**** BEGIN LOGGING AT Tue Nov 17 02:59:56 2020 Nov 17 07:19:30 good morning! Nov 17 07:36:16 good morning Nov 17 08:39:04 do you recommend me to use systemd in embedded systems? Nov 17 08:45:12 Hi all, Im stuck with a python exception raised while parsing a recipe created using devtool+recipetool Nov 17 08:46:08 it seems like recipes variables are not getting expanded Nov 17 08:46:57 Im getting a "RecursionError: maximum recursion depth exceeded while calling a Python object" Nov 17 08:49:30 kamel_b: "show us teh codez"s :) Nov 17 08:49:37 wyre: it depends. Nov 17 08:49:57 LetoThe2nd: yep here it is: https://pastebin.com/8m1igAxP Nov 17 08:50:31 actually a very basic recipe Nov 17 08:50:55 LetoThe2nd, what it does depend on? Nov 17 08:51:30 kamel_b: /me has no idea about the npm fetchers, sorry. might also depend on the version in use. Nov 17 08:51:36 yann: maybe you ? ^^^^ Nov 17 08:51:51 wyre: hum, on your usecase and your requirements? Nov 17 08:52:00 wyre: hum, on your usecase and your requirements? Nov 17 08:52:02 no I don't either, sry Nov 17 08:52:09 yann: k Nov 17 08:52:53 wyre: essentially, the question is like "does it make sense to use a car for transportation from a to b?". it depends. where is a? where is b? what do you want to transport? other constraints? same applies here. Nov 17 08:55:04 LetoThe2nd: yann: here is the log issue Im getting: https://pastebin.com/GBbUqEgc Nov 17 08:56:37 I couldn't figure out what does this recursion issue means Nov 17 09:00:01 kamel_b: does it work for another machine? Nov 17 09:00:56 kamel_b: the log suggests some strange behaviour concerning the cortexa9 tune. so the first thing i personally would try building the recipe for a trivial machine, like qemux86 or qemuarm Nov 17 09:02:51 LetoThe2nd: no I didn't, e it is con Nov 17 09:03:29 it is confusing as the defaulttune is set by the machine to armv7 Nov 17 09:04:11 kamel_b: then you have some pointers where to start digging. but cortex-a9 is a subtyoe of armv7a, so thats probably correct. Nov 17 09:22:21 LetoThe2nd: https://pastebin.com/zFeA1zrA Nov 17 09:22:34 same result for qemuarm Nov 17 09:24:21 kamel_b: then look at other recipes that use the same fetcher, read its documentation... maybe the autogenerated SRC_URI is wrong? i don't know. Nov 17 09:25:53 LetoThe2nd: the most weird thing here is that I have above 9/10 recipes generated with the same process devtool+recipetool and only 3 or 4 are behaving like this... Nov 17 09:26:42 LetoThe2nd: yes maybe i'll check that Nov 17 09:32:52 could I use a higher python version? Nov 17 09:33:27 or could I establish the python version in the build process? Nov 17 09:34:42 wyre: bumping python is usually quite problematic. if you need something thats not just a bump in the dot-dot range, then its usually easier to upgrade as a whole. Nov 17 09:36:38 what do you mean with "dot-dot range"? Nov 17 09:37:11 versions are usually like x.y.u Nov 17 09:37:29 erm, x.y.z - so changing something in "z" is dot-dot Nov 17 09:37:34 y is dot Nov 17 09:37:53 oh, I see Nov 17 09:38:06 well, I would like to upgrade to python 3.6 Nov 17 09:38:12 "from" Nov 17 09:38:29 you are lacking a lot of context, just liek for the systemd question earlier too. Nov 17 09:38:31 because the current version is 3.5.2 Nov 17 09:38:49 and f-strings are just supported since 3.6.x version Nov 17 09:39:09 then I'd say I'm in the dot-dot range, right? Nov 17 09:39:51 read and think again. Nov 17 09:41:26 LetoThe2nd, I think I don't get your point when you say "if I need something that's not just a bump in the dot-dot range" Nov 17 09:42:00 and when you say "to upgrade as a whole" you mean upgrading the whole yocto version? Nov 17 09:43:20 i wrote: versions are x.y.z. upgrading z is dot-dot, upgrading y is dot, upgrading x is... upgrading. so if you want to go from 3.5.something to 3.6.something, then it is?.... Nov 17 09:44:49 LetoThe2nd, ok, it's a dot upgrade Nov 17 09:45:08 which is usually problematic, you mean? Nov 17 09:45:51 wyre: even more so as yocto/Oe went straight from 3.5 to 3.7, and that even... let me calculate.... 