**** BEGIN LOGGING AT Mon Nov 02 02:59:59 2015 Nov 02 05:24:53 hi, I’m following the Intel Edison BSP user guide and trying to compile my first image. When I run `bitbake edison-image`, I get an error: Nothing PROVIDES 'edison-image'. Any idea what could be wrong? Nov 02 06:22:34 sqd: the imgae just isn't called like that :) Nov 02 06:22:50 sqd: which instructions are you following? Nov 02 06:28:05 LetoThe2nd: http://www.intel.com/support/edison/sb/CS-035278.htm BSP User Guide Nov 02 06:29:16 LetoThe2nd: already tried core-image-minimal instead, but that seems to generate an rpm-based system that meta-intel-edison/utils/flash/postBuild.sh can't process Nov 02 06:31:48 sqd: hm, this ./meta-intel-edison/setup.sh is something bsp specifc that i don't have experience with. Nov 02 06:32:28 LetoThe2nd: that just sets up an x86 qemu environment where i can run bitbake from, if i'm not mistaken Nov 02 06:33:16 sqd: no, usually such setup scripts set up all the needed layers and some minor (or not so minor) configuration stuff Nov 02 06:33:31 qemu shouldn't be involved at all. Nov 02 06:34:29 i see Nov 02 06:40:00 my first guess would be that some layer inclusion is missing Nov 02 06:40:48 let me have a look at the bsp :) Nov 02 06:41:08 cool, here's the meta-intel-edison repo if it's helpful http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel-edison/tree/ Nov 02 06:43:08 i'd prefer to go with the documentation, and that tells me to get edison-src from the intel server Nov 02 06:44:34 right. if i'm not mistaken that's the "Sources – Linux Source Files" link at https://software.intel.com/iot/hardware/edison/downloads Nov 02 06:44:51 (the only file that starts with edison-src) Nov 02 06:49:12 well it starts to build edison-image.... Nov 02 06:50:11 LetoThe2nd: ok, then something went wrong at my system, thank you for testing it Nov 02 06:50:53 sqd: the instructions seem to miss one step: after 3, before 4. i had to cd into out/linux Nov 02 06:51:20 because there are the build and poky directories that get referenced in 4. Nov 02 06:51:28 besides that, its running at the moment :) Nov 02 06:52:50 i'm trying again with a new download / dir Nov 02 07:05:29 LetoThe2nd: same here, thanks you! helped a lot Nov 02 07:05:48 sqd: have fun! Nov 02 07:28:25 Hello Nov 02 07:29:02 I need to build meta-toolchain-qt5 with graphics driver Nov 02 07:29:41 How to achive build meta-toolchain-qt5 with graphics driver Nov 02 07:37:23 panda_: any reason why you'd want meta-toolchain-qt5 instead of the standard image-derived sdk? Nov 02 07:56:34 because I need to GPU driver Nov 02 07:58:54 panda_: that doesn't answer my question. and a standard sdk would offer the whole library set available on the target. Nov 02 07:59:29 ok I send log file Nov 02 07:59:34 please wait Nov 02 08:03:29 first I run this command Nov 02 08:03:58 bitbake meta-toolchain-qt5 Nov 02 08:06:23 sorry, but would you mind reading what i wrote, and think about it for a moment? Nov 02 08:07:03 so again: "are there reasons wha you want meta-toolchain-qt5 instead of the standard sdk?" Nov 02 08:10:05 standard sdk not inclued qt cross toolchain Nov 02 08:10:16 for example qmake Nov 02 08:11:29 it would take me a few hours to verify that, but i doubt that thats correct. Nov 02 08:12:28 I see this error "qtwebkit requires libGLESv2.so but no providers in its RDEPENDS Nov 02 08:13:06 morning all Nov 02 08:13:46 panda_: so, is the problem actually related to the toolchain generation, or don't you have a working GLES driver for your platform altogehter? Nov 02 08:14:16 yes problem is related to toolchain generation Nov 02 08:14:29 my platform use GLES Nov 02 08:14:45 panda_: but you can actually build an image for your target with gles, and qtwebkit is working there? Nov 02 08:15:03 yes problem is related to toolchain generation Nov 02 08:15:41 I don't know how to set libGLESv2.so providers in toolchain Nov 02 08:19:13 yes my target wit works gles Nov 02 08:19:26 * panda_ slaps sdh11 around a bit with a large fishbot Nov 02 08:22:17 LetoThe2nd are you here ? Nov 02 08:22:43 panda_: i am Nov 02 08:22:57 ok Nov 02 08:23:09 I don't know how to set libGLESv2.so providers in toolchain Nov 02 08:23:34 my target platforms fine GLES driver Nov 02 08:23:35 have read that, but i don't have a proper answer as i still thing the