**** BEGIN LOGGING AT Thu Dec 19 03:00:12 2019 Dec 19 06:28:59 New news from stackoverflow: Bitbake: pass string with spaces to meson config Dec 19 07:48:24 diamondman: you get mybuild/conf/bblayers.conf as starting point. It is created autoatically the first time you run oe-init-build-env Dec 19 07:49:39 @mckoan The reason I was confused by that file being generated is I would expect bblayers to be checked into source control, but I will never check anything in build into source control. Dec 19 08:07:36 diamondman: the build system (bitbake) uses such file to know which are the layers you want to use. Dec 19 08:08:42 diamondman: it can't figure out which thery are. You have to add your layers to it. Dec 19 08:09:43 thanks, but I did know that. I managed to figure out how to override the conf templates with `TEMPLATECONF` Dec 19 08:29:05 diamondman: actually i think you're on the wrong path here Dec 19 08:29:23 bblayers is not meant to go into source control, as it usually contains absolute pathes Dec 19 08:30:01 diamondman: for a rundown of known good practises on that, please look at https://stackoverflow.com/questions/58863254/how-to-manage-meta-layers-for-a-yocto-project-build-configs-in-git/58865947#58865947 for starters. Dec 19 08:30:03 yeah I am editing bblayers.conf.sample in my own template directory now :) Dec 19 08:30:16 ooh happy for this link :) Dec 19 08:30:44 have fun Dec 19 08:34:29 thanks Dec 19 08:48:39 Hello. What is the correct way to add perl modules to my image? I'm confused with all the information I've found on the internet Dec 19 08:57:39 nacknick: IMAGE_INSTALL_append += "perl" in your distro.conf or image file? Dec 19 09:01:08 We use packages to define this Dec 19 09:02:53 as in, packagegroups Dec 19 09:04:09 wertigon: in local.conf under "CORE_IMAGE_EXTRA_INSTALL" Dec 19 09:04:38 local.conf is for your local distribution Dec 19 09:04:48 If you just want to pull a package... Dec 19 09:10:08 Does someone have some experience with booting / running from NAND QSPI Flash? Dec 19 09:11:55 GeneralStupid: ask the actual question and people will know if they can answer or not Dec 19 09:13:39 I have a question: if a package B is added in RDEPENDS of package A through the auto-RDEPENDS mechanism. If package B is rebuilt, that auto-RDEPENDS mechanism should be re-executed right? We have an issue in thud where we added ALL libs from package B to PRIVATE_LIBS but package A still RDEPENDS at packaging time on package B, which can't happen (there is no explicit RDEPENDS) Dec 19 09:17:59 and to complete the whole context. package A and package B provide the same SONAME libraries (hence the PRIVATE_LIBS for one) but one compiled "normally" and the other with libcpp and libc from clang (manually done, not with meta-clang). Dec 19 09:59:34 New news from stackoverflow: List packages which will be included in host Yocto SDK Dec 19 10:05:24 I had to change rust revision, how do I clean rust so it recompiles (meta-rust) it has a recipe called rust, and I did clean on that but nothing really happened Dec 19 10:05:36 it doesn't recompile rust Dec 19 10:06:05 qschulz: this is the actual question. For me it looks like that QSPI NAND is not really supported. Dec 19 10:07:37 GeneralStupid: There is no mention of which software you're trying to run, what tells you it's not supported, what's happening, what you want to achieve, etc. Dec 19 10:08:15 Have you tried without Yocto first, i.e. is it upstream support that is lacking/broken or is it Yocto doing some fancy stuff you want to fix. Dec 19 10:09:26 I doubt it's Yocto related, so you can head to your project's IRC channel. Most likely #mtd for the NAND