**** BEGIN LOGGING AT Mon Jun 29 02:59:57 2020 Jun 29 05:29:09 Hi guys. Do you know that can I compile "gstreamer1.0-vaapi" for imx8 with thud branch? Is supported? Because now I am getting error. Is this only for intel? Jun 29 06:41:44 beratiks: "getting error" is not exactly a helpful problem desctiption. Jun 29 06:47:12 Letothe2nd: co_compile error and give out a lot of stuff like syntaxs Jun 29 06:47:31 I thinks vaapi is not compatible imx8qxp Jun 29 06:47:44 thats a possible reason. Jun 29 06:48:47 mckoan: howdy! Jun 29 06:48:48 good morning Jun 29 07:24:47 RP: ping - for the weekly stats report, please change my livecoding url to https://www.twitch.tv/theyoctojester - thanks. Jun 29 07:26:28 Letothe2nd: updated Jun 29 07:26:40 RP: thx Jun 29 07:32:09 mckoan: morning Jun 29 10:14:28 kanavin_home: good news! rpm 4.16 supports sqlite for the database Jun 29 10:16:33 kanavin_home: https://fedoraproject.org/wiki/Changes/Sqlite_Rpmdb has details of f33 moving **** BEGIN LOGGING AT Mon Jun 29 10:19:15 2020 Jun 29 10:38:42 rburton: not quite, I think sqlite in 4.16 is still marked experimental, but ndb is considered stable. Bdb is deprecated (yay). Jun 29 10:39:08 if f33 are moving to sqlite then i'd consider that backend the best tested when f33 is released Jun 29 10:39:38 of course, I am just saying what rpm 4.16 release notes say Jun 29 10:40:58 let's keep track of fedora's rpm.spec options and settings then Jun 29 11:23:03 does anyone know why the crops/poky-container keeps prompting me for the sudo password (that I don't know)? looking at https://github.com/crops/poky-container/blob/master/sudoers.usersetup I thought it is set up to not prompt for sudo? Jun 29 11:24:01 nameclash: Which command are you trying to run by sudo? Jun 29 11:24:37 I'm trying to install a yocto sdk to /opt Jun 29 11:25:16 nameclash: That won't work. Those sudoers entries you're looking at only allow the commands which are explicitly listed Jun 29 11:25:42 ah ok Jun 29 11:26:05 You can use `docker exec -u 0 ....` to run a command as root inside an existing container, or you can write your own Dockerfile to create a new container image from poky-container with your extra dependencies installed Jun 29 11:27:53 ok, thx Jun 29 12:22:50 I have a bitbake process stuck after hitting disk-full condition - is there any way I can query the server process what it's still thinking it's doing ? Jun 29 12:23:44 yann, hold on Jun 29 12:24:37 yann, https://bugzilla.yoctoproject.org/show_bug.cgi?id=13843 Jun 29 12:25:10 yann, if the process is still alive you can attach with gdb and use "py-bt" Jun 29 12:28:43 kroon: I pasted the last lines output and the current process tree at https://pastebin.com/hrLAagfd - any guess as to which pid will be most useful to dig into ? Jun 29 12:30:20 yann, in my case there was a worker process left using all CPU, do you have any of those ? Jun 29 12:30:27 using all cpu that is Jun 29 12:31:07 no, there's one (see paste) but not doing much apparently Jun 29 12:31:33 Is it a good idea to build yocto images inside arch? Jun 29 12:32:15 Should I use a container? Jun 29 12:33:54 hm, debian's gdb would not have python support ? Jun 29 12:34:00 yann, sorry, i don't have a good idea on what to inspect then, maybe RP has some idea Jun 29 12:34:23 yann, you might need to install some additional package Jun 29 12:34:39 yann, python3 debug info iirc Jun 29 12:39:57 yann, yeah, at least that is required on Fedora 32 Jun 29 12:40:12 right, that's python3-dbg Jun 29 12:41:24 Any reason for having openssl 1.0.2 named openssl in meta-openssl102? We effectively lose the ability to have both openssl10 and openssl in the same build. Jun 29 12:48:02 I think this was never intended, and meta-openssl102 is meant