**** BEGIN LOGGING AT Tue Dec 06 03:00:00 2016 Dec 06 06:39:27 Hello. When are anonymous python functions executed in recipes? I am searching through the ${WORKDIR} for a file in an anonymous function, and set a variable needed in later stages based on whether the file exists or not.. For this to work, the anonymous functions must be executed after do_unpack, but I guess they are not? Dec 06 06:41:22 jonatan_: I believe they are executed after parse, not during task execution Dec 06 07:16:32 nrossi, That's what I suspect as well.. Thanks! Dec 06 08:01:44 good morning Dec 06 08:22:31 rburton: Got a yocto pyton issue, see https://lists.freedesktop.org/archives/xdg-app/2016-December/000491.html (read full thread) Dec 06 08:22:49 rburton: Any idea where pip gets the --sysroot thing from and how to get rid of it? Dec 06 08:41:44 rburton: Hmm, seems to be this one: http://git.yoctoproject.org/cgit.cgi/poky/commit/meta/recipes-devtools/python/python3_3.5.1.bb?id=3e03ca7e744a9c15eab2aa794a84add65806fabf Dec 06 08:45:31 rburton: Any chance of getting that into jethro? Dec 06 08:46:14 there is no recipe that builds mkfs.vfat? :S Dec 06 08:47:55 aV_V: dosfstools Dec 06 08:48:06 alexlarsson: sigh python. will look in a moment, got a egg and bacon sandwich to eat first ;) Dec 06 08:48:40 rburton: is there a way to override do_install in a bbappend, but call the overridden one? Dec 06 08:49:24 do_install_append? Dec 06 08:50:02 yes, _append and _prepend are best from a bbappend Dec 06 08:50:26 nrossi: thanks! Dec 06 08:53:01 hello, I wonder how you manage your wi-fi connections in your qt applications? are people using libconnman-qt from OE? Dec 06 09:00:34 so meta-fsl-arm is abandoned and meta-freescale is the new hotness... ? Dec 06 09:00:44 And https://freescale.github.io/ is just old and wrong now... Dec 06 09:00:56 Ok, it's meta-fsl-arm with krogoth and "Since morty meta-fsl-arm and meta-fsl-ppc must not be used." Dec 06 09:44:31 hey, does somebody know if there is a python scipy recipe for yocto? Dec 06 09:44:44 I just found the numpy recipe Dec 06 09:47:48 https://layers.openembedded.org/layerindex/branch/master/recipes/?q=scipi says no Dec 06 10:18:33 off-topic: is it possible to view the content of a script which is a .img file? Dec 06 10:19:02 what? Dec 06 10:19:44 "bootsrc.img" I can't mount it Dec 06 10:20:18 i guess you can always open it in a text / hex editor. Dec 06 10:20:33 correct filesystem type? Dec 06 10:21:42 well, hexdump helps :d Dec 06 10:22:29 graphiqs: is a 700B text file Dec 06 10:24:04 ah kk Dec 06 10:42:45 I want a kernel module to auto load during boot by the device tree compatible string. of_device_id struct is correct and the kernel module is build and placed in the /lib/modules/kernelversion/extra folder but never gets loaded during boot. What am I missing? Dec 06 11:19:35 aV_V, you mean it failed to mount with kpartx? Dec 06 11:23:36 eduardas_m: it can't be mounted because it the content type isn't the typical filesystem type Dec 06 11:32:10 aV_V, where is this actually from? the name sounds familiar, but I'm not sure where that really is Dec 06 11:34:07 eduardas_m: from my vendor preview-software Dec 06 11:37:25 How do I get uImage instead of zImage? Dec 06 11:40:27 I'm trying to install it on a hardrive, but my bootloader wants uImage so it throws me an error Dec 06 11:41:58 http://free-electrons.com/blog/uncompressed-linux-kernel-on-arm/ Dec 06 11:42:12 haven't tried myself yet, though Dec 06 11:51:11 KERNEL_IMAGETYPE did the job :) Dec 06 12:54:33 Is there a good way to make a list of all licenses used by software in a given image? Dec 06 12:55:44 jonatan_: the manifest should have extensive information on that Dec 06 12:57:16 LetoThe2nd, My image manifest only contains the package name, version and architecture of each package included in the image, but not the license.. Maybe I need to enable this somewhere? Dec 06 12:59:32 Ah, there's a licenses/ directory with the license for each package Dec 06 14:38:30 Hello, which package is including get/set rlimit ? Dec 06 15:16:34 I am having trouble packaging the qwt library I have built with qt5 Dec 06 15:16:38 http://pastebin.com/JsUPGK3r Dec 06 15:16:58 package contains non-symlink .so: qwt-qt5-dev path.... Dec 06 15:17:46 I tried installing the library and appropriate symbolic links in usr/lib just like I do with