**** BEGIN LOGGING AT Mon Oct 26 02:59:57 2020 Oct 26 03:45:05 moto-timo: i kinda feel k8s is a much better fit for overc then k3s ..... IMHO Oct 26 03:45:22 even KubeEdge could be a better fit Oct 26 03:46:43 k3s does some things in a funky way that makes it non-compat with various k8s things that just work on k8s Oct 26 05:25:23 @moto-timo, OutBackDingo: indeed https://docs.kubeedge.io/en/latest/components/kubeedge.html might be a better fit if you want k8s compatibility for you edge devices. As far as I understand you could have even smaller devices communicating to kubeedge via mqtt. Oct 26 05:46:45 RobertBerger: yeah now if we can get it in meta-virtualization :) Oct 26 05:47:33 WIP getting it running on overc, then look at making recipes for it Oct 26 06:32:26 @OutBackDingo: The doc does not say much about how to deploy it: https://docs.kubeedge.io/en/latest/setup/index.html Oct 26 06:34:19 @OutBackDingo: We only need some container runtime for it on the edge device? https://docs.kubeedge.io/en/latest/getting-started.html Oct 26 07:38:13 yo dudX Oct 26 07:39:27 good morning Dublin! Oct 26 07:42:29 i basically feel like this: https://youtu.be/BIikfdNIHQE Oct 26 07:46:47 LetoThe2nd: LOL Oct 26 07:47:12 * mckoan is trying to figure out how all this 'virtual' stuff works Oct 26 07:47:35 mckoan: hehe. thats easy. it just doesn't work at all. Oct 26 07:58:22 LetoThe2nd: what is the link to the conference? Oct 26 07:59:37 mckoan: schedule? Oct 26 07:59:48 https://events.linuxfoundation.org/embedded-linux-conference-europe/program/schedule/ Oct 26 08:05:14 LetoThe2nd: I'm in my schedule but I don't see any link to the conference Oct 26 08:15:52 mckoan: just click the ogo? https://events.linuxfoundation.org/embedded-linux-conference-europe/ Oct 26 08:19:57 paulbarker: so hopefully that was helpful and i did not ah... erm.... too much? Oct 26 08:20:48 LetoThe2nd: I thought that was a good way to start the conference day Oct 26 08:21:04 LetoThe2nd: looks like LF is going to send the actual link later by email Oct 26 08:21:10 paulbarker: :) Oct 26 08:21:30 mckoan: ah you mean the conference attendee link? yeah that should come in at 10:00 AM GMT, AFAIK Oct 26 08:30:58 is there going to be a general ELC-E via slack again? Oct 26 08:34:59 shoragan: afaik the LF is convinved their conference system is awesome enough this time so they don't need it. Oct 26 08:36:31 All the cool kids are already on irc Oct 26 09:22:47 I am using two different machines (qemu, imx8mmevk). For the machine imx8mmevk i added a bbapend to include the cp210x driver to the kernel (sources/meta-abc/recipes-kernel/cp210x/linux-imx_%.bbappend) which also contains 'COMPATIBLE_MACHINE = "^imx8mmevk$"'. But now i can't build qemu: 'ERROR: No recipes available for: /somedir/sources/meta-abc/recipes-kernel/cp210x/linux-imx_%.bbappend' Oct 26 09:22:51 any ideas? Oct 26 09:25:42 chris_ber: you should have two different BSP layers for two different architectures (meta-abcqemu, meta-abcimx8) Oct 26 09:31:27 mckoan: hum, nope. a bsp layer should have no effect if included, but building for another machine. Oct 26 09:32:31 chris_ber: it rather sounds like that meta-abc layer has a dependency on that imx bsp layer, that breaks when building for qemu? Oct 26 09:35:36 ok, thx. I create now 2 different additonal sources/layers that will be added to bblayers.conf depending og the machine Oct 26 09:38:01 ? Oct 26 09:39:13 well thats one option. the other would be using BBFILES_DYNAMIC (or similarly named, can't usually remember) in meta-abc. and the completely simple approch would be to just have the imx layer included even if you're building for qemu. if it is properly maintained, it should have no effect then. Oct 26 09:39:39 ahh ok Oct 26 09:55:58 shoragan LetoThe2nd: Yes there is a Slack channel for YP at ELCE: https://join.slack.com/t/theyoctoproject/shared_invite/zt-i5tjtidz-EgHG4cq4c7fLn_0PXT3~IQ Oct 26 09:56:12 Channel is #elce Oct 26 09:56:51 dreyna: for YP, sure. but AIUI, shoragan asked for a general, LF channel/slack Oct 26 09:57:41 Ah no. As you stated the LF is saving money with their platform's narrow chat channel Oct 26 09:58:07 dreyna, during ELC-NA, it was a really useful replacement for the hallway track Oct 26 09:58:09 * LetoThe2nd will try it out shortly! Oct 26 10:19:17 i can keep the booth in a secondary tab, thats ok Oct 26 10:20:07 Hi Oct 26 10:20:32 I ran menuconfig, ehich produced the config file as expected Oct 26 10:21:05 But the diffconfig does not produce the fragment.cfg as it should, what can be the issue here? Oct 26 10:24:07 glib datetime tests are always good at exposing differences between distributions Oct 26 10:24:29 the actual output would be nice in the regression report Oct 26 