**** BEGIN LOGGING AT Fri Apr 10 02:59:57 2020 Apr 10 04:22:57 zeddii: vmlinux image for linux-yocto/qemumips is whopping 196M Apr 10 04:24:19 well, you never know when you'll want all of wikipedia snapshotted into your zImage... Apr 10 04:30:02 khem, joking aside - what was the base branch for your kernel build? Apr 10 04:30:20 whatever master has Apr 10 04:31:06 and my rootfs is 4.2M Apr 10 04:31:18 ok, so v5.4.x I'm guessing... Apr 10 04:31:24 so ironical Apr 10 04:31:49 right vmlinux--5.4.27+git0+bee554e595_967a1708cb-r0.2-qemumips-20200409083339.bin Apr 10 04:32:49 thats unstripped btw. Apr 10 04:41:52 khem, do you have an approximate timeline for when qemu-mips swallowed a whale? Apr 10 04:42:49 * paulg is glad it isn't the v5.2 stream I've been feeding to zeddii Apr 10 04:45:35 no idea I was just looking at the deploy dir today and saw this Apr 10 04:57:24 well, I'm guessing this will be a slow weekend; if I have a chance, I'll look at v5.4 kernel MIPS builds..... Apr 10 04:59:19 mips is getting dangerously close to arch/alpha and arch/pa-risc status.... Apr 10 06:11:45 nod Apr 10 06:26:13 paulg: I think its been like this for a long time Apr 10 06:33:53 hy dotbkou ytrsd bkk f/me gootts sjg in eir Apr 10 13:16:28 is it possible to keep the sources and temporary build files for a recipe in the work directory? Apr 10 13:33:45 Hello. I built a yocto image with meta-qt5 and I am trying to run a Qt "app" using EGLFS that is located in /usr/share. When I try to runt he executable it just gives me "-sh: ./usr/share/app Permission Denied". In the do_install I gave executable permissions so I am not quite sure what the issue is. Apr 10 13:50:16 riz14: those executables that you mention is some sort of example for qt5 ? Apr 10 13:52:14 anyway I'm not sure about this, so maybe try to run /path/to/executable -platform eglfs (as root or tune some privilages on /dev/dri, /dev/card) Apr 10 13:53:00 you can check if qtbase is compiled with eglfs support by invoking: bitbake -e qtbase | grep ^PACKAGECONFIG Apr 10 13:53:09 check if PACKAGECONFIG contains eglfs Apr 10 13:54:41 Thanks Apr 10 15:11:44 In general, what is the best way to see what features and packages are responsible for making an image so large? Hopefully if I can get a list that breaks up the features by the size I can remove any unnecessary items. Apr 10 16:17:58 New news from stackoverflow: Errors including shared prebuilt libraries in petalinux Apr 10 17:50:22 it looks like you can't use python code in conf files, is that right? Apr 10 17:51:03 correct Apr 10 17:51:09 well, inline python Apr 10 17:51:57 i wanted to see if i could set SSTATE_MIRRORS dynamically based on wehter the URL was reachable or not Apr 10 17:52:42 i made a class that sends a head request to the url, but i cant figure out how to check just once instead of in each recipe :S Apr 10 17:57:00 that's going to be a tough call, we often re-parse metadata multiple times in the build process Apr 10 17:57:26 but i'd advise pursuing use of a ConfigParsed or BuildStarted event handler Apr 10 17:57:44 make a bbclass that adds a function that's an event handler and add your class to INHERIT in local.conf Apr 10 17:58:10 that said, pretty sure bitbake already does some sort of checking.. Apr 10 17:58:56 if the mirror is unreachable, then bitbake takes for *ever* to actually get to the build Apr 10 17:59:21 our mirror is reachable via vpn only, and sometimes people aren't or cant be on it Apr 10 17:59:42 so my thought was to make bitbake a little smarter and only set the variables if its reachable... Apr 10 18:03:28 i'd try ConfigParsed event, that should likely be early enough Apr 10 18:03:59 alternatively, use inline python, let it be expanded for every recipe, but cache the results somewhere.. but that'd have its own problems Apr 10 18:10:30 kergoth: yeah, i was not sure how to cache it. Apr 10 18:13:31 the hard part