**** BEGIN LOGGING AT Fri Feb 28 02:59:57 2020 Feb 28 06:20:42 hmm. my ath79 ubnt-bullet-m-ar7240's are flapping between 100Mbps and 10Mbps/half duplex: Feb 28 06:20:45 [296317.002429] eth0: link up (10Mbps/Half duplex) Feb 28 06:20:47 [296898.356445] eth0: link up (100Mbps/Full duplex) Feb 28 06:21:17 581 seconds! Feb 28 07:18:35 jow: thanks for taking care of that! Feb 28 08:06:54 jow: pong Feb 28 08:39:48 *yawn* Feb 28 08:39:58 * blogic splept for 15 hours last night Feb 28 08:40:03 and I am still tired Feb 28 09:02:57 hm, 6h37m is my average for past week Feb 28 09:19:20 ldir: jow: IIRC GitLab has runners for macOS Feb 28 09:23:41 ldir: jow: I'm already building my staging trees via GitLab for some time with openwrt-ci https://gitlab.com/ynezz/openwrt-ci/-/commit/43547239fc0c1604dc5ac92ebe324b048873ba71 Feb 28 09:24:18 wigyori: did you recently upgrade pielstick or did other modifications to the system? Feb 28 09:24:52 wigyori: can you please cd into one of its builddirs and see if you can execute "mkhash" or if it fails? Feb 28 09:25:29 ynezz: maybe but you still need an os x system to begin with Feb 28 09:28:38 "the builds will be able to do UI interactions, making it possible to run and test on the iOS simulator." Feb 28 09:28:51 so I assume, that it's just a matter of configuration Feb 28 09:29:05 the ios simulator only runs on os x Feb 28 09:29:11 besides we don't want to build for ios Feb 28 09:30:17 I mean, gitlab runner offers few modes of execution for jobs Feb 28 09:30:33 yeah sure, but you still need an os x system to install it on Feb 28 09:31:00 I assume lidr has some macOS hardware Feb 28 09:35:23 so it should probably be a gitlab runner running on macOS and configured with shell executor Feb 28 09:38:06 gitlab runner is another term for buildbot slave Feb 28 09:40:59 https://gitlab.com/ynezz/openwrt/-/blob/28daec298016e44ea4dd8154fa1d3acf26a3f0c4/.gitlab-ci.yml contains example configuration which builds the tree only for bcm27xx/bcm2708 Feb 28 09:41:29 putting . in front of the config disables it Feb 28 09:43:56 then its quite easy to schedule the builds with proper branch/.gitlab-ci.yml every once a while via GitLab pipeline scheduling, like it's for example done with Docker https://gitlab.com/openwrt/docker/pipeline_schedules Feb 28 09:46:01 I assume, that macOS build host would need some tweaks here https://gitlab.com/ynezz/openwrt-ci/-/blob/master/openwrt-ci/gitlab/target.yml Feb 28 09:46:26 I understand all that, but having a macbook run 24/7 as buildslave/worker/runner is not ideal Feb 28 09:46:42 is it needed? Feb 28 09:47:06 I mean, ldir can run it manually when needed Feb 28 09:47:09 if you don't want to see a lot of interrupted builds because the device was put to sleep or shut down Feb 28 09:47:15 then yes Feb 28 09:47:41 it would still be some improvement to current situation :) Feb 28 09:47:52 but yes, ideally there would be some dedicate hardware Feb 28 09:47:52 maybe Feb 28 09:48:07 maybe we could ask TravisCI for account? Feb 28 09:48:16 they've macOS runners Feb 28 09:48:34 ok Feb 28 09:49:13 ideally macOS runs under qemu :p Feb 28 09:49:32 dunno of virtualbox can run headless somewhere on server Feb 28 09:49:37 it can Feb 28 09:51:58 https://github.com/kholia/OSX-KVM Feb 28 09:52:05 qemu/kvm works too Feb 28 09:52:41 ah thanks for that link. I've been looking into that many many years ago but never finished that Feb 28 09:53:00 nice https://github.com/kholia/OSX-KVM/blob/master/fetch-macOS.py Feb 28 09:53:01 :) Feb 28 09:53:10 might pick it up again now, I've got plenty of RAM and most people at my client's company are running OS X Feb 28 09:53:24 could be very handy for some VPN troubleshooting Feb 28 09:53:45 guerby: ping Feb 28 09:55:41 https://www.macstadium.com seems like dedicated/managed macmini is $79/month Feb 28 09:56:19 less maintenance hassle Feb 28 