support in U-Boot/Kernel. Might also be worth asking help on #u-boot. Also if I recall correctly, it's also an NXP vendor BSP, so contact them. If it's BOOTROM issue, contact your supplier, etc. :) Dec 19 10:16:09 qschulz: Ideally i want to boot from it. but thats a NXP related question. So the idea is to initally boot from QSPI NOR and then, when Uboot is loaded i Chipselect the NAND Flash and boot from it. Yes i already work into that BOOTROM. Its extendible... Great... Dec 19 10:29:39 New news from stackoverflow: Where are bitbake python functions documented || Poky-sumo : nothing provides kernel-module-hello... error Dec 19 10:37:34 Ok, tested over and over to get psplash to work with my image Dec 19 10:37:47 It completely refuses to work -_- Dec 19 10:38:02 If I create a different image and convert it, it works however Dec 19 10:38:42 The problem so far is two-fold: Dec 19 10:38:55 1. psplash thinks red is supposed to be gray Dec 19 10:39:25 Has Someones already incurred in this error '#error "_PYTHON_INCLUDE_TARGET or _PYTHON_INCLUDE_NATIVE is not defined"'? Dec 19 10:39:48 2. psplash offsets the image by one pixel every time Dec 19 10:39:54 for each line Dec 19 10:40:11 It looks like the image is to blame here; Dec 19 10:40:56 is there a way I can check the .h image gdk-pixbuf generates? Dec 19 11:39:50 RP: ross/mut-good mostly passed on the ab. selftest failed in a test i added (removed from branch), armhost and arm-ptest failed in nspr-native (missing unpack) Dec 19 11:47:23 qschulz: oh and kudos for getting the helloworld compilation question on SO done. i read it, and the existing question was so off that i completely didn't feel like investing the time. Dec 19 11:51:48 s/existing question/existing answer already/ Dec 19 11:54:48 LetoThe2nd: :) Dec 19 11:55:23 qschulz: :) Dec 19 12:04:37 rburton: thanks. I'm trying to sort the bitbake-selftest issue that is plaguing builds :/ Dec 19 12:42:58 hi, I'm tring to add this directory to mc recipe (/home/root/.config/mc) and also set FILES_${PN} += "/home/root" but it still complain about QA during package Dec 19 12:43:03 any ideas? Dec 19 12:46:14 about what does QA qctually complain? Dec 19 12:46:35 LetoThe2nd: QA Issue: mc: Files/directories were installed but not shipped in any package Dec 19 12:47:06 this is bbappend: Dec 19 12:47:09 do_install_append () { Dec 19 12:47:09 install -d ${D}/home/root/.config/mc Dec 19 12:47:09 } Dec 19 12:47:09 FILES_${PN} += "/home/root/.config/mc/*" Dec 19 12:47:42 original issue is that if this path doesn't exists and I enable eradonly rootfs mc cannot be started Dec 19 12:47:54 *readonly Dec 19 12:48:33 not sure if FILES like the globbing Dec 19 12:48:40 why not just the directory? Dec 19 12:49:35 I tried also only "/home/root" same error Dec 19 12:50:28 and it does really complain about *that* specific file? Dec 19 12:50:58 ERROR: mc-4.8.22-r0 do_package: QA Issue: mc: Files/directories were installed but not shipped in any package: Dec 19 12:50:58 /home Dec 19 12:50:58 /home/root Dec 19 12:50:58 /home/root/.config Dec 19 12:50:58 /home/root/.config/mc Dec 19 12:54:24 hm Dec 19 12:54:48 and you have checked bitbake -e that FILES really looks the way you think it does? Dec 19 12:54:51 weird isn't it ;) Dec 19 12:54:55 yes I checked that Dec 19 12:54:57 path was there Dec 19 12:55:11 ok. no idea then, sorry. Dec 19 12:55:28 FILES_mc="/usr/bin/* /usr/sbin/* /usr/libexec/* /usr/lib/lib*.so.* /etc /com /var /bin/* /sbin/* /lib/*.so.