**** BEGIN LOGGING AT Wed Jul 02 03:00:00 2014 Jul 02 05:35:07 hi, ther is an ERROR in command "bitbake core-image-rt", it's: "ERROR: Multiple .bb files are due to be built which each provide virtual/kernel". how can I fix it? Jul 02 06:02:14 sm, Well I haven't build core-image-rt myself for quite a while. So this is from my old memory... core-image-rt has a DEPENDS on linux-yocto-rt, thus it will ask bitbake to build this. At the same time I guess that you have a machine conf specifying PREFERED_PROVIDER_virtual/kernel Jul 02 06:05:45 hi, is it possible to define a custom task from a image_types (a bbclass inheriting from image_types) which is executed after do_rootfs ? when my image is built my custom task is never executed. Ideally, my task needs to be executed after do_rootfs and after virtual/kernel:do_deploy. Jul 02 06:06:59 If I register my task in this way "addtask mytask after do_rootfs; [...] do_mytask[depends]="virtual/kernel:do_deploy" " it does not work Jul 02 07:27:04 <[Sno]> is it possible to modify distribution settings per recipe? Jul 02 07:27:41 <[Sno]> I have to build an image continaing java and xbmc - one needs x11 (xlibs, no server, no graphic) and one needs fb Jul 02 07:28:03 <[Sno]> but libx11 fails to compile when x11 is not in distro-features Jul 02 07:28:11 <[Sno]> (which is pointless) Jul 02 08:17:05 <[Sno] no, it's not possible. the purpose of DISTRO settings is especially to be applicable to all recipes Jul 02 08:17:42 <[Sno]> how can I build then an image with libx11, glxinfo (required by xbmc) and xbmc on fb? Jul 02 08:23:04 i don't know ;-) which doesn't mean that you can't, btw.. Jul 02 09:12:25 rperier, you can either define ROOTFS_POSTPROCESS_COMMAND += "mytask;" (note the semicolon) or add mytask after do_rootfs before do_build, the "before do_build" is necessary so the system knows the task *needs* to be executed Jul 02 09:25:27 Hello, juste a small question... When I build my core-image on Yocto Daisy, The hddimg file doesn't contains the initrd file. How to regenerate it ? Jul 02 09:28:47 Denwid: In fact, I want to build a initramfs builtin into the kernel image, so I have no choices. I need to do something like that Jul 02 09:29:20 Denwid: I will make a try, thanks Jul 02 09:52:24 rperier: are you trying to add a task to bundle the cpio in kernel? the framework is already there Jul 02 09:54:30 see /meta/conf/local.conf.sample.extended #294 Jul 02 09:55:10 <[Sno]> is there a way to build libx11 without x11 in DISTRO_FEATURES ? Jul 02 09:57:24 <[Sno]> see https://github.com/wolfgar/meta-stef/issues/1 for an example why Jul 02 10:14:11 we are using beaglebone linuxmainline 3.8.13 kernel and we want to integrate all the interfacing modules for ex:PWM ,UART,SPIDEV etc... these all in the format of dts file and located in kernel/firmware/capes directory and its compiled using dtc compiler also i am porting it manually to kernel at run time using uEnv.txt and writing into cape.9/slots file by following this link ( http://hipstercircuits.com/enable-serialuarttty-on-beaglebone-black/ Jul 02 10:14:11 ) am doing all the module like this. without using this method how can i integrate with the kernel to enable always. Jul 02 10:14:32 kindly any one help me please.. Jul 02 10:31:28 hi. i wonder, how do i include header files on the image? i've included the buildessentials packagegroup etc to allow compilation, but i can't figure out how to for example include the headers for Python... Jul 02 10:36:14 ionte: add dev-pkgs to IMAGE_FEATURES Jul 02 10:36:27 then you get the headers for everything installed Jul 02 10:36:41 if you want to be more specific then add eg python-dev to the image Jul 02 10:40:32 rburton: great! thanks! Jul 02 10:56:58 Hi yocto...i trying to boot my beaglebone black using kernel 3.8.13 while booting this image i am getting link mentioned error http://pastebin.com/qprXnryp Jul 02 10:57:14 please help me how to rectify this error.. Jul 02 11:00:01 please ......... Jul 02 