**** BEGIN LOGGING AT Thu Nov 08 02:59:59 2018 Nov 08 03:00:43 Missing or unbuildable dependency chain was: ['kernel-initramfs', 'virtual/kernel', 'initramfs'] Nov 08 03:00:43 ERROR: Required build target 'core-image-base' has no buildable providers. Nov 08 03:00:43 Missing or unbuildable dependency chain was: ['core-image-base', 'kernel-initramfs', 'virtual/kernel', 'initramfs'] Nov 08 03:00:43 Summary: There were 2 ERROR messages shown, returning a non-zero exit code. Nov 08 03:00:46 No that just means a variable Nov 08 03:01:06 do we need meta-initramfs Nov 08 03:01:08 INITRAMFS_IMAGE = "core-image-weston" Nov 08 03:01:18 ahhhhhh Nov 08 03:01:24 the name should be proper image name Nov 08 03:01:29 that you want to bundle Nov 08 03:02:20 and thats whats missing Nov 08 03:02:29 the secret sauuce to naming convetntions Nov 08 03:02:51 I think <> denotes a variable Nov 08 03:03:36 we are compiler people so we think whole world knows BNF Nov 08 03:04:19 so then bitbake core-image-weston Nov 08 03:04:33 after i set NITRAMFS_IMAGE = "core-image-weston" Nov 08 03:04:41 actually if you want to bundle core-image-weston then bitbake some other image Nov 08 03:04:45 khem: great lets see Nov 08 03:05:01 so itll inherit it Nov 08 03:05:02 bitbake core-image-base may be Nov 08 03:05:20 yes core-image-weston will get bundled into kernel Nov 08 03:05:47 which means remoove my dependency loop now :) Nov 08 03:05:56 right Nov 08 03:06:00 ughhhh Nov 08 03:06:03 LOOL Nov 08 03:06:33 ideally you dont want same images in initrd and real rootfs Nov 08 03:06:42 thats not usual Nov 08 03:07:12 ughhh Identifying dependency loops (this may take a short while)... Nov 08 03:09:05 add IMAGE_FSTYPES_pn-core-image-weston = "${INITRAMFS_FSTYPES}" in local.conf Nov 08 03:09:17 core-image-weston is not a common initrd image Nov 08 03:09:17 i get it so far now, buut now time to backtrack from my meta-build and remove the "dependency loops" as my image name is core-imaage-build-weston which inhherits core-image and core-image-weston ... YAY Nov 08 03:10:08 OutBackDingo: Do me a favor once it works write it down stepwise Nov 08 03:10:26 :P Nov 08 03:10:33 rewrite the book huuhh Nov 08 03:10:50 khem: done anything withh enncrypted rootfs ? hahhahhahahha Nov 08 03:11:34 yes but I have to leave now Nov 08 03:12:23 khem: if youu can catch up with me lateer, meantime lets see if i can get thhrough a initramfs **** BEGIN LOGGING AT Thu Nov 08 04:57:30 2018 **** BEGIN LOGGING AT Thu Nov 08 05:12:34 2018 Nov 08 05:32:55 khem: i dont see an sdimg ? i do see a core-image-weston-raspberrypi3-20181108052912.rootfs.cpio.gz Nov 08 06:26:24 khem: bootlooping with ERROR Image tooo large consider increasing CONFIG_SYS_BOOTM_LEN Nov 08 07:32:05 apears u-boot on the pi 3 doessnt like initramfs Nov 08 07:44:30 Moin Nov 08 09:19:20 New news from stackoverflow: version node not found for symbol Nov 08 10:41:13 what is the policy for backporting fixes to old releases? do you accept gcc-8 fixes for rocko? Nov 08 10:50:18 specifically, I'm getting alignment crashes in elfutils and python-native. these are fixed in sumo but not rocko. Nov 08 10:51:41 I'm using bbappend recipes to add the patches locally so it's not a problem for me. just wondering what the project policy is. Nov 08 11:08:27 Hi all! I'm trying to add a custom user to the target system. For this I added "useradd admin;" to the EXTRA_USERS_PARAMS var in local.conf.sample, but it fails. I've tried it with different parameters, finally it fails even with the username only. I've also tried groupadd with the same result. The error is " groupadd command did not succeed" without any other info. What am I doing wrong? Nov 08 11:09:36 zagor, gcc-8 was introduced with thud some months ago. I don't think it will be backported. Nov 08 11:44:15 hi, why do I need get a prompt when I boot up runlevel one, single user mode? Nov 08 11:44:38 ant_work: I mean when the host machine runs gcc-8 Nov 08 11:57:09 Hi all, what is the difference between standard and extensible SDK ? Nov 08 12:27:11 prabhakarlad: one is a fairly typical standalone toolchain, the other gives access to more tools like devtool