**** BEGIN LOGGING AT Thu Jan 30 02:59:57 2020 Jan 30 05:34:21 New news from stackoverflow: meta-swupdate layer fails to generate image in 32-bit Jan 30 06:56:57 ptest manual says "The test generates output in the format used by Automake: result: testname" -> isn't the ouput format completely dependent on how I write the run-ptest script? Is this format merely some suggestion or is it enforced somehow within ptest? Jan 30 08:45:22 stuom1: its parsed by our qa code Jan 30 08:45:40 stuom1: lets us collect results Jan 30 08:47:12 RP: so it is important only if I would be making a public package? Jan 30 08:49:30 i was planning to only run our own unit tests and output what our own QA prefers Jan 30 08:52:45 has anyone played around with ptest more? Maybe made some "summary" wrapper for ptest-runner? Jan 30 09:15:34 Hi folks. I was wondering how people using busybox as init_manager normally handled the symlinking of init scripts to rcS.d Jan 30 09:17:30 ehlo Jan 30 09:17:46 is it possible for a recipe to RDEPENDS on a package from the same recipe Jan 30 09:18:04 ie. RDEPENDS_${PN} = "${PN}-lic" Jan 30 09:18:14 amaury_d: sure Jan 30 09:19:09 qschulz: thats weird, I'm getting 'Nothing RPROVIDES ...' Jan 30 09:19:25 amaury_d: then that's your question and not the one you asked :) Jan 30 09:19:25 stuom1: only other consideration is whether you want to use ptest-runner Jan 30 09:19:43 amaury_d: this package does not exist :) Jan 30 09:20:06 but it exists on packages-split directory :/ Jan 30 09:20:16 and is populated with my licences files Jan 30 09:22:32 amaury_d: I'm wondering if -lic is not a special package. have you tried with something else? Otherwise, I'm passing your question to the next one because no idea honestly :) Jan 30 09:22:52 check the typos (missing $ or it's not PN-lic but something hand-crafted etc Jan 30 09:22:52 stuom1: the summary handler is resulttool Jan 30 09:23:15 stuom1: e.g. https://autobuilder.yocto.io/pub/non-release/20190829-10/testresults/testresult-report.txt Jan 30 09:24:09 qschulz: its not a typo, with another package it works fine Jan 30 09:24:20 you are probably right -lic packages muste be something special Jan 30 09:25:22 Thanks RP! Jan 30 09:25:37 nevermind I will put the files in the same base package Jan 30 09:26:47 amaury_d: why do you want the lic only for this package in your system? Jan 30 09:28:30 hi, I'm adding ROOTFS_POSTPROCESS command and call some python method from it which create directory in rootfs /mnt , but when call it like that for every parsed image I have directory created in /mnt (also added some debugging) but I'm not sure why this happens (done during parsing of recipes or?). thanks Jan 30 09:28:39 qschulz: oh you're right, I did not known that the lic packages are automatically installed on the rootfs Jan 30 09:29:00 do not know if it's special to our config or is a yocto default Jan 30 09:29:09 amaury_d: your config I think :) Jan 30 09:29:12 but it seems I do not need the RDEPENDS anymore Jan 30 09:29:14 amaury_d: they are not by default, but you can request the lic packages to be included. Jan 30 09:29:35 LetoThe2nd: what';s the variable for that, you member? Jan 30 09:29:48 qschulz: sure thing Jan 30 09:29:48 https://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#providing-license-text Jan 30 09:35:00 LetoThe2nd: hum I do not have these variables defined Jan 30 09:35:09 only LICENSE_CREATE_PACKAGE is set Jan 30 09:35:29 amaury_d: see, thats why the -lic package is created but nothing used for Jan 30 09:35:33 but the licenses are still included, I do not know the reason Jan 30 09:38:14 in the image? all of them? Jan 30 09:38:53 in that case, i'd try and find out what the variable actually triggers, probably just some globbing or virtual packagegroup. that might be triggered in another way too. Jan 30 09:38:57 yes installed on rootfs Jan 30 09:39:04 all of them it seems Jan 30 09:40:19 amaury_d: run bitbake with -g -u taskexp and see what's including your -lic packages Jan 30 09:44:47 qschulz: ValueError: Namespace GTK not available Jan 30 09:44:55 i'm on a ssh connection :/ Jan 30 09:45:42 when using -g, is there anymore a package-depends.dot ? Jan 30 09:45:54 task-depends and recipe-depends yes Jan 30 09:46:13 amaury_d: you should be able to have X forwarding over an SSH connection with -X right? Jan 30 09:46:31 i can try Jan 30 09:46:34 hum Jan 30 09:46:59 -g changed, its only the task-depends by now Jan 30 09:47:29 like i said, i'd rather just grep and find out what the variables actually trigger. Jan 30 09:47:44 too bad it was useful for me Jan 30 09:47:56 and if you really want to dig deeper, enable