11:05:20 AndersD:Are you there.. Jul 02 11:19:58 <[Sno]> I have a machine conf as in http://pastebin.com/LYSbDisk - and a meta-me/receipes-bsp/u-boot/u-boot-imx_2009.08.bb and meta-me/conf/layer.conf has BBFILE_PRIORITY_me = "8" Jul 02 11:20:30 <[Sno]> but very often the images (*.sdcard) I get have an unmodified u-boot prompt Jul 02 11:20:50 <[Sno]> how can I ensure which recipe is included? Jul 02 11:22:32 <[Sno]> how can I ensure that /lib/firmware/vpu contains vpu_fw_imx6q.bin instead of vpu_fw_imx6d.bin (which is now included) Jul 02 11:38:42 i am using layer meta-beagleboard/recipes-common/linux-mainline-3.8.13 kernel and same layer u-boot. Jul 02 11:53:43 Is there a bitbake function to find which task has been called? I would like to differentiate if "bitbake someimage -c populate_sdk" or "bitbake someimage" was called. Jul 02 12:25:06 ant_work: I want to build a initramfs into the kernel image (statically built) (throught the variable INITRAMFS_SOURCE in the .config of the kernel) Jul 02 12:25:15 as I understood this is not supported Jul 02 12:25:46 ant_work: correct me if I am wrong but we can only build separated initramfs images for now Jul 02 12:26:09 it is supported Jul 02 12:26:22 now that you say 'static', I do the same Jul 02 12:26:24 http://kexecboot.org/documentation/crosscompiling/oe-yocto Jul 02 12:26:45 it's a klibc-static cpio embedded in a separate kernel Jul 02 12:27:09 no modules Jul 02 12:27:27 mhhhh, interesting, I will look at it Jul 02 12:30:41 ant_work: thanks Jul 02 12:33:34 yw Jul 02 12:42:10 rperier: for the background, look at these : http://patchwork.openembedded.org/patch/56309/ http://patchwork.openembedded.org/patch/59093/ Jul 02 12:43:21 oe-core: 609d5a9 and 36faac8 Jul 02 13:26:06 ant_work: assuming I want to generate a flashable image from a image_types (a bbcclass which inherits from image_types) using this technic. is it possible too ? Jul 02 13:42:40 I wrote a recipe to build ffmpeg for arm, it builds ffmpeg ffserver and ffprobe but no ffplay any idea where this could come from, Jul 02 13:44:35 heres my recipe: Jul 02 13:46:07 http://pastebin.com/ZK1LCbkx Jul 02 13:46:15 rperier: you mean image A contains kernel A and is embedded in flashable image B? Jul 02 14:10:29 ant_work: I mean a kernel image "K" which contains a initramfs "IN" converted to a flashable image by a custom FSTYPE contained in IMAGE_FSTYPES, like IMAGE_FSTABLE = "myflashableimage" (which is my-layer/classes/myflashableimage.bbclass) Jul 02 14:12:31 (K contains IN statically built and K is converted into a binary image which is uploadable to a target) Jul 02 14:13:03 (no uboot on rockchip :) ) Jul 02 14:13:57 I don't understand that part... Jul 02 14:14:14 where does th ekernel reside?isn't raw flash? Jul 02 14:14:31 is it a nor or a nand? Jul 02 14:18:58 anyway yes, you can create another image-recipe EXT and populate it with that custom kernel K containing IN Jul 02 14:29:16 rperier: I think I understand now. You flash with rkflashtool? Jul 02 14:35:34 ant_work: no with the cmdline tool provided by rockchip Jul 02 14:36:55 ant_work: upgrade_tool Jul 02 14:38:02 I see, you can flash parameter, boot.img, system.img,.. Jul 02 14:38:38 yes, I only flash boot.img (kernel+initramfs) Jul 02 14:38:44 what is that darned .img format? Just some headers? Jul 02 14:39:37 (the mainline kernel has not internal flash memory drivers) Jul 02 14:39:44 (so I have no choices) Jul 02 14:43:30 I see on www.cnx-software.com some instructions about generating boot.img. Afais they use an external initramfs Jul 02 14:44:19 and a mkbootimg tool Jul 02 14:47:52 ant_work: yes with a kernel 3.0.36 not synced with upstream and the initramfs address depends on the proprietary bootloader Jul 02 14:48:14 build the initramfs statically into the kernel image depends on nothing :) Jul 02 14:48:22 right Jul 02 14:53:05 initially I thought you were trying to update on-device using a special kernel containing updater+image Jul 02 14:53:57 no :) Jul 02 14:54:36 ok, it's fairly similar to the situation I have with a new mips toy, you have to use proprietary flasher on PC Jul 02 14:55:18 apart this, you can build the single pieces with OE Jul 02 14:56:27 (there is a SD and not a flash on it fwiw, still applies) Jul 02 15:01:42 ant_work: there is a NAND flash and a sd reader Jul 02 15:27:39 question about PACKAGE_CLASSES, when set to package_rpm I am able to -c populate_sdk no problem, but if set to package_deb I get a target filesystem missing most packages. Jul 02 15:28:22 are you using multilibs? Jul 02 15:28:48 it's also possible deb and the SDK don't get along.. I know RPM and IPK are well tested.. I don't know if deb is for the SDK Jul 02 15:30:06 I'm not sure on the multilib front, but what you're saying is package_deb shouldn't be used then yes? Jul 02 15:31:09 I'm saying it's not used very much.. if it's not working, check the Yocto Project bugzilla to see if someone else has observed it.. if not.. please file a bug with your configuration details and bitbake commands so we can try to fix it Jul 02 15:31:18 (or feel free to fix it, if you can) ;) Jul 02 15:32:56 as far as the multilib, I say something on google about that from 2012, is it something that's meant to be configurable then? Does it follow the same conventions as debian? Jul 02 15:33:16 *I see* Jul 02 15:35:39 you would have a configuration in your system MULTILIBS = "..." Jul 02 15:35:54 if you don't you aren't using them.. multilibs can cause issues w/ deb (or at least have in the past) Jul 02 15:36:36 okay, thanks for the help Jul 02 15:37:17 (multilib is the mechanism we use to support multiple ABIs on the same machine.. i.a. x86 and amd64 installed at the same time) Jul 02 15:38:24 ... when it works. :) Jul 02 15:40:04 moin Jul 02 15:47:41 nerdboy: :) Jul 02 17:05:16 I run in a strange problem here. I want to extend the openssh package by adding a firewall script. So I create a folder in my own layer openssh/openssh_6.5p1.bbappend with content 'SRC_URI += " file://firewall-sshd.sh"' and the file openssh/files/firewall-sshd.sh Jul 02 17:06:01 Now running it leasds to " Unable to find file file://firewall-sshd.sh anywhere. The paths that were searched were:" All of these files are in the original meta layer and indeed the file is not there. Jul 02 17:06:21 It is analogue to a lot of other bbappend files I created, but somehow only with openssh it fails Jul 02 17:09:06 sounds like you missed FILESEXTRAPATHS Jul 02 17:09:08 or set it incorrectly Jul 02 17:10:48 kergoth: you are right. somehow it forgot to copy it. Looked for 15minutes for other issues.... Jul 02 17:11:21 FILESPATH is how bitbake seraches for file:// files, and FILESEXTRAPATHS is how you append to FILESPATH cleanly from appends Jul 02 17:11:32 so thats where to look if it cant' locate a file:// file, for future reference :) Jul 02 17:12:27 defective user in this case. totally forgot that snippet. thanks for the hint :) Jul 02 17:14:01 np Jul 02 17:14:27 I am still unconsidered if I should use bbappend to add firewall rules or if i should create a firewall-sshd package Jul 02 17:14:38 more a style question. Jul 02 17:15:35 I'm inclined to say the latter would be slightly cleaner, but its really personal preference at that point Jul 02 17:15:58 pro of separate packet are the ptests and that I don't have to care about software changes Jul 02 17:16:17 software changes in kind of versioning Jul 02 17:18:36 also consider impact on sstate reuse and binary packaging, how much can / will be shared between your builds with and without the firewall bits, or the possible re-run of openssh tasks when you change the firewall script to repackage it Jul 02 17:18:46 minor in this case, but good things to keep in mind Jul 02 17:36:18 I was wondering whether there are any plan to make wic create beaglebone/beagleboard images Jul 02 17:38:53 hi Jul 02 17:47:21 I might have asked it before: Is it possible to limit the amount of processes that