**** BEGIN LOGGING AT Tue Mar 13 03:00:03 2018
Mar 13 06:49:52 <learningc> I type make and I get following error: make: arm-none-linux-gnueabi-gcc: Command not found
Mar 13 06:50:00 <learningc> How do I get the env setup?
Mar 13 07:34:47 <robert_yang> is RP or ross around, please ?
Mar 13 09:56:16 <jww> Hello !
Mar 13 10:06:59 <jww> I'm trying to add u-boot support for a rpi3, from what I read I should just add RPI_USE_U_BOOT = "1" to local.conf , but it seems it do no add u-boot to my image and I don't see the u-boot.bin file in the boot partition.can somebody help please ?
Mar 13 12:39:00 <yocti> New news from stackoverflow: Bitbake meta-data adoptations in own layer - concept's limits <https://stackoverflow.com/questions/49256023/bitbake-meta-data-adoptations-in-own-layer-concepts-limits>
Mar 13 12:39:05 <paulbarker> jww: The u-boot.bin files is renamed to what the rpi3 expects to load
Mar 13 12:39:11 <paulbarker> I think it's kernel7.img
Mar 13 12:40:06 <paulbarker> Which branch of OE and meta-raspberrypi are you using?
Mar 13 13:09:09 <yocti> New news from stackoverflow: Bitbake meta-data adoptations in own layer - any concept's limits? <https://stackoverflow.com/questions/49256023/bitbake-meta-data-adoptations-in-own-layer-any-concepts-limits>
Mar 13 13:31:32 <cuero> Hi channel. I had a small question/contribution, did post on openembedded-core email list, but unsure if that was the right location. Is this channel alive ?
Mar 13 13:45:26 <jww> paulbarker: sorry I was out for lunch, I'm using rocko for both.
Mar 13 13:50:09 <jww> cuero: yep the channel seems alive :)
Mar 13 13:50:16 <cuero> I noticed that when using uninative (which is a default in poky), the class is modifying variables in the datastore after recipes are parsed but before the actual build is started. This cause the next build to reparse all recipes, but even worse (if understood correctly) is that it builds with recipes parsed with the wrong values for theses variables.
Mar 13 13:50:57 <cuero> hi jww, indeed, I just landed in during the lunch break :)
Mar 13 13:52:51 <cuero> I have a simple fix that is to change the uninative variables before the recipe parsing (it is a one line change), should I directly propose a patch as described here: http://www.openembedded.org/wiki/How_to_submit_a_patch_to_OpenEmbedded or should I discussed it before :) (like here )
Mar 13 13:54:19 <paulbarker> jww: RPI_USE_U_BOOT = "1" should work on rocko. Do you see u-boot when you boot the rpi3 with your image?
Mar 13 13:55:20 <jww> no I just see the yocto logo.
Mar 13 13:58:05 <paulbarker> do you see a file named uImage in the boot partition?
Mar 13 14:00:21 <jww> umm I cannot check now because I'm formatting the sd card, but I don't think so, I remenber seeing a zImage.
Mar 13 14:00:24 <paulbarker> if not could you share your local.conf file and the summary bitbake prints at the start of a build listing the layers and such?
Mar 13 14:01:03 <jww> okay, should I do that for a specific image build ?
Mar 13 14:01:31 <paulbarker> yes, probably best to focus on one image for now
Mar 13 14:08:58 <jww> here it is : http://dpaste.com/32EZR5F
Mar 13 14:09:43 <jww> I had troubles rebuilding the image I tried before ( I probably messed up things ), so I did it with rpi-hwup-image
Mar 13 14:11:26 <vladzouth> hi, i am newbie in yocto. I have added a package to my local.conf file. I have many kernel image( core-image-minimal, core-image-sato, core-image-w11) but i didn't know for what kernel image this configuration is done. How to know it?
Mar 13 14:11:59 <paulbarker> jww: ok. You're setting IMAGE_FSTYPES to only produce ext4.gz and tar.xz format images
Mar 13 14:12:52 <jww> paulbarker: it's wrong ? I use the .tar.gz to create the initial SD, and the ext4.gz is for swupdate.
Mar 13 14:14:41 <paulbarker> The tar.gz file will just contain the files for the main partition. The boot partition is separate on raspberry pi
Mar 13 14:15:31 <jww> I've taken inspiration from meta-rpi's script to create my sd card. this script manually copy stuff to the boot partition.
Mar 13 14:17:32 <paulbarker> ok. You'll need to handle quite a few things yourself in that case
Mar 13 14:19:03 <paulbarker> The dependency on u-boot is added for the "rpi-sdimg" image type only
Mar 13 14:20:27 <paulbarker> for your script you'll probably have to bitbake u-boot as well as rpi-hwup-image and then copy things over manually
Mar 13 14:21:25 <jww> umm I see.
Mar 13 14:22:20 <jww> I guess the best is to use rpi-sdimg to see what /boot looks like. and try to achieve the same ?