* /lib/udev /usr/lib/udev /lib/udev /usr/lib/udev /usr/share/mc /usr/lib/mc/* /usr/share/pixmaps /usr/share/applications /usr/share/idl /usr/share/omf /usr/share/sounds /usr/lib/bonobo/servers /home/root/.config/mc/*" Dec 19 12:55:50 isn't home dir somehow special and we cannot create files in it or so? Dec 19 12:59:20 Is it possible to use yocto with the meta-java layer on ubuntu 18.04? I'm facing some "os-check" problems during the build of the openjdk-8 recipe. Dec 19 13:00:03 nickes: "depends" Dec 19 13:03:05 Aaaah... Now I realise what I did wrong, psplash rejected a width of 513 pixels :D Dec 19 13:03:11 LetotThe2nd: on what? Dec 19 13:03:36 LetoThe2nd:sry, on what? Dec 19 13:04:08 nickes: primarily on the yocto/OE release in use. Dec 19 13:04:26 nickes: and, if openjdk checks it itself, maybe even for a good reason Dec 19 13:07:56 opennandra: When we add whole directories to a package we use the directory path ( FILES_${PN} += "/home/root/.config/mc" ), maybe that helps. Dec 19 13:08:36 LetoThe2nd: oh boy, I didn't thought that my build in a dependency hell. What would you suggest? Changing to a former os version e.g. ubuntu 16.04? What version(s) do you use? Dec 19 13:09:15 nickes: getting rid of java Dec 19 13:10:01 nickes: another option obviously building in a container Dec 19 13:10:06 LetoThe2nd: nice Dec 19 13:11:01 LetoThe2nd: yeah good idea, thanks a lot Dec 19 13:15:50 hpsy: I've tried that and it helps ;), thanks Dec 19 13:17:37 hpsy: huh... now whats the difference in what i said? Dec 19 13:18:11 LetoThe2nd: FILES_${PN} += "/home/root/.config/mc/" instead: FILES_${PN} += "/home/root/.config/mc/*" Dec 19 13:18:20 ther edoesn't exists any files in this directory Dec 19 13:18:22 just directory Dec 19 13:18:35 oh boy Dec 19 13:18:46 oh yes Dec 19 13:18:51 small stuping * :) Dec 19 13:18:55 thanks a lot for help Dec 19 13:18:59 LetoThe2nd> why not just the directory? Dec 19 13:19:13 i thought that was clear that you should point to the directory, without the trailing slash Dec 19 13:20:00 well overlooked it sorry Dec 19 13:20:37 opennandra: hint: often looking at other recipes help, seeing how they do things. Dec 19 13:20:59 LetoThe2nd: I'm looking a lot Dec 19 13:30:12 New news from stackoverflow: Different versions of yocto in one project || Build a recipe that depends on another recipe: Yocto? Dec 19 13:37:06 opennandra: You're welcome :) Dec 19 13:37:55 * kanavin is carpet bombing the oe-core list Dec 19 13:55:22 kanavin: what branch is the latest series in? Dec 19 14:41:58 maybe I'm working in some unholy scenario, but... Dec 19 14:42:08 Exception: NameError: name 'staging_package_populate_pkgdata_dir' is not defined Dec 19 14:42:30 I'm guessing this shouldn't happen Dec 19 14:43:36 RP: going to be late for triage, need to do the school run today Dec 19 14:44:43 it looks like a typo of package_populate_pkgdata_dir Dec 19 14:53:09 rburton: ok. Small issue is it means there will be about 3 of us :( Dec 19 14:55:03 RP: I also will not be there... we have a department meeting today Dec 19 14:55:44 * RP spent the day cancelling other things so I could be around for the meeting Dec 19 14:56:10 JPEW: ok, thanks for the headsup Dec 19 15:36:59 rburton, akanavin/package-version-updates, just pushed there Dec 19 15:39:59 thanks Dec 19 15:55:27 wow.. almost 300 new CVE entries overnight.. Dec 19 16:01:25 looks like most of them are MacOS related Dec 19 16:01:58 For some reason I was disconnected. Did someone answer my question before? Dec 19 16:02:25 I tried to add "custom" recipe for python 3.6.8 Dec 19 16:02:47 I have the source here: https://www.python.org/ftp/python/3.6.8/Python-3.6.8.tgz Dec 19 16:03:01 I've not seen anything.. a lot of people are out or on vacation right now.. so it might be slow getting responses to questions this time of year Dec 19 16:04:07 and I