**** BEGIN LOGGING AT Wed Jan 08 02:59:58 2020 Jan 08 03:11:00 root Jan 08 03:11:07 rooot Jan 08 06:51:16 New news from stackoverflow: Yocto Bsp Porting in non dts sdk Jan 08 07:15:40 I'm having a weird issue with the wget fetcher in bitbake. I want to download a file from a private bitbucket repository, and if I run wget with the username/password encoded in the url (i.e. https://user:pass@bitbucket.org/....) it works, but if it is supplied with --user and --password arguments it fails. The bitbake fetcher seems to extract the username/password from the url, and run wget using BOTH Jan 08 07:15:46 --user/--password arguments and having them encoded in the url, wich ultimately fails. Has anyone else had this issue? Jan 08 07:21:23 New news from stackoverflow: How to load image in sama5d27-som1-ek board using dataflash Jan 08 08:43:49 If i have a file like SRC_URI += "file://script.py;md5=dbeda8d8aa81df8611249a5b01414e03, why is the md5 ignored? I can put there whatever or change the file and I get no warnings? Jan 08 08:47:39 stuom1 hello, i think it's because you don't have any URL (local file) Jan 08 08:48:14 so file is managed by yourself ;) Jan 08 08:48:47 Ok so I dont even need the md5 sum there, who put it there...:P Jan 08 08:49:30 stuom1 I think so, more info: https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#new-recipe-fetching-code Jan 08 08:50:16 In fact the md5 is used to check if the file on the server side has been changed or not :) Jan 08 09:23:16 "ld: cannot find Scrt1.o: No such file or directory" -> how can I see what directories are looked from? Jan 08 09:29:38 ld complains about few files, I can find them all in recipe-sysroot but I don't know where ld expects them to be Jan 08 09:39:06 Hi! i am updating an old build from fido to thud atm and just realized, that systemd-networkd is now enabled by default in in the sytemd recipe. Is systemd-networkd now considered to be the default network manager in poky? Jan 08 09:39:25 (at least with systemd as init) Jan 08 09:56:37 fl0v0: i think thats more of a "if systemd is already around we can get it for little cost, so here we go" - but thats just my $.02 Jan 08 09:57:19 LetoThe2nd: ok, thanks :) Jan 08 10:00:13 i try to figure out if i should stay with connman or switch. Any expieriences if that matters at all? i just need a simple wired dhcp connection Jan 08 10:01:29 if thats all you need and if the interface is constant, no other config needed, then i'd pick networkd these days. like said, it comes at little cost. Jan 08 10:28:23 Hi. Is there a way to build an image without stripping "libc"? I tried to use INHIBIT_PACKAGE_DEBUG_SPLIT = "1" flag - did not work Jan 08 10:29:37 nacknick: which should not spliiting the image avoid stripping? Jan 08 10:31:15 nacknick: -> https://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#var-INHIBIT_PACKAGE_STRIP Jan 08 10:33:28 on the other hand, one could use the debugfs Jan 08 10:52:05 New news from stackoverflow: Change password in yocto squashfs image Jan 08 10:59:40 Hello, is there a special way to acces u-boot shell with poky on a rpi3 ? earlier when I had a problem and the board could not boot I pressed any key and had a prompt.but now it's fixed , I could not get the prompt even with smashing space key on power up. Jan 08 11:34:34 what can I do about "...do_image_teziimg, the basehash value changed from ... to .... The metadata is not deterministic and this needs to be fixed" Jan 08 11:59:50 stuom1: That usually happens when some variable changes between the time bitbake first parses a recipe and when it later tries to run a task Jan 08 12:00:21 E.g. if