Mar 13 14:23:03 <paulbarker> yea that's one option. You could look at sdcard_image-rpi.bbclass and see how it handles RPI_USE_U_BOOT
Mar 13 14:23:48 <paulbarker> it's basically one extra dependency (https://github.com/agherzan/meta-raspberrypi/blob/rocko/classes/sdcard_image-rpi.bbclass#L58)
Mar 13 14:23:56 <paulbarker> and some minor changes to the file copying (https://github.com/agherzan/meta-raspberrypi/blob/rocko/classes/sdcard_image-rpi.bbclass#L139)
Mar 13 14:24:16 <paulbarker> not too complicated to replicate as long as you're happy maintaining your own script
Mar 13 14:26:54 <jww> I wish to have a dual copy sd card to use with swupdate, I guess I have no other choice than a custom script.
Mar 13 14:27:15 <jww> thanks a lot for your help !
Mar 13 14:27:19 <paulbarker> I imagine someone has got meta-swupdate working on rpi before
Mar 13 14:28:16 <jww> yes I success with that, it's not too complicated.
Mar 13 14:28:18 <paulbarker> Have a look around, maybe ping the swupdate mailing list for specific advice
Mar 13 14:28:41 <paulbarker> And if you get dual-partition working contribute it back :)
Mar 13 14:29:11 <jww> I'll be glad to contribute :)
Mar 13 14:38:23 <vladzouth>  hi, i am newbie in yocto. I have added a package to my local.conf file. I have many kernel image( core-image-minimal, core-image-sato, core-image-w11) but i didn't know for what kernel image this configuration is done. How to know it?
Mar 13 14:40:04 <jww> vladzouth: I'm new too, but I think it apply on all images you build with that local.conf
Mar 13 14:41:08 <vladzouth> OK :) , thank you!
Mar 13 15:39:39 <yocti> New news from stackoverflow: Bitbake meta-data adoptations in own layer - any limitations in concept? <https://stackoverflow.com/questions/49256023/bitbake-meta-data-adoptations-in-own-layer-any-limitations-in-concept>
Mar 13 16:00:38 <rburton> zeddiii: are you at ELC?
Mar 13 16:01:04 <zeddii_home> nope. I was a budge casualty :D
Mar 13 16:01:09 <zeddii_home> budget even
Mar 13 16:02:52 <rburton> i've reason to suspect your kernel upgrades are causing do_package to take a lot longer
Mar 13 16:03:56 <zeddii_home> both v4.14 and .15 ? I haven’t noticed. but I can do some timings here.
Mar 13 16:04:26 <rburton> _ipkg only, interestingly
Mar 13 16:04:39 <rburton> linux-yocto:do_package_write_rpm 17.780000 29.400000 (11.620000)
Mar 13 16:04:39 <rburton> linux-yocto:do_package_write_ipk 15.470000 151.320000 (135.850000)
Mar 13 16:04:49 <rburton> before after durations (delta in brackets)
Mar 13 16:06:58 <rburton> thats 8e3839ce266 vs be2982a630d
Mar 13 16:10:47 <u1106> I use useradd.bbclass to add a new user to my image (it's also used by other recipes, so I need useradd, not extrausers). It nearly works, but the mode bits of the user's home directory I specify in the install command don't end up in the image. The home directory is always rwxr-xr-x. If I create another directory /usr/lib/test exacatly the same way as the home directory my mode bits end up in the image. Where does Yocto "normalize" protections of home dirs
Mar 13 16:10:47 <u1106> / how can I prevent that?
Mar 13 16:11:24 <kergoth> u1106: see fs-perms.txt in meta/files/
Mar 13 16:11:35 <u1106> in the ipk created during the build the protections are still correct
Mar 13 16:11:55 <u1106> kergoth: that sounds promising
Mar 13 16:12:22 <kergoth> if the ipk contents are correct, though, that implies a change occurring via the postinsts instead
Mar 13 16:12:32 <zeddii_home> rburton: I’ll fire up and build and see if there’s extra files, or some large files showing up that are slowing down ipk
Mar 13 16:13:11 <rburton> zeddii_home: buildhistory-diff suggests the packages are the same, so its very odd
Mar 13 16:13:27 * zeddii_home ponders
Mar 13 16:13:28 <rburton> but all the perf machines are seeing the same bump
Mar 13 16:14:15 <zeddii_home> in the packaging phase, it should all just be data. unless some of the new build outputs are triggering processing.
Mar 13 16:14:22 * zeddii_home ponders more
Mar 13 16:17:46 <u1106> kergoth: yeah, fs-pers.txt has only one line that matches grep -i home
Mar 13 16:17:56 <u1106>  /home                           0755    root    root    false - - -
Mar 13 16:18:06 <RP> zeddii_home, rburton: with the devsrc patch, I wondered if we might drop the openssl-dev dependency to get the patch in since we're not quite at 4.15 yet?
Mar 13 16:18:55 <zeddii_home> linux-yocto is there. it breaks without it on x86-64 in my testing.