need to run ./configure, make and make install in the target machine. Is that can be automatically by Yocto? Dec 19 16:04:27 nacknick: i'd just take a py3.6 recipe from a newer branch than you're using Dec 19 16:04:37 cross-building py is *not easy* and we have ~15 patches on it Dec 19 16:04:46 newer branches have ver 3.7 and above Dec 19 16:05:20 I have to use rocko branch Dec 19 16:20:53 zeddii: Are you around? Dec 19 16:52:49 RP: my world builds are now reduced by almost 2000 seconds after this cmake change Dec 19 16:53:07 rburton, we never had a 3.6 recipe. straight 3.5 -> 3.7 Dec 19 16:53:33 nacknick, I would suggest you rebase to a newer yocto release Dec 19 16:53:57 and talk to your management about why it's important to be close to upstream Dec 19 16:59:15 nacknick: if you need newer why not take 3.7? Dec 19 17:00:14 khem: is that just from the webkits or from other things too? Dec 19 17:00:32 khem: looks like there is some discussion on that patch to be had :/ Dec 19 17:03:46 khem: will that patch make it? I really like the idea of it, at least for us...If the build fails we hopefully can reproduce it on the developer machine and debug it there. Dec 19 17:09:17 I can't find the support schedule for releases. https://wiki.yoctoproject.org/wiki/Releases says 2.1 is on community-only support, how do I know when it's EOL? Dec 19 17:22:45 RP: I view it that the patch makes things better atleast we can now add/delete this option at recipe level or global level, the discussion is primarily if it should be on by default or off Dec 19 17:23:57 RP: I live in both worlds, as OE dev who is fixing errors on daily bases, I want these logs, as a user, I dont want them and users to dev ratio is 500:1 Dec 19 17:25:43 khem: we need a variable for it, then it becomes a question of the default Dec 19 17:25:59 we have a small team that works on browsers, who develop these browsers and they recommend to turn it off too Dec 19 17:26:50 RP: set EXTRA_OECMAKE_BUILD += "-v" in local.conf Dec 19 17:27:20 Which CMake change are you talking about? Dec 19 17:27:36 khem: I think it will need a specific variable so we can have a default, the recipes can change it and then ultimate the local.conf can override Dec 19 17:27:46 georgem: to turn off verbose logs Dec 19 17:27:51 ahhh. nice Dec 19 17:28:22 interesting that helps so much. Dec 19 17:28:24 RP: for specific variable, I was thinking should be pan component build sytems Dec 19 17:33:38 I kinda doubt user to dev ratio is 500:1. I think most OE/Yocto users inevitably get sucked into doing some dev to fix issues through the course of product development unless they have someone else on their team doing it. Dec 19 17:33:39 rangergord, community is effectively EOL, you might be able to get commercial support, but there is no upstream maintenance Dec 19 17:33:57 rangergord, generally a release is supported for 12 months, as of now Dec 19 17:35:11 heh. yeah community is a nice way of saying "here's the source, knock yourself out" Dec 19 17:35:48 kanavin, I'm new to Yocto, in general if a board supports Yocto 2.1, what are the chances of me being able to run a newer Linux kernel 5 years from now, if the manufacturer won't support it? Dec 19 17:36:19 rangergord: we do it routinely but requires some work Dec 19 17:36:42 depends how good upstream support is Dec 19 17:37:13 what are the best brands for upstream support? Dec 19 17:37:31 Well, kinda two camps on that. Intel and Other. Dec 19 17:37:45 and if we go with an Atom board, will that make the task a lot easier? Dec 19 17:37:47 For other I'd say the Freescale iMX line is quite good. Dec 19 17:38:15 RP: perhaps