you edit a recipe while bitbake is building or if you have a variable set to the date/time Jan 08 12:01:21 jwwww: Have you got u-boot enabled in your build and can you see any output from u-boot during the boot at all? Jan 08 12:08:20 LetoThe2nd: thanks Jan 08 12:30:08 is there a way to manipulate the content of a wic file? Jan 08 12:30:22 during creation. I always get ext4 images Jan 08 12:33:34 perdmann: sounds like you are mixing up steps Jan 08 12:35:30 exit Jan 08 12:36:59 LetoThe2nd maybe, but Rauc wants to have a wic file during bundle creation Jan 08 12:37:15 perdmann: yes, so what? Jan 08 12:37:39 LetoThe2nd: I want to have a jffs2 inside of it. But if i change IMAGE_FSTYPE i dont get a wic file Jan 08 12:38:38 perdmann: becuase IMAGE_FSTYPE and wic are kind of chained. if you set "wic" as the fstype, then the script gets run and bundled. Jan 08 12:38:57 perdmann: so if you want your wic to be a jffs2 internally, then you have to modify the .wks file Jan 08 12:39:13 oh Jan 08 12:39:48 like i said, you are "mixign up steps" Jan 08 12:40:09 is it valid to overwrite/append a task from the wic steps? Jan 08 12:40:51 Maybe i should look at the bundle scripts from rauc, for me it makes no sense to manipulate a wic file, which i dont need... Jan 08 12:40:55 no, wks files work differently Jan 08 12:41:19 as in, they are no recipes that can be appended. Jan 08 12:42:28 LetoThe2nd do you use rauc? am i using it wrong ? :D Jan 08 12:43:15 perdmann: i am not using rauc, hence i cannot comment. but it sounds legit that it consumes a wic file, as thats basically just a synonym for "binary image payload" Jan 08 12:45:54 did i understand you correctly that wic always have a ext4 inside? Jan 08 12:46:05 no Jan 08 12:46:12 wic can have whatever inside. Jan 08 12:46:24 (to my understanding, of course) Jan 08 12:46:25 Ah, then it makes absolute sense Jan 08 12:46:50 look at the correspondign wks script Jan 08 12:47:08 i think in livecoding session #8 there is also a (short) peek into one. Jan 08 12:52:31 New news from stackoverflow: getaddrinfo returns EAI_ADDRFAMILY on a distribution built with Yocto Jan 08 12:55:40 LetoThe2nd: I try to avoid libc from being stripped. INHIBIT_PACKAGE_STRIP = "1" did not work - anything else? Jan 08 12:57:27 nacknick: none right now, sorry. Jan 08 12:57:55 (y) thanks Jan 08 13:14:35 rburton, RESULTS - ptest.PtestRunnerTest.test_ptestrunner: PASSED (437.69s) Jan 08 13:14:35 \0/ Jan 08 13:14:40 that's for fast ptests Jan 08 13:14:47 I'll tackle the slow ones next Jan 08 13:27:04 kanavin: nice! :) Jan 08 13:27:38 RP: I'll send the patchbomb in a moment :) I had also addressed all of the things you mentioned over the holidays Jan 08 13:29:40 kanavin: I'll brace myself :) Jan 08 13:33:13 RP: sent! Jan 08 13:40:06 kanavin: i'd appreciate your feedback over the top of ross/glib. once the logic has been demonstrated to work i'll move it to the meson class itself. Jan 08 13:42:43 rburton, looks good to me, but can't really comment on the siteinfo magic as I have no idea about it Jan 08 13:43:22 it just looks at a number of variables to get nice names, typically used in meta/site/ Jan 08 13:43:25 for autotools Jan 08 14:25:09 paulbarker: yes I have u-boot in my build, and I see its output. Jan 08 14:45:00 kanavin: does look like a good series. I'll put it to the autobuilder, see how things look, thanks Jan 08 14:46:15 kanavin: are these on a branch somewhere? the ovmf patch line endings are managed via email :( Jan 08 14:46:37 RP, http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=akanavin/package-version-updates Jan 08 14:54:29 kanavin: thanks Jan 08 