to exclude 1.1 entirely from builds Jun 29 12:48:29 kroon: I've pasted all thread backtraces at https://pastebin.com/MbxeatC8 - without any clue on bitbake's inner workings I can't see what's pathological in there Jun 29 12:51:59 kanavin_home: it was never intended to support both versions of openssl in the same build? that's what you meant? Jun 29 12:59:01 yann, FWIW, I'd put that in a text file and attach it to the bug, or perhaps file a new Jun 29 12:59:26 yes, a new one could be worth Jun 29 13:05:01 yann: bitbake really doesn't like running out of space :( It can be hard to figure out where its locked up like that :( Jun 29 13:06:25 RP: so is it useful to open a new bugreport ? Jun 29 13:07:09 yann: Its probably worth documenting that there is an issue but realistically, I can't see anyone with time to look into that any time soon :( Jun 29 13:07:37 yann: there are other more serious issues that need attention and too few people to look at issues like that Jun 29 13:07:39 just adding this to kroon's report then Jun 29 13:08:30 yann: please make it a separate bug, its a different issue Jun 29 13:08:39 k Jun 29 13:08:42 multiple different issues in the same bug becomes a nightmare Jun 29 13:08:55 I appreciate there is potentially a connection here Jun 29 14:01:15 hi - i have my own BSP layer which specifies a version. When from another layers conf I do LAYERDEPENDS_mylayer = "core bsp:1" I receive an error message. When I remove the version from bsp I do not receive the error message Jun 29 14:01:32 Why is this happening? Thanks Jun 29 14:02:04 akane: https://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-LAYERDEPENDS Jun 29 14:02:30 bsp (=1) I guess? Jun 29 14:03:33 => qschulz - that fixed it - was looking at the wrong version of the mega manual - thanks Jun 29 14:06:51 akane: my pleasure :) Jun 29 14:14:46 qschulz: I do not know that for sure, but trying to mix openssl 1.0 and 1.1 quickly leads into trouble, as you are likely to run into a situation where both versions are attempted to be installed into a component sysroot, and the filenames clash Jun 29 14:14:58 (filenames of headers, .pc files) Jun 29 14:21:10 kanavin_home: the current situation is that I actually manage to build SW with a dependency on openssl10-native but once running extend_sysroot or something like that from the staging.bbclass, something attempts to install openssl headers into the sysroot-native which are obviously already there because of openssl10-native DEPENDS Jun 29 14:25:30 qschulz: yeah, so you need to stick to openssl 1.0, which is troublesome because its out of support Jun 29 14:27:41 yeah no, not going to happen for the reasons you stated :) I managed to remove the useless openssl10-native dependency (was needlessly brought up by some recipe) but I'm pretty sure it'll bite back soon enough Jun 29 14:28:36 now, I can understand it's impossible... but I really want to understand what's happening during this extend_sysroot function from staging class because it seems very weird to have something be appended to recipe-sysroot-native **after** it's been used for the recipe Jun 29 14:49:27 test Jun 29 14:50:15 THe YP booth at ELC is open! Several people have already stooped by. Come say hello and chat with us and visitors, and share about YP Jun 29 14:50:28 s/stooped/stopped Jun 29 15:05:06 dreyna: do we want to focus people to the booth or the slack channel? Jun 29 15:05:35 slack please Jun 29 15:05:56 yes, probably slack since it is more persistent Jun 29 15:05:59 the chat interface is awful. and i think only available during the "breaks" anyways, right? Jun 29 15:14:28 hi, has anyone managed to build chromium or chromium-ozone-wayland from meta-browser with SoC specific graphics, without