https://git.openembedded.org/openembedded-core-contrib/commit/?h=yoe/mut&id=488115e79805a250825a844e374da64a30bba4d6 Dec 19 17:38:47 I'm not 100% sure we'll be able to go with x86, it's what I'm pushing for as a pure SW guy...but if we have to go with ARM then suddenly my kernel requirements go up by a lot (due to the fact that I want to use .NET Core) Dec 19 17:39:24 georgem, thanks for the advice Dec 19 17:39:25 rangergord: Intel stuff is generally easier to keep going but there a bunch of other issues to deal with like UEFI, lack of devicetree, etc. Dec 19 17:42:12 rangergord, generally, choose a board where you can migrate to a newer yocto release Dec 19 17:42:37 yup Dec 19 17:42:38 RP: perhaps, I will just keep it enabled and since it offers me a way to disable it via config metadata, I will disable it globally for rdk and my distro Dec 19 17:45:58 newbie question, after compiling the kernel, where I can access the .config file for the kernel in yocto build system? Dec 19 17:49:06 I think this relentless sw upgrading with disregard for existing h/w, is quite irresponsible on world Dec 19 17:49:33 rostam: you can find it in build are of kernel Dec 19 17:49:45 depends how you have it setup but ... work directory -> machine directory -> linux-yocto -> version directory -> linux-*-build Dec 19 17:49:57 khem, because newer SW increases productivity, SW devs like me always want the latest Dec 19 17:50:11 rostam: bitbake -e virtual/kernel | egrep "^B=" will let you know your kernel's builddir Dec 19 17:50:24 there you go Dec 19 17:50:52 georgem, what kind of devices you work on that you need to routinely update kernels? Dec 19 17:51:42 &IMA Dec 19 17:51:46 &IMA Dec 19 17:51:59 rangergord: various utility and process automation systems Dec 19 17:52:08 rangergord:oh I know the whole spiel, but this even increasing demand for more h/w resources with new versions is like a hamster wheel Dec 19 17:56:18 and if I'm honest, as someone who had to skip all of Qt 5 until now (seriously, we were on Qt 4 since 2011...Qt6 comes out in 2020), it basically makes you obsolete. If you don't follow the SW world, you're left behind and become unhirable. On the company side, it becomes harder to hire people because no one wants to work with your old crap. Dec 19 17:57:57 rangergord: sometimes it just work company don't see the plus value of. If it takes 3 month to upgrade your software to Qt5, then you get 0 new feature to selle Dec 19 17:58:00 sell* Dec 19 17:58:20 Depends on the team size I guess. Dec 19 17:58:25 yeah depends Dec 19 17:58:27 RP: sent a v2, where default verbosity is kept as it is, I was hoping to improve the defaults Dec 19 18:01:45 I think people are over dramatizing the "unhirable" thing recently. I see people getting hired everywhere that don't have anywhere perfect experience for the job they're hired into. It's not always the most important thing. Dec 19 18:10:21 rangergord: lol!! a red green fan!! :-D Dec 19 18:11:21 georgem, sure but your value as a dev is reduced. Your experience in the tech required determines your value. Not saying you can't get hired at all, but you're worth less (not worthless). I can assure you that it takes a HUGE amount of time to recruit because devs avoid tech that either isn't super-interesting from a geek POV (like Rust/Haskell/F# hipsters), or stagnates their value. It took me 6 months to find a Dec 19 18:11:21 replacement for myself when I wanted to go on a sabbatical. Dec 19 18:13:11 tlwoerner: hehe always nice to see fellow fans. You're only like the 3rd person in a decade to comment though :) Dec 19 18:14:05 georgem, of course it didn't help that C++'s popularity totally tanked and