**** BEGIN LOGGING AT Tue Feb 10 02:59:58 2015 Feb 10 06:39:20 what is poky ? poky is not yocto? Feb 10 06:54:54 miandonmenmian: http://www.yoctoproject.org/docs/1.6/dev-manual/dev-manual.html#yocto-project-terms Feb 10 06:56:00 miandonmenmian: Some of the terms are mixed, but for simplicity poky is yocto Feb 10 06:57:15 the build system as in the layers and recipes? Feb 10 07:03:57 miandonmenmian: thats where it can start to become blured and confusing. But yes poky is a build system Feb 10 08:04:20 Hi Feb 10 08:29:11 RP: hey, my yocto-master builds started to be flaky. every second build my kernel fails to build and I get "cp: cannot stat `include/config/kernel.release" Feb 10 08:29:31 RP: is there any documentation how a kernel/linux bb file needs to be modified for the shared work directory? Feb 10 08:59:05 zecke: the idea was we'd minimised the modifications necessary. It sounds like some kind of race over the file, which task is failing? Feb 10 09:42:33 Can someone help me ? I still got an error while compiling linux-yocto : it complains it can't find some .patch files while doing do_patch(). FYI I use kernel 3.14 + preempt-rt for a beaglebone Feb 10 09:42:43 Here is the complete log : http://pastebin.com/9HKuYXeS Feb 10 09:42:54 My guess is that it has something to do with the latest update to v.3.14.29 Feb 10 09:44:09 morning all Feb 10 09:45:20 FrankSansC: so, presumably the file it's indicating genuinely does not exist? is it in some other path? Feb 10 09:47:16 I'm checking... Feb 10 09:49:16 Nope, the two patches files "arm_versatile_926ejs-kick-off-PrimeCell-RTC-PL031.patch" and "arm_versatile_926ejs-add-default-nor-flash-mtd-parts.patch" are nowhere to be found Feb 10 09:53:43 RP: http://paste.lisp.org/display/145718 Feb 10 09:53:55 FrankSansC: does your kernel recipe refer to these? Feb 10 09:53:57 RP: so compile succeeded, which starts do_compile_kernelmodules and do_shared_workdir Feb 10 09:57:15 zecke: at that point, do_configure should have run and the kernel should be configured. The only way that file might not exist if if something is reconfiguring the build? Feb 10 09:58:58 RP: I have a custom do_configure that copies a defconfig and runs oldconfig. Can this be an issue? Feb 10 10:00:27 RP: and I have ancient rules for do_compile_perf, do_install_perf. But these don't run configure the kernel tree Feb 10 10:01:29 bluelightning: Nope, these two patches are referenced here : poky/build/tmp/work-shared/beaglebone/kernel-source/.meta/patches/standard/preempt-rt/base/series Feb 10 10:05:48 zecke: the directories for source and build have been split out, perhaps its configuring in the source dir rather than the build dir? Feb 10 10:06:34 FrankSansC: which branch/release are you on currently? Feb 10 10:07:48 bluelightning: master branch of poky Feb 10 10:08:14 kernel 3.14 Feb 10 10:10:36 just running a quick test here Feb 10 10:12:27 and kernel type "preempt-rt" Feb 10 10:14:52 RP: I annotated the paste (http://paste.lisp.org/display/145718#1). It copies the .config to the build dir and it should generate the output in the build dir as well Feb 10 10:15:20 RP: my build failed tonight but I don't think there was a kernel change on either side. Feb 10 10:16:13 FrankSansC: how are you selecting that? Feb 10 10:16:23 (the preempt-rt kernel) Feb 10 10:18:42 bluelightning: I've got a "linux-yocto_3.14.bbappend" and I've added these two lines : LINUX_KERNEL_TYPE = "preempt-rt" and KTYPE = "preempt-rt" Feb 10 10:22:10 It was working/compiling very fine until a few days ago Feb 10 10:22:45 I haven't changed anything, just updated my poky repository Feb 10 10:24:49 FrankSansC: ok, could you please file a bug about this in our bugzilla? Feb 10 10:24:57 zecke: it looks correct. What does the compilemodules log output look like? Any sign it reconfigures? Feb 10 