**** BEGIN LOGGING AT Fri Jun 26 03:01:39 2020 Jun 26 03:40:14 build #422 of x86/generic is complete: Failure [failed kmodconfig] Build details are at http://buildbot.openwrt.org/master/images/builders/x86%2Fgeneric/builds/422 blamelist: Adrian Schmutzler Jun 26 04:03:07 build #424 of ramips/rt288x is complete: Failure [failed defconfig dltar] Build details are at http://buildbot.openwrt.org/master/images/builders/ramips%2Frt288x/builds/424 blamelist: Adrian Schmutzler Jun 26 05:07:18 build #361 of ath79/generic is complete: Failure [failed defconfig dltar] Build details are at http://buildbot.openwrt.org/master/images/builders/ath79%2Fgeneric/builds/361 blamelist: Adrian Schmutzler Jun 26 05:08:47 build #358 of samsung/s5pv210 is complete: Failure [failed defconfig dltar] Build details are at http://buildbot.openwrt.org/master/images/builders/samsung%2Fs5pv210/builds/358 blamelist: Adrian Schmutzler Jun 26 07:30:00 Hm… I sent a patch yesterday but it neither hit the mailing list nor patchwork. Should I resend, or wait? Jun 26 07:46:10 rsalvaterra: the lz4 one? Jun 26 07:46:32 jow: zstd, this time. :) Jun 26 07:46:42 So, yeah, it didn't hit the list at all. Jun 26 07:52:58 blogic: can we remove the getpid() check in procd.c to allow procd running with a different pid than 1? This is handy for container envs. Jun 26 07:53:06 dangole: -^ Jun 26 07:54:53 rsalvaterra: I checked the moderation queue and there was nothing in there Jun 26 07:56:43 Hm… I guess I need to resend, then. Jun 26 07:57:04 jow: do people have to resubscribe to the devel mailing list? There seem to happened some changes? Jun 26 07:57:04 dwmw2_gone: i'm 90% sure you are aware of this, but just to cover the last 10%, i noticed that git://git.infradead.org/linux-ubifs.git is giving me a "fatal: remote error: access denied or repository not exported: /linux-ubifs.git" Jun 26 07:58:19 aparcar[m]: only people subscribed since may / 2018 need to resubscribe Jun 26 07:58:28 jow: openwrt-devel is (or was) an open list, I'm sure. Jun 26 07:59:04 I never had to subscribe to send patches… :/ Jun 26 07:59:31 jow: can you manually add me? When I try to subscribe mailman spits out an error Jun 26 08:00:29 rsalvaterra: "Action to take for postings from non-members for which no explicit action is defined. Jun 26 08:00:34 ": "Accept" Jun 26 08:00:45 aparcar[m]: what error? Jun 26 08:02:10 jow: We're sorry, we hit a bug! Jun 26 08:02:24 licitly inhibited, but the webmaster can find this information in the Mailman error logs or the web server logs. [201~ been explicitly inhibited, but the webmaster can find this information in the Mailman error logs or the web server logs. [201~ Jun 26 08:02:52 jow: http://lists.infradead.org/mailman/subscribe/openwrt-devel or is that server outdated? Jun 26 08:03:07 no Jun 26 08:04:00 jow: Phew! I thought I had to subscribe to yet another list, thanks. It's much more comfortable for me to read the archives. Jun 26 08:07:55 time budget used up already again, bbl Jun 26 08:08:32 bbl? Jun 26 08:08:42 Be back later, I guess. Jun 26 08:09:25 ack Jun 26 08:09:28 ty Jun 26 08:10:36 I don't know how I still remember the jargon, after being away from IRC for about 17 years… :P Jun 26 08:15:01 :) Jun 26 08:16:21 ah crap, disk full Jun 26 08:19:22 aparcar[m]: where's your irc chops? :P Jun 26 08:19:46 russell--: yes. Do you have an up to date clone that I can recreate it from? Jun 26 08:24:15 Borromini: I'm afraid I'm to youg for this :( Jun 26 08:24:44 aparcar[m]: < 30? Jun 26 08:26:34 Borromini: jup Jun 26 08:26:50 :) Jun 26 08:27:13 I'm able-to-recognise-connection-problems-by-the-modem-handshake-sound old. Jun 26 08:30:27 hehehe Jun 26 08:31:06 i must be getting old because i like irc way more than those GUI chat thingies Jun 26 08:31:52 Oh, IRC improved *a lot*, since most idiots stopped using it. Jun 26 08:32:11 hrm Jun 26 08:32:16 the list should let people subscribe again now Jun 26 08:32:20 Please let me know if not Jun 26 08:32:38 rsalvaterra: they're now on discord or telegram or some other resource hogging web 4.88 chat platform Jun 26 08:33:07 dwmw2_gone: But we can still send mail without subscribing, right? Jun 26 08:33:09 forgot "hipster" in that sentence Jun 26 08:33:19 yes, but not when the server's disk is full :) Jun 26 08:33:34 At least, I think so. I normally set lists up that way, but the list admins can tweak that. Jun 26 08:33:40 stintel: Thanks for letting me know what platforms to avoid. ;) Jun 26 08:33:51 dwmw2_gone: I can't subscribe, or at least don't get any mail :( Jun 26 08:33:57 when? Jun 26 08:34:23 I tried to subscribe with a new mail address Jun 26 08:34:29 when? Jun 26 08:34:35 Your subscription request has been received, and will soon be acted upon Jun 26 08:34:43 when? Jun 26 08:34:44 oh actually I need approval now Jun 26 08:35:00 * dwmw2_gone goes to do something else. Jun 26 08:35:12 Ok, patch resent. :) Jun 26 08:35:22 dwmw2_gone: 39 seconds ago Jun 26 08:36:29 dwmw2_gone: ?? Jun 26 08:37:22 "Your subscription request has been received, and will soon be acted upon" sounds a lot like "duly noted". :P Jun 26 08:37:28 what address? Jun 26 08:38:03 mail@aparcar.org and spooren@hawaii.edu Jun 26 08:38:13 it's so busy it isn't accepting mail right now from the outside, now I've unblocked it. Jun 26 08:38:22 which one do you want subscribed? both? Jun 26 08:38:31 the first one please Jun 26 08:38:40 also to openwrt-adm please Jun 26 08:39:53 done Jun 26 08:40:11 the round-trip of confirmation emails was going to take a while but I think it should be functional now Jun 26 08:40:20 will look at offloading some of the SMTP work from that box. Jun 26 08:44:00 dwmw2_gone: thank you! Jun 26 09:10:09 dwmw2_gone: Pipermail isn't working yet, is it? Jun 26 09:44:46 No joy… second try also failed, I presume. :( Jun 26 09:45:25 rsalvaterra: you mean archives? Jun 26 09:47:09 Yeah, the archives, at least. Jun 26 09:47:22 I'll turn the dmarc wrapping crap off. Jun 26 09:47:38 I have no way to know if the email was received otherwise (well, Gmail hasn't complained, though). Jun 26 09:56:36 heh, turns out _my_ listemail problems were a lostgit-send emailconfig here. it doesn't complain, Ijust get some console lines later that "there's no esmtp config file" :) Jun 26 10:04:18 I'm sure mine is configured correctly, I sent two patches yesterday, only the second is missing. Jun 26 10:04:36 dwmw2_gone: i'm pushing what i have to a github repository, i generally fetch every week or two, so it's fairly recent Jun 26 10:07:05 ahha! found my gitsendmail config had beenconfigured only locally in one repo! Jun 26 10:12:38 russell--: thanks Jun 26 10:13:21 rsalvaterra: another message just arrived. Jun 26 10:13:49 Yay! Jun 26 10:14:25 Yep, it's my patch, and it's already on pipermail. Jun 26 10:14:43 don't you get your own messages back anyway? Jun 26 10:18:21 Sure, I receive a copy, as git-email automatically includes the sign-off email. Jun 26 10:22:08 Jeez, this is driving me nuts… Jun 26 10:22:13 root@heimdal:~# cat /proc/interrupts | grep gpio Jun 26 10:22:13 77: 876575 0 f1018140.gpio 14 Level 8-0071 Jun 26 10:22:14 root@heimdal:~# uptime Jun 26 10:22:14 11:20:56 up 12:08, load average: 0.00, 0.00, 0.00 Jun 26 10:22:14 root@heimdal:~# Jun 26 10:24:18 876575 interrupts in 12h08m. This is about 20 irq/s. On the GPIO line. Why?! Jun 26 10:31:06 hm, I get various openwrt-devel mails without subject Jun 26 10:33:35 dwmw2_gone: https://github.com/RussellSenior/linux-ubifs (10 branches) Jun 26 10:40:20 most recent commit date i have is 2020-05-17 Jun 26 11:15:44 russell--: thanks Jun 26 13:15:11 oh wow I like that simplified luci interface Jun 26 13:20:07 sintel: simplified LuCI…? Jun 26 13:20:14 see ML Jun 26 13:20:59 * rsalvaterra digs in the archives Jun 26 13:21:29 i would like to reprocude this attack on wpa2 with an openwrt router > https://hashcat.net/forum/thread-7717.html Jun 26 13:22:01 A WRT1900AC, that was an unfortunate choice… :P Jun 26 