**** BEGIN LOGGING AT Tue May 25 03:01:36 2021 May 25 04:19:23 theee song for my night: https://yewtu.be/watch?v=yJL5SE1i0u4 May 25 08:33:02 good YP-Summit's morning May 25 08:40:22 so... if I change layer priorities bitbake won't rebake unless I do_clean the package in question? May 25 08:57:49 x0n^: yes May 25 09:10:50 FYI people are asking for hands-on class credentials in Slack channel #training-room-summit-2021 May 25 09:27:02 mckoan: shit. how do I force it? I'v re built the entire recipy in question and forced a do_rootfs but the result still looks exactly like the failed first attempt May 25 09:29:22 all: Hi there. Do you have any experience with eSDK which does not "populate" all the sources from SRC_URI when doing `devtool modify linux-yocto`? If I run this command inside classic build environment, then it builds. Same thing for bitbake. However when I install the eSDK generated for my image, then I can not compile kernel, because of missing files. Any tips? May 25 09:32:46 x0n^: bitbake -C rootfs May 25 09:32:53 C uppercase May 25 09:36:37 mckoan: that apparent repacks rootfs. but ain't that the same as what I did by "bitbake -f -c do_rootfs "? May 25 09:55:14 rburton: https://autobuilder.yoctoproject.org/typhoon/#/builders/113/builds/1036/steps/12/logs/stdio May 25 09:55:36 yeah jonmason did a boo boo May 25 09:55:43 i'll quickly fix up May 25 09:56:04 curious how it got through CI! May 25 10:01:27 hooray checksum for noticing a licensing change \o/ May 25 10:15:12 RP: curiously, that exact same recipe worked in our CI May 25 10:16:28 NOTE: recipe trusted-firmware-a-2.5-r0: task do_populate_lic: Succeeded May 25 10:32:27 ok bug fixed May 25 10:40:07 I could really use some help if someone got the time. can't figure this out: https://termbin.com/3hi3 May 25 10:40:39 build command is: bitbake xen-image-minimal May 25 10:41:11 RP: i can't blame jonmason: he used the kas/CI I wrote to verify the upgrade, and it had a bug that meant it was ignoring some warnings May 25 10:43:08 relevant recipe: https://termbin.com/3hi3 May 25 10:44:51 oh, right. symptoms: u-boot not flashed onto image, cmdline.txt missing xen options May 25 10:45:33 I feel like I'm in a madhouse and bitbake got ALLL the mirros :/ May 25 10:49:55 this is what it looks like instead: https://termbin.com/rndo May 25 10:50:30 * x0n^ goes pray or something. appreciate any pointers! May 25 11:10:41 rburton: heh, bug in the CI is annoying :) May 25 11:54:51 Hello, I've got a raspberry pi yocto system that uses rauc (with uboot) for updating ext4 rootfses, however I'd like to also have two kernels, ideally on those updated rootfses. The kernel uses dt overlays, which reside on FAT /boot partition (as the kernel itself currently). My reasoning is that all those overlays should also be on rootfses to enable updating them in the future, am I correct? Could you give me some keywords that could be useful to May 25 11:54:54 research more? What about devicetree overlays which reside on /boot/overlays, and are loaded using raspberrys config.txt? May 25 12:29:03 Has anyone received their confirmation email for slack? Been waiting for ~10min, nothing in the spam/trash/inbox folders May 25 12:37:18 rburton: you figure out why it works for us and not RP? May 25 12:39:21 if anyone got karma to burn today, please check the backlog for my desperate cry for help. still praying for someone else too look at it because if I do it's just making my head spin at this point May 25 12:42:36 ok, the email to my work mail address took 15min and on my personal one, 1min don't know what happened, can only blame the bad mail provider :) May 25 12:46:46 jonmason: yeah, i broke it May 25 12:51:13 fix pushed to all branches so the CI will be busy for a bit May 25 13:10:32 whenenver I'm done with my fabulous home