buildhistory, rebuild the image and then see whats going on. Jan 30 09:48:06 amaury_d: you still get it as part of the buildhistory. Jan 30 09:48:23 LetoThe2nd: ah true, I always forget what's part of the buildhistory :) Jan 30 09:48:56 how does that works? should I inherit buildhistory on my image recipe ? Jan 30 09:49:16 amaury_d: https://www.yoctoproject.org/docs/3.0.1/dev-manual/dev-manual.html#enabling-and-disabling-build-history Jan 30 09:49:23 thanks Jan 30 09:50:34 !hello Jan 30 09:50:37 hm Jan 30 09:50:50 we need a factoid bot in here. Jan 30 09:51:00 i'm always struggling between all the yocto manuals :/ Jan 30 09:51:31 i could use the mega version Jan 30 09:53:45 LetoThe2nd: I thought you were the factoid bot Jan 30 09:55:22 https://twitter.com/TheYoctoJester/status/1222820538839633921 Jan 30 09:56:19 * RP wonders if the oe-core list is working? :/ Jan 30 09:57:21 LetoThe2nd: qschulz: I found the explication Jan 30 09:57:28 RP: its at least rather quiet, thats true. Jan 30 09:57:38 it seems that the -lic package is RRECOMMENDS by the main package Jan 30 09:57:51 LetoThe2nd: I've sent mails that haven't appeared :( Jan 30 09:57:58 I have not received spam from oe-core since yesterday, that is highly unusual ;) Jan 30 09:58:19 this is what the buildhistory reports Jan 30 09:58:53 strange that bitbake |grep -e '^RRECOMMENDS' returns me nothing Jan 30 09:59:33 bitbake PN -e Jan 30 09:59:57 qschulz: sorry yes I meant bitbake -e PN Jan 30 09:59:58 I have no mails from any ML since this morning Jan 30 10:00:27 thanks for your help, I did not know buidhistory feature, that seems a useful feature Jan 30 10:00:38 amaury_d: could be some core logic which is setting this RRECOMMENDS later, thus not explicit in the recipe? Jan 30 10:01:10 maybe but I tough even the core logic were listed in 'bitbake -e' Jan 30 10:01:38 thought* Jan 30 10:03:48 amaury_d: they expand the variables as much as they can when parsing the recipe, it does not run the tasks AFAIK Jan 30 10:35:16 New news from stackoverflow: Use a yocto layer's recipe or class without inheriting all bbappends Jan 30 11:58:30 RP: I have also sent mail to the openembedded-core list that has not been delivered. Looking at lists.openembedded.org, the last mail in any of the archives is from Jan 29 00:43:09 UTC... Jan 30 11:59:05 Saur: we'll need halstead when he's up :/ Jan 30 12:03:28 rburton: o/ Jan 30 12:03:40 awafaa: blimey Jan 30 12:03:58 honestly thought irc was below your pay grade these days Jan 30 12:04:27 I'm borderline grey beard ;) Jan 30 12:05:32 hi awafaa! :) Jan 30 12:05:52 aloha :) Jan 30 12:37:41 awafaa: i tend to rock a number 3-4 beard as if it gets too long the grey shows too much Jan 30 13:06:23 embrace the grey Jan 30 13:34:27 hello all. i have a question. i am using thud version for all layers. it has opencv 3.4 version. if i want to add a recipe from zeus version for opencv4.1. can i do it? if yes then what is the best way to do it? Jan 30 13:37:00 rizi: depends. either version bump everything to zeus, or backport the newer opncv version into your own layer Jan 30 13:40:21 ok. BTW your videos are amazing on youtube (y) learning alot from them Jan 30 13:42:55 LetoThe2nd: sorry what do exactly mean by "backport the newer opncv version into your own layer" Jan 30 13:43:39 rizi: 1) create a layer of your own Jan 30 13:43:52 rizi: 2) copy opencv recipes from zeus into it. Jan 30 13:44:03 rizi: 3) add your layer to thud build Jan 30 13:44:14 rizi: 4) hope, pray, and see what happends. Jan 30 13:44:31 i mean, your recipe has to go somewhere anyways. Jan 30 13:44:59 LetoThe2nd: haha thanx i will do just that Jan 30 14:08:10 How do I get yocto to push the /usr/include directory to my sysfs image? Jan 30 14:08:39 RP: did you happen to see what was in the SDE file? Jan 30 14:08:46 Basically I want headers so I can build stuff on my board. Jan 30 14:09:13 rob_gries, add the corresponding -dev packages to your image Jan 30 14:10:37 I have DISTRO_FEATURES_append += " dev-pkgs" in my local.conf. Wouldn't that do the same thing? Jan 30 14:12:38 rob_gries, looks like that is an image feature, not a distro features Jan 30 14:13:31 rob_gries, also you dont need both append and += Jan 30 14:16:07 kroon, the weird thing is that when I add dev-pkgs to the IMAGE_FEATURES list via the EXTRA_IMAGE_FEATURES directive my board refuses to boot properly. I realize that this is probably an issue unique to my specific target (qualcomm apq8053) and that the maintainer (qcom) should address it. Jan 30 14:16:27 rob_gries: or its just that the fs gets too large Jan 30 14:16:35 FS size is nominal Jan 30 14:17:46 The