application binaries in usr/bin Dec 06 15:18:25 apparently such an approach does not satisfy QA procedures Dec 06 15:20:02 these are the errors I am getting Dec 06 15:20:09 http://pastebin.com/SstxJRhV Dec 06 15:22:30 this is my error log: Dec 06 15:22:33 http://pastebin.com/7xgTCW1c Dec 06 15:34:17 How do you guys break dependency loops in Yocto? Dec 06 15:35:57 carefully Dec 06 15:36:16 one solution would be a bootstrap component Dec 06 15:36:23 I have added libavahi-qt5 as a patch to avahi recipe (which is a part of the upstream codebase). But now I get a loop at qt5 -> pulseaudio -> avahi -> qt5 (via the libavahi-qt5). How do I break the loop? Dec 06 15:36:38 SUCH FUN Dec 06 15:37:08 you could have a avahi-qt5 recipe that just requires the avahi.bb and only installs the qt5 library Dec 06 15:37:11 Make a new recipe (re)building avahi and cherry-pick only libavahi-qt5? Dec 06 15:37:22 heh, jinx Dec 06 15:37:27 ugly, but it works Dec 06 15:37:46 and you can require the existing recipe so you just need to rewrite a little bit - set PACKAGECONFIG to turn everything but qt5 off for example Dec 06 15:39:42 You need to trust and ensure that libavahi built in the first round is equal to the avahi build in the second, otherwise this fails in all colorful ways, I guess Dec 06 15:40:34 sveinse: thats why you start by requiring the bb file so you can't drift Dec 06 15:40:48 yes, I see that Dec 06 15:45:46 rburton: thanks Dec 06 15:56:44 YPTM armin is on Dec 06 15:58:37 YPTM Joshua is on Dec 06 15:58:51 YPTM Stephano is on Dec 06 15:59:30 YPTM is in-progress Dec 06 15:59:51 YPTM Sona is on Dec 06 16:01:19 YPTM Mark will be here soon Dec 06 16:02:22 YPTM: Saul here Dec 06 16:02:25 YPTM Randy is on Dec 06 16:02:43 YPTM: Manju is on Dec 06 16:02:56 * fray suggests Linux to Stephen ;) Dec 06 16:03:15 no critical IT mandated updates.. :) Dec 06 16:03:26 (at least ones that freeze you our of the system) Dec 06 16:03:43 eek YPTM Dec 06 16:04:33 YPTM: Richard joined Dec 06 16:04:39 (sorry I'm late) Dec 06 16:04:55 * fray notes https://github.com/Wind-River/wr-lx-setup Dec 06 16:04:58 YPTM ross on, ditto Dec 06 16:05:16 http://lists.openembedded.org/pipermail/openembedded-architecture/2016-November/000310.html Dec 06 16:05:29 "RFC BuildAppliance future" Dec 06 16:05:40 http://lists.openembedded.org/pipermail/openembedded-architecture/2016-December/000345.html Dec 06 16:05:45 the later is the build appliance Dec 06 16:06:07 YPTM: Stephen Dec 06 16:06:53 sjolley: not sure if you saw but Ross & Richard joined since you took attendance Dec 06 16:08:12 Thanks Joshua. Dec 06 16:08:49 sjolley links to setup program, for meeting minutes Dec 06 16:08:59 sjolley https://github.com/Wind-River/wr-lx-setup Dec 06 16:09:07 sjolley http://lists.openembedded.org/pipermail/openembedded-architecture/2016-November/000310.html Dec 06 16:09:15 later is the announcement email/thread Dec 06 16:12:29 http://lists.openembedded.org/pipermail/openembedded-architecture/2016-December/000345.html Dec 06 16:13:20 YPTM: is there any plan to integrate cve-check with CI/CD, and detect CVEs automatically? Dec 06 16:13:33 CVE check tool.. Dec 06 16:13:58 Thanks Dec 06 16:18:22 Sona: you're talking with rburton Dec 06 16:20:37 we just don't want to wait to the last minute to run the tool though Dec 06 16:21:12 doing it over time, and assigning the bug to the package maintainers is probably the only way to handle this.. (during a development cycle) Dec 06 16:21:40 fray: i'd like to see it run frequently yeah Dec 06 16:21:41 RP: agreed, recipe reporting system is exactly it Dec 06 16:27:35 YPTM is over. Thanks! Dec 06 17:28:42 hello, I'm working with the AM1705 processor and I'm trying to install the arago toolkit Dec 06 17:29:03 I'm following the instructions shown in http://arago-project.org/wiki/index.php/Setting_Up_Build_Environment Dec 06 17:29:25 Whilw building the following error is shown Dec 06 17:29:51 https://www.irccloud.com/pastebin/33eYAx8B/ Dec 06 17:30:53 can you please help me? Dec 06 17:35:03 cmos_dev_: you're using meta-qt5 from master but oe-core/poky from a release. checkout the right branch of meta-qt5 Dec 06 17:45:59 hello rburton Dec 06 17:46:05 how can I do that? Dec 06 17:51:33 well it depends on what version of the pieces you are using Dec 06 17:51:44 i suggest an arago specific support channel Dec 06 17:51:58 if you're