**** BEGIN LOGGING AT Tue Dec 17 02:59:58 2019 Dec 17 05:21:12 New news from stackoverflow: How Can I Read Video Using SAMA5D27-SOM1-EK1 kit Dec 17 06:23:21 I'm trying to make custom linux kernel configuration for amd64 where i spent 38 hours cherrypicking these based on googlefu, dmidecode, etc.. where i was told in ##linux to ask here so can someone help me with this? Dec 17 06:26:36 neofetch: http://ix.io/24Ez ;; dmidecode: http://ix.io/24EB Dec 17 06:26:46 config here http://ix.io/24Ey Dec 17 06:41:38 how do I start over completely with a specific recipe? I accidentally deleted couple files from build/tmp... with a bad wildcard and now I cannot get recipe to compile. Tried cleanall and force compile etc Dec 17 07:21:31 New news from stackoverflow: How to build hello world recipe in yocto **** BEGIN LOGGING AT Tue Dec 17 07:52:20 2019 Dec 17 08:12:05 what should I remove from tmp/ that bitbake would rebuild them and not complain that something is missing? Dec 17 08:12:34 I can remove whole build/ but i dont want that since rebuild would take half a day Dec 17 08:15:08 got this pdf, whats it mean really Dec 17 08:15:11 http://i.stanford.edu/pub/cstr/reports/csl/tr/76/111/CSL-TR-76-111.pdf Dec 17 08:15:35 A Distributed Algorithm for Dec 17 08:15:35 --r’l, Constructing Minimal Spanning Trees Dec 17 08:15:36 in Computer-Communication Networks Dec 17 08:15:56 Mr0rris0: hum, how come you think we do explain CS theory? Dec 17 08:16:34 cant say, just not sure if any if it applies to what i glanced over this bein about Dec 17 08:17:21 if not then thats something, i mean i dunno, i guess it informs me of something Dec 17 08:17:34 Mr0rris0: without atually looking at it, spanning trees in networks are usually about routing. Dec 17 08:19:26 stuom1: you can remove all of build/tmp/*, it should "rebuild" quite fast since it will use the sstate-cache Dec 17 08:23:47 @erbo Ok I'm trying that thanks. It gives me massive error spam for about every recipe that "The recipe XXX is trying to install files into a shared area when those files already exist". Is that expected? Dec 17 08:29:26 "ERROR: XXX do_package_write_ipk: If the above message is too much, the simpler version is you're advised to wipe out tmp and rebuild (reusing sstate is fine). That will likely fix things in most (but not all) cases. Dec 17 08:29:47 wiping out tmp is what I just did, causing the errors Dec 17 08:42:51 stuom1: wiping out, in the meaning of completely removing tmp? Dec 17 08:45:22 LetoThe2nd yes Dec 17 08:45:41 stuom1: strange, that is totally unproblematic Dec 17 08:48:45 ls Dec 17 08:56:11 Hi. I want to add package manager ("apt-get") to my rpi3-yocto image. After adding EXTRA_IMAGE_FEATURES += "package-management" and modify PACKAGE_CLASSES ?= "package_deb" in local.conf file, I guess that I need to add repositories to the image so I will able to use "apt" command. How to do it? I want to add recipes that available from Dec 17 08:56:12 openembedded.org afterwards Dec 17 08:57:29 nacknick: sorry, thats not how it works Dec 17 08:57:44 nacknick: openembedded does *not* provide package feeds. Dec 17 08:58:23 kanavin: I think with the feedback I've just sent that means you have feedback on most of the unmerged patches. I'm going to wait for a new queue now for those. Dec 17 08:59:30 letothe2nd: OK. So what should I do to be able to install packages after the image is running? Dec 17 09:00:31 nacknick: you'd have to set up your ownb package repository -> https://www.intel.com/content/dam/www/public/us/en/documents/white-papers/package-manager-white-paper.pdf Dec 17 09:01:20 nacknick: but usually its way better to think about your usecase first. if you want an interactive system to tinker on and install packages at random, then OE/yocto is just not what fits your needs. Dec 17 