mesa? Looks like several things expect mesa details like dri.pc.. or I've screwed up the build feature config. Jun 29 15:26:06 There is a yocto slack group? Jun 29 15:26:32 there's an ELC/OSS slack for the conference Jun 29 15:26:39 rabbit9911: not exactly, there is one for the currently ongoing ELC/OSS Jun 29 15:56:42 hmm, https://autobuilder.yoctoproject.org/typhoon/#/builders/60/builds/2081 is another weird failure. Something sending out SIGTERM to the wrong things? :/ Jun 29 17:40:18 booth is fairly quiet. And we've had a couple drive-bys that didn't stay long enough to engage Jun 29 17:43:04 moto-timo: there are too many channels for information Jun 29 17:43:32 RP: including this one today :) Jun 29 18:25:45 if someone has a qemu to hand, can you run "qemu-system-aarch64 --machine help" and pastebin the output? Jun 29 18:25:57 my qemu is the wrong side of a vpn right now Jun 29 18:27:25 rburton: https://pastebin.com/nZbjaUei Jun 29 18:30:42 thanks! Jun 29 18:30:56 rburton: no pb Jun 29 19:54:40 Some communities have moved to discord over slack because its free. Has anyone in the yocto community looked at discord? Jun 29 20:02:50 discord makes me wish for death Jun 29 20:09:26 zeddii: I'm quite happy with irc Jun 29 20:09:51 zeddii: So.. it reminds you of heaven? Jun 29 20:11:09 RP: agreed :D Jun 29 20:11:53 I just keep getting disconnected from IRC then I have to go back to the logs to see what was said. Just spoiled by systems like slack.. but slack is not free. Jun 29 20:12:17 zeddii: seriously, did you ever see any upstream fix for that make race kernel issue. We seem to be hitting it a lot. We need to try and reduce the number of intermittent failures as a matter of urgency as together they're causing problems :/ Jun 29 20:12:31 RP: no, nothing found yet. Jun 29 20:12:38 I was poking at it today though. Jun 29 20:13:08 zeddii: ok, cool, thanks. I spent half the weekend poking at other things and they're proving frustrating Jun 29 20:13:20 seems as soon as I fix one, something else pops up too :( Jun 29 20:13:50 I have that + the ARM32 bit go build failures to sort out before I can finish my v5.8 kernel stuff, so I'm feeling the same. Jun 29 20:41:16 RP: what is the race you are seeing ? Jun 29 20:52:56 khem: https://bugzilla.yoctoproject.org/buglist.cgi?quicksearch=AB-INT are our intermittent issues Jun 29 20:53:18 khem: or you mean the new ones we don't have bugs for? Jun 29 20:53:43 https://autobuilder.yoctoproject.org/typhoon/#/builders/60/builds/2081 and https://autobuilder.yoctoproject.org/typhoon/#/builders/87/builds/1085 puzzle us Jun 29 21:35:22 RP: I was interested in kernel one Jun 29 21:36:36 khem: https://bugzilla.yoctoproject.org/show_bug.cgi?id=13924 Jun 29 21:37:38 RP: is it happening on hosts with gcc10 natively Jun 29 21:37:54 its due to -fcommon being default Jun 29 21:38:45 I mean -fno-common being default Jun 29 21:39:06 so add -fcommon to host cc flags Jun 29 21:42:54 zeddii: can you reproduce/test that? Jun 29 21:44:22 I already have the kernel fix in the tree for that. Jun 29 21:44:34 we don't need to futz with flags Jun 29 21:45:19 IIRC you already have my latest. that being said, I can send you new SRCREVs and we can see if the fix holds. Jun 29 21:45:48 zeddii: I'm not sure we do. Steve was saying we needed to update some SRCREVs Jun 29 21:45:58 Perhaps this is where we're going wrong Jun 29 21:46:13 I have a stack of 6 -stable bumps that I've tested locally. I can send them out. Jun 29 21:46:53 zeddii: I think we need to ensure we bump meta-yocto-bsp too Jun 29 21:47:02 zeddii: but yes, please Jun 29 21:47:13 yep. otherwise, the reference will continue to show it. Jun 29 21:47:25 I'll confirm I have all the -f*common changes, and will bump them. Jun 