10:26:51 bluelightning: in which category should I fill the bug ? Kernel/linux-yocto ? Feb 10 10:27:58 there is a patch on oe-core ML for kern-tools (rt-preempt). Maybe this? Feb 10 10:31:09 FrankSansC: yes Feb 10 10:32:57 ant_work: interesting ! I'm going to try this Feb 10 10:38:32 that does look like it might help Feb 10 10:43:09 I see there is a layer called meta-jetson-tk1. Any known plans for more official support for tegra? Feb 10 10:49:20 bluelightning: that's definitely helping ! It's compiling again :) Feb 10 10:49:28 ant_work: works for your advice Feb 10 10:49:39 -works +thanks Feb 10 10:50:50 yw Feb 10 10:53:41 hi, how can you set different kernels for different images in the same checkout? seems like the kernel is selected for all images on a machine Feb 10 10:56:52 icanicant: you can't for the same machine, without separate configurations / TMPDIRs Feb 10 10:59:53 bluelightning: thanks, would be useful when building two very different products on the same hardware. Feb 10 11:00:22 bluelightning: i guess i can override in local.conf and just switch local.conf when switching between images/applications? Feb 10 11:01:13 icanicant: you can do that yes Feb 10 11:01:48 bluelightning: thanks Feb 10 11:30:00 Hey all Feb 10 11:30:55 If I'm creating a native recipe that makes a tool, e.g. a shell script that will be used elsewhere, these all end up (typically) in tmp/sysroots/x86_64-linux/usr/bin right? Feb 10 11:31:27 but that's not in the path for a normal build env but can be found by other recipes to use? Feb 10 11:42:23 pev, yes, scripts/binaries in the hosts sysroot can be used in other recipes. The hosts (build machine's) sysroot will be in your path Feb 10 11:42:49 s/hosts/native/ to improve the terminology Feb 10 11:43:37 AndersD: Thanks. That's the path when running bitbake though yes? I'd need to munge PATH manually to test by hand? Feb 10 11:44:20 Well, you can always use `bitbake -c devshell` to get a shell using the same environment as bitbake uses. Feb 10 11:44:46 Ah, of course... Doh! Feb 10 11:44:48 In this case, should be one of the recipes that uses your script Feb 10 12:59:34 Hi everyone Feb 10 13:02:02 I want to build an sdcard image for a IMX6, I would like to try with wic (as I need to flash 2 images on the same sdcard) but I don't know how to flash the uboot (as the uboot has to be flash in row mode just after MBR) Feb 10 13:25:33 RP: I will have a look Feb 10 13:28:19 Aurele__: Have you looked at how it's done in fsl-community-bsp? If you just create an sdcard image you can dd onto the board? Or do you mean for wic specifically? Feb 10 13:28:30 RP: it is a multimachine build as well. I first build for the BTS and then the BSC (one is armv5te the other x86) Feb 10 13:28:37 It certainly does all the uboot cleverness for you. Feb 10 13:28:49 pev, there is a bbclass for creating sdcard Feb 10 13:29:48 RP: in the end there is a build/include/config/kernel.release file inside what I think is the build directory Feb 10 13:30:36 pev, I will use bbclass for now as I don't know anything on wic, but I wanted to know if someone worked on this before starting to work with bbclass Feb 10 13:35:45 RP: the kernel.release and the run.do_shared_workdir file got created more or less at the same time Feb 10 13:35:50 RP: so something races Feb 10 13:37:42 RP: do_compile runs kconf/silentoldconfig again Feb 10 14:40:38 Hi, i'm trying to modify root password in a image recipe. Is there a common way to do that ? I'm trying with do_rootfs_append() but i got an ExpansionError. Feb 10 14:46:57 https://wiki.yoctoproject.org/wiki/FAQ:How_do_I_set_or_change_the_root_password Feb 10 14:51:11 paulg_: Thank you, that is the page I was searching for 2 days… Feb 10 14:56:55 odd, I put "yocto default root passwd" into google and it came up straight away.... ? Feb 10 15:02:38 