11:11:54 Is there someone around here who know more about bazel/tensorflow than me? https://pastebin.com/rzSSDAYp Oct 26 11:14:14 I already thought I routed it to my own tensorflow git repo where I fixed it, but it's ignored https://pastebin.com/P8x5AvLC Oct 26 13:08:44 Hello guys ! I need a tip ! I have three ARM64 platforms. I thought to create one yocto project to target the three different platforms, since they will differs more ore less only for devicetree and rootfs apps. In your opinion which is the better way to achieve that ? Oct 26 13:10:13 As far as I know I may define three different machine names, that are extension of a common machine (eg: zynqmp) and then use multiconfig to select the build of a specific target Oct 26 13:11:08 ilkappe: just making sure that you know of meta-arm :) Oct 26 13:11:40 rburton, thanks Oct 26 13:11:53 I'll look for it now Oct 26 13:13:01 Hi, while trying to use devtool (devtool modify virtual/kernel) I'm facing an error:ERROR: Execution of '/home/user/project/build/tmp/work/imx7d_pico-poky-linux-gnueabi/linux-fslc/5.4.71+gitAUTOINC+d4f3dcb65f-r0/devtooltmp-pl7y2yrg/temp/run.do_preconfigure.18190' failed with exit code Oct 26 13:13:01 2:/home/user/project/build/tmp/work/imx7d_pico-poky-linux-gnueabi/linux-fslc/5.4.71+gitAUTOINC+d4f3dcb65f-r0/devtooltmp-pl7y2yrg/temp/run.do_preconfigure.18190: 120: /home/user/project/build/tmp/work/imx7d_pico-poky-linux-gnueabi/linux-fslc/5.4.71+gitAUTOINC+d4f3dcb65f-r0/devtooltmp-pl7y2yrg/temp/run.do_preconfigure.18190: cannot open Oct 26 13:13:02 /home/user/project/build/tmp/work/imx7d_pico-poky-linux-gnueabi/linux-fslc/5.4.71+gitAUTOINC+d4f3dcb65f-r0/devtooltmp-pl7y2yrg/workdir/defconfig: No such fileWARNING: exit code 2 from a shell command.Would one of you be so kind to give me some pointers where to start resolving this? I'm on poky 3.1.3. Oct 26 13:16:27 hey rburton, I'm looking at meta-arm but it seems to me that it provides the recipes to build for an arm platform (which of course it is my case) Oct 26 13:17:06 the things, I already have the set of recipes to build the distro for the three platforms Oct 26 13:17:24 the thing is* Oct 26 13:17:51 but I don't want to have three separate projects Oct 26 13:19:11 ideally I want to build the three distros for the three platforms simply specify somewhere in someway which are the difference between the three of them and then launching bitbake Oct 26 13:19:34 but I don't know both if it is possible and how to do it Oct 26 13:27:17 Hi all! Oct 26 13:27:44 <xmp> I would have a question about a weird build issue for my recipe, but I am not quite sure if I am right in here? Oct 26 13:31:01 <xmp> Ok, so I'll just describe my issue. In case I am wrong here, please just tell me where to go. Oct 26 13:34:05 <xmp> So I got an error during populating a package to my SDK saying it cannot find a package e.g. "nativesdk-my-app". However, "my-app" builds just fine and so does the "nativesdk-my-app" target as standalone. The error arises during the populate_sdk task saying "Unable to find a match". The my-app recipe is pretty simple and los similar to all my other Oct 26 13:34:05 <xmp> recipes which build also fine. I don't know where to start searching. Oct 26 13:59:30 <moto-timo> FWIW: elago SD card USB adapter worked immediately on my Debian-10 machine vs uni reader which only worked on my Centos-8 machine Oct 26 14:02:02 <RobertBerger> @ilkappe: we have 3 things in Oct 26 14:02:32 <RobertBerger> @ilkappe yocto/oe which are orthogonal (independent of each other) Oct 26 14:02:49 <RobertBerger> @ilkappe: MACHINE, DISTRO, IMAGE Oct 26 14:03:48 <RobertBerger> @ilkappe: "they will differs more ore less only for devicetree and rootfs apps" Oct 26 14:04:10 <RobertBerger> @ilkappe: rootfs apps: is IMAGE in my imagination Oct 26 14:05:07 <clementp[m]> Hi, I have two boards very similar except the bootloader and the device-tree are different. Is it possible that yocto build only 1 Rootfs+Kernel and 2 bootloaders ? Oct 26 14:05:07 <RobertBerger> @ilkappe: you could build all device tree trees together and then install only one which is appropriate for the board Oct 26 14:06:31 <RobertBerger> @clementp[m]: I have one "layer collection" with all device trees and rootfs and a different layer collection for different bootloaders - independent Oct 26 14:07:13 <RobertBerger> @clementp[m]: actually I split the BSP in on layer for kernel/dtb and another layer for bootloader Oct 26 14:07:50 <clementp[m]> RobertBerger: is it closed source ? Oct 26 14:08:05 <RobertBerger> clementp[m]: of course NOT Oct 26 14:09:13 <clementp[m]> RobertBerger: could you point me the git repo just to check how you do it Oct 26 14:09:18 <RobertBerger> kernel: https://gitlab.com/meta-layers/meta-multi-v7-ml-bsp