Mar 13 16:19:06 <zeddii_home> no bio.h. same thing the full kernel build hits.
Mar 13 16:19:43 <RP> zeddii_home: ah, ok
Mar 13 16:19:54 <RP> shame as that would have been an easier fix...
Mar 13 16:20:28 <zeddii_home> indeed. I only added it in v3, since I ran into it when doing a “for arch in <all @# arches>; make scripts ..”
Mar 13 16:22:08 <RP> zeddii_home: conference and travel are making it a little tricky for me to know where anything is at :(
Mar 13 16:23:35 <u1106> kergoth: but that cannot be the one, because recusive is false, and my home dir in the image is not owned by "root root" (but the same way I installed it). Could you elaborate where/how to search in postinsts?
Mar 13 16:25:10 <lukma> Life on the channel :)
Mar 13 16:25:31 <zeddii_home> RP: yah. let’s just say, I’m not sad to be “not traveling” this week. I’ll see if I can reproduce the multilib issue this afternoon/tomorrow.
Mar 13 16:27:44 <rburton> zeddii_home: if you do a build before/after the upgrade with -C unpack virtual/kernel, you can use buildstats-diff to see the differences (assuming you've buildstats on)
Mar 13 16:28:24 <rburton> zeddii_home: https://pastebin.com/C3ZEhrPb
Mar 13 16:29:37 <zeddii_home> rburton: and that’s just my v4.14/v4.15 last round of -stable bumps ? for the before and after ?
Mar 13 16:33:20 <lukma> Can somebody correct my understanding:
Mar 13 16:33:20 <lukma> TOOLCHAIN_HOST_TASK="nativesdk-packagegroup-sdk-host packagegroup-cross-canadian-qemu nativesdk-XXX"
Mar 13 16:33:33 <lukma> Is the TOOLCHAIN_HOST_TASK for SDK (-c populate_sdk)
Mar 13 16:33:33 <lukma> on the other hand
Mar 13 16:33:51 <lukma> TOOLCHAIN_HOST_TASK_task-populate-sdk-ext="     meta-environment-extsdk-qemu     "
Mar 13 16:33:51 <lukma> is for eSDK - > which shall have the former included (or add its tools) ?
Mar 13 16:34:01 <lukma> In other words - if SDK_INCLUDE_TOOLCHAIN = “1”
Mar 13 16:34:01 <lukma> then -c populate_sdk_ext will consist of elements from TOOLCHAIN_HOST_TASK and TOOLCHAIN_HOST_TASK_task-populate-sdk-ext
Mar 13 16:34:01 <lukma> is this correct ?
Mar 13 16:59:59 <rburton> zeddii_home: yeah
Mar 13 17:40:47 * kergoth beats meta-mentor, meta-sourcery, meta-external-toolchain master into yocto compliance
Mar 13 17:41:26 <Crofton> kergoth, having fun?
Mar 13 17:41:34 <kergoth> I'm debating not registering meta-mentor-staging.. its entire purpose is a staging area for general purpose bugfixes intended for upstream layer submission, I'm not sure I care if those apply based on layer inclusion rather than some other variable. would be a pain to make it all conditional given its upstream bound anyway
Mar 13 17:43:00 <kergoth> Crofton: heh, not so much :) got sick of checking TCMODE and added it to OVERRIDES
Mar 13 17:43:10 <Crofton> heh
Mar 13 17:43:16 <kergoth> TCOVERRIDES = "tcmode/external:tcmode/${TCMODE}"
Mar 13 17:43:17 <kergoth> :)
Mar 13 18:40:15 <yocti> New news from stackoverflow: Yocto with Jethro 2.0 and meta-altera <https://stackoverflow.com/questions/48669762/yocto-with-jethro-2-0-and-meta-altera>
Mar 13 18:48:55 <otavio> kergoth: isn't meta-mentor-staging in same Git?
Mar 13 18:49:08 <otavio> kergoth: if so, it is a little confusing for it to be out
Mar 13 18:50:16 <kergoth> it is, yeah. that's a good point.. maybe that repo should be broken up
Mar 13 20:06:52 <clsulliv> zeddii, zeddii_home: make-mod-scripts needs a recipe maintainer, see here https://autobuilder.yocto.io/builders/nightly-oe-selftest/builds/905/steps/Running%20oe-selftest/logs/stdio
Mar 13 20:14:11 <zeddii_home> clsulliv: I’ll send a patch!
Mar 13 20:36:05 <clsulliv> zeddii_home: thanks!
Mar 13 21:10:43 <yocti> New news from stackoverflow: Yocto: Install different config files based on MACHINE type or target image <https://stackoverflow.com/questions/49265618/yocto-install-different-config-files-based-on-machine-type-or-target-image>
Mar 13 21:46:04 <CoRfr> On build nodes, what's your average sys CPU load? I feel like it's about 15~20% sy / ~80% user / <~1% iowait on my nodes.
**** ENDING LOGGING AT Wed Mar 14 03:00:01 2018