14:54:42 kanavin: ovmf isn't in this round but queued locally for the next Jan 08 14:55:02 kanavin: -next build fired (I included other patches so its quite the patchbomb) Jan 08 14:57:16 RP: cheers, but where are Tanu's alsa patches? Jan 08 14:59:09 kanavin: queued locally now for the next round Jan 08 15:00:33 i rebased mut here on top of alex's branch Jan 08 15:01:02 mut is pretty long now, depending on what next does i'll split up and post sane sections Jan 08 15:01:41 RP, seems like "TypeError: dump_sigtask() got an unexpected keyword argument 'overwrite'" is a failure everywhere :-/ Jan 08 15:03:44 kanavin: yes, fixing. My fault Jan 08 15:03:55 kanavin: looks like the other patches will make it in afterall Jan 08 15:04:58 kanavin: requeued Jan 08 15:07:35 rburton: had the libx11 locale failure again :( Jan 08 15:10:25 Hello guys, i'm currently on an old Yocto project based on custom Zynq7000 board, I have built opencv 3.1 with (imgcodec, png ...) but when I link my test application i got this error : libopencv_imgcodecs.so.3.1.0: undefined reference to `png_init_filter_functions_neon'. I have tried this patch Jan 08 15:10:25 https://github.com/opencv/opencv/pull/7643/commits/a50bc7d8248dbdf4cd89bf6bfb6efbeb974f79a5 but no chance, any idea to fix that ? Jan 08 15:25:16 oh I think found the issue, in run.do.compile from's u-boot build directory I see that bootdelay is -2. Jan 08 15:41:05 Anyone around that does CROPS? I'm trying to rebuild https://hub.docker.com/layers/crops/poky/debian-9/images/sha256-56ec06a8df354b4dc3ee6f467cd1c17f857e086957fccd622ff74df795f21531 (so I can extend the Dockerfile a bit) but I might not be calling build-and-test.sh right as the tests fail (and I haven't changed anything yet) Jan 08 15:44:40 It fails with "sudo: unknown user: pokyuser" but a manual test of the image I made says I'm "pokyuser" Jan 08 16:27:27 where should INCOMPATIBLE_LICENSE go? the image recipe or globally? Jan 08 16:33:16 milloni: we have it in conf/local.conf because we always build more or less the same SW and the licenses don't really change. That way, we're safe (we create conf/local.conf with a script, so shouldn't be touched by engineers). That's how *we* do it. Don't know if it's proper though Jan 08 16:33:43 ok, thanks Jan 08 16:34:02 i'll stick it in the distro config, it might be safer Jan 08 16:36:06 milloni: by *we* I mean the company *I* work for, not YP as a whole :) Jan 08 16:36:49 i figured :) Jan 08 16:37:37 yeah stuff like that should be in distro config Jan 08 16:37:42 as little in local.conf etc Jan 08 16:44:08 kanavin: first failures are showing up, rpm still doesn't look happy :/ Jan 08 16:44:36 RP: :( has been robust here Jan 08 16:47:32 RP, some ptests are accessing the network (e.g. busybox tries to reach www.example.com), should they be disabled? Jan 08 16:48:05 I can probably make it work, but it might fail in other environments where the internet is not accessible from the machine running the tests Jan 08 16:48:12 kanavin: probably not, its a good question though, we don't have a policy for that Jan 08 16:48:43 or it requires setting up a proxy etc. Jan 08 16:48:55 I guess it's best disabled, tests should be self-contained Jan 08 16:49:21 kanavin: If we're aiming for 100% pass in all cases it would be problematic, certainly Jan 08 16:49:33 I'm still not sure 100% pass is realistic Jan 08 16:49:58 RP, I kind of aim to drop @expectedfail from the ptest, but they might defeat me :) Jan 08 16:50:09 RP, at least I do get 100% pass on the fast ptests here Jan 08 16:51:30 kanavin: that is impressive in itself Jan 08 16:51:46 kanavin: I suspect we have