Oct 26 14:09:51 <RobertBerger> bootloader variation 1: https://gitlab.com/meta-layers/meta-u-boot-wic-bsp Oct 26 14:10:41 <RobertBerger> bootloader variation 2: https://gitlab.com/meta-layers/meta-u-boot-mender-bsp Oct 26 14:10:54 <RobertBerger> (this needs some updating) Oct 26 14:11:37 <clementp[m]> RobertBerger: awesome, thanks ! Oct 26 14:13:01 <RobertBerger> the special trick here is, that it's a multi-v7 (multi arch) kernel which runs on multiple boards Oct 26 14:13:14 <RobertBerger> Feel free to ping me Oct 26 14:14:33 <clementp[m]> RobertBerger: all my boards use the exactly same SoM, just the carrier board is different, so the device-tree Oct 26 14:14:48 <RobertBerger> So it's easy. Oct 26 14:15:00 <RobertBerger> You can build in one go different device trees and install the appropriate one. Oct 26 14:15:30 <RobertBerger> Just add to your machine config as many device trees as you like ;) Oct 26 14:21:24 <smurray> clementp[m]: you can add the extras to KERNEL_DEVICETREE, iirc Oct 26 14:21:47 <smurray> clementp[m]: generally done in the machine conf file Oct 26 14:23:34 <moto-timo> interesting that suse is buying rancher Oct 26 14:23:58 <moto-timo> https://www.suse.com/c/news/suse-acquires-rancher/ Oct 26 14:30:07 <clementp[m]> RobertBerger: smurray No because the default env of the bootloader need to be updated. fdt_file=boarda.dtb or ftd_file=boardb.dtb Oct 26 14:30:42 <smurray> clementp[m]: sure, but that's a different problem than getting the second dtb built Oct 26 14:30:59 <moto-timo> also, multiconfig can help with these different configurations... Oct 26 14:31:31 <clementp[m]> haaa you can do u-boot multiconfig ? Oct 26 14:31:33 <clementp[m]> didn't know about that Oct 26 14:31:56 <moto-timo> multiconfig is the sense of YP Oct 26 14:31:59 <smurray> moto-timo: not required for this, actually Oct 26 14:32:15 <smurray> clementp[m]: you can build more than one u-boot, UBOOT_MACHINE allows for that Oct 26 14:33:02 <smurray> clementp[m]: you can then have 2 different configs with the environments set differently Oct 26 14:33:31 <moto-timo> smurray: I'm not saying required... but I'm saying it might make sane build environment Oct 26 14:33:32 <smurray> clementp[m]: the other approach is to build one u-boot, but use scripting to determine which type of board it is Oct 26 14:33:43 <moto-timo> https://docs.yoctoproject.org/singleindex.html#building-images-for-multiple-targets-using-multiple-configurations Oct 26 14:34:04 <smurray> moto-timo: for just getting a second u-boot built, it's way overkill imo Oct 26 14:34:13 <moto-timo> anyway, I haven't touched dtb and uboot in a couple years so I should be silent Oct 26 14:41:31 <RobertBerger> @clementp[m]: bootloader and device tree are board specific, the rest does not need to be. Oct 26 14:44:01 <smurray> clementp[m]: actually, let me qualify that, I was thinking the logic around UBOOT_MACHINE/UBOOT_CONFIG may allow building more than one, but I might be wrong Oct 26 14:46:10 <tlwoerner> smurray: yes that's true. some platforms have multiple _defconfigs (e.g. one for emmc booting and one for sdcard booting) Oct 26 14:46:45 * moto-timo forces reboot with fat fingers on USB front port... sigh Oct 26 14:46:55 <smurray> tlwoerner: cool, good to know my pre-coffee brain isn't completely off ;) Oct 26 14:47:18 <moto-timo> smurray: I'm on 3rd espresso. how can you have not had coffee? Oct 26 14:47:39 <moto-timo> smurray: or was that a time-machine response ;) Oct 26 14:48:03 <smurray> moto-timo: heh, I try to keep to a couple of cups a day and just haven't gotten to the kitchen yet to make the first Oct 26 14:48:18 <smurray> moto-timo: too many tabs to keep track of Oct 26 14:48:31 <moto-timo> smurray: I just "solved" that problem (facepalm) Oct 26 14:54:57 <moto-timo> interesting.. trivy Oct 26 14:55:10 <moto-timo> https://github.com/aquasecurity/trivy Oct 26 14:58:20 <ThomasD13> moto-timo, i feel your pain ;) Oct 26 15:01:11 <armpit> is anyone on yocto slack? Oct 26 15:01:30 <armpit> there appears to be several yocto slack channels Oct 26 15:01:56 <moto-timo> armpit: we're mostly in the watercooler changnel... same slack at ELC Oct 26 15:02:03 * moto-timo still can't type Oct 26 15:02:25 <kergoth> there's a yocto slack? Oct 26 15:02:27 <armpit> the booth redirects to slack and closes the event one Oct 26 15:02:27 * kergoth yawns Oct 26 15:02:33 <kergoth> ah, elc Oct 26 15:04:00 <moto-timo> kergoth: sadly pay to play Oct 26 15:04:08 <moto-timo> kergoth: although much cheaper than normal Oct 26 15:04:09 <kergoth> ha Oct 26 15:04:14 * armpit stating to dislike these virtual events Oct 26 15:04:39 <moto-timo> The troll dinner last night was absolutely terrible Oct 26 15:04:42 <kergoth> half the point of a conference