29 21:47:38 I'll send it late tonight, so you can try it in your morning. Jun 29 21:51:02 zeddii: sounds good thanks Jun 29 21:51:07 I think that will help sakoman too Jun 29 21:56:20 this will be issue on all older kernels I believe. so using f32 as host will be PITA Jun 29 21:56:57 I looked at 50+ packages breaking builds when I turned it off in gcc10 by default. Jun 29 21:57:01 Lets others do the work :) Jun 29 22:05:57 * armpit hides Jun 29 22:39:09 sakoman: I'd ask kanavin_home if he remembers that ubuntu virgl error cause :/ Jun 29 22:40:28 RP: it is possibly ubuntu version related Jun 29 22:43:11 RP: I'll ping him when he returns Jun 29 22:45:55 Once you've rebooted I may try doing another couple builds to 16.04 and 18.04 to see if 16.04 still succeeds and 18.04 still fails Jun 29 22:46:36 sakoman: I won't reboot until my morning so feel free to run them Jun 29 22:46:45 well, just restart the controller Jun 29 22:46:49 (process) Jun 29 22:48:00 OK will do Jun 29 22:48:24 sakoman: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=1ec91031165f8801eeb5662aa98c73dc6be88cc1 maybe ? Jun 29 22:48:50 sakoman: look at the surrounding changes too :/ Jun 29 22:49:48 Ah yes, a possibility! Jun 29 22:57:26 RP: testing now on a 20.04 worker with those three patches. Fingers crossed . . . Jun 29 23:02:04 RP: Question on bugzilla bugs. A bug was reported on an earlier release. Currently it has a 3.2 milestone, however the code has been updated that its obsolete in 3.2, but its still an issue in the older release. What is the correct resolution? Jun 29 23:04:04 The actual release it was reported on is no longer supported even via community. Jun 29 23:08:40 dunno about "official policy" but I'd be personally wondering about whether there was an isolated commit that could be backported, or if a pkg-uprev was the only way to deal with the issue.... Jun 29 23:11:02 paulg: If that was at me, update is not likely the right word, completely reworked such that the failing recipe no longer even exists. Jun 29 23:12:35 jpuhlman, yeah - that is tougher - if it isn't even a patch to a pkg, but the original recipe is gone for the old pkg, then IMHO - if it was my project -- it would become "Won't Fix" Jun 29 23:13:24 unless an older release had the same issue on the older pkg, and say, it was a CVE, and that older release was still "maintained". Jun 29 23:14:16 jpuhlman, do you have a bugzilla link that you are able to share for more context? Jun 29 23:14:25 https://bugzilla.yoctoproject.org/show_bug.cgi?id=11351 Jun 29 23:15:23 * paulg reads Jun 29 23:15:40 Basically go-bootstap-native has been nuked from orbit and a binary replacement has been pulled in. It basically dosen't build the bootstrap any more we cheat. Jun 29 23:16:50 I haven't dug in to the history to see when the transision took place but the original was 2.3 Jun 29 23:18:18 ugh - I had the misfortune of dealing with go bootstrap stuff several years ago. Jun 29 23:18:40 there were no easy solutions back then. Jun 29 23:19:28 I'm pretty sure we can rule this out as NOT being a "single commit backport CVE instance".... Jun 29 23:19:31 Yeah its always been a bear. Jun 29 23:20:39 jpuhlman, ok - sorry to waste your time ; in this instance I'm pretty sure I've got no pro solution for you. :-/ Jun 29 23:20:48 maybe someone else reading can? Jun 29 23:21:12 * paulg normally hides in kernel cobwebs. Jun 29 23:23:29 No worries. Like I said for 3.2 its fixed, which is what the milestone is set to. Since is really a reorganization, moving it backwards is not likely correct either. Jun 29 23:32:06 I hate things with unreasonable and/or cross-compile unfriendly demands like "help2man". Jun 29 23:34:12 making me kill it with fire. https://patchwork.openembedded.org/patch/88533/ Jun 29 23:34:28 I sometimes wonder how many projects actually have users using man pages on target Jun 29 23:35:02 don't care. It still should be a working option. Yocto isn't all embedded targets. Jun 29 23:35:19 yeah, I know Jun 29 23:39:46 don't think I've ever worked on a project that used the doc packages, though Jun 29 23:41:20 smurray it's interesting, I keep getting requests for man pages, especially on network programs.. especially 'ip', route and similar.. Jun 29 23:41:38 smurray: More often then not its part of the --make-customer-happy option. Jun 29 23:41:44 seems like it's a short cut way to document the tool on the target for network devices without resorting to Cisco style IOS interfaces Jun 29 23:41:50 fray: heh, that is interesting Jun 29 23:42:04 a bunch of years back I worked on making an installable version of a "server" install based on yocto (SDK/self-hosting) Jun 29 23:42:28 fray: the large network equipment distro I worked on had its own CLI with its own help system Jun 29 23:42:33 I believe all the history is in https://github.com/OverC/meta-overc.git Jun 29 23:42:53 yeah, I looked at it one point Jun 29 23:43:17 I was focused on making the CLI env not suck - working git, working manpages etc. ; zeddii then took that and added gfx/X11 Jun 29 23:43:51 smurray I've been working with (as in a customer of) Ubiquity networks, ya I know not same level as Juniper or Cisco.. but they just use standard Linux tooling for command line -- their UI is all web based.. Jun 29 23:43:54 and a bunch of those changes fed back directly into oe/yocto - so IMHO time not wasted. Jun 29 23:44:25 commit deecd4314939268be94e3185239cd7e0c7e6dbf7 Jun 29 23:44:26 Author: Paul Gortmaker Date: Sat Feb 14 10:18:12 2015 -0500 Jun 29 23:44:26 git-manpages: remove recipe ; merged upstream Jun 29 23:44:52 overc commit that now everyone gets to use. Jun 29 23:46:16 making vim and/or "top" not suck - I believe those went to yocto/OE too? Jun 29 23:46:35 I can imagine the usecases. Was just an idle thought on my part given statements of there not being a lot of dev resources Jun 29 23:46:44 I'll retreat to my cave now Jun 29 23:46:45 on the uqiquity side I can run ip, route, ifconfig, iptables, etc and adjust things and my changes show up in the UI.. it's REALLY slick.. Jun 29 23:47:22 smurray, no - it is all good - I am sure you aren't the only one, and it is a good dialog. Jun 29 23:47:31 yup Jun 29 23:48:16 yocto does itself a dis-service if everyone thinks it is for flash install on non-x86 with 128MB ram. Jun 29 23:52:40 I don't disagree with that, but the project doesn't really market itself at all, and there are significant challenges once you require much beyond oe-core & meta-oe Jun 29 23:52:50 err, market itself for that at all Jun 29 23:53:37 yeah, I can't argue with that. Jun 29 23:57:00 fray: I sort of gave up on Ubiquity after they blew off all requests for source for their (now dropped) powerbar products. That kind of left me with a bad impression, even though I do hear good stuff re the access points Jun 29 23:57:25 s/Ubiquity/Ubuiqiti/ Jun 29 23:58:25 kinda thought people were smart enough now to toss a turdburger ftp site out there with un-compile-able code by now. Jun 29 23:58:54 lessons learned by Linksys on the WRT54G like 15y ago.... Jun 29 23:59:00 Ya, last ubquity stuff I used was years ago.. (10 or so).. Jun 29 23:59:12 the new stuff seems way better.. and frankly I got tired of building my own stuff that 'mostly' worked Jun 29 23:59:57 ooh! build 3.18 + binary radio blobs for my phone! I can't wait! Jun 30 00:00:11 :) Jun 30 00:00:32 yeah - want NO part of that time sink. Jun 30 00:00:46 heh, indeed **** ENDING LOGGING AT Tue Jun 30 02:59:57 2020