kernelspace? Nov 08 13:56:45 sadly not Nov 08 13:58:26 i monitored it with muninlite, which calculates memory useage by something like 'all-x-y-z' for the userspace.. and thus got wrong graphs... but after examining the ps output i was sure it was in kernelspace Nov 08 13:58:51 it would be nice if there were a way to see how much memory is used by each kernel module Nov 08 14:04:40 jup. but i guess it was wifi related Nov 08 14:05:05 simply because there was no special stuff enabled or used.. no iptables. simple routing only. Nov 08 14:05:24 so ath9k is broken? Nov 08 14:09:00 maybe was Nov 08 14:09:17 maybe still is Nov 08 14:09:35 r38381 works so far Nov 08 14:10:24 also it was a slow leak.. forcing to reboot about once every 1 or 2 weeks Nov 08 14:10:28 r38679 has the above memory leak on lantiq with ath9k Nov 08 14:10:35 took a month to eat up the 128mbyte ram Nov 08 14:10:42 oh, this is not that slow Nov 08 14:10:59 heh. then it should be much easier to debug Nov 08 14:11:08 how? Nov 08 14:11:43 dunno.. but atleast you do not have to wait weeks to see if its happening Nov 08 14:19:39 luka: sent Nov 08 14:47:43 DonkeyHotei: looks like its being used for slab Nov 08 14:48:35 ill start monitoring /proc/slabinfo Nov 08 14:48:51 ok Nov 08 15:12:40 hi blogic Nov 08 15:12:51 i miss you, why left #openwrt Nov 08 15:12:54 :P Nov 08 15:13:47 anyway, you are always welcome there, dont forget that ;) Nov 08 15:18:19 zajec: Nov 08 15:18:22 err Nov 08 15:18:38 ah, he isn't here anyway Nov 08 15:19:55 thanks for the heads up jow_laptop.. Nov 08 15:20:24 hey whats up up builder :D Nov 08 15:21:37 * oc80z was refferred here regarding procd not running init.d scripts when kernel parameter (from grub2) has console=/dev/ttyS999 (out of range, non-existing serial device) Nov 08 15:23:28 i cant recall, might have had the serial baud wrong at one time.. but it when it was "console=" then, kernel would boot, pingable, but no init scripts , dropbear etc... if returned to console=/dev/ttyS0 or 1, (both avail) its OK Nov 08 15:24:21 it looks exactly the same as the 2007 bug init.c from busybox.. (procd being the replacment) Nov 08 15:26:03 thats all -- i'll prob pull latest code and retry and post back with a better test-case of what ive seen.. may be even try /dev/ttyUSB999 too.. cheers, Nov 08 15:29:59 DonkeyHotei: looks like it is kmallocs :\ Nov 08 15:32:13 blogic -------------^ Nov 08 15:34:49 ? Nov 08 15:35:10 hi - do u know much about the procd code? Nov 08 15:35:58 yes Nov 08 15:36:49 hmm Nov 08 15:36:57 blogic: i need to leave, but JyZyXEL has info on kernel memory leaks on lantiq Nov 08 15:36:59 seems like when console=/dev/doesntexist -- procd cant open standard in/out and it wont continue init scripts.. Nov 08 15:37:09 so a x86 grub image does not boot when "console=" is passed? Nov 08 15:37:16 kk Nov 08 15:37:22 thx, i will look into it Nov 08 15:37:36 JyZyXEL: where does it leak ? Nov 08 15:37:47 the kernel boots, but i cant see (heh) -- but it looks like it may be failing on early console Nov 08 15:37:59 blogic: kmallocs seem to fill up the memory slowly but surely Nov 08 15:38:07 might have to map early standard out/error.. yeah grub2 yes blogic. Nov 08 15:38:24 they had to map it to /dev/null when this happened , regarding busybox init Nov 08 15:38:36 JyZyXEL: do you have wifi, dsl, tpai, usb loaded ? Nov 08 15:39:05 "as the 2007 bug init.c from busybox" whats this ? Nov 08 15:39:55 blogic: lots of stuff loaded: http://pastebin.com/raw.php?i=beK2zysz Nov 08 15:41:26 thanks blogic -- http://lists.busybox.net/pipermail/busybox/2007-July/062185.html Nov 08 15:46:39 blogic: could you apply my pach to trunk for openssl i sent to the ml? Nov 08 15:46:48 JyZyXEL: remove half and see if it still elaks Nov 08 15:46:52 JyZyXEL: remove half and see if it still leaks Nov 08 15:47:13 ill start with a default config and start reducing from there Nov 08 15:53:51 tripolar r38693 packages/net/openssh/ (6 files in 2 dirs) * [packages] openssh: update to 6.4p1 Nov 08 16:00:07 blogic: this might be specific to all targets,, kernel maps to console=null is passed into the kernel (any bootloader) -- kernel makes /dev/console and then uses passes bootloader stuff to init() which does the dupe of stdin/out/err --- then calls /sbin/init with cmdline.. so if its null , procd always uses (same case with busybox) console=null Nov 08 16:01:08 * oc80z cant program so, take above comment with salt. Nov 08 16:02:10 :-) Nov 08 16:17:29 blogic: did you notice the bug about fw_cutter not working on xrx200? Nov 08 16:18:35 JyZyXEL: yes Nov 08 16:18:41 working on it Nov 08 16:18:49 i fixed the 8970 last night Nov 08 16:18:56 or the day before ?! Nov 08 16:19:03 2 days ago Nov 08 16:19:59 i wish i had a dslam so i could test different DSL connections ;) Nov 08 16:20:33 i have a spare one Nov 08 16:20:39 but it does not work with lantiq HW Nov 08 16:20:40 :) Nov 08 16:20:50 only works with broadcom cpe Nov 08 16:21:35 would be interesting to experiment with different g997xtusecs values to see exactly what is supported Nov 08 16:22:38 hehe Nov 08 16:22:40 yes Nov 08 16:23:29 and if different firmwares have different support, be cause i have TP-links firmware and its slightly bigger than the W921V one Nov 08 16:24:50 there is support for the W921V now? Nov 08 16:25:19 roh: not had time yet Nov 08 16:25:47 the core functionality of owrt has a higher priority at the moment to board support Nov 08 16:26:00 but its possible?... i havent checked whats avail. documentation there is yet Nov 08 16:26:08 is VDSL much different than ADSL in terms of spec? Nov 08 16:26:19 Devastator: yes. Nov 08 16:26:27 i would really love ADSL/VDSL support from a opensource driver Nov 08 16:26:27 roh: should be possible to get the foss uboot running in no time Nov 08 16:26:32 roh: the rest is peanuts Nov 08 16:26:39 Devastator: im with you there Nov 08 16:26:42 no isdn/ dect/wifi though Nov 08 16:26:50 blogic: nice one.... well.. too bad Nov 08 16:26:57 (isdn/dect/wifi) Nov 08 16:27:00 there is TAPI support though! :) Nov 08 16:27:11 JyZyXEL: for what? Nov 08 16:27:17 i assume that would be for the two analogue phone sockets? :P Nov 08 16:27:20 w921v Nov 08 16:27:33 but not in owrt due to misisng firmware blobs Nov 08 16:27:35 JyZyXEL: forget those. nobody uses that stoneage stuff anymore. Nov 08 16:27:37 JyZyXEL unfortunately, as pointed in the ml, a group would have to sacrifice itself in order to create necessary specs for driver development Nov 08 16:27:39 and quite frankly i stopped craing Nov 08 16:27:46 s/itself/themselves Nov 08 16:27:59 as long as the dsl part of the boxes works i am happy Nov 08 16:28:30 blogic: hehe... i can understand. yep. dsl iss most important. i was rather asking because there are very few vdsl modems avail Nov 08 16:28:54 yep, mine is crap to say the least Nov 08 16:29:04 but on the other side, its expensive hardware, so i guess nobody buys it only as a modem/router Nov 08 16:29:06 there can't be many people using the ISDN and telephone ports, can there?! Nov 08 16:29:15 i was trying to buy a huawei, but can't find it Nov 08 16:29:16 JyZyXEL: everybody. Nov 08 16:29:27 oh noes :p Nov 08 16:30:02 JyZyXEL: tcom doesnt sell/rent out any other modems right now and they stopped selling isdn/analog also this year. only voip via adsl/vdsl now -> isdn/pots/dect to connect the phones is essential Nov 08 16:30:15 atleast for what the 'normal user' needs Nov 08 16:30:50 but i dont think anybody in here counts as a normal user ;) Nov 08 16:31:10 does anybody know pace? a company that makes modems as well Nov 08 16:31:12 roh: so even the customers who only want to use a regular analogue phone have to have this box? :p Nov 08 16:31:43 JyZyXEL: some box, but if they got vdsl, there is not much choice. adsl can be done with older/simpler ones. Nov 08 16:32:04 if you get a really fresh port (depending on region) you get adsl16/1 in annex J with voip Nov 08 16:32:12 as 'simple phone and internet' Nov 08 16:32:32 so not even a 'old fritzbox' does the job (modem wise) Nov 08 16:32:45 anyway, getting ready for my deutsch class, a bit ancious hehe Nov 08 16:36:43 man, like DSL wasn't hard enough but they have to get the telephony involved into it Nov 08 16:40:33 are the dsl drivers currently used by modem company manufacturers still