**** BEGIN LOGGING AT Wed Apr 14 02:59:56 2021 Apr 14 04:08:26 khem: I've queued the meta-mingw patch in master-next Apr 14 04:43:51 shouldn't adding a package RDEPENDS_${PN} be enough to install another package/recipe? Apr 14 08:01:58 yo dudX Apr 14 08:02:29 stupid question. what are places one could deposit a site.conf? are there any known best practises around? Apr 14 08:41:55 alex88: how do you expect to create the systemd service in your recipe? you're probably going to give a file right? Then that file is in SRC_URI. Not sure recipes without SRC_URI work out of the box, worst case scenario you noexec all the tasks that happpen before do_install? Apr 14 09:13:09 is there a reason why https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html does not list KERNEL_IMAGETYPES? Apr 14 09:13:48 i find only something on KERNEL_IMAGETYPE (without "S") Apr 14 09:19:30 qschulz: hey! hey you! how many beers do you already owe me? Apr 14 09:20:00 * LetoThe2nd counts days since threat, and points out that there is still no submission to the CFP Apr 14 09:30:45 LetoThe2nd: the main issue currently is I don't know how much time I need for slides I didn't write yet :p Apr 14 09:34:37 excuses excuses I know Apr 14 09:34:44 beer beer! Apr 14 09:35:16 i'm not much a fan of slides anyways. propose a show and tell := Apr 14 09:38:51 Suppose a company created a board but didn't provide any meta-layer or I create my own board, is there a guide/tutorial available on how to create a meta-layer like meta-raspberry from scratch? Apr 14 09:42:40 LetoThe2nd: I don't see myself explaining how overrides/operators are working without visual support :D Apr 14 09:43:37 qschulz: use legos. Apr 14 09:43:51 someone12983: theres a bsp document somewhere. qschulz ? Apr 14 09:57:31 LetoThe2nd: https://docs.yoctoproject.org/bsp-guide/index.html I guess? Apr 14 09:59:21 huh yeah. someone just left, though. Apr 14 10:06:41 derRichard: Don't think it was ever documented? Please send a patch so this omission is gone :) Apr 14 10:07:55 qschulz: well, it is documented in poky/meta/conf/documentation.conf Apr 14 10:08:04 KERNEL_IMAGETYPES[doc] = "The list of types of kernel to build for a device, usually set by the machine configuration files and defaults to KERNEL_IMAGETYPE." Apr 14 10:08:09 this is why i'm so confused Apr 14 10:11:16 derRichard: our variable documentation is not automatically extracted from variable's [doc] Apr 14 10:14:29 oh? Apr 14 10:14:32 i thought it is Apr 14 12:08:51 Hi I'm trying to build an image with "wic" and I get the error "ERROR: No boot files defined, IMAGE_BOOT_FILES unset for entry #1", has anyone experience with wic's raw mode? Apr 14 12:27:21 Is this a bug? I run: "wic create -D sdimage-stratix10-emmc.wks -o tmp_out/ --rootfs-dir tmp_rootfs --bootimg-dir tmp_uboot --kernel-dir tmp_kernel --native-sysroot /sbin" Apr 14 12:27:42 and get the debug output: "DEBUG: Kernel dir: tmp_uboot" Apr 14 14:03:05 It works now, but the sd-image has a different size compared with the one which is generated by yocto (the content is the same, so it is just empty space which is added by yocto, i guess?) Apr 14 15:46:24 qschulz, ok got it, thank you! Apr 14 15:58:06 i recently started to develop yocto linux for a embedded machine. i have to write it to sdcard everytime :( to check it. how can i run it in pc only like instructor is doing in youtube video? uname -a output is Linux phyboard-segin-imx6ul-2 4.14.93-bsp-yocto-i.mx6ul-pd19.1.0 #1 SMP Wed Feb 10 01:11:08 UTC 2021 armv7l GNU/Linux Apr 14 15:58:07 i am just started and its vast Apr 14 15:59:09 sahilgg: a) you can use qemu on the pc b) it wont' emulate the specific imx hardware Apr 14 16:00:43 so i am unlucky? for this particular hardware i have to write it to sdcard everytime to check the progress? Apr 14 16:00:46 sahilgg: to shorten the round trip, you can often make a board boot off nfs or such. Apr 14 16:01:59 boot Apr 14 16:02:00        1: nand Apr 14 16:02:00        2: spi Apr 14 16:02:01        3: mmc Apr 14 16:02:01        4: net Apr 14 16:02:02        5: back Apr 14 16:02:02 these are the bootloader options available in my machine. i use mmc to use sdcard image Apr 14 16:02:17 well net sounds promising, right? Apr 14 16:02:48 but its certainly baord specifc, so please consult the documentatino. Apr 14 16:04:20 booting 'net' Apr 14 16:04:20 could not open /mnt/tftp/none-linux-phyCORE-i.MX6UL: No such file or directory Apr 14 16:04:21 Booting 