image size is under 1GB and the partition is sized 2.1GB Jan 30 14:17:49 and what does "refuse to boot properly" mean? after all the headers alone won't do the trick, you'll need the compilers and all too. Jan 30 14:17:52 (hoping to catch more of the europeans...) what's that trick for resizing the root partition on bootup? Jan 30 14:19:09 IMAGE_OVERHEAD_FACTOR is one knob to turn Jan 30 14:19:32 kroon: I think that just sets the size when the FS is created Jan 30 14:19:58 JPEW, yeah, but I need to up that factor when I install debug packages Jan 30 14:20:15 kroon: Oh, sorry I thought that was in reply to tlwoerner :) Jan 30 14:20:45 well it kind of was, since I didnt get his question fully :-) Jan 30 14:21:52 LetoThe2nd, Sorry, I should be more specific. When building with the EXTRA_IMAGE_FEATURES = " dev-pkgs, tools-sdk" directive I get a valid image but after loading it the log shows this message "Stopping qti:" and reboots itself before getting to the logon prompt. Jan 30 14:22:21 LetoThe2nd, I have the complete boot log if you would like to see it. Jan 30 14:22:40 rob_gries: it never hurts to put it on a pastebin Jan 30 14:23:32 BTW, maybe a typo, but I'm not sure you're supposed to put a comma after dev-pkgs Jan 30 14:23:47 qschulz: thought the same. Jan 30 14:24:09 Oh yeah sorry. That was a typo. Jan 30 14:24:39 It's actually EXTRA_IMAGE_FEATURES = " dev-pkgs tools-sdk" in local.conf Jan 30 14:26:13 LetoThe2nd, I've posted the log on github gist https://gist.github.com/RobGries/5a7083ffdbbdc600638c38201b456c3b Jan 30 14:28:26 rob_gries: ok. sorry, but this doesn't look like anything i've ever seen before. Jan 30 14:29:13 LetoThe2nd, It's okay, thanks for taking a look. This seems to be less yocto-specific than it is qcom-specific. Jan 30 14:29:40 rob_gries: yup. plus, there's way too much android magic in there for me to properly comment. Jan 30 14:29:43 rob_gries: you have one partition which looks corrupted for Linux. Also most likely some qualcomm specific init script which is checking for something and then rebooting if that something isn't as expected Jan 30 14:29:57 LetoThe2nd: that's qualcomm for you :) Jan 30 14:31:31 LetoThe2nd: as we discussed i create a layer with opencv newer version bbfile. when i did bitbake opencv it has given 2 errors. any idea why Jan 30 14:31:46 qschultz, yeah I saw that too. You're talking about the mmcblk0p2 volume? Jan 30 14:32:00 ERROR: Nothing PROVIDES 'ade' (but /home/rizwan/src/ara-project/meta-riz/recipes-opencv/opencv/opencv_4.1.0.bb DEPENDS on or otherwise requires it). Close matches: Jan 30 14:32:10 rizi: *stares into crystal ball. doesn't see anything. checks that it is properly plugged in. still doesn't see anyhting.* Jan 30 14:32:31 qschultz, the odd thing is that when I remove dev-pkgs, the situation improves. As in the board boots and does not bootloop itself to oblivion. Jan 30 14:32:44 rob_gries: use the autocompletion for my nick :) Jan 30 14:32:46 rizi: well thats not clear about that one? you need to backport/copy "ade" too. Jan 30 14:33:12 rizi: or find a layer which provides this recipe for your version of yocto Jan 30 14:33:20 rob_gries: yes I was talking about that one Jan 30 14:33:31 qschulz, heh I didn't know I could tab complete users in hexchat. cool. Jan 30 14:34:07 rob_gries: qcom seems to be testing their system with only one configuration, i'm not *too* surprised it goes bad when changing something slightly Jan 30 14:34:31 ndec: you alive? Jan 30 14:34:46 qschulz: meta-oe provides opencv4.0.1 but it is fro zeus and all previous project is based on thud Jan 30 14:34:59 rizi: that is not what I said Jan 30 14:35:35 rob_gries: check that you are not over the max size of the partition with what you put into this partition Jan 30 14:36:23 qschulz, I'll check the partition map again. Is there a bitbake command to check the size of the rootfs before flashing? Jan 30 14:37:01 LetoThe2nd: i hope! Jan 30 14:37:16 rob_gries: you should have a file somewhere which explicit which file is going to which partition. Then look in tmp dir for a file named that way and you'll have the size? Jan 30 14:37:42 rob_gries: I guess you have a nand, so most likley some ubinize.cfg or something like this. Don't recall exactly Jan 30 14:37:53 ndec: hooray! can you maybe have a peek at rob_gries' problem? Jan 30 14:38:27 sure.. i am in a call right now, but will definitely have a look Jan 30 14:38:42 rizi: there's no ade in http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/opencv/opencv_4.1.0.bb?h=zeus.... where is this ade dependency coming from? Jan 30 14:39:00 qschulz, that's how I have been checking. And