09:01:55 nacknick: the OE way is to think and decide what one needs *before* building, and then create a custom image that exactly suits the usecae. Dec 17 09:04:42 LetoThe2nd: Thank you Dec 17 09:04:56 nacknick: have fun. Dec 17 09:05:27 nacknick: i think episode #2 gives a pretty good outline of how to do custom images: https://www.youtube.com/playlist?list=PLD4M5FoHz-TxMfBFrDKfIS_GLY25Qsfyj Dec 17 09:14:59 LetoThe2nd I had to remove build/deploy along with build/tmp to get rid of errors Dec 17 09:15:59 stuom1: ok. so you deliberately pulled deploy *out* of tmp? Dec 17 09:17:16 I didn't, but maybe somebody did :D Dec 17 09:17:22 thats very uncommon, and probably part of your problem. whenever somebody here says "wipe tmp", it is implicitly assumed that everything that is neither sstate nor downloads is removed. Dec 17 09:21:52 I think my local.conf is pretty much default what came with Toradex's sources, and yes deploy dir is separate Dec 17 09:22:09 go blame toradex Dec 17 09:22:16 I will ;) Dec 17 09:22:52 why would anybody ship you a local.conf for any reason other than hiding dirty hacks, no idea. Dec 17 09:29:03 When I run out of disk space, its not possible to start new tasks, not even 'clean'-task for a recipe in order to free up space. Are there any other way to salvage the build except wiping TMPDIR ? Dec 17 09:37:38 kroon: to be able to run bitbake, maybe modify BB_DISKMON_DIRS? https://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#var-BB_DISKMON_DIRS Dec 17 09:38:22 kroon: but I always just remove TMPDIR since with the sstate-cache it's pretty fast Dec 17 09:39:08 qschulz, ah.. I guess you mean temporarily lower the limits in order to run emergency cleanup tasks. sounds reasonable Dec 17 09:39:44 anyone seeing higher inode usage recently? Dec 17 09:40:49 I'm on 4M limit in TMPDIR with relatively small image Dec 17 09:52:10 has anybody tried to make a recipe for scipy using pip? Dec 17 09:52:57 stuom1: it has been asked several times, but i don't think there something out there. Dec 17 09:54:29 yeah all I find is different problems with it but no solutions Dec 17 10:09:24 hello! I have a board which can have two different flash sizes (4 and 8GB). I'm using wic to create a "data" ext4 partition. This data partition should have a differnt size for the 4 GB and 8GB flash Dec 17 10:09:57 lfa: 2 machine files which include all the common parts and just route to separate wks files Dec 17 10:10:01 What's the best practice to handle a situation like this? Generate two different wic images? Resize the partition dynamically on first startup? Dec 17 10:10:14 lfa: the wks files an also have a common include Dec 17 10:10:29 and therefor just haev different sizes. Dec 17 10:10:38 (thats the blunt, common aproach) Dec 17 10:11:11 Ok, so I should create two machines? Dec 17 10:11:23 the rest depends on your use case. Dec 17 10:11:59 it can be possible to resize on first startup. or maybe the only difference is in some for of data partition. or, or, or Dec 17 10:12:25 the 2machine approach certainly works, but there might be more elegant ones for your specific usecase. Dec 17 10:14:36 lfa: are you sure only the size changes? no page size? no block size? etc. If there's any chance that size is not the only change, different machines. If there is even the smallest chance the flash chip will change, different machines. (considering flash = nand device) Dec 17 10:14:45 considering/assuming Dec 17 10:14:51 ok, thanks. I will look into it. Two machines sounds a bit cumbersome, as really the only difference is the size of this partition (which is empty in the image btw) Dec 17 10:14:55 qschulz: it's an eMMC Dec 17 10:15:36 sorry, should have mentioned that :) Dec 17 10:16:05 lfa: or create two wic images with different sizes. Don't know if it's