updated? Nov 08 16:55:26 blogic: i can pastebin the code for you Nov 08 16:58:38 so you dont waste time, for the fix eh..10min or so eh.. Nov 08 16:59:45 (the code to fix it :D) Nov 08 17:05:17 ok Nov 08 17:05:19 show me Nov 08 17:16:12 Devastator: are you learning german? Nov 08 17:17:36 https://ezcrypt.it/Rz7n#KKz3np4HFT6EvNeHAUIsjbP6 Nov 08 17:17:38 yes sir Nov 08 17:17:49 my second class today Nov 08 17:17:52 that might be a quick way to fix it,, but u know the concept blogic Nov 08 17:18:02 moin moin Devastator :D Nov 08 17:18:46 within 2 years I might be able to ask questions in deutsch hahah Nov 08 17:19:52 Devastator: deutsch ist recht einfach ;) Nov 08 17:23:01 das ultimite driving machine Nov 08 17:23:10 JyZyXEL: before you start removing random stuff to pinpoint the memory leak, take out the dsl driver as the first thing to see if it's that. if it's not that, try wifi Nov 08 17:23:48 DonkeyHotei: yeah im gonna start with the DSL Nov 08 17:24:15 would be weird if its the wifi, ath9k should be the "best" wifi driver at the moment? Nov 08 17:28:45 blogic: i posted it above with ezcrypt Nov 08 17:35:17 tripolar the only think I know yet is how to ask "what is something in Deutsch", "Wie heibt das ... auf Deutsch?" Nov 08 17:36:41 Devastator: why are you starting to learn German? Nov 08 17:36:57 oc80z: thanks Nov 08 17:37:23 keep on keepin on bro Nov 08 17:38:10 Hauke I like the language and I intent to do some courses in .de Nov 08 17:38:47 tripolar einfach can be translated as "very easy", no? Nov 08 17:43:29 I think German is a difficult language at least when you compare it to English, German has different forms for the grammatical cases and different forms for times Nov 08 17:44:24 indeed Nov 08 17:44:53 english is one of the most difficult languages around Nov 08 17:45:00 i like challenges Nov 08 17:54:45 well Nov 08 17:55:20 if you dont know english, programming is going to be tough(er).. (i think...) Nov 08 17:55:35 just because, theres a lot (obviously im bias) of documentation done only in english Nov 08 17:56:58 learning Deutsch is being exciting so far Nov 08 17:59:40 unfortunately I'm slow learner hehe Nov 08 18:01:00 Hauke but I do understand what you say, my teacher was explaining anschreiben, prefix an means it should be written in something vertical, like a wall Nov 08 18:09:53 not to confuse with "Anschreiben" though Nov 08 18:11:38 hum, different meaning if uppercase, right? Nov 08 18:15:59 yes Anschreiben is a noun and means cover letter Nov 08 18:16:24 thanks for the tip Nov 08 18:16:55 I only know 3 words properly hehe Nov 08 18:17:27 Radiergummi, Wörterbuch and Kuli Nov 08 18:18:24 if writing is already complicated for me, you can imagine speaking Nov 08 18:23:20 <_trine> DonkeyHotei, I find English easy :) Nov 08 18:23:47 native speakers generally do Nov 08 18:24:09 <_trine> I have another openwrt problem Nov 08 18:25:47 <_trine> if I use reboot from my serial console when the dockstar has rebooted it has no eth0 Nov 08 18:25:59 <_trine> everytime Nov 08 18:28:02 DonkeyHotei: i think i have found something Nov 08 18:28:12 oh? Nov 08 18:28:54 DonkeyHotei: it is starting to look like killing the br2684ctl helps Nov 08 18:29:14 but that's in userspace Nov 08 18:30:33 indeed it is, but surely something running in the user space can cause what ever is happening in the kernel? Nov 08 18:31:02 that would be the dsl driver, no? Nov 08 18:31:32 i did remove the DSL stuff from the kernel, but i left all the atm stuff Nov 08 18:31:59 and it still leaked? Nov 08 18:32:11 yeah slab was accumulating like usual Nov 08 18:32:22 then it must be in the atm stuff Nov 08 18:32:29 but it stops when i kill br2684ctl Nov 08 18:32:36 DonkeyHotei: im thinking ltq_atm_vr9 Nov 08 18:32:59 seems highly likely Nov 08 18:33:23 other atm related stuff would be modules pppoatm and the br2684 Nov 08 18:33:28 except not limited to vr9, since i see the same thing on ar9 Nov 08 18:34:15 ar9 and vr9 are quite similar? Nov 08 18:34:35 yes, but different drivers i think Nov 08 18:36:15 the linux-atm codebase is so old its also hard to see there would be a bug like this in it Nov 08 18:37:22 i have