09:58:46 seems like TravisCI is running there Feb 28 09:59:00 https://www.macstadium.com/customers/travis-ci Feb 28 10:07:26 jow: I saw your email, I'm replying Feb 28 10:07:51 zorun: oh, thanks :) Feb 28 10:25:09 jow: any objections against this https://patchwork.ozlabs.org/patch/1230746/ ? Feb 28 10:25:28 I'll just adjust the commit message Feb 28 10:26:42 just noticed, that you've already solved the DMARC crap directly in gitolite, nice Feb 28 10:29:14 I'm asking myself what do we gain with that pkg-config -> pkgconfig change, it seems like both projects are not maintained, so why to bother Feb 28 10:29:27 referencing this https://patchwork.ozlabs.org/patch/1202266/ Feb 28 10:30:13 jow, pong Feb 28 10:38:41 guerby: nvm, zorun already replied Feb 28 10:38:46 sorry for bothering Feb 28 10:39:10 ynezz: in doubt I'd stick with the old one Feb 28 10:39:20 regression potential is quite high for such a low-level variant Feb 28 10:41:34 Sorry chaps, been AFK for a while. Re macos etc. I don't know what to say because I'm not sure I know what to do / what I want to do. What I have is a 2014 macmini running catalina 2.6 dual core 8GB ram with 1TB SSD doing very, very little. Feb 28 10:43:57 It would be good to get it doing 'something'. Yesterday's barely thought out scheme was to get it to build my staging tree for all targets, AIUI a 'phase1' style not quite buildbot Feb 28 10:47:12 https://www.anandtech.com/show/15554/amd-launches-ultralowpower-ryzen-embedded-apus-starting-at-6w Feb 28 10:48:07 time for an apu5 :) Feb 28 10:49:59 I can certainly offer it to you guys on a trial/test basis. And I can already tell you that arc700 doesn't build Feb 28 10:52:27 https://paste.ubuntu.com/p/dTq9pHPCfF/ Feb 28 11:02:13 ynezz: I'd need some guidance on the gitlab ci thing that's for sure :-) Feb 28 11:19:19 ldir: in short 1. create a gitlab.com account, 2. fork openwrt there, 3. setup gitlab runner (here I'm unsure, but it should be with shell executor), 4. add .gitlab-ci.yml and push that into your fork/branch (done in step 3.) Feb 28 11:20:25 ok, and this 'gitlab runner' it runs on my machine or theirs ? Feb 28 11:21:01 yours Feb 28 11:21:03 (I have step 1, just not done much with it :-) Feb 28 11:21:30 https://docs.gitlab.com/runner/install/osx.html Feb 28 11:21:36 ok, will see how I get on :-) Feb 28 11:22:08 gitlab.com is build master, your runner is build slave, master qeues jobs to slaves Feb 28 11:24:42 and I recommend you to disable the shared runners (they're all x86/64 based and slow for building OpenWrt) Feb 28 11:26:55 certainly not able to build OpenWrt under 2hrs, didn't tried it past that 2hrs timeout Feb 28 11:40:10 jow: haven't touched the host apart fro mthe normal upgrades recently - will check Feb 28 11:43:33 wigyori: in particular I am trying to debug this: https://downloads.openwrt.org/snapshots/faillogs/x86_64/base/iw/tiny/compile.txt Feb 28 11:43:57 wigyori: according to scripts/download.pl, that "Could not generate file hash" message appears when mkhash failed to produce the expected output Feb 28 11:48:22 wigyori: maybe you could get me the output of the following command: strace -s 1024 -f ./scripts/download.pl "/tmp" "iw-5.4.tar.xz" "a2469f677088d7b1070a7fbb28f3c747041697e8f6ec70783339cb1bc27a395f" "" "@KERNEL/software/network/iw" Feb 28 12:11:39 jow: not sure if buildbots are different, but running it gives me "Use of uninitialized value $ENV{"TOPDIR"} in concatenation (.) or string at ./scripts/download.pl line 38." and "sh: mkhash: command not found" Feb 28 12:22:29 yeah, my fault. you need to export TOPDIR=$(pwd) before Feb 28 12:22:39 its normally invoked by buildroot with an initialized env Feb 28 12:31:33 jow: after poking at it, TOPDIR not defined seems to be harmless, but I needed to add staging_dir/host/bin to PATH Feb 28 12:52:45 [ 75.855156] Generic PHY mdio-bus.0:1f:04: Master/Slave resolution failed, maybe conflicting manual settings? Feb 28 12:53:17 that was from an ifconfig eth0 down ; ifconfig eth0 up Feb 28 12:53:21 whut? Feb 28 12:56:20 weirdly, "ip link set eth0 down ; ip link set eth0 up" and i don't see the same thing Feb 28 13:28:00 blogic: it looks like the two commits for "enhance wifi reload" kill wifi on ipq806x (r7500v2) Feb 28 13:28:10 reverting them brings it back Feb 28 13:28:47 flyspray? Feb 28 13:31:51 that "Master/Slave" message comes from drivers/net/phy/phy_device.c and can be reached only if "phydev->supported & (SUPPORTED_1000baseT_Half | SUPPORTED_1000baseT_Full)", but this is a maximum 100Mbps interface. Feb 28 13:33:50 error messages from logcat look like: hostapd: Line 19: unknown configuration item 'radio_config_id' Feb 28 13:39:04 nmrh: details please ? Feb 28 13:39:14 ah Feb 28 13:39:22 nmrh: make package/hostapd/clean && make Feb 28 13:46:40 k, give me a sec Feb 28 13:48:45 blogic: i assume you want me to undo my reverts, then remake hostapd... Feb 28 14:05:12 slow "build box", its done now, brb Feb 28 14:08:14 blogic: after undoing reverts, clean package hostapd && make fixes it Feb 28 14:08:28 blogic: ty Feb 28 14:08:46 let me know if you want a bug report for this Feb 28 14:46:04 exit Feb 28 15:18:54 as there is no bug and it worked at the end i dont needs a bug report Feb 28 15:35:52 ynezz: ping - slowly moving forward with gitlab runner - compat request for you in target-build.mk :-) Feb 28 16:05:32 ldir: yep, patches welcome :) Feb 28 16:12:02 nice "Running on Kevins-MacMini.lan.darbyshire-bryant.me.uk..." Feb 28 16:14:18 it's not so nice since it goes kabooom shortly after :-) Feb 28 16:15:03 implementation detail Feb 28 16:17:19 lol! Feb 28 16:18:43 if you run it directly on the host it works? Feb 28 16:18:55 `xargs: illegal option -- r` looks strange Feb 28 16:20:08 `xargs -r` works just fine on obsd Feb 28 16:36:58 xargs -r is bogus on darwin :shrug: Feb 28 16:37:38 I suspect this is going to be a long trawl through things where I end up wondering how anything built in the first place Feb 28 16:41:12 but in the end... :) Feb 28 16:46:30 any ideas what all this "fatal: Invalid revision range ee53a240ac902dc83209008a2671e7fdcf55957a..HEAD" stuff is on about ? Feb 28 16:48:13 ah, ok, I guess it's a 'shallow clone' thing Feb 28 16:48:32 yes, scripts/getver.sh needs some silencer Feb 28 16:53:34 * xback closes his eyes, and presses THE button Feb 28 16:53:57 nothing happens Feb 28 16:54:51 xback: was it supposed to implement world peace? Feb 28 16:55:04 * xback shelters for a forecasted shitstorm (you should too!) Feb 28 16:55:19 or even better, make me a programmer? Feb 28 16:55:39 * ldir does not do a git fetch --all Feb 28 16:55:47 woo, christmas! Feb 28 16:57:33 xback: do the numbers 5 & 4 have anything to do with this? And is there space under your shelter? Feb 28 16:59:14 ldir: yeah Feb 28 16:59:36 There are 5 places, so room for 4 more people ;-) Feb 28 17:00:25 pah, it'll be fine! Feb 28 17:01:02 wigyori: I noticed that the build failure occurs on all builders Feb 28 17:01:16 something seems to have gone broken on the buildroot side Feb 28 17:03:53 so soon after my push? :o Feb 28 17:03:59 ;-) Feb 28 17:05:03 actually nevermind, I looked t the wrong logs Feb 28 17:05:16 the bahviour is still puzzling Feb 28 17:06:05 as is my typing today Feb 28 17:06:15 * ldir starts a build for x86_64 apu2 with bells & whistles Feb 28 17:22:53 oh, I have to set 'USE_TESTING_KERNEL' Feb 28 17:24:24 that failed quickly Feb 28 17:47:07 failing further along... much further along Feb 28 17:49:48 * ldir scratches head and gets splinters again https://paste.ubuntu.com/p/7FPDG8P2jx/ Feb 28 18:02:19 for a bit more context https://paste.ubuntu.com/p/fS9NNtsSjN/ - this is at the 'end' of the build process. Feb 28 18:26:46 mc Feb 28 18:26:52 