a world if intermittent failures to dive into :/ Jan 08 16:52:05 probably :-/ Jan 08 16:52:15 the failures I fixed were all deterministic though Jan 08 16:52:36 kanavin: right, I just know there are more painful problems :/ Jan 08 16:52:51 kanavin: e.g. the ones which depend on which other ptests were included in the image Jan 08 16:53:06 I fixed a lot of that Jan 08 17:10:09 rburton: we're using poky ATM, so no-go for us but good to know that's where it should be, thanks Jan 08 17:10:25 qschulz: cp poky.conf mydistro.conf, done Jan 08 17:10:41 dont use poky directly as you'll get bitten on upgrading if poky changes Jan 08 17:11:00 poky is an example and something QA can test Jan 08 17:12:51 rburton: I know. Not my choice, but I'll need proper convincing to change "something that works" before making us maintain our distro :) Jan 08 17:13:03 do you use systemd? Jan 08 17:13:07 rburton: (though size optimization could be one way to encourage :) ) Jan 08 17:13:10 no we do not Jan 08 17:13:54 the next release may well for poky so you'll find yourself using it :) Jan 08 17:14:09 as an example of something which will just change underneath you unless you own your own distro config Jan 08 17:14:28 also poky turns on wifi, bluetooth, pulseaudio, nfc... Jan 08 17:14:36 rburton: Unfortunately, upgrading is not a big concern ATM :/ Jan 08 17:15:39 (I mean, for the company, if it was up to me, we would be running 3.0 and testing 3.1 already. Budget. Priorities. etc.. you know the drill) Jan 08 17:16:12 yeah Jan 08 17:16:33 "maintaining your own distro" is not difficult, a distro .conf needn't have much in it at all, and you should really be aware of exactly what your distro is doing anyway Jan 08 17:16:44 poky's requirements aren't necessarily the same as yours.. Jan 08 17:18:40 kergoth: poky was a sane default back when the company started using YP. (we started with oe-classic fyi :) ) I have heard what rburton and you said, it's been on my secret todo list for months, but low prio :/ Jan 08 17:19:13 yeah thats fair Jan 08 17:19:22 my 'backlog' task list is ... quite long Jan 08 17:20:12 rburton: don't look at mine :) Jan 08 17:20:35 well, let's not make a "who's the biggest one" contest. That's nothing to be proud of :D Jan 08 17:20:55 kanavin: erm does perl-ptest need to have a copy of the perl tree in? Jan 08 17:21:22 rburton, I didn't get to perl ptests yet, so I don't know Jan 08 17:21:45 what I do know is that currently only VMS(!) and windows specific ptests fail, so I wouldn't want to mess with it too much ;) Jan 08 17:22:07 oh god it really does copy the source tree Jan 08 17:22:14 no wonder its so huge Jan 08 17:50:25 Im using PREFERRED_VERSION_lttng-modules = "2.10.10+git%" to select the devupstream version. When I try using devtool modify lttng-modules, it doesn't seem to care about the PREFERRED_VERSION. Jan 08 17:51:26 It fails on : Fetcher failure: SRCREV was used yet no valid SCM was found in SRC_URI Jan 08 18:55:10 hello all, question: When I added imagemagick to my local.conf, it pulls in the cantarell-fonts on my read-only rootfs. It throws an ERROR: do_rootfs: ... rootfs is read-only: [100-cantarell-fonts]. This script "100-cantarell-fonts" is in the /etc/rpm-postinsts and causes error during do_rootfs. What is best way to fix? Jan 08 19:02:26 mrk377: figure out why cantarrel postinst isn't running at rootfs time Jan 08 19:02:49 the rootfs log should give a hint Jan 08 19:04:09 Ok. Examining now. Jan 08 19:14:34 rburton: A bit miffed. When running cantarell-fonts scriptlet - sed: warning: failed to get security context of /home/.../../../poky-sumo-build/build/tmp/work/piu-poky-linux/core-image-piu-dev/1.0-r0/intercept_scripts-39843.../update-font-cache: No data available sed: warning: failed to get security context of Jan 08 19:20:23 It fails running update-font-cache so postinstalls for cantarell-fonts saves script in /etc//rpm-postinsts... Hmmm... I'm just talking it through. Must be some obscure security setting for update_font_cache. Do you concur? Jan 08 19:25:40 https://bugzilla.yoctoproject.org/show_bug.cgi?id=9155 Jan 08 19:25:40 Bug 9155: normal, Medium+, 2.1 M3, Martin.Jansa, RESOLVED FIXED, update_font_cache postinst failing with read-only filesystem Jan 08 19:40:00 Hi all! I'm trying to use bitbake-layers to fetch layers and add to bblayer using: bitbake-layers layerindex-fetch -b zeus meta-updatehub-raspberrypi. The layers were fetch and added to bblayer correctly, but are in master branch and not in zeus. I don't know if I'm missing something...Does anyone uses bitbake-layers layerindex-fetch command? Jan 08 19:49:08 Hi, I'm trying to add wic to my IMAGE_FSTYPE. IMAGE_BOOT_FILES has my firmware file (generated by another image_type class) but when I run bitbake do_image_wic can't find the file. Any ideas what I might be missing? Jan 08 19:50:18 my firmware image_type class is listed in IMAGE_TYPEDEP_wic Jan 08 19:55:48 kanavin: 3908 errors found filtered by "33ebcd4e074ae5355d7932535d141365d532d13a" is fairly impressive for the wrong reasons :/ Jan 08 19:59:59 kanavin: dropping some bits and retrying Jan 08 20:20:37 hello, it is possible to use variable which is kind of like PACKAGECONFIG in shell function? I'm struggling how to pass argument to python method which comes from loop of this kind of variable Jan 08 20:22:10 I have variable like VAR = "item1 item2" and then VAR["item1"] = "123" adn VAR["item2"] = "456" Jan 08 20:23:26 then I would like to use this is shell method like: if [ -n ${VAR} ]; then for i in ${VAR}; do j = "${@call_method(d, '$i)}'"; done Jan 08 20:23:40 but argument passed is not variable but string '$i' Jan 08 20:27:00 ${@} is expanded at parse time, before the shell is ever run Jan 08 20:27:30 kergoth: so is there some example how to use this in my specific case? Jan 08 20:27:51 the short answer is you can't do that with shell at all. you'd have to generate the shell you need from python Jan 08 20:28:22 ${@some_python_function_that_iterates_over_VAR_for_you(d)} Jan 08 20:29:06 kergoth: ok I can chaneg to iterate as you suggested but then I need to return to shell variables/list Jan 08 20:29:14 is this possible? like return string list? Jan 08 20:29:42 and then use it in shell function? Jan 08 20:30:39 make some_python_function_that_iterates_over_VAR_for_you() return the python code as a string Jan 08 20:31:37 kergoth: ok and then I can use this as list in shell right? it kind of return different results but I can live with that :D Jan 08 20:32:21 i don't understand the question. the function returns the shell code that'll be executed Jan 08 20:32:24 you can do whatever you want with that Jan 08 20:32:34 return shell commands directly, return a for loop, whatever you want Jan 08 20:32:43 it's just a string that's substituted befroe the shell function is run Jan 08 20:33:30 kergoth: ok fine Jan 08 20:33:33 I think I get it Jan 08 20:33:36 thanks Jan 08 20:45:00 berton: that's a known issue I'm afraid: https://bugzilla.yoctoproject.org/show_bug.cgi?id=7852 Jan 08 20:45:01 Bug 7852: normal, Medium, 3.99, paul.eggleton, IN PROGRESS DESIGN , bitbake-layers layerindex-fetch doesn't respect branch (-b|--branch) option Jan 08 20:45:35 ultimately not that hard to fix, it just needs to check out the correct