21 months ago? Nov 17 09:46:03 https://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/meta/recipes-devtools/python?id=e2c3247c233876ab090c9ce3d5325a6d46ab350f Nov 17 09:46:22 wyre: which indicates that you are on something totally outdated and out of support anyways. Nov 17 09:46:27 so current version of yocto has python 3.7? Nov 17 09:46:56 *sighs* Nov 17 09:47:29 please do at least a little homework yourself before abusing as as git/google frontend: http://layers.openembedded.org/layerindex/branch/master/layers/ Nov 17 09:47:40 *abusing us Nov 17 09:48:09 example: http://layers.openembedded.org/layerindex/branch/master/recipes/?q=python3 Nov 17 09:48:46 wyre: https://layers.openembedded.org/layerindex/branch/master/layer/meta-python/ Nov 17 09:51:13 sorry LetoThe2nd I'm new in yocto and I'm trying to understand the whole environment Nov 17 09:54:12 wyre: granted. but things that can easily be checked online with a few keystrokes - please do it yourself. Nov 17 09:55:10 wyre: so in a nutshell: whoever gave you the current setup, its time to complain to that person anyways because it is definitely outdated and doesn't receive bugfixes / security related patches. Nov 17 09:55:31 wyre: and while you're at it,demand a proper introduction/training. Nov 17 09:57:41 what do you recommend me for a proper introduction/training? Nov 17 09:58:14 wyre: the low-balling version is to watch and work along https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj Nov 17 09:58:18 wyre: there is some youtube tutorials on the Yocto Project channel you could have a look at Nov 17 09:58:19 this setup is part of the BSP from the manufaturer Nov 17 09:59:00 wyre: to get strarted properly, book a training by one of the usual yocto-savvy trainers - depends a bit on your whereabouts and preferences, though. Nov 17 09:59:06 wyre: Bootlin has some very wide but "shallow" introduction training to Yocto Project (CC-BY-SA IIRC, you can find it on their website) Nov 17 09:59:15 wyre: then its tme to complain to the manufacturer. Nov 17 09:59:21 and there are some people on this chan that are offering trainings as well Nov 17 09:59:56 like a certain jester.. Nov 17 10:00:08 finding a trainer should not be a problem once you've outlined your language/timing preference. or maybe even the hw manufacturer offers some. Nov 17 10:01:35 and what about reading resources? the main reference manual is enough? Nov 17 10:02:05 wyre: its a reference, not a training/tutorial. i discourage it. Nov 17 10:03:35 wyre: in the german region, maybe get in touch with robert berger. italian, its mckoan. french, bootlin is a good choice. english, probably robert berger, chris simmons, or the linux foundation themselves. (apologies to all i didn't mention now, those were the most prominent ones on my mind) Nov 17 10:04:26 LetoThe2nd, what about the Yocto Project Quick Start? Nov 17 10:05:35 wyre: to return to the earlier analogy: now you're standing at the front door of the driving school and ask "isn't it enough to read the cars manual? um... or maybe the short version of road legislation?" Nov 17 10:06:22 wyre: if you want to properly drive, then knowing both of those suffficiently well is needed. but neither