**** BEGIN LOGGING AT Wed Dec 23 02:59:56 2020 Dec 23 05:58:01 I have often log lines like Dec 23 05:58:01 NOTE: recipe updatable-image-app-1.0-r0: task do_populate_sdk_ext: Started Dec 23 05:58:01 Bitbake still alive (5000s) Dec 23 05:58:01 in my nightly (AUTOREV) builds. Is there any chance to configure a watchdog for such a process? Dec 23 06:06:08 last lines in log.do_populate_sdk_ext are: Dec 23 06:06:09 DEBUG: Executing python function copy_buildsystem Dec 23 06:06:09 NOTE: Skipping layer /build/buildbot/sources/meta-security/meta-tpm, already handled Dec 23 06:06:09 NOTE: Generating sstate task list... Dec 23 06:20:11 sno: I see something like this too, you have to kill bitbake Dec 23 06:20:26 so I put a mac time on builds like 12 hrs and let it kill it Dec 23 06:21:28 Guest5532: I wish there would be something more automatic :) Dec 23 07:35:58 morning Dec 23 07:46:54 good morning Dec 23 08:23:14 I want to add y patches to yocto-kernel-cache... which mailing list should I post to? Dec 23 08:23:22 https://git.yoctoproject.org/cgit/cgit.cgi/yocto-kernel-cache/tree/00-README says, I should talk to linux-yocto@yoctoproject.org but the page seems outdated Dec 23 09:33:03 dv|2: probably https://lists.yoctoproject.org/g/linux-yocto? Dec 23 09:33:23 please send a patch for the README also if that is the correct mailing list :) Dec 23 09:40:49 I have a weird issue that the Esdk fails (sdk is built normaly), and it seems to be connected to INITRAMFS image. Aynone had issues like that? | error: | linux/linux-imx_4.19.35.bb:do_fetch) failed with exit code 'setscene whitelist' Dec 23 09:44:47 qschulz, It could be more convinient if I'd be able to do git push. Am I? Dec 23 09:45:45 dv|2: that's not how contribution to open-source projects works usually :) Dec 23 09:46:17 If it is really a long series of patches or one big patch, you could probably request write access to some -contrib git repo Dec 23 09:47:10 qschulz, On github it is very convinient: I can fork the project, do my patch and ask for review and merge... Dec 23 09:51:37 how much time does it usually take to process and show my e-mail to the group? Dec 23 09:54:24 dv|2: in some cases (very long patch series), you can probably push to a personal github repo and send a mail to give the URL of the repo with the branch. Up to the maintainers to accept or not Dec 23 09:55:05 dv|2: you need to register to the mailing list before sending mail IIRC and then it's just there, you can check in the archives here: https://lists.yoctoproject.org/g/linux-yocto/messages?start=12:2020:-60 Dec 23 09:56:00 give it tops an hour after you've been successfuly registered. If you don't get a mail to register (confirmation mail), the link does not work or you cannot send a patch series, just let us know here Dec 23 09:56:28 dv|2: worth a read probably: https://docs.yoctoproject.org/dev-manual/common-tasks.html#making-changes-to-the-yocto-project Dec 23 09:56:47 it gives you the whole process Dec 23 09:57:40 qschulz, OK. thanks. But I think if you setup the github-like workflow for the merges it will make life easer. BTW, I've already registered and posted. Still not appear in the messages list. Dec 23 09:59:59 dv|2: it's been discussed already, might come back on the table one day too (or maybe even worked on right now, can't talk in behalf of the official team) Dec 23 10:00:23 dv|2: what was the email address you used? Dec 23 10:00:31 dvorkin@tibbo.com Dec 23 10:00:31 (the mailing list's) Dec 23 10:00:46 oh, linux-yocto@yoctoproject.org Dec 23 10:00:50 yeah no Dec 23 10:01:04 you need to resubscribe IIRC: linux-yocto+subscribe@lists.yoctoproject.org Dec 23 10:01:11 linux-yocto@lists.yoctoproject.org is the new address Dec 23 10:01:26 we changed to a different backend a few months ago IIRC Dec 23 10:03:15 qschulz, sent the e-mail to linux-yocto+subscribe@lists.yoctoproject.org . No reply yet Dec 23 10:04:30 dv|2: or.... maybe: https://lists.yoctoproject.org/g/linux-yocto/join ? Dec 23 10:04:54 I did it long time ago so don't really remember how I did unfortunately :/ Dec 23 10:06:02 I already logged in. I see two lists: yocto and linux-yocto. Dec 23 10:06:54 ok, then send the mail now to the correct mailing list address and monitor the archives Dec 23 10:10:40 Hi guys... I use a custom class to set some of the variables in my recipes... Now I do bb.warn() in case I detect some condition in this class... but the problem is that this message from bb.warn() statement floods console output with like 20 statements Dec 23 10:10:58 is it possible to somehow ratelimit or print this statement only once? Dec 23 10:11:22 wooosaiii: The issue is your class will be parsed for every recipe which includes it Dec 23 10:11:45 Is bb.warn() in an anonymous python block in the class? Dec 23 10:13:03 paulbarker: well I have a class and a method... I call this method from recipe to set PV variable... Dec 23 10:14:06 wooosaiii: Ok. That may be a little trickier. You'll need to separate the checks you're doing and the bb.warn() call from the method to generate a string for PV Dec 23 10:14:25 The checks and the warning can be placed in an event handler which fires only once Dec 23 10:14:30 For an example see http://git.yoctoproject.org/cgit/cgit.cgi/meta-virtualization/tree/classes/sanity-meta-virt.bbclass Dec 23 10:14:48 paulbarker: thanks :) Dec 23 10:15:31 The "bb.event.SanityCheck" event may actually be the right one for you if what you're doing is checking validity of the configuration Dec 23 10:47:31 qschulz, doesnt see my e-mail in the group yet... Dec 23 10:50:15 did you check that you're registered with the email address with which you're sending the mail? did you triple check the mailing list address too? Dec 23 10:59:34 qschulz, probably git send-email -1 --to ... doesn't really work... How can I pack the patch into my thunderbird and send? Dec 23 11:01:55 dv|2: try git send-email with your own mail address (or a different one you own) first to check it works correctly (correct credentials, it does not end in spam folder or is caught by spam filter of your email provider (happened to me a few times already)) Dec 23 11:02:20 I don't use Thunderbird so I cannot help with that :/ Dec 23 11:04:03 qschulz, No, it doesn't send. Looks like I need to dump my patch, create e-mail and send it from my e-mail client. everything is not convinient... Dec 23 11:04:35 how can I easily do that? Dec 23 11:09:59 dhooo.. made it finally. Please, think about github-style merge requests... Dec 23 11:26:07 anybody has the experience about openbox vs matchbox-wm ? which one requires less resources? Dec 23 11:28:34 it's probably going to be hard to have people online for the next week or so with the holidays season. If you don't have answers to your questions, we're not actively avoiding you so please ask them again next week or even in January :) Dec 23 11:29:28 qschulz, NY is the best time for pushing your work to public :) Dec 23 13:41:23 RP: for how long is the current merge window open? Dec 23 13:49:45 tlwoerner: for what? M2? Dec 23 13:56:18 RP: ah, found it: https://wiki.yoctoproject.org/wiki/Yocto_3.3_Schedule Dec 23 14:21:43 i want to tweak the CFLAGS of just one package. TARGET_CLFAGS_append_pn-package = " -DDEBUG" should work? Dec 23 14:22:16 it kicks off building a whole bunch of packages instead of just Dec 23 14:22:25 (looks like all it's upstream dependencies) Dec 23 14:22:31 s/it's/its Dec 23 14:23:22 tlwoerner: dates are also in the weekly status report Dec 23 14:23:55 tlwoerner: that should work and only recipes depending on that package should rebuild Dec 23 14:24:23 tlwoerner: btw, I worry one of you patches might have resulted in https://autobuilder.yoctoproject.org/typhoon/#/builders/79/builds/1668 Dec 23 14:24:39 tlwoerner: I don't have "proof" of which patch caused this yet though, its something in -next Dec 23 14:27:12 RP: the package in question is