My search engine show me a mailing list entry with a similar solution but without the do_root_fs() function declaration. Feb 10 15:04:43 (duckduckgo) Feb 10 15:21:36 hi Feb 10 15:21:52 after one day reading the yocto docs i came here to cry ! Feb 10 15:24:00 i came up after without knowing how to set it up so i can cross compile Qt "5.4" on my IMX.6 device, have not been as frustrated as i am since 10 years or more :( Feb 10 15:25:18 ohmy: perhaps you can outline the problem you are having? Feb 10 15:25:37 bluelightning: Hello Feb 10 15:29:45 bluelightning: Sorry for looking so upset, i'm just stuck to undersatnd the installation doc, it looks so complicated Feb 10 15:30:06 which installation doc is that just so I'm following? Feb 10 15:32:04 bluelightning: the official one http://www.yoctoproject.org/docs/1.7.1/adt-manual/adt-manual.html but since i'm reading other documents in parallel, for my specific imx6 device i can't get the right steps Feb 10 15:32:49 I'm not entirely sure that that is the method I would suggest for building Qt itself Feb 10 15:33:59 here's something perhaps relevant to your target platform: http://wiki.wandboard.org/index.php/Building_Qt5_using_yocto_on_Wandboard Feb 10 15:34:01 bluelightning: the docs are spread out different piece of docs, did not find any strainght doc explaining how to set up yocto itself and prepare it for Qt building. Feb 10 15:34:30 our getting started document is this one: http://www.yoctoproject.org/docs/current/yocto-project-qs/yocto-project-qs.html Feb 10 15:34:56 bluelightning: i'm not expecting to find someone that have already made all the job for me, but i've been using ltib (and other similar tools) for ling time now and i still did not figure out how to start with yocto Feb 10 15:36:59 ohmy: the quick start guide is the one you want, if there's something you can't follow in there by all means ask questions Feb 10 15:38:02 bluelightning: ok, i'm a new user Feb 10 15:41:28 bluelightning: the docs says git checkout -b dizzy origin/dizzy for example than source oe-init-build-env than bitbake adt-installer Feb 10 15:41:50 yes Feb 10 15:42:31 bluelightning: at this step the Build Configuration choose TARGET_SYS = "i586-poky-linux" and MACHINE = "qemux86" while i want to an armv7 toolchain. Feb 10 15:43:33 the thing is that all the documentation is constructed in that way, you run different command, and later in the documentation you'll find that you have to set some env variables/modify some files to select the right target etc Feb 10 15:43:46 that is the default yes... you can add layers on and set MACHINE to build for something different Feb 10 15:44:19 bluelightning: as i said, i'm new, at this level i do not (the target audience) do not know what layers are ! Feb 10 15:44:45 so you've started with the ADT documentation, I'm not quite sure how you got there Feb 10 15:44:45 Hi all ! Feb 10 15:45:38 ohmy: the problem is it's a little difficult for us to make generic documentation that also covers everything needed for specific hardware platforms that aren't directly supported by the core platform e.g. imx6 Feb 10 15:47:57 bluelightning: I completely agree, i've been using Windriver Linux since long time and they've been trying to seel their "layer" concept, we ended up getting completely confused for a lot of reasons that are out of topic. Feb 10 15:48:21 bluelightning: and i agree about how hard is to write documentation Feb 10 15:50:36 I'm generating a few different images for my board and have a native tool that combines them to an archive. What's the best way to achieve that in the build system, are there any obvious example recipes I can adapt? (i.e. pulling images from deploy dir and doing stuffand spitting out again...) Feb 10 15:51:21 I'm scratching my head a bit now...! Feb 10 15:54:23 YPTM: armin is