run do_rootfs? Right now both image rootfs are build at the same time if I execute "bitbake image1 image2" Jul 02 17:53:29 volker-: only via the usual mechanism to control how many processes run at once — BB_NUMBER_THREADS. but you can't exert control at a per-task level, no Jul 02 17:53:49 you could presuambly add a lockfile so bitbake would *try* to run both at the same time ,but would be prevented from it Jul 02 17:54:09 not quite the same thing since bitbake wouldn't be awareof the influence on its scheduling, but you could force the matter that way, i guess Jul 02 17:54:14 kergoth: when I compile etc. it is usually fine to have it parallel. do_rootfs is really IO heavy, therefore I would prefere only this serial Jul 02 17:54:17 i'm assuming your disk load is getting high? Jul 02 17:54:19 * kergoth nods Jul 02 17:54:38 bitbake's scheduler doesn't let you do resource bound control like that.. yet. it's been on the wishlist for years Jul 02 17:54:53 A feature for extraction and packaging could be nice too, but that IO load is half as bad as the image step Jul 02 17:56:20 I can't add NOHDD and NOISO in my image recipe? Jul 02 17:57:17 err... my friend the quote symbol is required... Jul 02 18:04:38 the deploy/image directory has a file called README_-_DO_NOT_DELETE_FILES_IN_THIS_DIRECTORY.txt that states to run "bitbake -c clean TARGET". is target here the image name? Will it delete all the packages it build? Jul 02 18:04:53 Right now I ignore the warning and only delete old images to free up disk space Jul 02 18:09:11 mainly that warning is just to indicate that bitbake won't automatically recreate everything in there if you rerun bitbake, as it has its stamps, afaik Jul 02 18:11:00 just tried the clean command with the image name -- the files still exists Jul 02 19:18:25 anyone tried booting a USB core-image-sato for genericx86 from master recently? Seems udev isn't mounting usb block devices properly anymore. With debugshell enabled, I see nothing mounted except the root once it gives up. Jul 02 19:51:32 blloyd, there is an open bug Jul 02 19:51:50 blloyd, something I'm planning on looking at, but have been prevented by various things Jul 02 19:52:14 blloyd, if you have time to debug, please post your findings to the bug: Jul 02 19:53:13 blloyd, https://bugzilla.yoctoproject.org/show_bug.cgi?id=6492 Jul 02 19:53:14 Bug 6492: normal, Undecided, ---, dvhart, NEW , [Generic x86_64] USB not mounted Jul 02 19:57:15 <[Sno]> stef/wolfgar? Jul 02 19:57:57 dvhart_: I'd commented on that one, since I had found. It was x86_64. I just found same applies to x86 too. Jul 02 19:58:19 and for x86, it's happening during init script too for live boot. Jul 02 19:58:22 OK, we can just update the bug subject Jul 02 19:58:29 probably a udev issue Jul 02 19:58:36 or ... possibly a udev issue Jul 02 19:59:48 <[Sno]> I have a machine conf as in http://pastebin.com/LYSbDisk - and a meta-me/receipes-bsp/u-boot/u-boot-imx_2009.08.bb and meta-me/conf/layer.conf has BBFILE_PRIORITY_me = "8" Jul 02 19:59:49 <[Sno]> but very often the images (*.sdcard) I get have an unmodified u-boot prompt Jul 02 19:59:50 <[Sno]> how can I ensure which recipe is included? Jul 02 19:59:51 <[Sno]> how can I ensure that /lib/firmware/vpu contains vpu_fw_imx6q.bin instead of vpu_fw_imx6d.bin (which is now included) Jul 02 19:59:57 np. I had found and commented there, but was checking in case someone may have known a work-around. Pretty sure it is genericx86 image based, as I hadn't seen it on my own atom-pc based machines. But I fired up this to see if I had introduced a different bug and got a surprise when it was even less functional. Jul 02 20:00:17 <[Sno]> is there a way to build virtual/libx11 without x11 in DISTRO_FEATURES ? Jul 02 20:00:46 blloyd, have you tried the meta-intel intel-core2-32 MACHINE? Jul 02 20:00:59 blloyd, I'd be interested to know if it has the same issue for you or not Jul 02 20:01:15 (daisy or master branch) Jul 02 20:02:06 no I haven't. Thank