Nov 08 12:37:30 ok, anyyone know whats thhe secret to initramfs size on a PI ?? the build keeps failing Nov 08 12:38:56 OutBackDingo: it sounded like your build worked but it failed to boot? Did you try doing what the error said? Nov 08 12:40:59 RP: https://pastebin.com/tnLTNcQT Nov 08 12:41:06 its seems the disk is ful;l Nov 08 12:41:13 its not creating images Nov 08 12:41:31 ive even had to remove u-boot Nov 08 12:43:35 where thhe initramfs image should be soore-image-minimal Nov 08 12:43:45 core-image-minimal Nov 08 12:44:50 OutBackDingo: you could try INHERIT += "rm_work" Nov 08 12:46:26 RP: inherit it where ? Nov 08 12:46:39 iver already had to set The initramfs size 245760(K) overrides INITRAMFS_MAXSIZE: 131072(K) Nov 08 12:47:47 OutBackDingo: I guess it depends if your image is full or your host system, its hard to tell from that pastebin Nov 08 12:48:00 RP: the host disk isnt the issue Nov 08 12:48:08 its when its trying to create the images Nov 08 12:48:38 /dev/mapper/newdebian--vg-root 885G 284G 556G 34% / Nov 08 12:48:53 OutBackDingo: sorry, my advice won't help then Nov 08 12:49:45 OutBackDingo: it sounds like its trying to write the whole rootfs into the initramfs which would be why its out of space Nov 08 12:50:05 hmm, maybe not Nov 08 12:50:40 RP: and thats what im thinking.... when it shhould only be trying to write core-image-minimal Nov 08 12:51:06 OutBackDingo: best guess is the do_rpi_sdiimge itself doesn't know to include the initramfs size in its size of the sd image it builds Nov 08 12:52:16 RP: thank you :) Nov 08 12:53:01 my config https://pastebin.com/c9zTk7dA Nov 08 12:53:16 and im thhen bitbake core-image-weston Nov 08 12:57:13 i cant believe its this difficult to generate a weston image with initramffs Nov 08 13:04:15 OutBackDingo: likely something simple but I'm not able to page in the right data to help :( Nov 08 13:17:43 OutBackDingo, do you want to have core-image-weston as initramfs? Nov 08 13:21:55 fsdun: no i want core image minimal as initrams with encryptfs and ssh Nov 08 13:22:13 so i can encrypt rootfs and then boot to weston Nov 08 13:30:23 OutBackDingo, do you have SDIMG_SIZE set?. could you paste /home/dingo/INITRAMFS/rpi-build/tmp/work/raspberrypi3-poky-linux-gnueabi/core-image-weston/1.0-r0/temp/run.do_image_rpi_sdimg.24666 Nov 08 13:34:17 fsdun: https://pastebin.com/7FDkrgaf Nov 08 13:36:40 fsdun: SDIMG_SIZE ? Nov 08 13:40:16 i literally think... its trying to pack it all inside initramffs where i only want core-image-minimal in initramfs Nov 08 13:41:15 if i literally strip my config down it build core-image-base withh core image-minimal as initramfs Nov 08 13:42:10 itll complete i can dd the sdimg and it boots, though no networking... as soon as i tryy to add thhings for the config to get weston, ort my custom packages it fails Nov 08 13:45:30 what's the size of /home/dingo/INITRAMFS/rpi-build/tmp/deploy/images/raspberrypi3/bcm2835-bootfiles/ ? Nov 08 13:46:33 I think it's failing in line 141 of you pastebin which is line 121 in sdcard_image-rpi.bbclass Nov 08 13:46:46 du -h /home/dingo/INITRAMFS/rpi-build/tmp/deploy/images/raspberrypi3/bcm2835-bootfiles/ Nov 08 13:46:46 13M /home/dingo/INITRAMFS/rpi-build/tmp/deploy/images/raspberrypi3/bcm2835-bootfiles/ Nov 08 13:47:20 i cant be the only one whos done this Nov 08 13:47:31 LOL :) Nov 08 13:49:07 are you aware that IMAGE_INSTALL_append also applies for core-image-minimal? Nov 08 13:49:55 fsdun: it shhould be what, to not install / append to initramfs but to install in rootfs ? Nov 08 13:52:36 core-image-minimal is a normal image which also uses IMAGE_INSTALL (as core-image-weston does), but is packaged as INITRAMFS_FSTYPES. Nov 08 13:53:28 I'll probably need to different image recipes. Nov 08 13:53:38 *facepalm* Nov 08 13:53:43 You'll probably need to different image recipes. Nov 08 13:54:26 OutBackDingo: set INITRAMFS_MAXSIZE = "262144" Nov 08 13:54:51 in local.conf, the image that is being bundled is big Nov 08 13:56:56 khem: i know you said youve done this beforee but i really cant see how Nov 08 13:58:10 does a filesystem exist for eeproms? Nov 08 13:59:50 yates: Does cpio fit your