it's now the age of Java/C#/PHP/Javascript/Python Dec 19 18:14:23 well, not PHP Dec 19 18:16:29 rangergord: red green and sctv are the best! schitt's creek is pretty good too; it's filmed at two locations i know quite well :-) Dec 19 18:34:00 What is the correct task to use to run a command that manipulates the rootfs? Specifically, I want to run depmod after installing a config file in /etc/depmod.d/ and installing some modules. Dec 19 18:39:13 zeddii: I just sent a patch for kernel-yocto.bbclass... I'm not sure if there is a reason that the BSP was omitted from patching or if it was just an oversight? Dec 19 18:40:50 yah. it absolutly can’t do that. Dec 19 18:40:58 so I’ll nak the patch on the list and explain there. Dec 19 18:41:25 zeddii: OK, that's fine, but how do I apply the patches then? Dec 19 18:41:57 what you have there would reapply all the patches that are already part of the tree. bad things. Dec 19 19:59:58 * zeddii sobs Dec 19 20:00:16 someone just mentioned meta-debian as valid option. Dec 19 20:00:29 * zeddii means no offence if the meta-debian folks are here, just not my cup of tea. Dec 19 20:16:26 zeddii: valid for what Dec 19 20:17:51 as a viable option for a supportable rootfs and a bridge to debian instead of yocto. Dec 19 20:18:29 there must be a reason why people want that Dec 19 20:19:11 yep. I can't speak for others, but why I'm hearing it is the same as I was saying in Lyon at the meetings. Dec 19 20:19:31 zeddii: have you looked at isar? Dec 19 20:19:39 it's well done in my experience Dec 19 20:20:02 yah. Kevin and the gang were talking about it at ELCe, so I checked it out. Dec 19 20:20:14 if you are going that way (debian), I'd agree. Dec 19 20:20:59 ah, you're looking to move *away* from debian? Dec 19 20:21:49 not exactly. just going around the wheel of debian vs yocto again, and heard meta-debian as a half way. I wouldn't go half way myself, chose and go for it. Dec 19 20:22:21 but that's just my point of view, I can't say I've looked into the options lately. Dec 19 20:23:32 * kergoth nods Dec 19 20:23:38 kergoth: does isar use prebuilt archives ? Dec 19 20:23:42 i feel like you should either go debian or go oe/yocto, not somewhere in between Dec 19 20:24:08 kergoth I have said exactly the same thing multiple times.. Dec 19 20:24:23 If you want binaries and don't -ever- want to customize or build, Debian is fine.. Dec 19 20:24:24 khem: it builds root filesystems from debian package feeds, using bitbake to do the work, with classes, uses debootstrap/etc, but has classes to create packages from recipes using dpkg tools and add them to the feeds for use in image construction Dec 19 20:24:35 s/do the work/wrap the tools/ Dec 19 20:24:41 if you want binaries, but might need to customize (or plan to) and you MUST be able to rebuild.. OE/Yocto is the only answer../ Dec 19 20:24:48 but don't mix and match, then you get the worst of both Dec 19 20:24:51 kergoth: thats neat actually, Dec 19 20:25:15 isar is okay if you just want to build on debian, but it won't give you the ability to rebuild existing upstream components, though presumably they could add a class to use apt-get source or something.. Dec 19 20:25:21 * kergoth nods Dec 19 20:25:29 kergoth: so one can take a package modify and rebuild it Dec 19 20:25:44 ?ah that answers Dec 19 20:25:52 so its basically an image builder Dec 19 20:26:08 to spin out custom images Dec 19 20:26:18 it does have classes that let you fetch your own sources and build a package out of the debian/ bits in the source tree and add it to the image Dec 19 20:26:22 but for the most part yes, afaik Dec 19 20:27:22 it's a nice way to go if you want