the board I'm using has an eMMC Jan 30 14:39:15 ndec: awesome, thanks! Jan 30 14:40:56 qschulz: i have no idea Jan 30 14:41:47 rizi: how did you backport the recipe? Jan 30 14:43:01 qschulz: created a layer. added opencv_4.1.0.bb file(copied). did bitbake opencv Jan 30 14:43:41 honestly I don't know, it does not look very explicit where the dependency is Jan 30 14:43:57 though, there's actually an ade recipe in the same directory as opencv 4.1.0, so maybe just take this one? Jan 30 14:44:18 I hope you took the patches for the recipe as well? Jan 30 14:44:55 no i have not. =# Jan 30 14:45:25 qschulz:how to do that? Jan 30 14:46:01 rizi: you have an opencv directory at the same location as the recipe, take those as well Jan 30 14:46:13 http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/opencv/opencv?h=zeus Jan 30 14:46:21 and this one: http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-support/opencv/ade_0.1.1f.bb?h=zeus Jan 30 14:46:34 put them in the same dirs in your layer as they are in meta-oe layer Jan 30 14:47:34 qschulz: ok got it. i try that. thankyou Jan 30 14:48:18 after putting patches i will just try bitbake opencv again? Jan 30 14:48:42 yes Jan 30 14:48:54 i'm actually surprised it didn't complained it couldn't find the files Jan 30 14:48:58 or did you ignore the warnings :) Jan 30 14:49:25 qschulz: i ignored them. silly me Jan 30 14:49:39 5 of them Jan 30 14:49:54 rizi: and there are 5 patches :) Jan 30 14:50:26 making sense already (y) Jan 30 14:56:32 RP: It seems the mails to openembedded-core has started to trickle in. I saw the patch from you with the memoriam, and was sad to learn about Scott's demise. :( Jan 30 15:00:56 Saur: ? Jan 30 15:02:27 LetoThe2nd: RP sent a patch for OE-Core with a memoriam for developers who are no longer with us, and it includes Scott Rifenbark. Jan 30 15:05:45 * LetoThe2nd didn't get it so far. Oh, sad. Jan 30 15:06:08 New news from stackoverflow: apt sources.list for embedded yocto linux? Jan 30 15:06:54 JPEW: on the YP call the other day you mentioned you were doing some scraping of the autobuilder. Do you have your current scripts posted somewhere? Jan 30 15:08:11 I've been thinking too, do we track build failures over time by error message or something else? Jan 30 15:08:55 sstate mirrors are awesome. that is all. Jan 30 15:10:20 Saur: right, I mentioned it here yesterday evening too. Its very sad :( Jan 30 15:14:55 RP: any context that is being "public"? Jan 30 15:15:27 LetoThe2nd: Scott lost a battle with cancer that has been going on for a while Jan 30 15:16:44 that's a nice patch Jan 30 15:16:47 RP: i see. a loss for us. Jan 30 15:25:59 i am creating a custom image of automotive grade linux - the version i'm working with is using thud, but i've found i need a newer meta-qt5, from warrior. AGL's layers include patches that are specific to 5.11 and don't apply in 5.12 (so there's a qtwayland_%.bbappend that includes patch files). in my layer i'm trying to do SRC_URI_remove = "files://0099_qtwayland_no_evdev.patch" to remove the patch that doesn't apply, but Jan 30 15:25:59 it's not working (it still tries to apply the patch). tried making my layer higher priority but no dice... Jan 30 15:26:24 fury: file:// and not files:// Jan 30 15:26:29 ooof Jan 30 15:26:33 thanks! Jan 30 15:26:46 it's clear i need more sleep :P Jan 30 15:33:20 qschulz: i have added patches this time no warning signs. but still same error messages. ERROR: Nothing PROVIDES 'ade' (but /home/rizwan/src/ara-project/meta-riz/recipes-opencv/opencv/opencv_4.1.0.bb DEPENDS on or otherwise requires it). Close matches: Jan 30 15:33:21 dependency chain was: ['opencv', 'ade'] Jan 30 15:36:11 rizi: read again what I sent you earlier. Two things you had to do Jan 30 15:54:52 rizi: found it? Jan 30 15:56:29 qschulz: yeah. actually i made one blunder. i was using bb file for opencv.4.1.0 master version. that is why it was asking for ade dependency Jan 30 15:57:15 i recommend grabbing the ade recipe too Jan 30 15:57:30 because otherwise opencv just goes and downloads it during configure anyway, which is stupid Jan 30 15:58:17 but it is not in dependencies? why wd it need ade? Jan 30 16:00:29 rizi: http://cgit.openembedded.org/meta-openembedded/commit/meta-oe/recipes-support/opencv/opencv_4.1.0.bb?id=adfe0b63443d8d2b18428afcb922aca422858890 Jan 30 16:02:23 rizi: it seems like master wouldn't be such a bad idea :) Jan 30 16:02:23 ah got it . thankyou:] Jan 30 16:02:50 just take ade as well or have a bbappend for removing ade from PACKAGECONFIG ifyou don't need it Jan 30 16:03:39 ok got it. if zeus does not work then i go tho master:] Jan 30 16:05:09 