network project, I'm gonna check out joyent's triton May 25 13:10:56 am still somewhat partial to solaris-based OSs May 25 13:11:10 shame that omnios went under May 25 14:01:47 oops, please ignore. I do not endorse joyent(tm) triton(r) products any more than the next guy May 25 14:04:17 on the other hand, if someone were be so kind and checked my earlier problem with layer/recipe precence (see https://termbin.com/3hi3), you'd make me seriously happy. I'd even stop stalking about joyent(tm) smartos(cc-by-sa) for a bit May 25 14:09:11 that link 404s here May 25 14:09:18 there's a Yocto Project conference today and tomorrow so expect much longer than usual delays in support May 25 15:10:55 smurray: sorry. I'll repost. thanks qschulz. is there a webcast? May 25 15:12:57 @dash1tfvck: the link points to some configurations May 25 15:13:22 @dash1tfvck: what exactly is the problem? what do you try to do? May 25 15:17:38 RobertBerger: I've reorganized my thoughts: May 25 15:17:49 problem: trying to build rpi-xen image with "bitbake xen-image-minimal" May 25 15:17:49 relevant recipe: https://termbin.com/3hi3 May 25 15:17:49 related recipes: https://termbin.com/3hi3 May 25 15:17:49 symptoms: u-boot not flashed onto image, cmdline.txt missing xen options May 25 15:17:49 this is the output as of now:: https://termbin.com/rndo May 25 15:21:02 also, the first link should be https://termbin.com/jdei May 25 15:37:56 Hi RP, Summit seems to be going well, I have a multilib-32 question, should the site-config file from an SDK with 64-bit and Multilib-32 have the same sizeof values, is this correct? May 25 15:42:12 sgw1: that does sound strange and probably not right May 25 15:43:19 Seems that it's been that way for a long while now, but no real bugs filed, which means no one has really noticed maybe? May 25 15:43:36 sgw1: I suspect it isn't heavily used May 25 15:54:24 oh, the summit is attendance-only, no recording? May 25 15:57:45 its being recorded and will be on youtube May 25 15:58:00 nice May 25 15:58:22 reading thor May 25 15:58:34 s/thor/through some 2020 talks now May 25 15:59:58 when a project has hardening docs that's usually a good sign IMO :) May 25 16:00:20 I still hate bitbake's guts May 25 16:18:35 dash1tfvck: I'm sure bitbake loves you too :) May 25 16:32:53 aren't you migrating to libera? 🤔 May 25 17:35:33 this link appears not to be working now. is there a different one for the afternoon sessions? https://urldefense.com/v3/__https://www.cvent.com/api/email/dispatch/v1/click/l5vw8rrw7lc74k/yk5kyb5l/aHR0cHMlM0ElMkYlMkZsaW5hcm8tb3JnLnpvb20udXMlMkZqJTJGOTQwODc5MTM5ODQlM0Zwd2QlM0RNekZNTlVScU5tVjJlbkJzWldrelREWndVVUUzZHowOSZYMkhHZzNyb0RLQll5UTgyJTJCdk5LYVdiclhuMGtxNzdJWWxMTklMQ2clMkJnVSUzRCZQcmVzZW50YXRpb24rUm9vbQ__;!!JvSy May 25 17:35:33 GAA!WX9S_C6wb25qS6XzpRw9LeDhgZ7lAz48mrGP4u_PjnHEm43RUXa5gEsy-YAQqgA$ May 25 18:01:46 yates: presentation room or training room? May 25 18:12:59 JPEW: https://gitlab.apertis.org/pkg/maynard This is a desktop shell client for Weston based on GTK. May 25 18:13:15 Got it. Thanks May 25 18:26:54 kanavin: specifically with the ptest images, I think we should remove graphics and switch that to minimal May 25 18:27:51 RP: right, I will look into that May 25 18:28:08 kanavin: I'm sure I did work on this a while back somewhere May 25 18:29:40 kanavin: if you go down the maynard path, I'm interested in collaboration May 25 18:31:56 moto-timo, I guess you could start by quickly throwing together a recipe and image, and boot that in qemu and play with the UI? May 25 18:33:09 moto-timo, I think we could adopt a hybrid approach, and use core-image-weston for the bulk of the testing, but also provide core-image-maynard for the eye candy May 25 18:33:57 kanavin: fair enough May 25 19:07:14 dl9pf, are you sending emails to the layer