meta-toolchain-qt5 route isn't the best choice. Nov 02 08:23:43 s/thing/think/ Nov 02 08:23:59 please give me advice Nov 02 08:24:14 how to create full sdk Nov 02 08:24:42 - cpopulate_sdk when you build your image Nov 02 08:24:52 -c populate_sdk Nov 02 08:25:25 I try that command online create lib and c toolchain Nov 02 08:25:49 not inclued full qt sdk and toolchain Nov 02 08:26:51 i don't know it, then. Nov 02 12:32:31 hm. git on yocto daisy seems broken Nov 02 12:58:57 qknight: broken how? Nov 02 12:59:38 bluelightning: like so: Nov 02 12:59:40 no changes added to commit (use "git add" and/or "git commit -a") Nov 02 12:59:40 root@t4240rdb:/squirrel/nixpkgs# git diff pkgs/development/compilers/sbcl/bootstrap.nix Nov 02 12:59:43 Aborted Nov 02 12:59:46 git: xdiff/xutils.c:358: xdl_hash_record: Assertion `*ptr == '\n'' failed. Nov 02 12:59:48 root@t4240rdb:/squirrel/nixpkgs# echo $? Nov 02 12:59:51 134 Nov 02 13:00:01 hmm, not good Nov 02 13:00:45 daisy is no longer officially supported by the project though unfortunately Nov 02 13:01:25 bluelightning: ah ok Nov 02 13:01:45 bluelightning: i don't think that this is mission critical Nov 02 13:01:50 bluelightning: just wanted to report Nov 02 13:01:53 ok Nov 02 13:02:08 we'd probably take the patch if you were able to come up with a fix Nov 02 13:08:00 bluelightning: https://dev.openwrt.org/changeset/37239 Nov 02 13:08:24 bluelightning: i probably should try that patch Nov 02 13:08:51 https://dev.openwrt.org/ticket/13611 Nov 02 13:25:32 qknight: interestingly this got brought up last year on the mailing list: Nov 02 13:25:40 http://lists.openembedded.org/pipermail/openembedded-devel/2014-September/098024.html Nov 02 13:25:50 I wonder if this is still unfixed on master Nov 02 13:36:20 hi, i've tried to compile btusb for edison by changing .config and running `bitbake virtual/kernel -c compile_kernelmodules`. i copied btusb.ko to the edison and tried to load it, but got this error: "modprobe: ERROR: could not insert 'btusb': Exec format error". does this mean my cross-compilation will not work for the kernel precompiled by intel, or can i fix it? Nov 02 13:36:28 dmesg says: [ 354.273889] btusb: version magic '3.10.17-yocto-standard SMP preempt mod_unload ATOM ' should be '3.10.17-poky-edison+ SMP preempt mod_unload ATOM ' Nov 02 13:59:11 bluelightning: https://dev.openwrt.org/browser/packages/net/git/patches/200-disable_fasthash.patch?rev=37239 fixed the issue Nov 02 14:49:18 Hi, I built a toolchain which I use for cross compiling, but the binaries it produces point to the wrong program interpreter and won't run unless I symlink the proper interpreter to the one the binaries are looking for, I'm having a hard time figuring out why this is Nov 02 14:49:55 then you are likely using the wrong ABI settings, or a compiler that was not configured properly. If it's an OE built-toolchain, did you source the environment files before using it? Nov 02 14:50:05 sometimes the default does not match what the environment files setup Nov 02 14:53:04 I sourced the environment yes, and I built the toolchain myself for some custom work Nov 02 14:53:40 I'm not sure yet how the ABI settings come into play, but my project requires -mfloat-abi=softfp Nov 02 14:54:26 Where would I find any docs on how to fine tune the toolchain? Nov 02 14:56:45 This maybe a long shot, but is there maybe an OE built-toolchain for the Freescale sabrelite board? I have not come across it if there is Nov 02 15:16:46 my connection crashed... did i miss a reply to my question about compiling the btusb module? Nov 02 15:19:28 sqd: I don't think so but it would appear there is some sort of mismatch between the kernel you're building and the kernel on the device Nov 02 15:21:15 bluelightning: yes.. i'm now compiling the whole edison-image with the adjusted kernel config, but a bit worried about putting it on my edison if the module alone already mismatches Nov 02 15:29:31 awww, 2.0 docs link to gcc5 changes but not the "porting to" guide Nov 02 15:29:39 the latter would be super useful Nov 02 15:29:57 joshuagl: damn, if I'd known I'd have got scott to include it Nov 02 15:30:13 do you want to talk to him or should I? Nov 02 15:30:14 bluelightning: pretty sure I filed a bug report suggesting it Nov 02 15:30:18 https://gcc.gnu.org/gcc-5/porting_to.html Nov 02 15:31:16 I