rizi: zeus will work, it's just downloading stuff it shouldn't Jan 30 16:05:37 zeddii: is there a patch in linux-yocto that we can apply to upstream kernels so "-c menuconfig" works? (i.e. ncurses issue) Jan 30 16:06:00 my host has all the "ncurses*-dev" packages installed Jan 30 16:07:01 qschulz: yeah got it Jan 30 16:34:52 tgamblin: I will post them Jan 30 16:41:14 btw, i found the auto-resize on bootup thingy: https://github.com/96boards/96boards-tools/blob/master/resize-helper Jan 30 16:49:00 tlwoerner: FWIW, you can do it with regular sfdisk these days now that is has gpt support Jan 30 16:49:10 tlwoerner: useful if you are trying to avoid GPLv3 :) Jan 30 17:05:18 tgamblin: https://gist.github.com/JPEWdev/00dc3dd790ff00c64e87c90a96ba1572 Jan 30 17:06:16 tgamblin: Enter the number from the dump on the AB and it will fetch, run diffoscope and popup the results in firefox Jan 30 17:06:34 New news from stackoverflow: Two different recipes names providing same package with different versions. How can one force a default one? Jan 30 17:20:04 JPEW: did you see my note about the power supply? Jan 30 17:20:13 tlwoerner: Ya, that is interesting Jan 30 17:21:54 i guess the ASUS power supply i had been using (which, technically, isn't recommended, i.e. USB PD and QC3.0) was starting at the wrong place Jan 30 17:22:53 qschulz: after bitbake opencv do i need to just add layer into my previous thud build and do a bitbake for previous image or should i create a new image? what is best way Jan 30 17:31:32 I buil qemux86-64 image and my coreutils are build without xattr. why? Jan 30 17:33:38 rizi: you need to add your layer (bitbake-layers add-layer is your friend). Then you add opencv to your image. Either you create one, or you use another one with a bbappend, or if it's for quick debugging, add your package to the image from local.conf directly Jan 30 17:34:56 RP sent QA email Jan 30 17:37:06 dv|2, check your DISTRO_FEATURES Jan 30 17:37:31 look at the recipe to figure out what needs to be set Jan 30 17:37:53 armpit, thanks! Jan 30 17:41:42 qschulz: if i create a new image. then in newimage.bb file in require if i write previous image then it includes all the stuff from that image? Jan 30 17:42:11 rizi: it should yes Jan 30 17:45:57 qschulz: a follow up for sequence of steps: 1. add layer. 2. bitbake previous image 3. create newimage.bb with IMAGE_INSTALL += "opencv" Jan 30 17:46:39 rizi: if you bitbake the previous image, you bitbake the previous image, it does not impact the new image. Jan 30 17:47:40 qschulz: thankx:] Jan 30 17:47:41 rizi: https://www.youtube.com/watch?v=nqHylLP2NmA Jan 30 17:50:28 qschulz: that is great (y) Jan 30 17:52:45 halstead: hi, I registered for the docs ML with my poky ML account, and right after login it said that I didn't have a password set (I do have one) and that a mail will be sent to set one up. I only received a welcome mail to the docs ML. I don't have a screenshot, I'm sorry but I wanted to let you know that this password warning seems wrong to me Jan 30 17:53:40 qschulz, That does sound like an error. I'll check the logs. Jan 30 17:54:06 qschulz, Can you PM me your account e-mail address? Jan 30 17:58:36 halstead: How are the mailing lists on lists.openembedded.org doing? There were no mails delivered yesterday, and today, only a few seem to have trickled through... Jan 30 18:01:01 Saur, I'm still investigating. A runaway log filled the volume. It's cleared now. It appears only a single message was discarded in error and the rest were refused. Messages should appear as re-delivery is attempted. Jan 30 18:01:22 Saur, I expect it would catch up faster though so I'm continuing to look at it. Jan 30 18:01:33 halstead: ok, my bad for that one. I didn't have a new password since we moved to the new system of ML. I might have been redirected to a page where it didn't show that the password has to be set but I'm not sure so please ignore. Thx! Jan 30 18:01:53 halstead: Ok, thanks. Jan 30 18:02:37 No problem. Thanks for the update qschulz. Jan 30 18:10:28 Hello. I have created a custom bistro by creating a separate layer "meta-mydistro" in poky. If I want to add more packages, should I add CORE_IMAGE_EXTRA_INSTALL += "packagename" in build/conf/local.conf OR should it be inside. meta-mydistro/conf/distro/mydistro.conf. ? Jan 30 18:10:55 kiwi_29: you should create your own image recipe and add to IMAGE_INSTALL there Jan 30 18:12:21 kiwi_29: local.conf is... local and you shouldn't rely on people using the one you share (if you do share it). Distro.conf is for distro (there is no DISTRO in the name of the variable, that's usually a good