13:22:04 is anyone familiar enough with hostapd to tell me if it is even possible? Jun 26 13:24:30 stintel: it's just the main page that's simplified, it seems. Jun 26 13:25:56 I'm not a GUI guy, but I have to agree it does look neat. Jun 26 13:26:05 sadly the artice and any other source i could find is very vague about what software/devices are affected. it only states, that pmkid is included in wpa handshake message 1. Jun 26 13:27:16 rsalvaterra: I'm a bit weird (if it comes to GUIs) ... I prefer mostly CLI, but also prefer a full-blown DE (using KDE Plasma) Jun 26 13:28:10 stintel: It's not just you. ;) Jun 26 13:30:32 When I'm using the computer to say, browse the web, I like a nice, consistent desktop environment. KDE is too bloated for my taste, though (I stopped using it after 3.5). I'm a classic GNOME 2 fan, so I obviously fell in love with MATE, a couple of years ago. Jun 26 13:31:15 I've been using gnome for a long time Jun 26 13:31:44 switched to KDE when 4 became usable again Jun 26 13:32:45 I'm not exactly fond of Qt… Jun 26 13:36:07 It's like that quote I read many years ago, "GNOME vs KDE in 50 words or less"… Jun 26 13:36:21 "If you believe reimplementing the C++ STL in C is a bad idea, you'll probably like KDE. If you believe the C++ STL is a bad idea, you'll probably like GNOME." Jun 26 13:41:08 too many negatives.... is this because g* libs ar eall c++ functionality in C, so if you're against that, you like kde, and kde is all plain c++, so you're agianst that you like gnome? Jun 26 13:43:15 karlp: It's just a joke, nothing more. ;) Jun 26 13:44:46 I'm not against C++ (even though I don't like it), but my problem with GTK vs Qt is the libraries. If you have a GTK-based desktop and you install a Qt app, you'll have to pull in a boatload of dependencies (and vice-versa). Jun 26 13:48:08 I don't have enough programming experience to have an opinion of this Jun 26 13:55:55 rsalvaterra: just trying to understand it :) Jun 26 13:56:15 the deps are only a problem the firsttime :)you end up with both the library sets sooner or later. Jun 26 14:21:33 ahhh, lua versioncompat is awesome: https://paste.jvnv.net/view/HxIMi Jun 26 14:25:59 Speaking of Lua… is it possible to use the JIT as a drop-in replacement for the interpreter? Jun 26 14:28:24 not really. Jun 26 14:29:44 Well, that's a bummer… IIRC, LuCI still uses the interpreter, right? Jun 26 14:38:31 sure, Jun 26 14:38:41 it's not like "lua sucks onlylua jit is worthusing" Jun 26 14:38:49 also, my quirk above is just a REPL issue: https://paste.jvnv.net/view/jFIxh Jun 26 18:20:44 zorun you mentored the students doing the simplified interface right? Jun 26 18:23:39 aparcar[m]: yes Jun 26 18:25:41 zorun: great work! Jun 26 18:25:53 Hope they got a A+ ;) Jun 26 18:29:57 thanks, they were the ones doing the work though, not me ;) Jun 26 18:30:58 we are really hoping for code review (because I know the code is not 100% fine), I managed to have them for a little longer so that they can improve based on feedback Jun 26 18:47:53 I'll test it, can't really review it Jun 26 18:58:17 testing is good also Jun 26 19:03:27 new to openwrt development... and trying to understand something... I have successfully built an application (mosquitto_pub) via the overall build system, but when I go into the directory (build_dir/target-mips_24kc_musl/mosquitto-ssl/mosquitto-1.6.10/client) and type the same gcc command that was used to create it, I'm seeing a linker error: "ld: Jun 26 19:03:28 warning: libssl.so.1.1, needed by ../lib/libmosquitto.so.1, not found (try using -rpath or -rpath-link)", though the library is in the first -L directory... anybody understand what's happening? Jun 26 20:11:10 i think i have a pretty noob question, as i'm not very familiar with the build system. there are some patches i would like to apply, but they are only available for another kernel version. obviously copying them into the patches-XX direcotry doesn't work out of the box, as they refer to