'net' failed: No such file or directory Apr 14 16:04:21 booting 'net' failed: No such file or directory Apr 14 16:04:22 Booting failed with: No such file or directory Apr 14 16:04:22 Press any key to continue Apr 14 16:04:23 this is the output of it. documentation is incomplete ;--; i have to call him to ask the doubts and notepad file is almost trash that he gave me . Apr 14 16:05:26 well booting procedures are highly board specific. so, consult the vendor or whomever provided you with the board. Apr 14 16:07:36 got it :) Apr 14 16:08:24 there are several tutorials on how to boot from NFS/FTP, I remember writing one several years ago... but I just found this one https://community.nxp.com/t5/i-MX-Processors-Knowledge-Base/Boot-I-MX6q-SABRE-over-the-Network-using-TFTP-and-NFS/ta-p/1129169 Apr 14 16:08:38 maybe, this can help you on setting the needed variables Apr 14 16:08:59 if you're using u-boot, the process will be quite the same for imx6ull or imx6q Apr 14 16:09:31 @sahilgg Apr 14 16:10:24 or you don't have internet access, maybe? Apr 14 16:15:03 thanks @angolini i will try to follow this, it never came into mind to boot via network. luckily i have internet for development of this product. i waste time to solve tiny tiny mistakes of mine while building it  xD Apr 14 16:15:48 i have another doubt. i am able to ping google.com with nice ping like 5ms , able to git clone too. but i am unable to install python modules via pip, it gets retrying error i extended the timeout too by using --timeout=120 but it does not do it's work. any way around for thiss? is this good practice to install via pip install inside the image? or Apr 14 16:15:49 i have to include it local.conf only all the pip modules? Apr 14 16:38:05 I have a task that run a script and i want to have it re-run if the script changes, is it possible? Apr 14 16:45:46 oh that's a tricky situation. following Apr 14 16:46:41 is wayland/x11 needed to display boot messages? Apr 14 16:47:28 because I have a rpi4 and with this config https://gist.github.com/alex88/2c17db9a02a2149c8e91a5d363fdb226 I get no hdmi output Apr 14 16:50:04 oh it seems that enabling uboot and uart worked! Apr 14 17:14:53 anyone have tried to install remote.it service or similar service to get remote ssh to yocto device for support ? Apr 14 17:41:02 halstead: ping in pm Apr 14 18:00:38 RP: I was looking at analysing the gcc11 reproducibilty fails you reported is there some data on diffoscope output that is accessible ? Apr 14 18:01:57 this is what you sent my way https://autobuilder.yoctoproject.org/typhoon/#/builders/118/builds/159/steps/12/logs/stdio Apr 14 18:52:38 Hi Folks, I am looking for some help with cmake, which in the past I avoided! Don't judge the fact that install is trying to do some compilation, but I think that's part of the problem. Ceph tries to build some python binding C code during the install phase and it's not finding the core libraries and crt related objects. Apr 14 18:53:11 It seems to be searching in the recipe-sysroot-native area instead of the recipe-sysroot directory Apr 14 18:59:05 I'm working on a recipe to create a programming package for my image. Basically it just packages my rootfs image and a few scripts into a zip file. I'm basing my image off of some of the swupdate recipes. I'm running into an error in staging.bbclass during my do_install. The stack trace is below: Apr 14 18:59:06 ERROR: device-programming-0.1-r0 do_install: Error executing a python function in exec_python_func() autogenerated: Apr 14 18:59:06 The stack trace of python calls that resulted in this exception/failure was: Apr 14 18:59:07 File: 'exec_python_func() autogenerated', lineno: 2, function: Apr 14 18:59:07      0001: Apr 14 18:59:08  *** 0002:extend_recipe_sysroot(d) Apr 14 18:59:08      0003: Apr 14 18:59:09 File: '/home/cdgarren/Projects/device/bsp/sources/poky/meta/classes/staging.bbclass', lineno: 552, function: extend_recipe_sysroot Apr 14 18:59:09      0548:                    dest = newmanifest[l] Apr 14 18:59:10      0549:                    if l.endswith("/"): Apr 14 18:59:10      0550:                        staging_copydir(l, targetdir, dest, seendirs) Apr 14 18:59:11      0551:                        continue Apr 14 18:59:11  *** 0552:                    staging_copyfile(l, targetdir, dest, postinsts, seendirs) Apr 14 18:59:12      0553: Apr 14 18:59:12      0554:    bb.note("Installed into sysroot: %s" % str(msg_adding)) Apr 14 18:59:13      