tell), so you're left with image recipes :) Jan 30 18:14:14 I see.. I have IMAGE_INSTALL_append in meta-mydistro/conf/distro/mydistro.conf. where I am adding splash screen to custom distro. Is that the right place for the splash screen? Jan 30 18:14:44 IMAGE_INSTALL... IMAGE... in a distro. That is not correct no :) Jan 30 18:14:46 RP: Is there anything that would prevent anonymous python functions from running before a task? Jan 30 18:15:14 kiwi_29: https://www.youtube.com/watch?v=nqHylLP2NmA Jan 30 18:15:39 RP: Or better yet, are they even supposed to run before each task? empircally, they do in my test, but the documentation says they only run at parsing time Jan 30 18:15:52 qschulz thanks.. will check image recipes Jan 30 18:16:05 LetoThe2nd: I think we'll need to write up some partnership agreement because I bring a few people to your videos :D not right that you'd get all the beers Jan 30 18:18:36 JPEW: cheers Jan 30 18:22:12 JPEW: aren't all recipes parsed before a task is being run? (those not in the parsing cache or whatever its name is?) Jan 30 18:26:23 qschulz: Ya they all should be running during the "Parsing ..." phase before bitbake starts executing tasks, I'm referring to "parsing" before each task run Jan 30 18:29:41 I'm trying to add functions to the ROOTFS_POSTPROCESS_COMMAND in two seperate files. The first file calls the second file with an IMAGE_INSTALL_append. The function in the first file gets added, but the function in the second file does not Jan 30 18:30:11 grep'ing for that flag shows that nothing is hard setting the variable. Any ideas? Jan 30 18:31:00 IMAGE_INSTALL is for packages Jan 30 18:31:10 Sorry yeah, bad terminology Jan 30 18:31:13 packages do not touch the rootfs hence can';t use ROOTFS_POSTPROCESS_COMMAND Jan 30 18:32:17 ROOTFS_POSTPROCESS_COMMAND += should be used in both files (or at least the last one setting this varibale Jan 30 18:32:36 all functions are separated by a semi-colon => ';' Jan 30 18:33:01 and you want your first file to "call the second file" by including or requiring it (you want to require it) Jan 30 18:33:12 Ahhhh. That makes sense. Thanks a million. That should definitely be documented by that variable though Jan 30 18:34:48 The second 'file' is a package. Should I require it instead of IMAGE_INSTALL'ing it? Jan 30 18:37:36 the second file is a recipe (there's no file for packages, a file is usually a recipe (or a conf file for machine and distro; or an inc) and creates packages (or a full image)) not a package :) Jan 30 18:37:53 You cannot have ROOTFS_POSTPROCESS_COMMAND in a recipe other than an image recipe Jan 30 18:38:16 what do you want to do exactly with that ROOTFS_POSTPROCESS_COMMAND in this (non-image) recipe? Jan 30 18:39:04 You're most likely willing to do something the way you shouldn't, so let's tackle that first. afk for a while (maybe until tmrw but I'll catch up if ever) Jan 30 18:40:06 My bad thanks for the info! Jan 30 18:41:04 hey i have a quick question about a change i need to make to one of my recipes but im not sure where to make it. I am trying to produce a fitimage of my kernel. I have enabled the fitimage.bbclass. It seems to be working and generating a fitimage but not in the location my linux kernel recipe expects. If i create a symbolic link between where my Jan 30 18:41:04 kernel recipe expects and where it is build( one directory up and over) i can restart the build and it completes. NOTE: make HOSTCC=gcc -isystem/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/include -O2 -pipe Jan 30 18:41:05 -L/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/lib -L/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/lib Jan 30 18:41:05 -Wl,-rpath-link,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/lib -Wl,-rpath-link,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/lib Jan 30 18:41:06 -Wl,-rpath,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/lib -Wl,-rpath,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/lib -Wl,-O1 -Wl,--allow-shlib-undefined Jan 30 18:41:07 -Wl,--dynamic-linker=/mnt/sharedwork/.../build-mci/tmp/sysroots-uninative/x86_64-linux/lib/ld-linux-x86-64.so.2 HOSTCPP=gcc -E zImage CC=arm-poky-linux-gnueabi-gcc -mno-thumb-interwork -marm -fuse-ld=bfd Jan 30 18:41:07 -ffile-prefix-map=/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/git=/kernel-source/ LD=arm-poky-linux-gnueabi-ld.bfd Jan 30 18:41:07 CONFIG_INITRAMFS_SOURCE=/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/build/usr/mci-core-initramfs-sama5d2-xplained-sd.cpio Jan 30 18:41:08 ./Makefile Jan 30 18:41:11 include/generated/compile.h Jan 30 18:41:15 /mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/temp/log.do_bundle_initramfs.26551) Jan 30 18:41:39 I either need to define a variable i dont know about or something Jan 30 18:42:01 but im not quiet sure where that would be. Jan 30 18:42:05 thanks Jan 30 18:45:12 actually the if image is in the same directory. just under a longer name Jan 30 18:49:22 can devtool (or something) help do a remote gdb debug from host to target? Jan 30 18:49:56 qschulz: I'm trying to create directories for nginx. Why the recipe doesn't create them already I'm not sure. I was following a guide that had this as a fix Jan 30 18:56:53 qschulz: everything worked. i now have opencv 4.1.0 version. but a module "dnn" is missing. how can i add that here? Jan 30 19:03:56 sorry my client disconnected Jan 30 19:04:05 if anyone answered my question Jan 30 19:04:14 could they repeat what they said Jan 30 19:05:05 Does packagegroup-core-buildessential contain both the compiler toolchain and includes? Jan 30 19:11:22 Is Rick Liu here? Jan 30 19:20:43 Hi , I have SC_P_MIPI_CSI0_CLK_P 0xC0000041 pin. when I make dts file I get error as syntax. What must I do? Jan 30 19:28:04 beratiks: not enough info for us to help. "I have xxx" where? you're doing what with it? "I make dts file" where? for which reason? "I et error as syntax" can we get the error? Jan 30 19:29:38 Nathan22: create the directories for nginx in the nginx recipe? use a bbappend, most likely you'll need a do_install_append in there which is creating dirs required by nginx. Might need to tweak FILES_${PN} to include those dirs. Jan 30 19:31:00 I was confused about that. If nginx requires these directories to run it should be part of the upstream recipe right? Jan 30 19:31:50 Nathan22: first thing is to check if those directories aren't in other packages by any chance (tmp/deploy/ipks/ and ucompress them to check what's inside) Jan 30 19:32:11 Nathan22: there might be other tool like oe-pkgdata-util but I've never used it /me shrugs Jan 30 19:32:16 qschulz: you are right. I will write my problem clearly Jan 30 19:32:36 multiconfig question: does 'do_fetch[mcdepends] = mc:machine1:machine2:image2:do_rootfs' only establish a dependency between the steps of the package, or also the existence of files? i.e. woould Image1 only be built when Image2 rootfs task is executed? e.g. if Image2 is manually deleted, and ./bb Image1 is run, it would not know that Image2 needs to be rebuilt for inclusion in Image1? Jan 30 19:32:43 halstead: if YP is going to be doing GSoC, we could probably use a mailing list for that too, but it's not urgent. we should coordinate with ndec to see what he thinks Jan 30 19:34:19 Nathan22: second thing is to find out if there isn't a PACKAGECONFIG for it or something weirder (I don't know how nginx is compiled... configure options? some weird variables used that you have to set? etc...) Jan 30 19:34:55 tlwoerner: I'll wait to hear from ndec about it. Jan 30 19:35:29 Nathan22: if you're not using rm_work, you can check in ${WORKDIR}/image (WORKDIR is the workdir of the recipe) if the directory is here. might be faster than unpacking the ipks Jan 30 19:35:41 Nathan22: i'm off for the night, gl, see you tomrrow Jan 30 19:35:55 I will look into it, thanks again! Jan 30 19:36:08 scripts/verify-bashism question: how long should this take and should it generate incremental output, or one report at the end? I stopped it after 15 minutes, image is ~5000 steps. Jan 30 19:39:34 qschulz: 1) I have nothing under tmp/deploy/ other than images. 2) There is several references to PACKAGECONFIG in the recipe but I haven't used that before so not quite sure what it does. Will have to look into further. 3) Nothing in the nginx recipe dierctory for image. Jan 30 19:53:48 Hello Guys, Jan 30 19:54:12 Is there anyone here who knows about generating fitimages ? Jan 30 19:59:21 kukela_cd: A little bit Jan 30 19:59:29 hey Jan 30 20:00:31 i think i have pretty much everything working. it is just that when i build. It does not create a symbolic link. if i add the link manually it does finish Jan 30 20:01:06 kukela_cd: Which symbolic link? Jan 30 20:01:07 i was wondering if there is a variable you have to set so that it know the name of they symbolic link Jan 30 20:01:46 fitimage-... to fitimage Jan 30 20:01:54 NOTE: make HOSTCC=gcc -isystem/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/include -O2 -pipe -L/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/lib Jan 30 20:01:54 -L/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/lib -Wl,-rpath-link,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/lib Jan 30 20:01:55 -Wl,-rpath-link,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/lib -Wl,-rpath,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/usr/lib Jan 30 20:01:55 -Wl,-rpath,/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/recipe-sysroot-native/lib -Wl,-O1 -Wl,--allow-shlib-undefined -Wl,--dynamic-linker=/mnt/sharedwork/.../build-mci/tmp/sysroots-uninative/x86_64-linux/lib/ld-linux-x86-64.so.2 HOSTCPP=gcc -E zImage Jan 30 20:01:56 CC=arm-poky-linux-gnueabi-gcc -mno-thumb-interwork -marm -fuse-ld=bfd -ffile-prefix-map=/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/git=/kernel-source/ LD=arm-poky-linux-gnueabi-ld.bfd Jan 30 20:01:56 CONFIG_INITRAMFS_SOURCE=/mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/build/usr/mci-core-initramfs-sama5d2-xplained-sd.cpio Jan 30 20:01:57 ./Makefile Jan 30 20:02:00 include/generated/compile.h Jan 30 20:02:04 /mnt/sharedwork/.../build-mci/tmp/work/sama5d2_xplained_sd-poky-linux-gnueabi/linux-at91/4.9+gitAUTOINC+29796588eb-r0/temp/log.do_bundle_initramfs.26551) Jan 30 20:02:14 kukela_cd: pastebin next time please :) Jan 30 20:02:25 oh sorry i didnt know Jan 30 20:03:20 the files themselves are sitting there Jan 30 20:03:39 they just dont seem to be taking on the default name Jan 30 20:05:00 kukela_cd: "sitting there" in your deploy dir? Jan 30 20:05:33 no in the build dir under arch/arm/boot/ Jan 30 20:06:26 it names them fitImage.bak fitImage-mci-core-initramfs fitImage-mci-core-initramfs.bin Jan 30 20:06:37 if i link to the other name Jan 30 20:06:46 the build completes Jan 30 20:08:41 kukela_cd: Hmm, not sure. The last thing I did with FIT images I only had to add these two lines: https://git.yoctoproject.org/cgit/cgit.cgi/meta-rockchip/tree/conf/machine/include/rk3399.inc#n13 Jan 30 20:09:42 kukela_cd: (Line 13 & 14) Jan 30 20:12:56 ill try to add it to my machine configuration file and see that happens Jan 30 20:13:14 i added something similar to kernel recipe bbappend Jan 30 20:13:25 thanks Jan 30 20:13:30 kukela_cd: np Jan 30 20:13:52 kukela_cd: Are you using a stock kernel recipe, or is it from your BSP? Jan 30 20:19:08 it is from my BSP Jan 30 20:19:22 it is an atmel sama5d2 Jan 30 21:03:12 I'm using the a5d27 and the mainline kernel works for me. Jan 30 21:07:26 New news from stackoverflow: Yocto Bitbake fsl-image-qt5-validaiton-imx always error Jan 30 21:25:37 RP: very sad to hear about Scott, he was always so helpful to me Jan 30 21:27:46 still shocked Jan 30 21:34:37 aehs29_: He was one of a kind, lots of good memories Jan 30 21:41:28 JPEW: parsing would have to happen before a task runs Jan 30 21:41:37 the MEMORIAM thing is nice, i was thinking about treke after hearing about scott Jan 30 21:55:47 RP: Hmm, I have *no* idea how YOCTO #13763 would be possible then... Jan 30 21:56:37 RP: It's like the SOURCE_DATE_EPOCH variable keeps the value from the original recipe parse and doesn't get updated after that Jan 30 21:57:33 RP: The value on the inital parse will be 0, since no __source_date_epoch.txt file exists. You can see this if you 'bitbake -e' a recipe that hasn't been unpacked yet Jan 30 21:59:08 JPEW: right, it should have a value after that Jan 30 21:59:23 JPEW: I've been afk but wanted to look at that bug further as it seems odd Jan 30 21:59:59 kergoth: I've thought of treke on a few occasions. The license stuff he did was quite ahead of its time Jan 30 22:00:00 RP: Ya a second set of eyes would be good. The only slightly out of the ordinary thing I've see is that it uses multiconfig Jan 30 22:24:21 RP: I finally managed to stay awake long enough today to have a look at that mips failure you sent (https://autobuilder.yoctoproject.org/typhoon/#/builders/102/builds/238/steps/8/logs/step1c) Jan 30 22:24:29 but I can't reproduce it on the same config here Jan 30 22:24:55 https://pastebin.com/0aNGWQRp Jan 30 22:25:11 is there a full log of those test sessions, somewhere that I'm not seeing it on the autobuilder ? Jan 30 22:25:33 because that error shows that the asm dirctory isn't setup right, and that implies make scripts/prepare failed, but yet it worked here. Jan 30 22:26:32 I wonder if something OOM'd and I'm just not seeing it in the ab log Jan 30 22:26:53 anyway. ZZZzzz for me. I'll try and look into it more as time allows. Jan 30 23:20:55 zeddii: https://autobuilder.yocto.io/pub/repro-fail/mipslogs/ Jan 30 23:21:13 zeddii: I've saved the builddir for now Jan 31 02:09:49 zeddi: is there a patch that you apply to kernel sources that allows "-c menuconfig" to work? Jan 31 02:11:56 oops, zeddii ^^ Jan 31 02:35:38 tlwoerner, he is traveling so not sure what timezone he is in **** ENDING LOGGING AT Fri Jan 31 03:00:42 2020