did! https://bugzilla.yoctoproject.org/show_bug.cgi?id=8079 Nov 02 15:31:17 Bug 8079: enhancement, Medium, 2.1, srifenbark, ACCEPTED , Improve migration guide Nov 02 15:31:18 2.1 milestone Nov 02 15:31:41 so you did ... I missed reading it, sorry Nov 02 15:31:48 np Nov 02 15:31:59 I'll ping Scott and we'll see if we can at least get the changes in Nov 02 15:32:48 excellent, thanks bluelightning Nov 02 16:02:27 Is there a way to dump all bitbake variables in real-time while (*.conf and *.inc) files are being parsed??? Nov 02 16:05:21 mrk377: not really, bitbake -e is what we have for that, but it includes history of how each var was set Nov 02 16:17:23 would toaster do that? Nov 02 16:23:55 armpit: missed that, sorry. Would Toaster do what? Nov 02 16:24:30 the above question "Is there a way to dump all bitbake variables in real-time while (*.conf and *.inc) files are being parsed???" Nov 02 16:24:47 out of the box not AFAIK.. Nov 02 16:24:58 however you might be able to write an anonymous function that does that Nov 02 16:25:34 armpit: no, sorry. Nov 02 16:25:44 belen2, k Nov 02 16:25:56 is wasn't my question Nov 02 16:26:25 recall something with vars and toaster I saw at ELCE Nov 02 16:26:46 armpit: you will get the global variable history, but once the build completes Nov 02 16:26:52 belen2, fray thank Nov 02 16:27:15 k Nov 02 17:09:26 Hi all! Nov 02 17:09:26 Does anybody know how i can see all the current revisions used by recipes that use AUTOREV in an overview? Nov 02 17:11:11 i need to freeze the current state of the build so i can get back to this revision if needed. Nov 02 17:12:13 Is there a better way than going through all recipes and define the specific revisions? Nov 02 17:12:20 my intel edison (running yocto) doesn't report the wifi card anymore in ifconfig, and it also doesn't show a bluetooth interface in hciconfig after doing 'rfkill unblock bluetooth'. it worked earlier today. did i break something in the hardware, or is there something i can do software-wise to bring things back? Nov 02 17:14:37 Hi. How do I tell yocto to build older recipe version? In my case: gnupg_1.4.7 not gnupg_2.1.1 Nov 02 17:16:03 PREFERRED_VERSION_gnupg = "1.4.7" Nov 02 17:16:24 * armpit bah.. fray beat me Nov 02 17:19:35 fray, I tried that earlier, did not seem to work Nov 02 17:23:56 i used that line on recipe that depends on gnupg Nov 02 17:23:56 is that correct? Nov 02 17:23:57 *in recipe that depends Nov 02 17:23:58 no, you need that in distro config Nov 02 17:23:58 ah ok Nov 02 17:23:58 thanks Nov 02 17:23:59 preferred version are global.. Nov 02 17:23:59 recipes do not affect global scope Nov 02 17:24:02 fl0v0: one easy way to do this is to enable buildhistory and use buildhistory-collect-srcrevs Nov 02 17:24:22 bluelightning: thanks! i will look into this Nov 02 17:24:32 gatisp: note that that assumes that there is actually a recipe present corresponding to that version Nov 02 17:26:02 bluelightning, yes, that is the behaviour I would expect :) Nov 02 17:26:47 gatisp: great, just mentioning as there are some folks who assume it'll magically build that version in the absence of a recipe ;) Nov 02 17:27:51 bluelightning: hmm, i notice that the commented recipes are still emitted even when no srcrev lines are, is that intentional? (buildhistory-collect-srcrevs). e.g. for non-autorev recipes when -a isn't passed Nov 02 17:30:08 kergoth: recipes or architecture triples? Nov 02 17:30:08 oh, right, indeed, it's the header i'm seeing, just with nothing under it Nov 02 17:30:36 I think it was laziness on my part when I did the implementation, but it might be nice to fix it so it didn't print those Nov 02 17:33:33 yeah, minor enhancement, if there are no srcrevs to print in that triple, hide the header. Nov 02 17:33:33 k Nov 02 17:33:39 was just cuirous, hadn't used it before Nov 02 17:34:38 * kergoth digs into why a repo with one commit accessible is still 1.1g, must have missed a ref somewhere, repack -ad didn't shrink it Nov 02 18:10:36 VAR ??= foo (takes lower precedence that ?=) When is this used (example)? Nov 02 18:15:08 Got it. Found an example of usage. Nov 02 18:35:43 with ?=, the first variable definition wins, since it's immediate, set right now if not already set Nov 02 18:35:51 ??= is lazy, so the last ??= wins **** ENDING LOGGING AT Tue Nov 03 02:59:58 2015