bill? :p Nov 08 14:00:52 khem: why is it big, if i only want core-image-minimal to be initramfs with encryptfs to unlock mmcblkk0p2 and then boot to weston ? Nov 08 14:01:07 so whhats the secret that im missing hhere Nov 08 14:01:20 I thought you were using core-image-weston **** BEGIN LOGGING AT Thu Nov 08 14:02:16 2018 Nov 08 14:05:01 khem: i fails to buildd the image Nov 08 14:07:53 OutBackDingo: I used wic last time i did an initramfs boot with the raspi bootloader. One initramfs rootfs image packed onto the filesystem of p1 (via IMAGE_BOOT_FILES), and one disk image. In your case the disk image could include your encrypted root on p2 of the disk. Just a suggestion if the sdcard image type doesn't pan out for you. Nov 08 14:08:09 I see you have IMAGE_INSTALL_append which could be bloating the image quite a bit Nov 08 14:09:11 SD card has two paritions one for boot files where kernel lives and its relatively small Nov 08 14:09:35 now when you pack initramfs into kernel the size of bootfiles will grow too Nov 08 14:09:57 so you might need to adjust the size of boot partition Nov 08 14:10:25 something like BOOT_SPACE ?= "40960" makes is 40M Nov 08 14:10:39 khem: thats the default Nov 08 14:10:46 you might need to set it bigger than that Nov 08 14:10:57 whats the size of cpio.xz Nov 08 14:11:37 I have recipeFoo that DEPENDS on recipeBar-native. I thought that if I did a "cleanall" on recipeBar-native, it would cause recipeFoo to get rebuilt the next time. Well, it doesn't. But how can I enforce it? Nov 08 14:12:58 jofr: if nothing changed then it should not since signatures wont change Nov 08 14:13:17 clean both Nov 08 14:17:41 khem: It is changing, but all the changes are in recipe-sysroot-native because recipeBar-native is.. well.. native. :p Nov 08 14:21:05 I think I managed to fix it by changing recipeBar to a non"-native", BBEXTENDing native and defining a FILES_ section Nov 08 14:22:35 khem: must be the appends Nov 08 14:22:46 i have a weston image now im testing Nov 08 14:26:18 khem: nope blank screen, not booting Nov 08 14:29:08 khem ok wait... i can ssh into it Nov 08 14:29:25 devtmpfs 171.3M 0 171.3M 0% /dev Nov 08 14:29:25 tmpfs 304.8M 92.0K 304.7M 0% /run Nov 08 14:29:25 tmpfs 304.8M 48.0K 304.7M 0% /var/volatile Nov 08 14:29:36 its booted to tmpfs not rootfs on sdcard though Nov 08 14:29:50 guys Nov 08 14:29:53 ughhhhh Nov 08 14:37:26 noow, hhot to get it to continue to boot rootfs Nov 08 15:10:29 Hi , I have built and SDK, and now when extracting I get this ERROR: Sstate artifact unavailable for package, any thought what could be missing ? Nov 08 15:56:48 RP: hmm, now that getVar defaults to expansion, indexing and .get() should be expanded, so we should be good to change things like (d.getVar('FOO') or '').split() to d.get('FOO', '').split(), more like a regular mapping, yes? Nov 08 16:38:28 kergoth: I guess in theory but my preference is to leave getVar as a) its clearer to read somehow and b) the patch churn and conflicts such changes generate make life a nightmare Nov 08 16:39:23 true. i also wonder how it would affect checksum generation, i don't remember if the signature generation checks indexing and .get and friends Nov 08 16:39:46 probably *should*, but.. Nov 08 16:40:30 definitely easier to do ast analysis when we don't have to differentiate dicts from datasmart objects Nov 08 16:42:43 Hmm, I wonder if there'd be an interest in https://github.com/MentorEmbedded/meta-mentor/blob/master/meta-mentor-staging/classes/win_sdk_cull.bbclass going upstream Nov 08 16:45:30 kergoth: getVar is a much nicer expression for the magic ast code to spot Nov 08 16:46:31 kergoth: that might be a good one for meta-mingw Nov 08 16:46:38 ah, good point Nov 08 16:47:07 kergoth: I agree (on win_sdk_cull.bbclass). We do something like that already... but I can't find it atm Nov 08 16:47:18 ah, k. will add to my list to check Nov 08 16:47:21 thanks Nov 08 16:48:24 I should probably see if anyone is interested in our self extracting python script for mingw SDKs some day too.... Nov 08 16:55:42 JPEW: yes, as a patch to meta-mingw :) Nov 08 16:55:52 mingw really needs to get some maintainers who