to build debian based filesystems for your devices with some level of customization, just not the flexibility of oe/yocto Dec 19 20:27:27 I think multiarch has grounded itself well Dec 19 20:27:55 https://github.com/ilbers/isar/blob/master/doc/technical_overview.md Dec 19 20:28:04 kergoth: how needs ferrari, while camry can take you from A to B Dec 19 20:28:24 yeah, it's good enough for a lot of folks. depends on your requirements, really Dec 19 20:29:08 * zeddii nods Dec 19 20:29:24 https://mgc-images.imgix.net/embedded-software/MEL_Omni_OS-12C8EFDB.png?q=60&fit=max&w=1440 - note the upper left box Dec 19 20:29:41 I think cross building somehow does not stick with developers unless the workflow is wrappen in native workflow Dec 19 20:29:51 Mentor now has two embedded linux products, flex is oe/yocto based, omni is isar based, to provide both for folks that need an OSV Dec 19 20:30:06 that's true Dec 19 20:30:25 kergoth and I think that is a reasonable thing (two different systems for two differetn users) Dec 19 20:30:27 so its again arm vs x86, where do you debug your problems ? Dec 19 20:31:03 but I think Yocto can reach a bunch of people asking for Debian [or more likely Ubuntu] by having a 'binary distro'... problem is it takes resources to do that work, otherwise we have most, if not everything we need already Dec 19 20:32:00 agreed Dec 19 20:32:32 * zeddii nods Dec 19 20:32:47 I'm that student in the front row that nods at everything :D Dec 19 20:33:17 * fray segues into https://wiki.yoctoproject.org/wiki/Binary_Distro_Process Dec 19 20:33:43 havn't sent that out officially yet.. but if anyone has comments, I'd like to address them before I do.. (Was planning to send something out on Monday) Dec 19 20:33:48 (or Friday) Dec 19 20:34:42 Doing that process would enable someone to build a binary distro for quick start.. and then allow the person to transition to an eSDK environment where they could augment their private package feeds, all the way up to a full Yocto Project environment where they can customize everything Dec 19 20:34:54 wise plan. send it out when everyone has gone for the holidays, and then write it down in stone after :D Dec 19 20:35:17 I'll read it though, and break your process Dec 19 20:35:22 sounds like a nice idea, will have to read it Dec 19 20:35:33 fancy. diagrams. Dec 19 20:35:41 * zeddii whistles Dec 19 20:35:51 this is just talking about the binary distro bit.. I don't have pretty diagrams showing the transition yet.. it's in my TODO.. in my copious spare time Dec 19 20:36:16 bah. cut out sleep. Dec 19 20:36:17 there is only one fancy diagram.. this stuff is pretty much simple for anyone who has done it before.. but people keep saying it's not possible or they don't know how to do it Dec 19 20:36:50 looks good on my read so far, but having done a lot of it before .. I'm probably not the target :D Dec 19 20:38:18 fray: what BSPs would that binary distro support? Dec 19 20:38:50 answer is whatever BSPs are enabled Dec 19 20:38:53 this is about a process.. Dec 19 20:39:20 the configuration becomes specific to the distribution.. but it does mean that BSPs would need to use common tunes and such to make things common Dec 19 20:39:41 yah. exactly, this isn't for someone that is trying to wring the most out of their ISA Dec 19 20:39:50 but neither is debian, so that's ok. Dec 19 20:39:54 it COULD be, but that would be very inefficient Dec 19 20:40:18 this process very well could also be for -one- board as well.. but again, less efficient from a management standpoint Dec 19 20:40:22 agreed. jonmason and I were defining some baselines in the fall and need to pick it up again. Dec 