an idea which i will test Nov 08 18:37:53 i started br2684ctl back up to see if the slab starts accumulating again Nov 08 18:43:03 Anyone have experience getting Avahi up and running for advertising services? Nov 08 18:43:16 <_trine> http://pastebin.com/XD6pwS62 Nov 08 18:43:41 <_trine> huh huh Nov 08 18:45:02 aww _trine you are out of memory too? Nov 08 18:45:19 <_trine> looks like it Nov 08 18:45:29 <_trine> could this be zram Nov 08 18:55:33 <_trine> whenever I send reboot from cli in my serial console eth0 never comes back Nov 08 18:55:46 JyZyXEL: which slabinfo line did you see the growth in? Nov 08 18:56:39 DonkeyHotei: it was the kmalloc- lines, i think -128 and -8192 were the worst Nov 08 18:56:52 but also increase in the ones between those Nov 08 18:58:09 how quick was the growth? Nov 08 19:00:23 <_trine> what does this mean root@OpenWrt:/# [ 164.086124] [sched_delayed] sched: RT throttling activated Nov 08 19:01:08 i haven't really measured it accurately yet, something like 300Ki per 5 minutes Nov 08 19:07:31 we may have more than one leak Nov 08 19:12:21 i don't have anything connected to my DSL port so that might affect it too, perhaps if there was a connection this would not occur Nov 08 19:15:33 JyZyXEL: i tried turning off ppp and it's actually growing faster in the -128 than you said, but no growth in -8192 Nov 08 19:16:11 i don't have anything connected either, but i highly doubt that could fix it Nov 08 19:36:26 tripolar r38694 packages/net/ tor/Makefile tor-alpha/Makefile * [packages] tor[-alpha] add fix to compile with SSP support Nov 08 20:09:30 <_trine> I am not allowed to buy a nexus 5 from Google at U-Spying-Association unless I give them copies of my driving license and passport Nov 08 20:09:50 <_trine> unbelievable Nov 08 20:28:35 not buying will probably put you on a watchlist too :) Nov 08 20:51:48 build #436 of orion is complete: Failure [failed compile_8] Build details are at http://buildbot.openwrt.org:8010/builders/orion/builds/436 Nov 08 21:08:18 <_trine> jow_laptop, but would you let them have a copy of your passport? Nov 08 21:09:15 <_trine> Google is just a private company what right do they think they have to demand that Nov 08 21:10:00 <_trine> Google employees won't even tell you their second name Nov 08 21:11:10 _trine: I probably wouldn't Nov 08 21:11:20 <_trine> nor me either Nov 08 21:11:38 especially since its illegal in germany to make (non-offically signed) copies of id documents Nov 08 21:12:22 this includes scanning, fotocopying etc. Nov 08 21:12:43 <_trine> it makes me laugh when Americans always accused the Chinese of backdooring hardware now we really know who the culprits are Nov 08 21:15:43 the Americans are just scared that the Chinese do the same as they do ;-) Nov 08 21:16:02 <_trine> If we handed over copies of our documents we would probably many trines and jow_laptop's turning up in Pakistan and Afghanistan Nov 08 21:16:15 <_trine> find ^^^ Nov 08 21:18:23 <_trine> Julian Assange Bradley Manning and Edward Snowden are some of the bravest people I have ever heard of in all my life Nov 08 21:19:24 <_trine> they knew the consequences before they did what they did but still went ahead,, I applaud them Nov 08 21:21:34 <_trine> and jow_laptop if you knew more about me you would not think to ask if I am on there watched list you would know :) Nov 08 21:24:07 <_trine> and now you also are on their watch list for knowing me lol Nov 08 21:40:27 <_trine> I have had to revert back to an older revision because Trunk runs really badly on my dockstar Nov 08 21:53:34 this is irc fool Nov 08 21:54:02 ? Nov 08 21:54:16 why is trine angry at u? Nov 08 21:54:38 <_trine> whats U Nov 08 21:55:01 an upercase vowel? Nov 08 21:55:19 blogic r38695 trunk/package/system/procd/Makefile * procd: update to latest git head Nov 08 21:55:35 <_trine> could be Nov 08 22:17:37 <_trine> jow_laptop, this MK908 reports an isane RF power level of 1.6 watts and its as deaf as a post Nov 08 22:17:44 <_trine> insane Nov 08 23:27:06 just got back from class, wow Nov 08 23:27:43 my brain had to kill some processes to not panic :P **** ENDING LOGGING AT Sat Nov 09 02:59:58 2013