is the mingling and networking, watching a bunch of videos isn't quite the same thing Oct 26 15:06:49 <moto-timo> kergoth: more than half... and the dinners and hallway track and F2F OED{A,E}M Oct 26 15:06:58 * kergoth nods Oct 26 15:07:07 <moto-timo> kergoth: and yet you and I have never managed to be in the same physical time and space Oct 26 15:07:12 <carlsb3rg> how would I go about running chmod u+rwx /home/someuser after /home/someuser has been created? Oct 26 15:07:50 <kergoth> yeah, that's my fault, for a couple years i avoided travel when my son was super young, then just kept missing the registration deadlines :) Oct 26 15:08:12 <moto-timo> kergoth: I also had several years without sufficient income to fund travel to EU Oct 26 15:08:33 <carlsb3rg> I tried to add a task after useradd_base.bbclass' perform_useradd but /home/someuser isn't created yet Oct 26 15:09:24 <tlwoerner> kergoth: i'd say it's more like 100% of a conference is the mingling etc Oct 26 15:10:02 <tlwoerner> if we ever get back to in-person conferences we should just skip all the "talks/papers" nonsense and just stick to mingling (lol) Oct 26 15:10:31 <kergoth> hah. just a big gym or set of hallways with people b/s'ing for a few days. i'm on board Oct 26 15:11:45 * moto-timo still can't explain why I never attended FOSSDEM Oct 26 15:11:45 <tlwoerner> lol Oct 26 15:12:18 <tlwoerner> moto-timo: because you keep spelling it incorrectly? Oct 26 15:12:26 <tlwoerner> (lol) Oct 26 15:13:36 <LetoThe2nd> hehehe Oct 26 15:13:52 <moto-timo> tlwoerner: step one... ? Oct 26 15:14:06 <moto-timo> lol Oct 26 15:14:41 <moto-timo> my grammar and spelling has completely atrophied because of social media... Oct 26 15:14:53 <moto-timo> and vocabulary Oct 26 15:15:32 <tlwoerner> and getting up at 4am with creme stout handy? Oct 26 15:16:21 <moto-timo> tlwoerner: until it is evening somewhere, I'll stick to espresso with home roasted beans Oct 26 15:17:01 <kergoth> you roast yourself? nice. i keep meaning to try that. thankfully we have a local roaster / coffee shop in walking distance here Oct 26 15:17:29 <moto-timo> kergoth: Dave J Cobbley the younger Oct 26 15:17:38 <moto-timo> kergoth: my partner in home brew crimes Oct 26 15:18:06 <moto-timo> kergoth: but green coffee beans are quite reasonably priced Oct 26 15:20:53 <sakoman> moto-timo: be careful, I started with home roasting about 15 years ago and ended up growing coffee in Hawaii Oct 26 15:21:07 <kergoth> haha Oct 26 15:21:16 <moto-timo> sakoman: also involved MOVING to and BUYING property in Hawaii Oct 26 15:21:19 <kergoth> is that supposed to be a warning or enticement? Oct 26 15:21:32 <moto-timo> sakoman: not a trivial amount of "be careful" Oct 26 15:21:47 <moto-timo> sakoman: although I have a finca in Costa Rica that COULD be growing coffee again... Oct 26 15:21:52 <sakoman> kergoth: take it whichever way you like ;-) Oct 26 15:22:00 <moto-timo> travel when? yeah... not this year Oct 26 15:22:15 <kergoth> I've wanted to move to hawaii ever since I got married there as a destination wedding, but sadly our family would never be able to afford to visit :) Oct 26 15:23:36 <sakoman> moto-timo: I marvel at how cheap green coffee is Oct 26 15:24:03 <sakoman> It is a lot of work! Oct 26 15:24:34 <tlwoerner> kergoth: and that's a problem… how? (lol) Oct 26 15:25:17 <moto-timo> sakoman: after a tour of the co-operative that processed the beans from our finca (when it was still growing beans)... I was SHOCKED Oct 26 15:25:37 <kergoth> tlwoerner: haha Oct 26 15:25:41 <moto-timo> also macadamia nuts are incredibly hard to process Oct 26 15:25:59 <moto-timo> by the time you crack the shell, you have made macadamia nut butter Oct 26 15:27:01 <sakoman> moto-timo: green coffee tease: https://photos.app.goo.gl/s2qWp1efY2aLHinj8 Oct 26 15:27:09 <tlwoerner> kergoth: one of the many benefits of having our farm is the ability to say "sorry, we can't visit for <event>, somebody's got to feed the horses!" (lol) Oct 26 15:27:22 <moto-timo> little known fact: piles of macadamia branches will attrack rats which will attract fer-de-lance, the most poisonous snake in central amaerica Oct 26 15:27:26 <kergoth> hah, can't argue with that excuse Oct 26 15:27:41 <moto-timo> lol @ tlwoerner Oct 26 15:28:28 <sakoman> tlwoerner: works both ways too - when they visit hand them a pitchfork and have them help clean the stalls :-) Oct 26 15:28:53 <moto-timo> sakoman: it's amazing how few revisits from he annoying relatives you get with that simple technique Oct 26 15:29:00 <moto-timo> ^the Oct 26 15:29:05 <sakoman> moto-timo: re: mac nuts, like everything you need the right tool to do the job Oct 26 15:29:39 <moto-timo> sakoman: indeed... we had very old non-producing trees and the market