will make up for proper, hands-on training/experience. Nov 17 10:06:48 LetoThe2nd: Bootlin mainly gives their training in English BTW :) Nov 17 10:07:22 wyre: you can either make up with a lot of time, trial and error (student way) - or just cough up the coins and get started properly (way of those who do this for a living) Nov 17 10:07:40 qschulz: about everybody also does english, i know. Nov 17 10:08:37 * LetoThe2nd is out for some time. earning coins, y'know. Nov 17 10:11:48 Offtopic: Some of LetoThe2nd's messages are italic in my irc client (Thunderbird). What irc feature is this? Nov 17 10:13:40 dleppich: IIRC, you can surround words with * or ** or _? Nov 17 10:13:57 Thanks Nov 17 10:14:08 *one star*, **to star**, _underscore_ ? Nov 17 10:14:14 well nope :p Nov 17 10:14:39 hmm, no italic ^^ Nov 17 10:19:50 LetoThe2nd: wyre we usually provide YP training in Italian and sometimes in English Nov 17 10:23:23 wyre: you are using MicroGea aren't you? Nov 17 10:48:52 mckoan, yes I am Nov 17 10:49:09 imx6ull, in fact Nov 17 10:51:48 I was wondering if could I build the current version (3.2, I think) for this hw Nov 17 11:03:07 wyre: IIRC the manufacturer provides a quite old version of the BSP/YP Nov 17 11:04:20 mckoan, that's because I was thinking to build a most recent version Nov 17 11:04:56 but I think I have to take into account the main script provided by the manufacturer (engicam-setup-environment) Nov 17 11:05:39 wyre: no, it's not so straightforward Nov 17 12:21:12 is there anyone who maintains the meta-arago layer here? Nov 17 12:25:01 * LetoThe2nd summons denix Nov 17 12:26:46 I did send an email regarding a change in `SRC_URI` in a `tensorflow-lite` but it bounced back to me Nov 17 12:40:55 wyre: this whole setup from Engicam is so outdated, that it would be really difficult to build 3.2 for it. they do not have any upstream support provided for their layer, only uprevving `pyro` :O Nov 17 12:42:03 zandrey, what do you mean with "uprevving `pyro`"? Nov 17 12:42:58 wyre: they did some patches on the `pyro` branch in their (what appears to be) public meta layer Nov 17 12:43:12 latest one dated 16. Oct Nov 17 12:45:29 i personally would not commit on updating this layer, since the step from `pyro` to `dunfell` is not a single step, but rather a leap. Nov 17 12:49:23 actually, looking at the github of engicam - there is an impression that they are working on towards bringing something, which resembles later yocto versions. Nov 17 12:50:12 unfortunately, all their latest forks are targeting NXP vendor BSP... latest being 5.4.47-2.2.0. Nov 17 12:52:04 wyre: i guess you might talk to them as your vendors to figure out plans regarding their upcomming updates (if they plan some). but beware: you most probably would receive a modified version of NXP BSP, which is not encouraged to be used in production and serves as rather Demo. Nov 17 13:09:33 Hey folks, I was wondering what repo management tool you use? the repo-tool by google? What alternatives are there? Nov 17 13:11:29 emrius: many people use kas from what I heard Nov 17 13:11:45 we use git with our home made scripts to fetch layers to given revisions Nov 17 13:14:16 qschulz: Thanks! kas looks interesting. Yeah, I also had a bash script for that as well. So, I'm guessing I wasn't completely off there :) Nov 17 13:18:39 RP: Just reading now, no problem. I'll rebase this morning and send a fixup as required. Nov 17 13:19:46 RP: is this in master-next ? I just pulled poky master and don't see any updates. Nov 17 13:20:49 never mind. one of my boxes got no updates, the other did. Nov 17 13:20:55 I'll figure out what's up with that later. Nov 