another state of the source. Jun 26 20:15:05 what is the right procedure to make patches for the other kernel version? i assume i could somehow apply all the original patches manually? ( how would i do that?) and then i could edit the files manually and make a new patch from the diff? Jun 26 20:25:33 hbug: https://openwrt.org/docs/guide-developer/build-system/use-patches-with-buildsystem#adding_or_editing_kernel_patches Jun 26 20:26:22 anyone here familiar with the golang integration in OpenWrt? I'm having trouble with ldflags Jun 26 20:29:14 PaulFertser: thanks. wasn't aware there was a prepare make target. overlooked it somehow Jun 26 20:29:42 hbug: and then quilt, yes. It's not too hard to get used to. Jun 26 20:30:22 is there a guide which make targets exists and what they do? or is this generic procedure i'm just not used to? Jun 26 20:35:51 hbug: I guess everything like that is supposed to be part of https://openwrt.org/docs/guide-developer/ Jun 26 20:58:47 Just found out that the Linksys devices were renamed according to their common names a few months ago. Sysupgrade complains now. Should be safe to use sysupgrade -F ? Jun 26 21:16:20 adrianschmutzler: hey any updates on the JSON patch for 19.07? Jun 26 22:00:35 lnslbrty: it should. Jun 26 22:00:49 lnslbrty: which board is it? we could add it to the compatible list Jun 26 22:01:13 anybody know why running mips-openwrt-linux-musl-gcc outside of openwrt's make system might cause missing library warnings (although the library is there)? Jun 26 22:03:52 bemo: you mean the staging_dir warning? Jun 26 22:04:19 no (had that at first too, before setting the STAGING_DIR to my toolchain directory) Jun 26 22:04:36 I used some of the toolchains outside of openwrt. but not lately. what's your warning? Jun 26 22:04:47 "ld: warning: libssl.so.1.1, needed by ../lib/libmosquitto.so.1, not found (try using -rpath or -rpath-link)" Jun 26 22:05:29 (I'm essentially copying-and-pasting the gcc command that was used by openwrt to successfully build mosquitto_pub binary... but when *I* try it, I see the above ld error (for libssl and libcrypto) Jun 26 22:06:02 followed by a bunch of undefined symbols from those libraries Jun 26 22:06:17 ("undefined reference") Jun 26 22:06:42 was wondering if there might be some other env setting I'm missing Jun 26 22:07:25 i think that's ld_library_path. i only used the toolchain together with autotools. Jun 26 22:08:00 during build? Jun 26 22:08:40 bemo: just to understand your motivation. why can't you use the openwrt buildsystem? Jun 26 22:08:43 (I know about LD_LIBRARY_PATH for runtime...) Jun 26 22:10:24 technically, I'm trying to debug a different openwrt tree (based on 18.06.2) where the openwrt build is failing for the same reason... when I managed to reproduce it with the tip of openwrt mainline by manually using the command-line option, I thought maybe that could point me in the right direction for why the openwrt builds are failing in my Jun 26 22:10:25 other tree. Jun 26 22:11:51 do you know BUILD_LOG=1 or CONFIG_BUILD_LOG? Jun 26 22:12:02 is openwrt master also broken? Jun 26 22:12:17 real goal: to move from "mosquitto-client-nossl" to "mosquitto-client-ssl"... (in the 18.06.2 reference tree I have, which has been changed from stock mainline) Jun 26 22:12:27 no... openwrt 18.06.2 proper builds correctly Jun 26 22:12:53 I compared the package directories on both, and they're the same... so seems it's somewhere deeper into the openwrt build system it's failing Jun 26 22:13:40 bemo: try `make package/mosquitto/{clean,compile} BUILD_LOG=1` and there will be a logs/ directory in the openwrt directory Jun 26 22:13:43 (and tip of openwrt master also builds fine... *unless* I try to run the command manually -- and I don't know why that would fail) Jun 26 22:14:28 can you upload the compile.txt of your package. so I can have a look on it. Jun 26 22:15:29 sure... still new to this... let me figure out where to upload it for you Jun 26 22:16:07 (wasn't