owners or the stable branch maintainers or both? May 25 19:09:47 armpit: whatever email is in the layerindex for the respective layer's maintainer . May 25 19:10:07 armpit: I assume this may or may not match with owner/stable branch maintainers ;) May 25 19:11:25 e.g. https://layers.openembedded.org/layerindex/branch/master/layer/meta-oe/ lists khem May 25 19:12:08 Is #yocto moving to irc.libera.chat? May 25 19:12:16 aka even https://layers.openembedded.org/layerindex/branch/dunfell/layer/meta-oe/ lists khem May 25 19:12:39 I can't know better at this stage (from the layerindex data) May 25 19:12:51 armpit: you raise a good point May 25 19:13:39 It comes down to the original design of the tool and layer maintainers self-reporting to some extent May 25 19:18:07 e.g. maintainer might be in the “layer” ORM and not the “branch” ORM, so having stable branch maintainer(s) might require a database migration… May 25 19:18:40 Also, stale data is just a reality May 25 19:20:44 dl9pf, khem will just pass stable issues to me in the end May 25 20:08:21 v2d: I don't think any decision has been made yet May 25 20:09:07 JPEW: Okay, thank you. May 25 20:40:23 during todays talk by Tom King on yocto summit, he showed a glimpse of their CI/CD setup - do you folks have any hints about setting up Jenkins for yocto CI? May 25 20:40:52 e.g. example docker builders, Jenkinsfiles etc May 25 20:43:37 nevermind, I see that tomorrow at 6:45 PM UTC is talk about "Gitlab CI for dummies" :) May 25 21:33:06 wow. this is a gold mine: https://pretalx.com/yocto-project-summit-2020/talk/W3YWW7/ May 25 21:44:34 @rburton May 25 21:44:58 rburton: Unfortunately, maynard only works with the AGL compositor, not weston :( May 25 22:03:56 JPEW: :( May 25 22:43:42 zeddii: https://errors.yoctoproject.org/Errors/Details/585528/ seems 5.12 needs some changes for selftest any ideas ? May 25 22:45:08 khem: 5.12 isn't staying, it was just to see if the AB intermittent hangs persisted May 25 23:07:24 RP: ok May 25 23:15:43 JPEW: really? That surprises me, I'd have expected them to keep plain wayland client compatibility and not require agl-shell-desktop May 25 23:16:34 smurray: https://gitlab.apertis.org/pkg/maynard/-/blob/apertis/v2022dev2/client/maynard.c#L775 May 25 23:21:35 JPEW: yeah, looks like one of the earlier commits straight out flipped it to depending on the agl-shell protocols. In theory that'd not be too bad as a dependency, as it only requires weston. Probably less interesting to the general community, though May 25 23:21:46 khem: is that setuptools upgrade patch still problematic for meta-oe? May 25 23:22:05 smurray: Ya too bad :( May 25 23:22:07 RP: no, I fixed the package May 25 23:23:16 khem: cool thanks. I'll stop blocking that patch then May 25 23:23:18 khem: I do try to help if I can :) May 25 23:23:34 RP: https://git.openembedded.org/meta-openembedded/commit/?id=728e5b08e20f26e077d61f2857c31b20ff8162f4 May 25 23:23:42 JPEW: it's a very significant fork from the old version, I suspect outside of whoever is keen to use it in Apertis there'll be no community May 25 23:24:11 RP: master-next barring the kernel upgrade issue is all good in meta-oe CI May 25 23:24:21 khem: I just pushed a new version :) May 25 23:24:56 ok cool onto next cycle then May 25 23:25:05 zeddii: btw, 5.12 has qemuppc kernel module issue and defconfig warnings. Just for reference May 25 23:25:11 haversine patch got merged upstream too . nice May 25 23:25:30 khem: I try and keep a gentle trickle of things flowing in rather than huge steps May 25 23:27:08 yeah thats good. Actually gives some breathing space for other layers too keep in sync with core May 25 23:32:56 RP: btw esdk question came up, can we ship 32bit host tools along for a 64bit SDK host May 25 23:33:27 issue is that its editing the ldso to point