there is cache invalidation. which is almost always the hard part, really... :) Apr 10 18:13:41 ex precisely when do we re-check sites outside our control Apr 10 18:13:52 so i'd avoid that approach anyway Apr 10 18:16:08 kergoth: i think it would be sufficient to rechek mirror reachability everytime a build is started? Apr 10 18:16:51 thats what i was thiking with configparsed. it's not exactly one time, but it's close, adn BuildStarted won't work, as thats when actual tasks are going to run, long after sstate Apr 10 18:20:29 okay, i hooked bb.event.ConfigParsed and it looks like it works Apr 10 18:20:57 the only downside is that it seems to cause a whole reparse based on the changing variables Apr 10 18:21:31 for example if you can reach the vpn -> set SSTATE_MIRRORS. but if the next build vpn is off, and can't reach mirror, SSTATE_MIRRORS is not set and all recipes are reparsed.. Apr 10 18:24:47 hm.. actually i take that back. seems fine now after a few toggles of vpn on/off. Apr 10 18:30:40 kergoth: do you know why bb.event.ConfigParsed fires twice? Apr 10 18:37:12 mischief: it can fire even more than that, iirc, if multiconfig is in use Apr 10 18:38:56 no big deal. i added a bb.note in my event handler and it appears more than once. Apr 10 18:40:22 still better than every recipe eh? :) Apr 10 18:40:38 it's possible there's an event even earlier than that, but then you woudln't have SSTATE_MIRRORS available Apr 10 18:40:56 uff, now i noticed a different problem Apr 10 18:41:17 it looks like setting d.appendVar("INHERIT", " own-mirrors") in the event handler does nothing Apr 10 18:41:36 i guess i can set premirrors myself Apr 10 19:14:51 RP: I lost ssh access to the ptest-runner2 repo, I'm working on Yocto #12604 Apr 10 19:39:29 Hi everyone! I'm building an image with yocto for a raspberry cm3. I'm want to compile a qt5 application with yocto in order to put my application in the image. It's a big qt5 application developped by me. Usually I need to perform a qmake command then a make. Here; I'm trying to write a recipe to do the same thing but I have a problem during the "do compile" task :( Apr 10 19:39:49 problem is : Could not find qmake spec 'linux-oe-g++' Apr 10 19:40:14 I included this file into my recipe: require recipes-qt/qt5/qt5.inc Apr 10 19:42:56 I would say copy an existing qt5 app recipe and make changes to cutomize it for your app Apr 10 19:45:20 unfortunately, the recipes I found on internet didn't help me :/ I can't find a recipe exemple with a qmake, than a make command Apr 10 19:45:37 then* Apr 10 19:50:18 alimon, you may want to talk to halstead Apr 10 19:52:17 alimon, I'm PMing you. Apr 10 19:58:52 alexb3600: look into something in meta-qt5 itself Apr 10 20:42:22 alexb3600, welcome to the club, I think we're trying to solve a very similar problem Apr 10 20:43:10 I'm trying to write a recipe for a 3rd party project that uses its homebrew configure script and homebrew build engine Apr 10 20:44:37 the configure script internally uses qmake to configure the project and the samples in meta-qt5 are all vanilla qt projects with a .pro file in their root Apr 10 20:45:27 that's why I chose to inherit qmake5_base instead of qmake5 to join the game one level below Apr 10 20:47:17 the problem with the makespec I solved by passing -spec ${QMAKE_MKSPEC_PATH_NATIVE}/mkspecs/${OE_QMAKE_PLATFORM} to qmake, where the last variable expands to linux-oe-g++ Apr 10 20:48:58 however the project's configure step fails in a later step with qmake not being able to find feature force_asserts Apr 10 20:49:35 that's where I'm stuck at the moment but maybe you won't run into the same problem ;) Apr 10 22:19:25 armpit: RP just tried again my fault the identity wasn't added, everything is fine, sorry for the noise halstead Apr 10 22:20:01 No problem. :) Apr 10 22:27:06 alimon, glad it was resolved **** ENDING LOGGING AT Sat Apr 11 02:59:57 2020