using dylan then git checkout dylan in the meta-qt5 repository Dec 06 17:52:05 nobody answers in that channel Dec 06 17:52:12 (this is a problem with the arago setup) Dec 06 17:52:24 is arago still alive? Dec 06 17:52:43 I'm working wiuth an ARM9 device Dec 06 17:52:48 the example is for 1.4 which is *old* Dec 06 17:52:58 so it seems that arago is the only way to go Dec 06 17:53:06 that is what tech support told me Dec 06 17:53:10 they're wrong :) Dec 06 17:53:24 where the example says dylan, use 'morty' Dec 06 17:53:41 to use something from this year instead of three years ago Dec 06 17:54:03 rburton: yes, it's still alive and well. is poky still alive? Dec 06 17:54:36 denix: you have users complaining that nobody answers in the support channel ;) Dec 06 17:54:52 cmos_dev_: AM1705 is an old device and has been deprecated Dec 06 17:55:22 rburton: I cannot be everywhere at once Dec 06 17:55:46 rburton: also, most questions happen in the middle of my night - everyone expects I'd be around 24/7 Dec 06 17:55:47 denix: sure, wasn't intended as a dig - just curious if it was still being used by TI Dec 06 17:56:14 quick start saying 'dylan' suggests it hasn't been updated for a while. Dec 06 17:56:18 denix: this device is shown as with an ACTIVE STATUS in TI website Dec 06 17:56:49 http://www.ti.com/product/am1705?keyMatch=am1705 Dec 06 17:57:22 otherwise it should be labeled as NRND Dec 06 17:57:52 cmos_dev_: is there any software provided on that page? Dec 06 17:58:11 rburton: dylan was probably the last release that supported that device Dec 06 17:58:47 denix:yes Dec 06 17:59:06 but when I asked at the forum E2E the told me to use arago instead Dec 06 18:01:03 cmos_dev_: that's incorrect. got a link? Dec 06 18:02:57 rburton: we have very limited resources on this and I cannot cover everything myself Dec 06 18:06:31 here https://e2e.ti.com/support/dsp/omap_applications_processors/f/42/p/539396/1966911#1966911 Dec 06 18:11:38 cmos_dev_: well, according to that thread, there are no references to arago distribution, nor oe/yocto. a standalone arago toolchain is mentioned as one of the ways to build u-boot... Dec 06 18:12:57 denix: can you please tell me how can I install arago so it can be used with the AM1705]? Dec 06 18:18:05 Please take a look at http://www.ti.com/tool/linuxezsdk-sitara Dec 06 18:18:24 cmos_dev_: you'd have to use one of the older releases from http://www.ti.com/lsds/ti/tools-software/processor_sw.page (list on the bottom), as newer code base doesn't support that platform natively. Dec 06 18:18:26 it says in a big picture yocto project compatible Dec 06 18:20:02 denix: can you please give me which one of list I need to dowload? Dec 06 18:20:14 cmos_dev_: yeah, check the date of that release! May 2010... you can still use that particular SDK release, if you want. Dec 06 18:20:25 denix: can you please give me the name of the releaset I need to dowload? Dec 06 18:22:49 cmos_dev_: I thought your platform was AM1705 - there's a line on linuxezsdk-sitara page with it, called LINUXSDK-AM17X and "Get Software" button next to it Dec 06 19:08:13 Is there a way to have yocto/bitbake check that a build is reproducible (ie: check that if I build the same package twice the built package is the same)? Or use it to check this? Dec 06 19:10:42 Not that I'm aware of, you'll have to hand-roll this. Dec 06 19:11:01 There is buildhistory, which has a list of files you could diff for modifications, but it doesn't contain checksums or anything Dec 06 19:11:41 Also note your build environment may change between builds due to the use of a shared sysroot area. There is work going on to change this. Dec 06 21:38:33 anyone have a recommended os update mechanism for x86-based yocto systems? it appears many of the existing solutions assume uboot, which appears to not have great x86 support (is this true?) Dec 06 21:39:11 any experience with mender, swupd, or swupdate would be appreciated Dec 06 21:49:31 miceopede: swupdate could work if you come up with a way to check a boot pointer of some kind in grub or whatever bootloader you're using and boot different kernel, rootfs copies based on it. Dec 06 21:51:30 Basically you just need a way to check whether slot A or B is active and use kernel/rootfs from that slot and then you need to have swupdate switch the active slot after the upgrade succeeds Dec 06 