supported already though Dec 17 10:16:16 but usually, different hardware = different machines Dec 17 10:18:10 its the better safe than sorry appraoch. Dec 17 10:20:45 well.. there are other hardware differences too, but all these are handled during "runtime". I followed the approach the imx6 sabresd boards took. 1 machine, supporting different variants (e.g. iMX6 Solo vs Quad) Dec 17 10:21:26 The only remaining difference during build-time would be the eMMC size Dec 17 10:25:49 lfa: do à la RPi with resizing of the partition at runtime? Dec 17 10:28:00 qschulz: thanks for RPi pointer. Will look at how they do it Dec 17 10:35:35 lfa: I don't know if they do it in Yocto though, but raspbian and others do for sure Dec 17 10:36:02 fairly simple, grow the partition table then grow the file system Dec 17 10:37:13 systemd actually has a service built in to grow file systems, but it can't do partitions too. probably worth sending a patch to them to do the partition table too... Dec 17 10:52:15 New news from stackoverflow: Unable to start bitbake server Dec 17 11:34:37 anybody using chromium? Any way to get rid of that "getrandom indicates that the entropy pool has not been initialized" message at startup? Dec 17 11:35:05 I mean it doesnt start before it gets some entropy (mouse movement etc) Dec 17 11:40:42 get some entropy in your pool Dec 17 11:41:01 a lot of boards have a hardware rng you can hook up Dec 17 11:43:12 hi, what should i look at if fetching of an ssh+git source fails? if i run the git ls-remoto command with all the variables set by the fetcher in a normal shell, it looks fine, but bitbake complains about being unable to parse the HEAD revision Dec 17 11:46:43 stuom185, rburton, I got a stalling sshd on boot once, nowadays I think sshd RRECOMMENDS "rng-tools" in order to help with the entropy Dec 17 11:46:52 right Dec 17 11:46:55 its a brute force hammer Dec 17 11:48:37 i've been meaning to remove that and get BSPs to add it if they need it Dec 17 11:49:17 thanks, I'll look into rng-tools Dec 17 11:50:37 first look at your hardware Dec 17 11:51:08 there is /dev/hwrnd but what to do with it Dec 17 11:51:08 you might just need to pass a kernel argument to get it working nicely Dec 17 11:51:13 what is the hardware? Dec 17 11:51:22 hwrng i mean Dec 17 11:52:16 toradex colibri imx8x Dec 17 11:59:02 http://main.lv/writeup/kernel_dev_hwrng.md is a useful resource Dec 17 12:26:18 stuom185: assuming https://github.com/torvalds/linux/blob/master/drivers/char/hw_random/imx-rngc.c is the driver then maybe a fix like https://github.com/torvalds/linux/commit/62f95ae805fa9e1e84d47d3219adddd97b2654b7 is appropriate? Dec 17 12:38:13 rburton thanks for looking into it. I will take a note of these if I will need them later. It looks like it's way over my head though, I was hoping more like some --dont-ask-about-entropy flag for chromium haha. Dec 17 12:38:46 the problem is deeper: you have *no* entropy on boot Dec 17 12:38:53 rng-tools will bandaid it Dec 17 12:40:03 So far only chromium seems to care about that. Is there something bad about rng-tools? Dec 17 12:41:18 no, its just that for some hardware it isn't needed Dec 17 12:41:30 (if the kernel does the right thing) Dec 17 12:41:44 thus that omap patch: one line to tell the kernel to use the driver out of the box Dec 17 13:47:12 ndec: *sigh* andreea skipped my typo in the mail - can you edit a tweet she made? Dec 17 13:53:32 LetoThe2nd: let me check Dec 17 13:54:00 ndec: s/4:00/4:30/ Dec 17 13:58:24 LetoThe2nd: i couldn't edit, but added a corretion.. Dec 17 13:59:49 ndec: thanks and sorry for the confusion this time :( Dec 17 14:16:15 LetoThe2nd: 1630 is new start time? Dec 17 14:22:42 Crofton|road: it was always the start time, but it seems that this was only in my head Dec 17 14:22:55 Ah :) Dec 