on Feb 10 15:57:52 hmm, someone at WR finally changed the on hold music Feb 10 15:58:39 lol Feb 10 15:58:53 that music last time was aweful.. Feb 10 15:59:06 * fray is on another meeting.. I may not make the call Feb 10 15:59:22 it was kind of like a sales presentation video soundtrack from the late '80s Feb 10 15:59:28 YPTM: Ready-Access Number: 8007302996/9139049836 Access Code: 2705751 Feb 10 15:59:35 Morning all: YPTM: Saul is on Feb 10 16:00:15 * armpit wonders if the music is available on iTunes Feb 10 16:00:40 YPTM: Paul Eggleton is on Feb 10 16:00:40 YPTM: Stephen has joined Feb 10 16:00:50 YPTM: Michael here. Feb 10 16:01:12 YPTM: Richard joined Feb 10 16:01:26 YPTM: Denys is on Feb 10 16:01:32 YPTM: Ross joined Feb 10 16:02:07 bluelightning: 4 simple letters: dash Feb 10 16:02:33 let's just say I opened a can of worms Feb 10 16:02:56 YPTM: Joe Mac here. Feb 10 16:06:35 YPTM: Bruce Ashfield on the call. Feb 10 16:06:39 sgw_: can we not just force that script to bash ? Feb 10 16:06:45 bluelightning: Giving up for today, trying annother day, thanks for the discussion Feb 10 16:07:01 ohmy: anytime Feb 10 16:09:09 YPTM: cristian.iorga joined Feb 10 16:10:59 RP: tried that, still seems to fail, I also found that lsb_release will give interesting input to our oe/lib checking code Feb 10 16:17:10 YPTM is over. Feb 10 17:53:02 BSP documentation says to use bcfg in EFI shell to add a device to the boot manager. Unfortunately this command isn't available in EFI shell on my MinnowBoard MAX. Why would that be? Feb 10 17:54:23 dvhart: ^ Feb 10 17:54:33 -ENODVHART :) Feb 10 17:54:40 :-) Feb 10 17:54:51 bcfg is an optional part of the UEFI specification as I understand it Feb 10 17:55:01 so it is there for MinnowBoard, but not on MinnowBoard MAX Feb 10 17:55:14 the BIOS menu however will allow you to specify boot order Feb 10 17:55:19 (which MinnowBoard did not have) Feb 10 17:55:25 oh Feb 10 17:55:25 so from the shell, just type exit Feb 10 17:55:44 dvhart: thx! Feb 10 17:55:54 then navigate to the boot manager configuration, etc etc. Be sure to save changes via the on screen menu, and not F10 - which doesn't work, even though it says it does Feb 10 17:55:55 grmble Feb 10 17:56:00 does anyone know if yocto supports mklibs? Feb 10 17:56:33 excuse me while I go cough up a lung Feb 10 17:56:45 dvhart: nooooooooooo Feb 10 17:56:48 dvhart: You can't be sick :) Feb 10 17:57:51 * ulf` throws cough drop at dvhart Feb 10 17:58:05 realBigfoot: yes, AFAIK we run mklibs as a matter of course when generating an image - see meta/classes/image-mklibs.bbclass Feb 10 17:59:40 dvhart: edit initialization failed Feb 10 18:00:24 bluelightning: nice thanks Feb 10 18:20:22 I just got screwed by avahi-ui because it was settting S to avahi _after_ sourcing ("require") avahi.inc Feb 10 18:20:41 once I moved it _before_ the require avahi.inc it then works. Feb 10 18:20:59 I just can't figure out how this ever worked before for anyone else.... :-/ Feb 10 18:21:19 at what point is ${S} expanded? Feb 10 18:21:34 every variable is expanded when it's used Feb 10 18:21:39 unless it's forced earlier by e.g. a := Feb 10 18:22:05 this is how gmake works, and that's where we got it from Feb 10 18:22:35 paulg_: sounds like avahi.inc must be expanding it at parse time with a := or so Feb 10 18:22:48 OK, so I'm still at a loss why this worked yesterday and broke today ; when nobody touched the pkg since Oct 2014. Feb 10 18:23:44 don't you just love problems like that? :) Feb 10 18:24:18 yeah, totally. :-/ Feb 10 18:24:44 Guess I'll just send a patch even though I've NFI what the root cause was. Feb 10 18:29:05 RP: okay. The file gets created during compilekernelmodules Feb 10 18:29:41 RP: I used fstat to look at the file and inode number. This is a 3.10er kernel Feb 10 18:41:26 hey all, I am trying to add a