in CR had already dried up by then... Oct 26 15:29:43 <sakoman> I let the local pros do the work and just but shelled mac nuts Oct 26 15:29:43 <tlwoerner> sakoman: yep! lol. i never turn down extra hands for mucking or processing wood Oct 26 15:30:18 <moto-timo> sakoman: and the harvest of coffee pretty much was a wash with paying our caretaker (plus the fertilizers and so on) Oct 26 15:30:34 <moto-timo> sakoman: not a loss, but not a retirement plan by a long shot Oct 26 15:30:53 <tlwoerner> had a "city friend" come help pile wood once. at one point he says "by the way, what do you do with the manure?" i said "what do you think you're standing on?!" Oct 26 15:31:06 <sakoman> :-) Oct 26 15:31:32 * moto-timo points them at the manure spreader they are about to drive around the pasture before lunch Oct 26 15:32:00 <moto-timo> very important that it be BEFORE lunch Oct 26 15:32:07 <sakoman> moto-timo: I can't even imagine what I would have to charge if I wanted to sell my coffee and make a living at it Oct 26 15:32:25 * tlwoerner doesn't share the tractor duties, too much fun (lol) Oct 26 15:32:29 <moto-timo> sakoman: makes that consulting fee look cheap ;) Oct 26 15:32:45 <moto-timo> smart @ tlwoerner Oct 26 15:33:15 <sakoman> But every time I have a cup I'm happy to do the work, it's worth it Oct 26 15:33:57 <RP> sakoman: CVE stats look better this week, thanks! Oct 26 15:34:53 <sakoman> RP: yes, and master now looks correct too :-) Oct 26 15:35:44 <sakoman> I'll keep spending one morning a week on CVE's till I get to diminishing returns Oct 26 15:39:08 * moto-timo keeps meaning to attack a couple CVEs a week Oct 26 15:41:09 <tlwoerner> moto-timo: https://twitter.com/meghamohan/status/1319988640534155265/photo/1 Oct 26 15:42:37 <RP> sakoman: yes, I wonder what the issue was Oct 26 15:43:03 <RP> sakoman: you managed about 50 a week so shouldn't be long, right? :) Oct 26 15:43:28 <caiortp> Hi All, is is possible to register to yocto dev summit withtou ELCE /OSS registration? Oct 26 15:43:49 <tlwoerner> dreyna: ndec: ^^ Oct 26 15:43:51 <sakoman> RP: I changed the time of day that I run the report, moving it away from midnight Oct 26 15:44:02 <paulbarker> caiortp: It should be possible yes Oct 26 15:44:06 <ndec> caiortp: yes! Oct 26 15:44:10 <sakoman> And I also now use separate sstate for master/dunfell Oct 26 15:44:26 <RP> sakoman: right, so we don't know exactly what it was now! :) Oct 26 15:44:31 <caiortp> @nde Oct 26 15:44:35 <sakoman> Let's see if those two changes eliminate some of the strangeness we've seen from time to time Oct 26 15:45:21 <ndec> caiortp: https://events.linuxfoundation.org/yocto-project-summit-europe/register/ Oct 26 15:46:02 <caiortp> ndec, great thanks a lot! Oct 26 15:46:09 <moto-timo> tlwoerner: guilty Oct 26 15:46:19 <sakoman> RP: We may never know which of the two did the trick, but if things work going forward I'm ok with that :-) Oct 26 15:46:44 <RobertBerger> @RP: BTW it looks like with master some pseudo/golang? permissions were fixed. Still testing, but so far it looks good. Oct 26 15:46:50 <RP> sakoman: true, I just like to understand problems so I stand a better chance they don't recur! Oct 26 15:48:22 <sakoman> I suspect data base update time, since when I reran the test later in the day last week master looked correct Oct 26 15:48:55 <RobertBerger> #RP: I didn't even dare to mention those ;) But bitbake -c cleanall <some golang recipes> didn't work before. Needed to have sudoer permissions to remove build artifacts. Oct 26 15:51:54 <caiortp> ndec my entry point was https://www.yoctoproject.org/yocto-projectr-summit-virtual-europe-2020/ and the link send to the the event page, tks for the link Oct 26 16:01:50 <moto-timo> RP: nice shirt :) Oct 26 16:02:30 <moto-timo> RP: I also like the red accept stripe on the walls Oct 26 16:04:36 <moto-timo> ^accent Oct 26 16:10:45 * tlwoerner is glad RP is "yocto compatible" :-D Oct 26 16:13:00 <moto-timo> we should get "built by bitbake" shirts or something like that Oct 26 16:13:37 <moto-timo> RP: nice to "see you" even if only on a video chat Oct 26 16:13:46 <armpit> and take over that "We built this city" song too Oct 26 16:16:10 * LetoThe2nd notes, that it was the barcelona shirt. 