17 14:23:43 Dodn't forget to mention https://pretalx.com/oe-workshop-2020/ please Dec 17 14:23:53 Crofton|road: yeah i've got it mentally noted. Dec 17 14:24:38 I'll try and watch and remind you if you slack off Dec 17 14:26:06 "if"? Dec 17 14:29:03 RP: yes, I'm going to work on those one by one - thanks Dec 17 15:27:36 public service announcemente: live coding session about to start in 3 minutes :) Dec 17 15:27:55 https://www.twitch.tv/yocto_project Dec 17 15:41:18 So, trying to change image on my psplash, but nothing I do seems to be working, what am I doing wrong? Dec 17 15:41:47 I have gotten an image with psplash up and running and now I want to replace the image Dec 17 15:42:00 But merely replacing it seems to not do anything Dec 17 15:49:30 Everyone get on twitch and say hi! Dec 17 15:55:12 fray: LetoThe2nd needs some firecracker animations! Dec 17 15:59:35 YPTM: Scott Murray has joined Dec 17 16:07:02 wertigon: did you change the png, or the generated source Dec 17 16:07:21 wertigon: you need to re-generate a header from the png Dec 17 16:09:05 I did regenerate the .h file Dec 17 16:09:18 And copied it Dec 17 16:09:41 Thinking a cache problem somewhere, but nvm... Time to go home for today Dec 17 16:09:58 I'll do a clean rebuild during the night and see if that solves my problem :) Dec 17 16:10:50 My problem btw, image showing up is a bit large, so I want to shrink it somehow Dec 17 16:13:11 Anyway; see you tomorrow! Dec 17 16:18:58 YPTM minutes: https://docs.google.com/document/d/1ly8nyhO14kDNnFcW2QskANXW3ZT7QwKC5wWVDg9dDH4 Dec 17 16:19:29 RP, bug 13714 for the cento7 gcc uninative issue Dec 17 16:19:30 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=13714 normal, Undecided, ---, unassigned, NEW , add gcc to buildtools for Centos7 uninative Dec 17 16:21:24 rburton, how does one actually build a devupstream variant of a recipe? Dec 17 16:21:42 I tried lttng-modules-devupstream, devupstream-lttng-modules, nothing seems to work Dec 17 16:22:04 lttng-modules-devupstream:target, devupstream:target-lttng-modules Dec 17 16:25:52 Crofton|road: was that enough advertising? Dec 17 16:26:02 Yes Dec 17 16:26:11 having the link in chat should be great Dec 17 16:26:54 Crofton|road: for reasons unknown we seemed to only have a handful of viewers today. no idea why. Dec 17 16:30:12 actually thought that finding out why things happen is an interesting topic for most users ;-) Dec 17 16:33:32 Maybe the time issue? Dec 17 16:33:44 holidays? people are already slacking off Dec 17 16:34:18 yeah timing is really a problem for me these days. Dec 17 16:35:49 anyways. off then! Dec 17 16:36:02 I'm half asleep and unproductive today Dec 17 16:46:20 kanavin: it makes an identical recipe clone with a different PV Dec 17 16:46:30 kanavin: so set PREFERRED_VERSION Dec 17 16:46:51 oh, I see Dec 17 17:50:42 armpit: I took the qemuppc64 machine conf from one of your layers, it does not boot linux-yocto 5.2 though ,should it ? Dec 17 17:52:26 I have not touched it in years so I have not idea what state it is in Dec 17 17:53:23 khem: ask zeddii about mips64 ;-) I'd hope ppc64 is in better shape Dec 17 17:53:57 armpit:ok Dec 17 17:54:05 RP: yeah heh Dec 17 18:06:42 rburton, usage of devupstream badly breaks 'devtool upgrade' Dec 17 18:06:50 it gets confused between the two variants Dec 17 18:26:23 that's odd, it should know what's preferred and what isn't Dec 17 18:26:35 keying off of filename only perhaps Dec 17 20:50:09 what's the right way to override /etc/sysctl.d/99-sysctl.conf from procps? Dec 17 20:52:39 mischief: via a bbappend provide your own and add it to do_install Dec 17 20:53:20 see also recipetool appendfile Dec 17 23:54:25 New news from stackoverflow: How to install scipy on yocto warrior [closed] **** ENDING LOGGING AT Wed Dec 18 03:00:35 2019