17 13:21:09 zeddii: thanks. I just missed the v2, sorry Nov 17 13:21:26 meh. if stones were thrown for that, I'd be dead long ago. Nov 17 13:24:43 emrius: git submodules works too. Nov 17 13:29:25 mcfrisk: Thanks, that's a good idea. and it's lightweight Nov 17 13:49:46 emrius: kas++ from my side Nov 17 13:50:43 (also did a session on it) Nov 17 14:07:20 RP: fixup sent. Nov 17 14:08:14 and I see why I thought my box wasn't updating, my commit of the same shortlog was happily rebasing, since it is in a completely different part of the class .. and I just thought the original hadn't made master yet :D Nov 17 14:08:39 zeddii: thanks Nov 17 14:21:42 LetThe2nd: Great! Going to check that out right away. Nov 17 14:35:51 LetoThe2nd: You said you made a session on kas. Is there a youtube video for that session? I coulnd't find it based on the video names. Nov 17 14:36:05 https://www.youtube.com/watch?v=KJHJlOtTdaE Nov 17 14:36:20 ty Nov 17 14:36:25 dleppich: #12 in the playlist Nov 17 14:36:42 too bad the "kas" was hidden behind dots :D Nov 17 14:37:37 ... finally looks like kas can help me get rid of my cancerous yocto project folder yam I have ended up with... Nov 17 14:38:03 jam=yam Nov 17 14:40:06 it almost certainly can, yes. Nov 17 15:00:13 hello Nov 17 15:02:04 Hey, my name is Zygmunt; I'm working on the open harmony project. Hello everyone :) Nov 17 15:02:27 howdy Nov 17 15:09:59 after Randy's ELCE talk on Rust, I started learning Rust. I don't want to let the meta layer "rust", if you know what I mean! Nov 17 15:11:52 sorry had to make the pun because it fits well with the meta layer terminology Nov 17 15:11:58 my main thing with rust is that cargo is totally badly named. it should have been "oxide" or "oxidate" Nov 17 15:12:47 better yet, corrosion! Nov 17 15:13:04 that sounds pretty destrucive. Nov 17 15:14:03 I am so going to enjoy the live coding session if you do a rust example Josef ! Nov 17 15:14:20 * LetoThe2nd won't do one in the foreseeable future Nov 17 15:15:00 Maybe we should ask Randy to do a guest live coding session Nov 17 15:16:49 paulbarker already did videos on that, just not live Nov 17 15:18:01 shan1: https://www.youtube.com/watch?v=N5UPFo4JiTs Nov 17 15:18:32 (y) Nov 17 15:18:33 Complete with a bunch of figuring things out as I go Nov 17 15:18:54 I'll be following up, got another video ready to go I just need to write the description and publish it Nov 17 15:19:49 i need to prepare sample code for next week :( Nov 17 15:28:34 how can i append to a variable previously set with ?= (BLAB ?= "thing") Nov 17 15:30:01 Should be BLAB_append = "foo" or BLAB += "foo", when I'm not totally wrong Nov 17 15:31:35 tepperson: a ?= needs _append, AFAIK. the += would replace it if i'm not mistaken. Nov 17 15:32:13 ok. if its possible to make the whole dumpsterfire of 2020 more dumpsterfirey, then this is it: https://youtu.be/zu3k2PJumfI Nov 17 15:32:25 I think I read something that the += works differently in local.conf / recipes. Not quite sure Nov 17 15:32:58 dleppich: find some inspiration in session #15 :) Nov 17 15:33:02 all of these are set in recipes Nov 17 15:33:19 dleppich: the point of recipe vs. .conf is visibility. Nov 17 15:33:31 the point of += vs. _append is evaluation stage. Nov 17 15:33:58 LetoThe2nd: Thanks for the clarification. Hmm.. Much to learn I still have Nov 17 15:34:35 dleppich: easiest is to just tinker a bit around with such to get the feel. just like i did in that session. Nov 17 15:34:42 LetoThe2nd: you're incorrect about += :) Nov 17 15:35:03 qschulz: happens. please correct hard and fast! Nov 17 15:35:07 FOO ?