psplash, which should be pretty far out on the dependency tree (i.e. towards the leaves) but the TARGET_CFLAGS_append-pn-psplash line above kicked off builds of gcc-runtime rpm-native glib-2.0… iow "root" packages Dec 23 14:42:37 tlwoerner: bitbake-diffsigs on the gcc-runtime package maybe to identify what triggered a rebuild? Dec 23 14:44:34 tlwoerner: It shouldn't, I don't understand it. diffsigs should say why as qschulz says Dec 23 14:53:47 hmm… diffsigs didn't bring up anything. and then it went away. i must have ended up making some other change Dec 23 19:25:14 Is gitlab.com constantly slow for everyone or is it just me? Can't remember that page ever being fast. Dec 23 20:33:44 Is someone using kas in the vendor provided docker image? Dec 23 20:53:39 RP: i take it you narrowed the problem down to 52a3a3fd4e105058781d23918550d3d552c6f83c ? Dec 23 20:54:13 RP: the problem is, there *is* a bug in that code which i fixed. if that caused a "bug" that might mean that some software was relying on the bad code :-( Dec 23 21:19:06 how do i go about re-creating a build from the AB? on my humble home machine? Dec 23 21:19:51 is there a relatively easy way to get the conf/local.conf conf/bblayers.conf and know the SHAs of all the layers used? Dec 23 22:00:41 tlwoerner: I don't honestly know if it is that, I reverted it for this run to see Dec 23 22:01:09 tlwoerner: the reproducer should be oe-selftest -r runtime_test.Postinst.test_postinst_rootfs_and_boot_sysvinit Dec 23 22:01:28 tlwoerner: and/or or-selftest -r runtime_test.Postinst.test_postinst_rootfs_and_boot_systemd Dec 23 22:01:47 tlwoerner: I have another build running but it hasn't gotten to these tests yet Dec 23 22:04:16 tlwoerner: I will admit I'm trying to avoid powering on build machines and debugging things so I'm a bit "lazy" Dec 23 22:13:18 https://hub.docker.com/u/crops What's the difference between crops/poky and crops/yocto? Dec 23 22:23:59 RP: no worries. i was reading the commit and associated bugzilla report that lead to the change that i "fixed", it feels like a bit of a rabbit hole, setting pi_dir to point to the wrong place was intentional :-S Dec 23 22:25:45 RP: if i understand it correctly, if pi_dir is wrong then the image's postinsts won't get run, only the package manager's Dec 23 22:25:55 tlwoerner: I have not looked into it at all Dec 23 22:26:55 in bash, one doesn't break out of a case statement with "break", so when i saw that i figured it was a mistake. when i then noticed that pi_dir was being left set to the wrong location, i was sure it was wrong Dec 23 22:27:12 but it turns out that was the "fix" for something else, lol Dec 23 22:27:54 it feels like something that could do with a little more TLC, but not today :-) Dec 23 22:35:46 tlwoerner: it may need some comments or something for sure... Dec 23 22:36:27 is there a way to create a fifo as part of a recipe? Dec 23 22:36:53 i.e. create a fifo in the image Dec 23 22:41:21 tlwoerner: that could be interesting. I have no idea if it would work Dec 23 22:41:57 okay, i'll keep poking at this one. i'd like to see a nice clean solution too Dec 23 22:42:24 (that works on sysvinit) especiallly since the first boot is a slow one, and feedback would be a good thing Dec 23 22:42:52 i like the tweak to print the task that's running above the progress bar, i think it's a nice touch Dec 23 22:43:22 i'm looking into adding that same behaviour into the systemd bootup as well, although i don't know that it'll work as well (being parallele) Dec 23 22:43:27 parallel Dec 23 23:48:48 RP: doing a mkfifo in a recipe doesn't work, it causes do_package_qa to hang (it's probably trying to read the file?) Dec 23 23:49:03 i.e. the file contents or something Dec 23 23:50:04 is there a way to tell do_package_qa to skip a file, or, ideally, it would skip fifos on its own :-) Dec 24 00:15:44 found it; i can make the fifo if i patch insane.bbclass **** ENDING LOGGING AT Thu Dec 24 03:00:14 2020