**** BEGIN LOGGING AT Tue Mar 17 02:59:57 2020 Mar 17 03:20:02 Oh great. transmission uses a lot of memory when using libcurl compiled with mbedtls... Mar 17 05:53:19 build #262 of x86/legacy is complete: Failure [failed pkgbuild] Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Flegacy/builds/262 blamelist: Martin Blumenstingl , Tomasz Maciej Nowak , Luis Araneda , Yanase Yuki , ?lvaro Fern?ndez Mar 17 05:53:19 Rojas , Mathias Kresin , Sungbo Eo , Michael Heimpold , Adrian Schmutzler , Hans Dedecker , Kevin Darbyshire-Bryant , Robert Marko , Petr ?tetiar , Rozhuk Ivan Mar 17 05:53:19 , Aleksander Jan Bajkowski Mar 17 07:20:10 mangix: my position has always been to have as few subtargets as possible Mar 17 07:20:44 mangix: but in the end its up to the others to decide Mar 17 08:18:18 jow I think there is two points here. The Kernel is getting two big for some of the routers and it's a shame to hold back the performance of the other routers because of the wrt1900ac-v1. Mar 17 08:22:42 What do others use here for irc bouncer provision? Is there any service/server you would recommend? Mar 17 08:23:07 preferably inexpensive Mar 17 08:24:50 I'm using znc on my openwrt router ;) Mar 17 08:26:46 ldir: i have experience with znc as well. just as a user though, not with setup. Mar 17 08:28:05 I run znc on my own router but occasionally it gets rebooted for test purposes and I'd like to avoid the disruption to the group. Mar 17 08:28:28 ldir: I've an edis.at vps for 5eur/months Mar 17 08:28:47 occasionally using it for various vpn tests too Mar 17 08:29:45 and that simply runs irssi in a screen session Mar 17 08:32:06 What's the possibility of znc running somewhere on the openwrt infrastructure - it's a wild speculative question?! Mar 17 08:33:09 if you or someone else volunteers to take care of it Mar 17 08:33:26 firing up a digitalocean instance is simple enough Mar 17 08:36:38 ok - let's take this offline for the moment Mar 17 08:50:45 Hi is the forum the Appropriate place for job / contract requests? Mar 17 09:03:11 ldir: matrix ftw! Mar 17 09:03:41 putting jokes aside, I'm also irssi+screen positive Mar 17 10:14:06 I'm considering writing a small driver (for mikrotik devices, but I think it can be generalized) that would present the radio calibration data in sysfs (thus removing the need for extraction scripts): is this a bad idea? It seems rather obvious to me so I'm thinking there must be a reason why it hasn't been done already? Mar 17 10:14:43 (on these devices the cal data is stored on flash) Mar 17 10:29:07 systemd, oh systemd, what the fu*k are you doing? Mar 17 10:30:06 enlighten us? :) Mar 17 10:30:21 stintel: bees and flowers, you know the story Mar 17 10:30:38 day for into the apocalypse, still no zombies Mar 17 10:30:43 *day 4 Mar 17 10:31:08 I actually just ditched Docker because it is still not compatible with unified cgroups, and I refused to build systemd with hybrid-cgroup option Mar 17 10:32:41 I'm trying to get znc as a service on a debian virtual box. It says "service not running" but, lo there's a znc process running just fine(tm) Mar 17 10:32:44 last thing I need to fix is that network is brought down before iSCSI targets are disconnected, causing a hang afterwards. Mar 17 10:33:21 https://wiki.znc.in/Running_ZNC_as_a_system_daemon Mar 17 10:34:53 blogic, stintel: feelings about my proposal above? Mar 17 10:35:10 * f00b4r0 being stranded at home with very little "work" to do has time to spare ;) Mar 17 10:35:31 f00b4r0: snap! Mar 17 10:35:46 hehe Mar 17 10:35:52 haven't really followed. for me nothing changed, I work remotely most of the time :P Mar 17 10:35:59 stintel: snap Mar 17 10:36:07 but fulltime so not much openwrt time :( Mar 17 10:36:13 heh Mar 17 10:36:14 stintel: only difference, kid home whole day and fridge/kitchen full of food Mar 17 10:36:22 f00b4r0: hmm Mar 17 10:36:28 why not use a script ? Mar 17 10:36:47 like where is the claib data located ? Mar 17 10:37:50 claib - that's a new one Mar 17 10:38:00 it's stored compressed on flash. Current scripts wear the flash by (repeatedly, afaict) extracting the caldata to overlay Mar 17 10:38:43 a small driver would remove the need for (slow) scripts, reduce flash wear and declutter. Sounds ideal, no? :) Mar 17 10:39:24 f00b4r0: reading does not wear Mar 17 10:39:29 writing does Mar 17 10:39:36 and write will only happen upon first boot Mar 17 10:39:42 at every boot Mar 17 10:39:48 ?! Mar 17 10:39:52 then the script is broken Mar 17 10:40:02 and or it should just use /tmp Mar 17 10:40:14 regardless of that you could go further and just handle it via dts Mar 17 10:40:26 so that ath9/10k can just be pointed at the dts node Mar 17 10:40:38 same as mtd-eeprom works Mar 17 10:40:42 blogic: how do you deal with compression? Mar 17 10:41:01 i was thinking of doing it via dts but my impression was that to uncompress you need an intermediary driver Mar 17 10:41:25 yes Mar 17 10:41:31 how is it compressed ? Mar 17 10:42:36 varies from machine to machine. RLE encoding with/without LZO Mar 17 10:42:55 kernel should be able to decompress that Mar 17 10:45:30 more precisely (from my current understanding): a specific mtd segment contains (among other things) at some random but identifiable offset RLE/LZO ERD data, which itself contains calibration Mar 17 10:45:58 currently this is handled by package/utils/rbextract/src/rbextract.c Mar 17 10:48:14 and what I aim at getting rid of is mostly stuff like hotplug.d/firmware/10-ath9k-eeprom. If it can be done via DTS it should, I guess :) Mar 17 10:50:47 (that one doesn't write at every boot) Mar 17 10:53:25 another key benefit is that it would save precious flash space Mar 17 13:12:53 can some1 help me with makefile? i need to change it so it will access source code from local disk Mar 17 13:13:43 Guest79320: https://openwrt.org/docs/guide-developer/packages#working_on_local_application_source Mar 17 13:40:17 f00b4r0: isn't that the previous way of extracting caldata? Mar 17 13:41:14 when I looked at porting the hap-acĀ², I think I stumbled into custom kernel-space that decoded the caldata and exported it through sysfs Mar 17 13:41:50 let me find it again Mar 17 13:53:45 f00b4r0: got it, it gets exported as /sys/firmware/routerboot/ext_wlan_data Mar 17 13:54:08 target/linux/ar71xx/files/arch/mips/ath79/routerboot.c Mar 17 13:54:41 well, it seems to be done that way only for ar71xx Mar 17 13:56:27 I think rbextract was introduced in ath79 because maintaining this in kernel space was a pain Mar 17 13:56:44 newer mikrotik devices have really weird caldata encoding Mar 17 14:30:03 zorun: i would very much prefer to either have it entirely done via DTS or in kernel space, versus having the overhead of doing the dance in user space and wasting flash space. I'll take a better look at the code, to see if it can be made upstreamable. rbextract.c apparently builds on the ar71xx code which is rather ugly. Mar 17 14:30:11 I'm hoping it can be cleaned up Mar 17 15:04:56 f00b4r0: fully agree, the routerboot.c thing was not very clean in the first place Mar 17 15:05:54 f00b4r0: especially because now mikrotik devices spread over several targets (ar71xx & ath79, but also ramips and ipq40xx) Mar 17 15:06:06 if this can be made generic & working in DTS, that would be awesome Mar 17 15:07:11 zorun: agreed. I need to get a better understanding of the other targets (I'm only familiar with the atheros ones), and I need to figure out if there's enough plumbing already in place or if some extra glue is required. Mar 17 15:07:18 luckily I now have time to do that :) Mar 17 15:08:43 :D Mar 17 15:09:34 the problem with confinement is the backlog of "things to do when I will have time" Mar 17 15:10:01 this, and maintaining the various infrastructure that gets hit by a surge in traffic Mar 17 15:14:48 zorun: indeed. My ISP kindly provides a visualization tool for their network backbone load. It's an interesting watch :) Mar 17 15:18:23 ah, that looks like K-Net :) Mar 17 15:18:40 I like their public AS-stats Mar 17 15:42:00 zorun: k-net it is. Best ISP i've had the opportunity to use so far Mar 17 15:42:09 way ahead of the big players in .fr Mar 17 15:42:20 quality of service (and customer service) wise Mar 17 17:07:34 new ath10k-ct firmware and driver posted to mailing list, testing is appreciated! Mar 17 17:41:21 greearb__: a minor nitpick if you do not mind, the official project name is OpenWrt :) Mar 17 17:42:27 wasn't when folks started accepting patches some time back... Mar 17 17:43:00 next time I have too much time I can change my script Mar 17 17:46:36 Probably just nobody noticed or deemed necessary to say. Mar 17 17:48:55 nod Mar 17 17:49:03 http://git.openwrt.org/562c3db31d84832110e3c17a7f171e4d2a02140d Wed Feb 23 21:06:26 2005 Mar 17 19:36:14 build #290 of x86/geode is complete: Failure [failed pkgbuild] Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Fgeode/builds/290 blamelist: Martin Blumenstingl , Roger Pueyo Centelles , Tomasz Maciej Nowak , Leon M. George , ?lvaro Mar 17 19:36:14 Fern?ndez Rojas , Adrian Schmutzler , Mathias Kresin , Sungbo Eo , Michael Heimpold , Luis Araneda , Hans Dedecker , Rozhuk Ivan , Robert Marko , Petr ?tetiar Mar 17 19:36:14 , Aleksander Jan Bajkowski Mar 17 20:34:15 hiI've updated today all of my router packages and now l2tp ipsec tunnel is not bring upi've getting error regarding /usr/sbin/pppd: unrecognized option ''can someone help me please? **** ENDING LOGGING AT Wed Mar 18 02:59:58 2020