= "a" followed by FOO += "b" results in FOO = "a b" Nov 17 15:35:14 Does it needs a _space_ ? BLAB_append = "_space_thing" Nov 17 15:35:25 FOO += "b" followed by FOO ?= "a" will result in FOO = "b" Nov 17 15:35:30 It's also very difficult to override _append (in a bbappends or other downstream consumer) Nov 17 15:35:37 faba_: _append needs a space, = doesn't Nov 17 15:35:57 I'm seeing behavior where =+ and += override the variable instead of appending Nov 17 15:35:59 _append is appending to a space delimited list, so you must provide the space Nov 17 15:36:13 tepperson: depends on the ordering of the parsing Nov 17 15:36:30 otherwise we would almost have only _append in code Nov 17 15:36:32 i hereby propose this to be the soundtrack for the next time we meet in person. somebody shall pick us up at the airport and blast it all along: https://youtu.be/-A7BPXEBh1Q Nov 17 15:37:57 tepperson: could you share your code and what you're expecting? Nov 17 15:41:20 qschulz: i can throw something together in a bit Nov 17 15:41:30 tepperson: also.. are you using overrides? Nov 17 15:41:47 FOO_machine += "something" won't probably do what you think it is Nov 17 15:41:58 s/is/does Nov 17 15:42:34 qschulz: what does that do then? Nov 17 15:43:48 for a raspberry pi, KERNEL_DEVICETREE ?= " \ Nov 17 15:44:34 I want to do KERNEL_DEVICETREE_raspberrypi4-64 += "overlays/blabla.dtbo" in a recipe Nov 17 15:45:05 tepperson: 1. why in a recipe? Nov 17 15:45:19 2. you want KERNEL_DEVICETREE_append_raspberrypi4-64 Nov 17 15:45:42 FOO_machine will replace FOO when building for machine Nov 17 15:45:49 += does not matter in that case Nov 17 15:46:06 because it applies after FOO_machine is resolved not after FOO Nov 17 15:46:07 how i love it. i can just talk rubbish, confuse people,a nd get away with it while qschulz does the hard part of cleaning up my mess and actually explains stuff! Nov 17 15:46:24 * LetoThe2nd hi5es qschulz Nov 17 15:46:32 lol Nov 17 15:46:40 * qschulz low5ses LetoThe2nd Nov 17 15:46:53 * LetoThe2nd fistbumps qschulz Nov 17 15:47:11 * qschulz dodges LetoThe2nd fist fight attempt Nov 17 15:47:24 * moto-timo makes popcorn Nov 17 15:47:35 * LetoThe2nd stops fighting and proceeds to post-fight drinking instantly. Nov 17 15:47:47 qschulz: i want a particular overlay only active for a particular image recipe Nov 17 15:48:07 tepperson: join qschulz for the chant, then. Nov 17 15:48:16 (he's really good at it!) Nov 17 15:48:29 tepperson: repeat after me (cue the music LetoThe2nd ) Nov 17 15:48:47 * LetoThe2nd drumroll.s Nov 17 15:48:53 tepperson: recipe data is local (to the recipe), conf data is global Nov 17 15:49:03 or do you want me to actually find you some music? Nov 17 15:49:05 LetoThe2nd: too late, you ruined it Nov 17 15:49:13 i always do that. Nov 17 15:50:35 tepperson: since KERNEL_DEVICETREE is used in the kernel recipe, you can't modify it from anywhere else than a configuration file or the kernel recipe itself Nov 17 15:51:22 tepperson: so two choices, either create a new machine for it, or a new distro Nov 17 15:51:52 the rationale behind this is that adding an overlay is meaning that you actually have different HW Nov 17 15:52:01 s/is meaning/means Nov 17 15:52:11 differrent HW requires different machine Nov 17 15:52:36 qschulz: would recommended method mean creating a new (layer)/conf/machine/blablab.conf ? Nov 17 15:53:00 qschulz: please say "it depends (TM)" Nov 17 15:54:07 tepperson: new machine conf file requires a new (layer)/conf/machine/blablab.conf, a new distro conf file requires a new (layer)/conf/distro/blablab.conf Nov 17 15:54:35 tepperson: but it might be overkill, what is your overlay representing? are you sure you can't just add