branch - the only tricky thing is what to do if the layer already exists (checking out could be tricky if there are local changes) Jan 08 20:48:00 bluelightning: I forgot to search in bugzilla, thanks! Jan 09 01:48:32 hey, I bumped a machine from warrior to zeus recently and weston doesn't start Jan 09 01:48:46 thing is, in weston@root.service, the XDG_RUNTIME_DIR is not set Jan 09 01:49:07 I presume something in systemd changed when it's used without PAM, but I can't figure out what ... any ideas ? Jan 09 01:55:42 khem: ^ Jan 09 01:55:50 khem: I see you did something there about non-root users Jan 09 02:05:51 Marex: yes the patches were to accomodate non-root users Jan 09 02:06:13 keeping root user working as it is Jan 09 02:07:11 khem: did you test it without PAM ? Jan 09 02:08:59 khem: without PAM, XDG_RUNTIME_DIR is not set ; weston-start script checks of XDG_RUNTIME_DIR being zero and adjusts it accordingly Jan 09 02:10:42 Marex: it should be defined by weston-init package IIRC Jan 09 02:10:56 are you including weston-init in your image ? Jan 09 02:11:00 khem: it's not :) Jan 09 02:11:01 khem: yes Jan 09 02:11:15 if I undo just the change in weston.service (resp. weston@.service), things work again Jan 09 02:12:04 khem: what happens in my case, without having "pam" in distro features, XDG_RUNTIME_DIR is not set => weston-start would detect that and set it to default Jan 09 02:12:21 khem: but with your change, where you launch weston directly (instead of through weston-start), this logic is gone Jan 09 02:14:29 right it expect it in /etc/default/weston Jan 09 02:14:43 khem: that file does not exist Jan 09 02:14:52 khem: it worked in YPRR 2.7 warrior even without that file Jan 09 02:15:02 see weston-start , there's a test for XDG_RUNTIME_DIR being empty Jan 09 02:15:16 yeah I see that perhaps it should be added to weston-init Jan 09 02:15:51 Marex: yes certainly, I know it worked Jan 09 02:16:04 khem: now it doesn't , and we know why ... what do we do ? Jan 09 02:17:09 but I think its a bug you have stumbled on Jan 09 02:17:18 khem: why would it be a bug ? Jan 09 02:17:45 new logic was so that we can easily add/override to weston init mechanism Jan 09 02:17:45 khem: anyway, will you deal with it or shall I ? Jan 09 02:17:50 from other layers Jan 09 02:18:11 khem: sure, I can override stuff, but now there's default which has nothing set, so there's nothing to override either ;-) Jan 09 02:18:13 can you try a fix to create this file in weston-init recipe Jan 09 02:18:52 which file ? Jan 09 02:19:29 btw you can override the user by setting WESTON_USER env variable and running weston-start , running weston directly instead seems odd Jan 09 02:20:07 and add https://github.com/YoeDistro/openembedded-core/blob/master/meta/recipes-graphics/wayland/weston-init/weston-start#L70-L80 logic to it Jan 09 02:21:21 khem: errr, all right, I lost you, but that's because it's 3am-ish here and I'm tired Jan 09 02:21:29 khem: which file should I augment with that logic ? Jan 09 02:23:45 Marex: sleep dude, I will fix it Jan 09 02:25:24 khem: I am jetlagged, no stress Jan 09 02:25:35 khem: just uh, slowly tell me what to try :) Jan 09 02:26:08 khem: also, no stress, I have a workaround and I can work on a fix too, either way works Jan 09 02:51:44 khem: btw is JaMa still around ? Maybe he's asleep Jan 09 02:52:42 khem: with meta-qt5, he wants something like "if (DISTRO_FEATURES contains (qt5 and wayland)) DEPENDS += " qtwayland " " in gstreamer1.0-plugins-good_%.bbappend recipe Jan 09 02:53:20 how does the patch submission work for that metalayer ? **** ENDING LOGGING AT Thu Jan 09 02:59:58 2020