you for that suggestion. Not sure core2 will boot on the hardware I'm using though. It's fully i586, neither are full i686 (well depends on the definition of i686 for one...) Jul 02 20:02:40 Any idea what core2 is targetted for? I could have sworn core2 was late i686. Jul 02 20:03:06 'core2' is litterly 'core2'... Jul 02 20:03:09 core2 is targeted for "core2" and later :-) Jul 02 20:03:22 basically all Atom CPUs (and core2 before that) Jul 02 20:03:29 yeah, I can't remember where core2 fit in the timeline. Jul 02 20:03:30 so yeah, that can break on your system. Jul 02 20:03:39 i686 is Pentium Pro 'or so'.. i586 is Pentium, Quark and similar CPus.. Jul 02 20:03:40 oh, late 90s iirc Jul 02 20:04:04 I thought 'core' was early 2000's.. and core2 was a bit after.. Jul 02 20:04:10 yeah, one of the gems I get to play with is Geode LX... Jul 02 20:04:18 hrm... yeah, I think I'm off a decade.... checking Jul 02 20:04:22 ya thats going to be i686 Jul 02 20:04:23 I'm terrible at dates Jul 02 20:04:56 core2 introduced July 2006 Jul 02 20:04:57 Geode LX is i686 minus NOPL (which technically wasn't in the original specification for i686) Jul 02 20:05:30 So core2 supports MMC and SSSE3 Jul 02 20:05:34 (MMX) Jul 02 20:05:42 so not a good choice Jul 02 20:05:46 which will be a kernel panic for me. Jul 02 20:05:50 yup Jul 02 20:05:55 Hun 'Core' was January of 2006.. I thought there was a larger gap Jul 02 20:06:07 so scratch that idea :-) Jul 02 20:06:52 that was during new processor every 6 months phase... Jul 02 20:07:05 apparently.. Jul 02 20:07:12 I ignore 'core' since it was only 32-bit.. ;) Jul 02 20:09:13 I'm kinda stuck with 32 bit. They still tend to be more power efficient than 64 bit due to performance verse power trade-offs made. Jul 02 20:10:32 and some of those old 32bit chips are rather nicer about working in a noisy (physical shaking or poor power) environment. Jul 02 20:11:10 blloyd, do you know when it worked and when it stopped working? Jul 02 20:11:51 I've never tried genericx86 before, so no. My own changed behavior after my last pull, which pulled maybe 2 weeks of changes in. Jul 02 20:12:03 Right, OK Jul 02 20:14:07 First thing I checked was if my proposed oe-core had made it in (it hasn't). Then checked if my changes were the culprit, which is why I can tell it's udev not mounting and not the script unmounting wrong or something else I may have done. Jul 02 20:15:14 So, who is the unlucky soul that has to follow oe-core for committing to master? I can't keep up with that mailing list. I think it would be a full time job just reading it for patches. Jul 02 20:17:57 that would be RP and others Jul 02 20:24:17 i want to add a DEPENDS to python on tk, but I don't want nativesdk... Jul 02 20:28:28 someone good with bugzilla willing to tell me how to find the list of bugs I am currently watching (cc me is on)? Jul 02 20:28:57 use the advanced search, search for your email address, save the query? Jul 02 20:30:20 thanks. quick search on my email gave zaro bugs. Jul 02 20:33:25 blloyd: that works, as i've got that query. there's an explicit field for CC list. Jul 02 20:46:58 thanks dvhart_ and rburton. I missed it first time I had peaked through the advanced. Figured it was something easy I was overlooking. Jul 02 21:06:53 is anyone else running on i586/i686 era hardware? I was wanting to compare notes about connmand. My log files on multiple machines end up looking like http://pastebin.com/4sxEApr2 a lot. Jul 02 21:08:32 (ok usually I am not running debug so only get the {add} and {del} repeating thousands of times a second) Jul 02 22:38:23 it seems like an image is not getting rebuild if you modify ROOTFS_POSTPROCESS_COMMAND Jul 02 23:56:50 Is Qt Creator inlucded in the Yocto SDK? Jul 03 00:16:35 no Jul 03 00:33:44 hey, any Intel Galileo makers/hackers in here? Jul 03 02:28:06 behanw: how was your canuck day? Jul 03 02:58:38 mranosta1, Legen... wait for it... dary. **** ENDING LOGGING AT Thu Jul 03 03:00:00 2014