it all the time and not care about it? Nov 17 15:55:19 LetoThe2nd: not sure about the "it depends" part? Nov 17 15:55:38 qschulz: it represents some i2c and spi devices Nov 17 15:55:56 qschulz: i'm pretty sure that the recommended way of solving his problem depends on the specific circumstances/requirements. Nov 17 15:57:08 with kas you can have a local "meta-custom" and carry quite a bit of stuff in a light weight layer Nov 17 15:57:29 but LetoThe2nd is right... "it depends (TM)" Nov 17 15:57:29 should be meta-kustom Nov 17 15:57:41 zeddii: it should be Nov 17 15:59:48 YPTM: Jan-Simon joined Nov 17 16:00:04 zeddii: ++ Nov 17 16:00:06 can you do weird things like name the layer meta-(y) up Nov 17 16:00:14 YPTM: I'll be joining a few mins late Nov 17 16:00:16 (actualy smiley) Nov 17 16:00:27 tepperson: if the devices aren't here, the device drivers won't probe so it's fine to have them all the time? except if you have conflicts between devices or if part of your product is having the same hw but disabling some feature depending on the "image" (~artifact) you want to build.. in which case it's probably more a distro you want? Nov 17 16:00:41 so all in all... "It depends (TM)" Nov 17 16:01:44 tepperson: vendors already do weird stuff, don't give them ideas :) Nov 17 16:01:49 qschulz: ok i'l need to look into the distro side as I am not familiar with that yet. Nov 17 16:02:36 tepperson: i take it you have already watched the youtube playlist a couple of times? it goes best with sufficiently large number of beers, i'm being told. Nov 17 16:02:45 note that distro is the biggest change you can make in Yocto. This means that having different distros mean enormous build times Nov 17 16:03:08 (more or less as much time as building from scratch your current distro) Nov 17 16:13:10 LetoThe2nd: Im now have the confirmation that there is an issue with npmsw fetcher that inhibit correct variable expansion "sometimes" Nov 17 16:14:01 kamel_b: i see. at least report a bug for it, even better prepare a patch :) Nov 17 16:14:15 I still don't know why (I tried increasing the recusion limit) and if it could be related to some tasks order Nov 17 16:14:43 LetoThe2nd: yep I'll stick on it for sure Nov 17 17:03:28 I missed the yocto meeting. I hope y'all behaved. Nov 17 17:18:23 word is they deployed some mall cops in your absence... Nov 17 17:18:38 my doorbell just rang, maybe they are here for me. Nov 17 17:39:20 zeddii: we assigned all the qemu CVEs to you Nov 17 17:40:05 sweet! Nov 17 17:55:19 we should strong consider mapping clenall to cleanall, would save me a lot of typos. Nov 17 17:55:57 or cleenall Nov 17 17:56:06 +! Nov 17 17:56:10 +1 hah. Nov 17 17:56:17 how ironic I typo'd +1 Nov 17 17:59:06 0: libvirt-6.3.0-r0 do_cleansstate - 3m57s (pid 896555) Nov 17 17:59:11 that's a long clenall ! Nov 17 18:00:51 cleansstate is for chumps Nov 17 18:00:59 0: libvirt-6.3.0-r0 do_cleansstate - 5m47s (pid 896555) Nov 17 18:01:06 still going!! Nov 17 18:01:11 libvirt and I are long time enemies. Nov 17 18:03:23 0: libvirt-6.3.0-r0 do_cleansstate - 8m17s (pid 896555) Nov 17 18:03:25 wheee Nov 17 18:03:30 finally finished. Nov 17 18:17:23 I wonder if packagegroups should have EXCLUDE_FROM_WORLD set by default Nov 17 18:42:39 EXCLUDE_FROM_WORLD += "covid-19" ? Nov 17 19:02:31 Hello. I am seeing "The following required tools (as specified by HOSTTOOLS) appear to be unavailable in PATH, please install them in order to proceed: cpp g++ gcc rpcgen" but these tools are all in the PATH. :-/ Nov 17 20:52:29 moto-timo: Let me know if there's anything interesting about Tekton at KubeCon... I've been playing around with it and it's pretty cool Nov 17 20:56:37 Hello.. I am getting lots of these errors. " Exception: FileNotFoundError: [Errno 2] No such file or directory: './deploy-source-date-epoch/deploy-source-date-epoch/__source_date_epoch.txt' " . The build is happening in side Docker container on a macbook Nov 17 20:56:42 any idea Nov 17 21:25:57 kiwi_29: What version are you using (e.g. master, dunfell, etc.)