bitbake recipe for Apache CouchDB, when I go to configure though I get the error: hecking whether we are cross compiling... configure: error: in; and configure: error: cannot run C compiled programs. Feb 10 18:41:39 Anyone see something like this before? I've been trying to search google Feb 10 18:42:08 that's typical. AC_TRY_RUN() or AC_RUN_IFELSE() macros can't be used when cross-compiling, as yo ucan't run the binaries it just compiled. Feb 10 18:42:21 if it's doing it just as a sanity check, you can surround it with a check of the cross_compiling variable Feb 10 18:43:01 if it's doing it to get real information to use in the build, then you'll have to ensure that it's wrapped in use of the configuration cache (AC_CACHE_CHECK? something like that), so that we can pre-supply a cache value to tell it about the target platform, since it can't get the information itself Feb 10 18:46:29 so my other question is though, EXTRA_OECONF I supply --host, but it still tells me: If you meant to cross compile, use `--host'. Feb 10 18:46:58 that one is harmless Feb 10 18:47:03 and yo ushould *not* be pasisng —host in EXTRA_OECONF Feb 10 18:47:05 we pass all that already Feb 10 18:47:15 ah, okay Feb 10 18:49:01 it's called extra for good reason, it's just extra things. e.g. —enable/--disable/--with/--without options Feb 10 18:49:39 so I am trying to grep for the AC_TRY_RUN or AC_RUN_IF_ELSE, and I cant seem to find them Feb 10 18:50:42 let me trace through the error log Feb 10 18:50:47 checking configure.in/configure.ac is insufficient, as it can be in a .m4 file as well Feb 10 18:50:58 there are possibly other causes, but that's the most common Feb 10 18:59:20 kergoth, do you have any recommendations for hunting down which ones would be causing this? I found AC_RUN_IFELSE in several files, however they are surrounded by AC_CACHE_CHECK Feb 10 18:59:40 difficult to say without seeing the full do_configure log Feb 10 19:01:45 kergoth, doesn't look like theres an abundance of information: https://gist.github.com/Matulis/11ef8b93e13d6e59a0fd Feb 10 19:03:22 oh, i see, i misunderstood the initial message Feb 10 19:03:50 i'd suggest reading config.log in the build directory Feb 10 19:13:25 sigh, meta-selinux seems like it's always broken one way or another Feb 10 19:13:45 not only is it still missing version bumps, but it appends to lxc without a LAYERDEPENDS indicating a requirement on meta-virtualization Feb 10 19:13:54 either it need sthe dep, or it needs to move the bbappend into a layer-specific area Feb 10 19:26:12 armpit: FYI the current build on the AB is dizzy-next: https://autobuilder.yoctoproject.org/main/waterfall Feb 10 19:27:08 k Feb 10 19:36:34 Does relocate_sdk.py know how to resize the sections / binary to make room if necessary, or is it dependent upon recipes-core/glibc/glibc/relocatable_sdk.patch? Feb 10 19:36:49 (Dealing with an external toolchain that likely doesn't have it applied) Feb 10 19:37:18 if it doesn't, maybe I could ship a static patchelf for it to use instead Feb 10 19:37:20 hmm Feb 10 20:09:18 about one build in 10, I get this... Feb 10 20:09:21 | make[1]: *** No rule to make target '/home/paul/poky/build/tmp/sysroots/intel-corei7-64/usr/lib/gcc/x86_64-poky-linux/4.9.1/include/stdint.h', needed by 'perlmini.o'. Stop. Feb 10 20:09:21 | make[1]: Leaving directory '/home/paul/poky/build/tmp/work/corei7-64-poky-linux/perl/5.20.0-r1/perl-5.20.0' Feb 10 20:09:21 | Makefile:37: recipe for target 'perl' failed Feb 10 20:09:54 restart and it goes away ; so we are missing a build dep on whoever populates the sysroot with that header, I guess. Feb 10 20:10:33 I typically build with -j20 and || pkgs = 20. Feb 10 21:01:23 paulg_: there is an open bug on that Feb 10 21:01:58 paulg_: https://bugzilla.yoctoproject.org/show_bug.cgi?id=7287 Feb 10 21:02:01 Bug 7287: normal, Medium+, 1.8 M3, richard.purdie, NEW , perl build race with gcc for the target Feb 10 21:02:16 * paulg_ goes to look. Feb 10 21:04:38 kergoth: relocate_sdk won't help for that unfortunately but I did make patchelf work with the uninative class for something like that Feb 10 21:10:10 armpit: patch sent for dizzy yocto tiny issue Feb 10 21:13:50 sgw_, thanks. will queue it up Feb 10 23:58:28 Hi guys! Question. When using vardeps on do_install, shoudn't this function be called everytime the variables defined in vardeps changes? Feb 10 23:59:17 I have an install function and want to do stuff based on a variable called BUILD_TYPE Feb 10 23:59:23 do_install[vardeps] += "BUILD_TYPE" Feb 10 23:59:43 For whatever reason this doen't work as expected. Is there something I miss? Feb 11 00:03:52 BUILD_TYPE is defined in distro file Feb 11 00:05:48 THere must be something I don't understand or just ran into a very annoying bug Feb 11 00:07:03 i know bluelightning can help :) Feb 11 00:10:36 Grrr. Default mailing list settings are still mangling CC lines via garbage reply-to insertions. Feb 11 00:11:00 the reply to my coreutils series from fray never made it in my inbox. Feb 11 00:11:34 dvhart and I went through this a while ago and IIRC, we thought we'd got the defaults set to sane (i.e. non mangling) settings. :-/ Feb 11 00:14:25 dvhart, do you recall what we got changed at the time? It was so long ago, I'm goldfishing on the whole thing; aside from thinking you were involved in the discussion... Feb 11 00:16:36 agherzan: hi... what's up? Feb 11 00:16:51 hrm.... don't recall off the top of my head, I did think that got fixed though. Feb 11 00:17:15 paulg: you can change that in the settings for your subscription, I have Feb 11 00:18:16 bluelightning: i'm pretty confused with vardeps. I want to do stuff in do_install based on a custom variable defined in distro file. For whatever reason the install is not forced when i modify the custom variable Feb 11 00:18:37 bluelightning: check above for the concrete example Feb 11 00:19:16 to be a little more specific i inject that distro variable in distro features Feb 11 00:19:31 so the checks in do_install all done checking DISTRO_FEATURE Feb 11 00:19:38 FEATURES* Feb 11 00:19:40 bluelightning, ok -- that may be true but IMHO the default setting should not violate the principle of least surprise. i.e. I've never come across this behaviour in 20+ years except for here. Feb 11 00:20:15 bluelightning: do_install[vardeps] += "DISTRO_FEATURES" Feb 11 00:21:09 ... and per discussion with dvhart, I _thought_ we all agreed a while back to make a global change to fix this... Feb 11 00:21:52 can't immediately find it in the mail archives, so perhaps it was done here in IRC? Feb 11 00:23:04 bluelightning: http://pastebin.com/8AFxb4FZ Feb 11 00:23:13 bluelightning, btw - I stuck you on the cc of the coreutils patches since you'd guided me on how best to fix the same thing with git manpages. Feb 11 00:23:22 (in case that wasn't obvious...) Feb 11 00:35:27 agherzan: I don't know - as far as I can see that should work just fine Feb 11 00:38:17 for whatever reason it doesn't... bluelightning Feb 11 00:38:26 anybody can help me with any hints? Feb 11 00:55:59 agherzan: maybe look at bitbake-dumpsig on the siginfo/sigdata file for the task? Feb 11 01:08:13 bluelightning: how do get the right sig file from staging? Feb 11 01:08:21 i find multiple Feb 11 01:08:33 for install on the respective package Feb 11 01:09:34 agherzan: it'll be the newest one Feb 11 01:13:34 bluelightning: ok - let me try] Feb 11 01:14:58 heading to bed, if you don't figure it out send a mail to the list Feb 11 01:16:21 ...and watch the CC line! :-) Feb 11 01:30:11 what are best practices for packaging up prebuilt binaries in recipes? **** ENDING LOGGING AT Wed Feb 11 02:59:58 2015