to uninative but the binary is 32bit so it does not load May 25 23:33:41 and uninative is 64bit May 25 23:33:56 so it is trying to use 64bit ldso May 25 23:35:26 second question came up, how can we avoid shipping source info in sstate, is there a single setting somewhere ? May 25 23:35:30 khem: basically, no, esdk struggles with 32 bit on 64 bit May 25 23:36:02 khem: "source info"? you mean in the hash siginfo? May 25 23:36:11 yeah thats what I thought, so perhaps we can get esdk to leave the binaries alone which are not 64bit May 25 23:36:37 khem: you can't do that as uninative wouldn't work and that means the sdk is distro specific May 25 23:36:39 RP: the sstate contains all objects and debug info and srcs May 25 23:37:14 khem: there is currently no filter option. You can set SSTATE_DIR to a different directory on a per recipe or class basis though May 25 23:37:59 at least I think it might work. May 25 23:38:14 in some cases no symbols are to be exported and in some cases symbols can be exported and in general cases normal sstate is ok May 25 23:39:03 So are you guys moving from Freenode -> Matrix like everyone else? May 25 23:39:04 so for third parties application SDK is reliable way then not esdk May 25 23:39:58 khem: we could do with better class support around this. You'd need some kind of "binary only" recipe to make sstate work in that context May 25 23:40:09 matthewcroughan_: I already have reserved #yoctoproject on matrix in case we ever decide to use it 🙂 currently its only me in there May 25 23:40:13 it can be done but doesn't exist today May 25 23:40:40 RP: yeah May 25 23:40:45 matthewcroughan_: many are going to libera or oftc rather than matrix May 25 23:40:56 which just makes the problem more complex May 25 23:41:39 Yeah it does. I see a lot moving to Matrix though. So that's quite nice. May 25 23:41:43 Matrix is a great thing. May 25 23:43:19 matthewcroughan_: we'd lose a load of our "old timer" people, maybe even me. Depends if we care about them or the newer tech more May 25 23:43:44 * RP doesn't have an opinion on matrix other than "not another protocol" May 25 23:45:38 Well, can always be bridged nicely. There is that. May 25 23:45:44 The bridging between Matrix/IRC is great. May 25 23:45:57 Yeah.. I like IRC a lot too. May 25 23:46:23 https://github.com/hifi/heisenbridge May 25 23:46:29 Sorry if this is offtopic :) May 25 23:47:02 matthewcroughan_: not off topic, just no good answer :/ May 25 23:47:45 JPEW: https://autobuilder.yoctoproject.org/typhoon/#/builders/89/builds/3482 mingw failure :( May 25 23:48:02 I think heisenbridge is a good enough answer. It would be acceptable and bring the two communities together. May 25 23:48:14 And not be annoying to those on IRC, which is the key bit. May 25 23:48:36 The bridge is stateless and requires no configuration. May 25 23:48:42 Ok, I'll try to take a look tomorrow May 25 23:52:46 or maybe matrix-appservice-irc... May 26 00:25:20 do you guys use deploy keys on github to clone private repository from bitbake? May 26 00:43:15 khem / RP: i'll look into both as I gear up for 5.13, since I'm doing all arches on it, with the thought it might be the release kernel for the fall. Those will be the same there, and I'll solve it for both versions at the same time. May 26 00:56:06 Thanks to my deploy key, I can `git clone git@github.com:company/repo.git` for the kas shell, but bitbake fails to fetch the sources. May 26 00:56:45 how should I format the SRC_URI for my private github repo + deploy key? May 26 01:50:49 I use matrix-appservice-irc May 26 02:03:08 I need whatever it is, to be something I can plug znc into. I haven't looked into it. May 26 02:03:20 I just connect to znc, znc talks to all my networks. May 26 02:03:23 * zeddii searches **** ENDING LOGGING AT Wed May 26 03:00:34 2021