? Nov 17 21:32:03 JPEW dunfell Nov 17 21:37:01 zeddii: cleansstate iterates the sstate directories. If they're large... Nov 17 21:37:08 its really not a fast thing Nov 17 21:40:01 kiwi_29: Can you post the full backtrace somewhere? Nov 17 21:46:00 JPEG --> https://pastebin.com/S0rKFgnW Nov 17 21:46:11 JPEW (sorry for autocorrect) Nov 17 21:51:12 blerg our IT blocks pastebin.... Nov 17 22:04:12 let me try something else Nov 17 22:04:33 kiwi_29: I was able to see it Nov 17 22:06:19 ok Nov 17 22:06:27 kiwi_29: That's... intetesting Nov 17 22:12:02 kiwi_29: It looks like the file is being removed and that's racing with the deploy? Nov 17 22:14:51 looks like...but I saw there was a talk about this behaviour before Nov 17 22:15:20 https://www.mail-archive.com/openembedded-core@lists.openembedded.org/msg128249.html Nov 17 22:16:47 kiwi_29: I think that's a different issue that was already fixed Nov 17 22:16:54 I see Nov 17 22:17:43 kiwi_29: Can you reproduce it easily? Maybe try running `bitbake -DDD ...` Nov 17 22:18:27 kiwi_29: It really does look like some weird race though.... Nov 17 22:21:29 yes..it happens everytime Nov 17 22:21:38 I will do -DDD and send pastebin Nov 17 22:26:42 JPEW https://pastebin.com/5wtQsVHG with -DDD Nov 17 22:28:47 I created a case-sensitive APFS volume on my macbook and am sharing this volume to docker. The output of build goes to this volume Nov 17 22:29:22 kiwi_29: Hmm Nov 17 22:29:28 because yocto builds were not happening on case-insensitive APFS volume Nov 17 22:29:59 I think there should be a log.do_deploy_source_date_epoch in the failed recipes log directory, can you post that? Nov 17 22:37:43 JPEW https://pastebin.com/GYbPWMS2 Nov 17 22:47:14 kiwi_29: That appears to be a different error? Nov 17 22:47:28 possibly related though.... Nov 17 22:47:50 Let me check other failing recipes Nov 17 22:53:36 JPEW this logs looks different.. https://pastebin.com/vqQqNqbv Nov 17 22:59:36 zeddii: hmm: https://autobuilder.yoctoproject.org/typhoon/#/builders/15/builds/2996 Nov 17 23:12:58 kiwi_29: Ya, weird Nov 17 23:14:14 kiwi_29: It really looks like the file is being deleted Nov 17 23:14:31 hmm Nov 17 23:15:18 Well, it's weird because the directory enumeration (e.g. ls, readdir()) is reporting the file is there, but when a process goes to open the file it is "not" there Nov 17 23:16:02 This happens twice; once in the sstate_hardcode_path function (where it is not fatal) and once again in sstate_report_unihash (where it is fatal) Nov 17 23:16:28 The fact that it is happening the same way in two functions that execute sequentially makes me think it is not a race condition Nov 17 23:17:32 kiwi_29: I'm pretty sure this is some weird quirk of APFS.... if you can figure out what it is, that might be useful Nov 17 23:20:58 I see... yes.. I have to solve it ..so will report what I find Nov 17 23:25:00 Hi everyone. Quick question. When using devtool for the kernel or u-boot, does running 'bitbake -c menuconfig linux-yocto' perform the menuconfig in the devtool workspace? Nov 17 23:26:18 JPEW: will do Nov 17 23:51:08 Okay I think I answered my own question. Looks like it will create the .config in the normal workspace and then symlink to it in the devtool workspace. **** ENDING LOGGING AT Wed Nov 18 03:02:03 2020