8yrs ago. Oct 26 16:16:34 * moto-timo notes that was one of many conferences I wish I had attended and most remember me attending Oct 26 16:16:47 <moto-timo> nearly everybody thinks I was at Dublin Oct 26 16:16:49 <moto-timo> lol Oct 26 16:17:11 * RP hopes he's YP Compatible Oct 26 16:21:22 <clementp[m]> smurray: looks like UBOOT_MACHINE allow only one machine Oct 26 16:22:33 <tlwoerner> clementp[m]: yes, but in the source code there can sometimes be more than one to choose from, for a specific board/target Oct 26 16:22:54 <clementp[m]> @RobertBerger I agree with you that the Kernel RootFS doesn't need to be board specific, but I don't know what the easiest way to build 2 bootloaders and 1 kernel. Maybe multiple layer is a bit overkill Oct 26 16:22:56 <tlwoerner> clementp[m]: or you could write your own _defconfig and include it as a patch in SRC_URI then select it from UBOOT_MACHINE ;-) Oct 26 16:24:07 <clementp[m]> tlwoerner: still I need to build 2 times Yocto, so this will generate 2 bootloaders and 2 Kernel/Rootfs Oct 26 16:25:01 <clementp[m]> Either I use 2 layers, and I build 3 times Yocto, one time for each bootloader and one time for the Kernel/Rootfs like RobertBerger suggest Oct 26 16:28:25 <clementp[m]> Ha ok tlwoerner got it Oct 26 16:32:26 <smurray> clementp[m]: do your boards have some way to tell which variant it is, like a gpio pin or pins? If so, then a little bit of scripting in u-boot could be used to pick which dtb file to use Oct 26 16:33:10 <clementp[m]> smurray I thought about that but no I can't :( Oct 26 16:33:45 <clementp[m]> Or I could write the information into the efuse and then read them again Oct 26 16:33:55 <clementp[m]> But still not a good idea I think Oct 26 16:37:44 <smurray> clementp[m]: so one option is multiconfig as moto-timo mentioned earlier, though I'd probably just do the extra u-boot build in the second multiconfig. Oct 26 16:40:42 <moto-timo> smurray: that's loosely what my half-baked suggestion was Oct 26 16:41:02 * moto-timo inserts magic >here< Oct 26 16:43:37 <smurray> moto-timo: I actually have a potential usecase in AGL to build multiple u-boots for use with the same kernel + image, when I looked last week my initial glance suggested the u-boot config stuff would allow it, I'm a little bummed out if it'll take using multiconfig, esp since the deployed u-boot files all have the board names in them Oct 26 16:43:54 <clementp[m]> UBOOT_CONFIG[xx] one for board_A and one for board_b Oct 26 16:44:03 <clementp[m]> first time I call yocto to build board_A + rootfs/kernel Oct 26 16:44:13 <clementp[m]> second time I call only to build Board_B u-boot Oct 26 16:47:11 <moto-timo> and you should be able to make second build mcdepends on output of first build Oct 26 16:49:45 <carlsb3rg> I'm trying to add a custom file in /etc/sudoers.d/ but get this message "file /etc/sudoers.d conflicts between attempted installs of custom-script-1.0-r0.i586 and sudo-1.8.31-r0.i586" - I realize what the problem is I guess, but unsure about the cure Oct 26 16:49:59 <moto-timo> most of my multiconfig usage currently is building host + guest OSes for virtualization (and containers) Oct 26 16:52:14 <smurray> moto-timo: it's not clear to me if the second config needs a mcdepends if it's essentially the same except for the UBOOT_CONFIG Oct 26 16:54:08 <kergoth> carlsb3rg: you might need your FILES to list the files *under* sudoers.d, but not list sudoers.d itself? then only one package owns the directory Oct 26 16:55:26 <moto-timo> smurray: yeah... I haven't really thought it through enough...sorry for bringing up vapor ware Oct 26 16:55:52 <carlsb3rg> I only have FILES_${PN} += "${sysconfdir}/sudoers.d/sudo_the_user" not the actual sudoers.d directory Oct 26 16:55:59 <moto-timo> smurray: I've been using it a lot over the past year though for many interesting solutions Oct 26 16:57:24 <smurray> clementp[m]: so the meta-renesas BSP definitely builds more than one u-boot by listing them in UBOOT_CONFIG, I'm trying to work out if that's via some modification they've made to the u-boot recipe or if it should work in general Oct 26 17:01:51 <smurray> clementp[m]: okay, I think I was correct in that you can do it, but wrong about the variable. You can specify more than one config in UBOOT_CONFIG, I see it in other BSPs Oct 26 17:02:17 <smurray> clementp[m]: so you can define UBOOT_CONFIG[foo] and UBOOT_CONFIG[bar], and have UBOOT_CONFIG = "foo bar" to build both Oct 26 17:02:24 <smurray> clementp[m]: I'd suggest trying that Oct 26 17:19:47 <carlsb3rg> oh...to add a file to /etc/sudoers.d/ you have to make sure that you create /etc/sudoers.d/ with the same permissions as the sudo package... Oct 26 17:38:05 <d32> Hi everyone, I'm unsure if my question I've posted earlier today made it onto the chat. So trying not to sound impatient I just wanted to check if maybe one of you has picked it up and has a clue for an answer. Thanks :) Oct 26 17:38:14 <rburton> khem: oooh scanview integration is neat (catching up on sessions I missed) Oct 26 17:40:04 <zeddii> wth is a scanview Oct 26 17:40:07 <khem> rburton: yeah its quite useful too :) thanks for watching session Oct 26 17:40:54 <khem> we found few bugs in libcurl Oct 26 17:43:23 <khem> zeddii: scan-view is a static analyser