sure if a modified openwrt tree was something people would want to try to help me debug... which is why I tried translating it to something I *can* reproduce from the unmodified openwrt tree) Jun 26 22:16:15 LD_LIBRARY_PATH is also taken into account link-time. Jun 26 22:16:53 But for a native linker only. Jun 26 22:23:46 lynxis: Marvell Armada 380/385 Jun 26 22:24:57 not sure the best place to upload a log, but you can find it here: http://siphon.net/logs/compile.txt Jun 26 22:27:38 lynxis: let me know if you have any issues getting to it Jun 26 22:34:56 lynxis: it's a different toolchain, so I was originally thinking there might be something wrong with it... so I started comparing it against the upstreamed 18.06.2, which worked fine with a different toolchain... then I decided to compare the individual build commands, which are similar, and realized I see the same problem with the "good" tree when Jun 26 22:34:56 I do that (and a different toolchain)... so thought that might help me figure out what might be wrong, since I know a pristine tree works. Jun 26 22:35:51 ... and eventually tried the tip of the tree as well... with similar results (fails manually)... so now I think there's an environment problem... just not sure what that might be. Jun 26 22:36:29 bemo: if you change the toolchain, you should do `make clean` before. can you find libssl under your staging_dir? Jun 26 22:36:57 each of these trees started with their own toolchain (never switched in the middle) -- I have three trees. Jun 26 22:38:47 yes, I can find libssl (and libcrypto)... both are in the path of the first "-L" argument. Jun 26 22:39:21 there are both .so and .a files in that directory Jun 26 22:43:36 ... which is why I started down the "run it from the command-line to see why it's failing" method... and it appears it fails in the good case as well... so started scratching my head. Jun 26 22:45:15 (so figured it was time to ask the experts). :) Jun 26 22:47:11 lynxis: btw, thank you for taking a look! Jun 26 23:09:41 lnslbrty: Armada is only the SoC. Which board (or product) are you using? Jun 26 23:11:51 bemo: I guess you also checkd if libssl.so.1.0.0 is present in the directory. maybe you've to recompile your libssl. unsure why ld isn't happy with your libssl there. can you share our package/.../mosquitto/Makefile (and patches etc.)? Jun 26 23:14:14 yes, libssl.so.1.0.0 is present (and it looks legit), and libssl.so is a symbolic link to it Jun 26 23:16:39 siphon.net/logs/Makefile Jun 26 23:17:21 lynxis: it's the same as the stock 18.06.2 tree's Makefile as well Jun 26 23:19:20 bemo: maybe it's related of changing between ssl and -no-ssl. openwrt doesn't track well InstallDev. maybe cleanup your staging/target-* directory from all libmosquitto files. Jun 26 23:19:48 I tried doing the same switch with the stock trees and it works fine Jun 26 23:22:35 lynxis: and since I can reproduce the same behavior if I build mosquitto-client-ssl in the stock openwrt tree and then cd into the client directory and run that one command that fails, and it fails for those as well, although it builds fine as part of the overall build... so I'm confused. Jun 26 23:23:37 (sorry... that sentence was kind of hard to parse) Jun 26 23:32:48 lynxis: and if you think it will help, I can start fresh and try again (I'm pretty sure I had originally started down the ssl path, and ran into this problem, and then went to the nossl version and it actually builds correctly) Jun 26 23:39:37 bemo: I'm out of ideas atm. sorry. i would try a make clean, make oldconfig and ensure the staging_dir/target* is empty (it should be deleted by make clean). Jun 26 23:40:52 lynxis: thanks for the help! I keep finding myself drawn back to the fact that in a working tree (stock openwrt tip and 18.06.2), it builds fine... and if I manually run the command I get the same failure, which feels like it's a clue (just not sure what kind of clue). :) **** ENDING LOGGING AT Sat Jun 27 02:59:57 2020