oops Feb 28 19:24:16 Hauke: as kernel 5.4 now uses the upstream driver - are you planning on patching support for the ath79 gmac configuration into the upstream driver? Feb 28 19:24:30 If I'm not mistaken, this is currently missing Feb 28 19:26:46 Also: the mpc85xx target currently has some issues left, which i already have patches for. Just testing them rebased for generic and p1020 Feb 28 19:30:00 blocktrron: I added some patche to make it work with my device Feb 28 19:30:06 it probably needs some more fixes Feb 28 19:30:31 In the end I would like to use the upstream driver instead of maintaining or own version Feb 28 19:49:58 Hauke: +1 for using the upstream driver Feb 28 20:05:13 build #248 of mpc85xx/p1020 is complete: Failure [failed images] Build details are at http://buildbot.openwrt.org/master/images/builders/mpc85xx%2Fp1020/builds/248 blamelist: Mark-MC Lee (???) , John Crispin , Xu Wang , Koen Vandeputte , David Bauer @david-bauer.net>, Christian Lamparter , Sungbo Eo , Adrian Schmutzler , Roger Pueyo Centelles , Kevin Darbyshire-Bryant , ?lvaro Fern?ndez Rojas , Jo-Philipp Wich , Hauke Mehrtens Feb 28 20:40:56 Invalid compression, supported are: [...] Feb 28 20:41:09 apparently a bad mkimage invocation Feb 28 21:39:16 testing 5.4 kernel today got me "Kernel panic - not syncing: Unexpected DSP exception" on ath79 ;-( Feb 28 21:39:30 and a reboot :P Feb 28 21:48:49 jow: yes, there's a patch from the applied micro target which works around the issue Feb 28 21:48:59 Will push that with more fixes shortly Feb 28 22:11:13 aarrrghh I don't understand. macos' xargs doesn't understand -r option. But why do I only see the build barf when trying to run either on a gitlab-runner OR if I manually try to build the gitlab based repo. Feb 28 22:11:55 it makes no sense. Feb 28 22:13:38 why don't my normal builds try to invoke 'xargs -r' at some point and barf? Feb 28 22:18:59 different paths, different xargs? Feb 28 22:35:25 light is slowly dawning - looks like it comes down to build option Automatic removal of build directories which I don't have selected but buildbots appear to do so... as a result xargs is used to tidy up... and macos xargs blows up. Feb 28 23:29:37 build #81 of x86/generic is complete: Failure [failed kmods] Build details are at http://buildbot.openwrt.org/openwrt-18.06/images/builders/x86%2Fgeneric/builds/81 blamelist: Koen Vandeputte , Jo-Philipp Wich , Petr ?tetiar Feb 29 00:22:26 build #127 of kirkwood/generic is complete: Failure [failed updatefeeds] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/kirkwood%2Fgeneric/builds/127 Feb 29 01:00:18 build #128 of kirkwood/generic is complete: Success [build successful] Build details are at http://buildbot.openwrt.org/openwrt-19.07/images/builders/kirkwood%2Fgeneric/builds/128 Feb 29 01:40:08 build #270 of mpc85xx/generic is complete: Failure [failed images] Build details are at http://buildbot.openwrt.org/master/images/builders/mpc85xx%2Fgeneric/builds/270 blamelist: Mark-MC Lee (???) , John Crispin , Koen Vandeputte , David Bauer , Christian Feb 29 01:40:08 Lamparter , Steven Lin , Adrian Schmutzler , Kevin Darbyshire-Bryant , ?lvaro Fern?ndez Rojas , Eneas U de Queiroz , Jo-Philipp Wich , Hauke Mehrtens Feb 29 02:15:16 anyone know how to export a variable globally? Feb 29 02:34:30 build #280 of mvebu/cortexa72 is complete: Failure [failed pkgbuild] Build details are at http://buildbot.openwrt.org/master/images/builders/mvebu%2Fcortexa72/builds/280 blamelist: Mark-MC Lee (???) , John Crispin , Koen Vandeputte , David Bauer , Christian Feb 29 02:34:30 Lamparter , Steven Lin , Adrian Schmutzler , Kevin Darbyshire-Bryant , ?lvaro Fern?ndez Rojas , Eneas U de Queiroz , Jo-Philipp Wich , Hauke Mehrtens Feb 29 02:41:31 jow: can you look at https://github.com/openwrt/packages/issues/10731 ? **** ENDING LOGGING AT Sat Feb 29 02:59:59 2020