use that stuff Nov 08 16:56:59 RP: Ya I've noticed. We use it pretty heavily, I *might* be convinced to voluteer for that ;) Nov 08 16:58:42 * RP quicks fires up an editor on the README Nov 08 17:03:36 :) Nov 08 17:04:13 self extrasking mingw SDKs? confused Nov 08 17:04:44 we just create a zip file using the post processing stuff.. and the user extracts the zip and it 'all just works' Nov 08 17:05:08 (seems to be what our customers have expected at least...) Nov 08 17:25:12 fray: There are some post processing steps that we have found necessary (just like on Linux SDKs) Nov 08 17:25:51 fray: I don't remember what atm, we've been using it for years Nov 08 17:51:02 New news from stackoverflow: AOT compile system assemblies using Yocto meta-mono Nov 08 18:25:20 armpit: fixed the 2.4.4 artefact publishing and resent the qa email with help from halstead Nov 08 18:26:04 kergoth: did you want to take 12987? Nov 08 18:26:18 yeah i can do that Nov 08 18:26:52 kergoth: its yours! (thanks0 Nov 08 18:27:32 np Nov 08 18:27:58 sorry i havent' had more time to help with core work recently, been busy at work and at home. looking to improve that again soon Nov 08 18:28:51 kergoth: np, I've been having my own sets of challenges too... Nov 08 18:29:14 * RP is getting frustrated with doing QA stuff all the time Nov 08 19:01:37 Hi. I have a layer that I use together with yocto rocko, sumo and to-be thud. In there I have a linux-yocto_4.12.bbappend, but in oe-core thud there is no longer a linux-yocto_4.12 recipe to append to, which results in errors. Is there a way to have a .bbappend apply only if the recipe exists, and not error otherwise ? Nov 08 19:04:17 maybe some trickey with the layer.conf ? Nov 08 20:19:42 kroon see the concept of dynamic-layers https://github.com/YoeDistro/meta-wpe/tree/master/dynamic-layers Nov 08 20:19:59 layer.conf https://github.com/YoeDistro/meta-wpe/blob/master/conf/layer.conf Nov 08 20:24:17 you should really be using branches Nov 08 20:35:08 i'm rather annoyed i had to homeroll checking for shell parameter expansion modifiers because pysh never got around to implementing them.. Nov 08 20:35:10 * kergoth rolls eyes Nov 08 20:35:40 ended up pretty easy though: https://github.com/openembedded/bitbake/compare/master...kergoth:yocto-12987 Nov 08 20:48:20 kergoth: not bad Nov 08 20:48:52 kergoth: as ever I worry about performance and memory storage size of the objects but I'm paranoid Nov 08 20:50:39 yep, that was my first thought after doing it as well, we do a hell of a lot of shell and python parsing, any changes to it affect performance. no statistically significant impact on parse time, still need to test a few builds Nov 08 20:51:19 unexpanded ${} in shell is likely pretty rare to begin with, so the # of times it even has to enter this block is pretty low Nov 08 20:52:26 there were 12 instances where modified_expansions was non-empty across all of oe-core and meta-oe Nov 08 20:58:32 kergoth: fair enough. A non-obvious issue is that even an empty set() has memory storage size. Where things were rare, I did notice a big difference elsewhere by using hasattr() and leaving the entry missing Nov 08 20:59:02 huh, that's interesting, this would be a good candidate Nov 08 20:59:08 i'll look into t hat, thanks Nov 08 20:59:31 kergoth: http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=a9dc2ac9e00e3bbea872c8d22568e3acfb9ba3d4 - the removes Nov 08 21:02:52 i'm getting spoiled using f-strings in my other projects, looking forward to the next python 3 version requirement bump :) Nov 08 21:25:05 kergoth: Is that 3.5? I'm looking forward to getting to that Nov 08 23:22:08 New news from stackoverflow: Using PREMIRRORS in Bitbake configuration Nov 09 00:23:48 what is the name I need to use to pull in the busybox httpd into an image? I have tired httpd, busybox-httpd, and lighthttpd and all of them fail because it can't find a provider for it Nov 09 00:47:46 demonimin: lighttpd is in meta-openembedded/meta-webserver Nov 09 00:48:00 as is nginx and others Nov 09 00:48:16 whooops he left Nov 09 00:48:21 wrong guy Nov 09 00:48:36 disregard **** ENDING LOGGING AT Fri Nov 09 02:59:58 2018