19 20:40:38 the requirements is a list of things that need to be considered and either accepted, defined or specifically stated as not applicable Dec 19 20:40:56 zeddii: does linux-yocto support qemuppc64 ? Dec 19 20:40:57 I still owe fray a review of that.... Dec 19 20:41:23 zeddii: I see you nodding at everything I might get past it Dec 19 20:41:31 lol. :D Dec 19 20:41:33 is there a qemuppc64 yet? last I checked it STILL didn't work for anything but Power9? (Power LE..) not 'powerpc' Dec 19 20:42:10 it did, or at least I thought it did at one point. but I don't see it now. this wouldn't be the only time something went poof (riscv did a couple of times). I'll look into it. Dec 19 20:42:23 lol Dec 19 20:42:28 I swear I remember booting it. Dec 19 20:42:41 How can I run depmod against my rootfs after everything has been copied into place, but before it's packed into an image? Dec 19 20:44:31 dfrey, it is already run in several places. are you building out of tree modules ? Dec 19 20:45:09 fray: if you dont need kvm it should Dec 19 20:45:10 zeddii: Yes. I'm using linux-backports Dec 19 20:45:42 my condolences ;) Dec 19 20:45:52 I know... Dec 19 20:46:11 what release of yocto ? Dec 19 20:46:15 2.7 Dec 19 20:46:23 sorry, 2.5 Dec 19 20:46:54 have a look at kernel-module-split.bbclass, and see if it is inherited, anything packaged with it should have a postinst that runs depmod -a Dec 19 20:47:13 and the image bbclass has a call to depmodwrapper as well. Dec 19 20:47:38 Should my backports recipe inherit kernel-module-split? Dec 19 20:49:37 I'm not sure. I haven't had to support an out of tree module build for a while. but at the least, you can see how it arranges to run depmod. Dec 19 20:51:05 zeddii: Thanks for the hints **** ENDING LOGGING AT Thu Dec 19 21:08:38 2019 **** BEGIN LOGGING AT Thu Dec 19 23:03:52 2019 Dec 19 23:57:13 RP: fun bug: bitbake [recipe] succeeds but nothing in deploydir, or any tasks for it run at all in fact Dec 20 00:30:06 hm what makes recipe-sysroot-native fill my arm .so dir with x86 libs? Dec 20 00:30:34 Ad0: the clue is in the -native bit Dec 20 00:30:42 its native code for the build host to run Dec 20 00:31:16 ok it tries to link the ARM built stuff to those .a files Dec 20 00:31:23 and fails because of "wrong format" Dec 20 00:31:34 sounds like your makefiles are broken Dec 20 00:31:42 yeah I want to track those down Dec 20 00:31:45 RP: bug retracted: i can't ready Dec 20 00:32:20 Ad0: presumably this is a recipe you wrote, probably for code you wrote? Dec 20 00:32:26 RP: (read, not ready) Dec 20 00:33:20 rburton, no :/ Dec 20 00:33:25 what recipe? Dec 20 00:34:04 meta-iotedge -> iotedge-daemon Dec 20 00:34:27 https://github.com/Azure/meta-iotedge/issues/22 it already has an issue registered Dec 20 00:34:37 I wanted to check if it's recipe related or makefile related Dec 20 00:35:19 bah its all rust Dec 20 00:35:29 could be either. Dec 20 00:35:33 ok thanks Dec 20 00:35:50 maybe MS didn't notice as they use disable-static-libs, so there are no .a files in recipe-sysroot-native to link against? Dec 20 00:36:12 try building something else using rust and seeing if that works Dec 20 00:36:25 hm yes I will try to find something Dec 20 00:36:42 I did an earlier version in warrior, that worked Dec 20 00:36:48 so it's clearly something local to that layer Dec 20 00:37:01 but rust breaks the code and vice versa between versions Dec 20 00:37:07 I can never seem to catch a brake lol Dec 20 00:37:13 break* Dec 20 00:38:42 if I clean the recipes and build it all over again, how did I log all the build stuff? I guess I could catch it there. **** ENDING LOGGING AT Fri Dec 20 03:00:11 2019