21:52:37 georgem: this would be implemented as some sort of custom postinstall script after swupdate completes to switch the grub entry? Dec 06 21:55:11 well, if you have a separate common boot partition with the grub configuration in it you could probably have a swupdate postinstall script just change that. Just keep in mind what could happen if power is lost at any point in that process and if you could end up with an empty grub configuration for instance. Dec 06 21:57:40 georgem: got it. is a 'mv' (rename()?) atomic? could this be implemented as a simple mv new_grub.cfg grub.cfg? Dec 06 21:58:39 i suppose it depends on the filesystem Dec 06 21:58:41 miceopede: I'm not certain. Could even depend on the filesystem. I'd have to do some research to know for certain. Dec 06 21:59:58 i see. how does uboot solve this problem? there are two configuration partitions? Dec 06 22:04:01 miceopede: swupdate is setup to store that boot pointer in the u-boot environment. From swupdate docs "Is it safe to change U-Boot environment ? Well, it is, but U-Boot must be configured correctly. U-Boot supports two copies of the environment to be power-off safe during a an evironment update. The board’s configuration file must have defined CONFIG_ENV_OFFSET_REDUND or CONFIG_ENV_ADDR_REDUND. Check in U-Boot documentation for these Dec 06 22:04:01 constants and how to use them." Dec 06 22:05:07 i see. i'll look into how hard it is to bring up uboot on a general x86-based system... Dec 06 22:05:41 looks like it's been ported to a quark, maybe edison. i'm trying to target both atom and a regular desktop-based x86 pc Dec 06 22:06:00 georgem: thanks for your help. any last sage advice? Dec 06 22:07:20 miceopede: I think you may be able to run u-boot from EFI but I haven't done that. On our Atom E3845 based system we're running u-boot natively but not very many x86 platforms support that. You may also check out this presentation on update technologies: https://youtu.be/pdHV9H9nZks?list=PLbzoR-pLrL6pRFP6SOywVJWdEHlmQE51q Dec 06 22:08:16 georgem: i used to work closely with matt porter. i'll contact him. thanks! Dec 06 22:08:25 OSTree looks promising but getting it all working looks like quite a chore. Dec 06 22:08:27 Cool Dec 06 22:23:34 on the bbb, is 5 volts on the ring or tip of power cable? Dec 06 22:23:36 cder Dec 06 22:27:33 Crofton: i would guess it's +5v center positive Dec 06 22:27:38 usually ground is on the ring Dec 06 22:27:59 i'd check with a meter though, i've blown my fair share of boards by using the wrong polarity Dec 06 22:32:18 "WARNING: Host distribution "ubuntu-16.04" has not been validated..." Dec 06 22:32:49 Given the above is an LTS, I wonder why we still exclude it. I've been using it a crapload and not had issues... Dec 06 22:33:32 paulg: i'd make sure it's not a case mismatch. they lowercased the string coming from the lsb release, not sure if they also lowercase the sanity tested distros var when checking it.. Dec 06 22:33:34 * kergoth shrugs Dec 06 22:46:04 I try to create a custom image type and use IMAGE_TYPEDEP_mytype = "tar.xz", but it seems not to follow that Dec 06 22:46:59 When IMAGE_CMD_mytype gets executed, the *rootfs.tar.xz file is not there yet... Dec 06 22:47:12 This happens on morty... Dec 06 22:48:40 I do inherit image_types... Dec 06 23:02:22 kergoth, I'll check that... Dec 06 23:02:26 miceopede, thanks, that is correct I meant to type that in another channel :) Dec 06 23:24:33 Found it, the image is ready, but only in IMGDEPLOYDIR, not (yet) in DEPLOY_DIR_IMAGE Dec 06 23:37:30 kergoth, ya you are onto something ... Dec 06 23:37:49 meta-poky/conf/distro/poky.conf has ubuntu-16.04 but the regex matching somehow fails Dec 06 23:38:23 but the case in the warning matches the entry in the above file... Dec 06 23:38:30 more debugging required... Dec 06 23:41:54 curious. "-e" shows... Dec 06 23:41:57 SANITY_TESTED_DISTROS=" overc-1.0 \n Ubuntu-12.04 \n Ubuntu-14.04 \n Ubuntu-14.10 \n Fedora-20 \n CentOS-6.5 \n Debian-7.* \n Debian-8.* \n openSUSE-project-12.3 \n openSUSE-project-13.1 \n " Dec 06 23:42:11 ignore the overc ; that is from a local append. Dec 06 23:43:37 oh wait. I bet the append isn't doing a += Dec 06 23:45:37 ya, my fault ; local dain-bramage. Dec 06 23:46:08 sometimes you need -e to remind yourself of local wreckage. **** ENDING LOGGING AT Wed Dec 07 03:00:00 2016