that comes with clang for free :) Oct 26 17:47:34 <zeddii> ahah Oct 26 17:53:06 <rburton> khem: how tragic is a scanview over all of poky? Oct 26 17:53:41 <khem> rburton: never run it on whole image Oct 26 17:53:48 <rburton> that bad eh Oct 26 17:54:36 <khem> selectively on important libs Oct 26 17:55:35 <rburton> sounds like you're wimping out of fixing the warnings to me Oct 26 17:55:39 <rburton> are you chicken? Oct 26 18:02:00 <khem> :) Oct 26 18:03:26 <manuel1985> What's the relation between the <build-dir>/conf/local.conf and recipes, distro conf and machine conf files _from the technical POV_? It seems I can set any variable I can set just about ANYWHERE also in <build-dir>/conf/local.conf. Is <build-dir>/conf/local.conf prepended to everything else? Oct 26 18:03:40 <moto-timo> rburton: public shaming FTW Oct 26 18:04:04 <rburton> manuel1985: bitbake.conf will show you the parse order Oct 26 18:04:26 <rburton> everything global is parsed, then the data store cloned for each recipe which then overlays its own values Oct 26 18:07:23 <manuel1985> rburton: I see, thank you Oct 26 18:07:27 <ad__> hi after a recent upgrade to dunfell, seems imx6 vivante xorg driver is missing. if i try to add f86-video-imx-vivante i get : xf86-video-imx-vivante was skipped: incompatible with machine Oct 26 18:08:50 <rburton> ad__: http://git.yoctoproject.org/cgit/cgit.cgi/meta-freescale/tree/recipes-graphics/xorg-driver/xf86-video-imx-vivante_6.4.0.p0.0.bb?h=master#n65 <-- presumably you're not using one of those machines Oct 26 18:09:05 <rburton> if your have a custom machine that does support it then use a bbappend to change COMPATIBLE_MACHINE Oct 26 18:10:20 <ad__> rburton, well in my custom machine i have "MACHINEOVERRIDES =. "mx6:mx6q:"" Oct 26 18:10:31 <khem> rburton: clang-tidy is neat too Oct 26 18:10:43 <rburton> ad__: COMPATIBLE_MACHINE doesn't care about overrides, it's literal string matching on MACHINE Oct 26 18:10:54 <rburton> so just add a bbappend Oct 26 18:11:28 <ad__> rburton, thanks a lot. doing that Oct 26 18:24:56 <ad__> rburton, mm in my bbappend i added COMPATIBLE_MACHINE = '(imx6q-mymachine)' but still getting same error Oct 26 18:25:15 <rburton> try just unsetting it :) Oct 26 18:33:29 <khem> rburton: where are we with optee-test issue ? Oct 26 18:33:39 <khem> http://errors.yoctoproject.org/Errors/Build/111424/ Oct 26 18:34:08 <khem> it works ok with clang btw. so thanks for that :) Oct 26 18:36:40 <ad__> rburton, : ok there was more recipes with same name involved :) Oct 26 18:37:13 <ad__> thanks a lot for super support :=) Oct 26 18:38:12 <khem> RP: master-next needs something like http://sprunge.us/WZlrme or you gonna see lot of red on AB Oct 26 18:40:52 <RP> khem: thanks, not got around to testing that yet! Oct 26 18:41:56 <RP> khem: squashed in Oct 26 19:11:19 <rburton> khem: we're at goddamnit optee stage Oct 26 19:17:10 <khem> this is only package showing red in my CI Oct 26 19:22:40 <RP> agaikova: FWIW I added a fix for the valgrind musl issue to gatesgarth-next as we need the fix urgently for 3.2 rc2 Oct 26 19:26:42 <fenrig> Hi how do I use python3-native from openembbed? I've added the RDEPENDS="python3-pyopenssl-native" but when I execute the recipe that uses python3 it complains about pyopenssl missing. So I assume it uses the python of the host Oct 26 19:28:51 <RP> fenrig: did you inherit python3native ? Oct 26 19:29:11 <fenrig> no Oct 26 19:29:13 <fenrig> so just Oct 26 19:29:21 <fenrig> inherit python3-native ? Oct 26 19:30:04 <fenrig> oh ok I found the bbclass Oct 26 19:32:21 <fenrig> @RP: thx dude Oct 26 19:32:42 <fenrig> RP: it's working like it should now Oct 26 19:51:18 <manuel1985> Which tools do you usually use to checkout all the repositories, a typical yocto project consists of? I'm using googles repo tool, but dislike it. I'm looking for alternatives. Oct 26 19:52:24 <khem> manuel1985: git submodules Oct 26 19:54:09 <manuel1985> You have your build directory outside of the poky repo, don't you? Oct 26 19:55:05 <manuel1985> I would like to keep them inside poky, but that would need me to change the .gitignore of that repo. Oct 26 19:55:56 <kergoth> kas, repo, submodules, myrepos, etc. lots of options, but what works best for you depends on your usage and workflow. Oct 26 19:56:12 <kergoth> manuel1985: or just don't use poky directly. Oct 26 19:56:34 <kergoth> poky is basically bitbake+oe-core+meta-yocto Oct 26 19:56:41 <kergoth> so youc an put those wherever you like Oct 26 19:57:32 <khem> I think monorepo is the way to go usually Oct 26 19:57:47 <zeddii> cd layer; git pull Oct 26 19:57:50 <khem> whatever you use to set it us like kergoth said is own taste Oct 26 20:00:44 <manuel1985> kergoth: Thanks for the tools you mentioned, will