0555:    bb.note("Skipping as already exists in sysroot: %s" % str(msg_exists)) Apr 14 18:59:13      0556: Apr 14 18:59:14 File: '/home/cdgarren/Projects/device/bsp/sources/poky/meta/classes/staging.bbclass', lineno: 152, function: staging_copyfile Apr 14 18:59:58 Does anyone know the fix to this? I've found a few instances of it around the web, but none have any solution. Apr 14 19:04:15 cdgarren: might be easier if you include what your recipe is doing Apr 14 19:04:19 Look at the inputs to the call of that python function either the function itself isn't being called right, or had bad parameters Apr 14 19:04:35 that would all come from the recipe.. put the recipe in a pastebin and it'll be easier to look at Apr 14 19:06:17 https://pastebin.com/raw/2yqWAEPn Apr 14 19:06:29 rburton recipe is here Apr 14 19:37:40 I'm also a little confused about what the sysroot is really doing in this recipe. My recipe doesn't use the libgcc-initial package that seems to be part of the issue here. Is it being brought in by the image I'm zipping up? Apr 14 19:58:27 khem: there should be, yes. Look at the log output and then find the matching non-release directory Apr 14 19:58:42 khem: it follows the same pattern each time Apr 14 20:23:40 RP: where should look those logs dont say much Apr 14 21:10:05 should I set PACKAGE_CLASSES = "" in the distro conf to avoid the _ipk tasks? Apr 14 21:14:29 khem: do you have a build link handy? Apr 14 21:14:39 yes Apr 14 21:14:47 https://autobuilder.yoctoproject.org/typhoon/#/builders/118/builds/159/steps/12/logs/stdio Apr 14 21:15:44 khem: Non-reproducible packages will be copied to /srv/autobuilder/autobuilder.yoctoproject.org/pub/repro-fail/oe-reproducible-20210410-74b39g5q Apr 14 21:16:28 how do I copy them out Apr 14 21:16:44 khem: https://autobuilder.yocto.io/pub/repro-fail/oe-reproducible-20210410-74b39g5q/packages/diff-html/ Apr 14 21:17:55 khem: binaries are also at that url Apr 14 21:18:10 ok cool Apr 14 21:18:21 well, https://autobuilder.yocto.io/pub/repro-fail/oe-reproducible-20210410-74b39g5q/packages but you get the idea Apr 14 21:18:23 it will be cool to dump the url in logs Apr 14 21:18:44 khem: its nice it does this at all ;-) Apr 14 21:18:50 patches welcome Apr 14 21:32:47 ok. `bitbake core-image-sato` "just worked" with instructions on ubuntu 20.04 where it did not on fedora 34 (and maybe fedora 33) so I'm good to go but maybe some things are broken. I'll try again and submit a bug if I can reproduce again. :) Apr 14 21:37:24 RP: gcc11 switches to dwarf5 for default debug info and I think these diffs are related to that, I have send few dep fixes for valgrind are also related to same Apr 14 21:43:33 khem: I saw the valgrind fixes, thanks, I'll queue those Apr 14 21:44:14 hopefully will see something for gcc one too Apr 14 21:45:25 I think kergoth had seen similar issue when work-shared was involved gcc did not map paths correctly Apr 14 21:45:54 perhaps kergoth can add a bit to jog my memory here Apr 14 21:46:32 khem: I think his problem was different, the shared sources were meaning the sources in the debug package were not being populated correctly Apr 14 21:46:57 yeah, the shared sources weren't copied. just needed a couple additional debug and macro path maps and an addition to package.bbclass to allow pulling sources from those alternative paths. i'll see about submitting it in the near future Apr 14 21:47:06 yes I rememberr Apr 14 23:57:48 is there a way to keep enough debug info for backtraces to work in binaries? Apr 15 00:07:58 how do you enable a custom systemd .mount file at build time? Apr 15 00:09:26 SYSTEMD_AUTO_ENABLE and SYSTEMD_SERVICE are use for package recipes, but here I'm adding custom unit files to systemd-conf_%.bbappend Apr 15 00:10:10 should I also add the symlink manually to these systemd-conf files? Apr 15 00:26:33 vdl: just make your own recipe? Apr 15 00:34:59 mischief: I thought all systemd conf should go into systemd-conf, but I think you're right Apr 15 00:40:56 vdl: no.. only modify that recipe if you need to change the files that systemd-conf installs Apr 15 00:41:27 (in my project, we delete upstream networkd units for example) Apr 15 01:59:51 RP: kergoth The gcc11 reproducibility issue is due to use of git SHA1 id in S, It would not happen for proper releases, however I think the check to replace debug-maps can be fixed Apr 15 02:00:08 and made mode robust, I Will send a patch **** ENDING LOGGING AT Thu Apr 15 02:59:56 2021