look into them first. Oct 26 20:01:11 <khem> git subtrees is also an option Oct 26 20:01:13 <kergoth> there are lots of tools for a more monorepo type setup, the one poky uses, git-subtree, or then ones that don't relate to the scm at all, like peru Oct 26 20:01:15 <kergoth> yeah :) Oct 26 20:02:50 <kergoth> https://github.com/apenwarr/git-subtrac is quite interesting. it uses submodules, but actually stores the subrepo objects in the main repo, which means you can always clone and set it up independent of the state of actual upstream repositories Oct 26 20:05:02 <kergoth> oh, also https://github.com/ingydotnet/git-subrepo Oct 26 20:05:08 <kergoth> (haven't tried that one) Oct 26 20:23:02 <khem> yeah my method is clone all the repos to somewhere I can control and use git submods Oct 26 20:23:19 <khem> cuts the chase pretty early on both producer and consumer side Oct 26 20:23:53 <sgw> RP Morning! Is there a replacement for Distrodata bbclass that you removed about 2 years ago? Package manifest and build history don't quite provide the right info. Oct 26 20:24:14 <khem> some projects dont respect public branches or may rewrite transfer etc. so you need to offset that Oct 26 20:33:52 <RP> sgw: you can query directly with tinfoil Oct 26 20:34:06 <RP> sgw: I think we updated some of the selftests to show how to do that? Oct 26 20:36:49 <sgw> RP: Ok, that means a separate script then right Oct 26 20:38:10 <RP> sgw: yes, but you'd need one anyway Oct 26 20:51:40 <rburton> RP: are we waiting on a fix from bluelightning for the buildhistory thing, or is that revert in next hitting master? Oct 26 21:01:00 <RP> rburton: revert hitting master. Its not a simple fix Oct 26 21:08:31 <bluelightning> right, I have to redo it, it can wait Oct 26 22:28:03 <RP> JPEW: around? Oct 26 22:28:45 <RP> https://autobuilder.yocto.io/pub/repro-fail/oe-reproducible-20201022-tf6b08oj/packages/diff-html/ from https://git.openembedded.org/openembedded-core-contrib/commit/?h=stable/dunfell-nut&id=e555c1e3d77d2cf2e758626444a0af500e178b6b which I think could be explained and fixed with something like http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/t222&id=a124b0305fd5f679bf2b0a16c10e94e8405748db Oct 26 22:28:58 <JPEW> Ya, I'm here Oct 26 22:29:25 <RP> JPEW: sakoman is running into it in dunfell, I think we'll have the same issue in master. hashequiv isn't accounting for timestamps leading to misleading matches? Oct 26 22:30:33 <RP> JPEW: I'm concerned that I reverted that in my local test queue so there was some issue, I just can't remember what Oct 26 22:32:19 <JPEW> Hmm, I'm not sure. It' the timestamps in internal the package files themselves, which seems more like a problem with the packaging tools? Oct 26 22:33:09 <JPEW> Oh, maybe the internal files are different because of the timestamps of the actual files Oct 26 22:33:11 <sakoman> The timestamps in the package files match the new upstream tarball Oct 26 22:33:37 <sakoman> master should have the same fetch issue, since both master and dunfell are at netbase 6.1 Oct 26 22:34:04 <JPEW> Ah, it thinks they are equivalent because it doesn't know about the timestamps? Oct 26 22:34:09 <JPEW> is the theory? Oct 26 22:34:30 <sakoman> Yes, I think that is RP's theory Oct 26 22:35:25 <sakoman> The new upstream tarball has the same version number, same contents (modulo changelog) but different timestamps on the identical files Oct 26 22:36:31 <sakoman> and of course a new tarball filename Oct 26 22:37:27 <RP> JPEW: hashequiv is saying they're the same, reproducible builds is saying they're not Oct 26 22:38:15 <RP> sakoman: I think I remember what the patch needs. I think we care about timestamps in some tasks but not others where we only care about the content Oct 26 22:38:23 <JPEW> sakoman, RP: Ya, that seems reasonable Oct 26 22:38:38 <RP> my patch applies for all tasks whereas it only probably should apply to some subset Oct 26 22:39:20 <RP> I suspect I never worked out the subset Oct 26 22:39:52 <JPEW> Anything that is encoded in the sstate needs to be in the hash caculation Oct 26 22:41:13 <RP> at the very least it needs to do this for do_package Oct 26 22:41:35 <JPEW> ya Oct 26 22:43:42 <RP> JPEW: I'm too tired/fried to think straight about that right now... Oct 26 22:44:03 <JPEW> Ya, fair. I need to hang up today too. I'll try to look tomorrow Oct 26 22:44:11 * RP is slightly concerned he looked at that in January Oct 26 22:44:47 <RP> JPEW: thanks, I just wondered if a second pair of eyes might see (or remember) something I don't Oct 26 22:45:16 <JPEW> The theory sounds correct at least Oct 26 22:47:33 <sakoman> Thanks RP, JPEW! Oct 26 22:49:43 <RP> JPEW: thanks. Will think some more tomorrow too **** ENDING LOGGING AT Tue Oct 27 02:59:57 2020