**** BEGIN LOGGING AT Wed Oct 03 02:59:59 2012 Oct 03 07:27:14 good morning Oct 03 08:09:29 morning all Oct 03 08:49:45 | The directory that should contain system headers does not exist: Oct 03 08:49:45 | /home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/genericarmv8/home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/genericarmv8/usr/include Oct 03 08:49:57 why it adds path to path? Oct 03 09:26:39 morning all Oct 03 09:28:33 bluelightning: do you have idea why gcc for target wants to find headers in wrong dir? Oct 03 09:28:38 | The directory that should contain system headers does not exist: Oct 03 09:28:41 | /home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/genericarmv8/home/hrw/HDD/devel/canonical/aarch64/build/tmp-eglibc/sysroots/genericarmv8/usr/include Oct 03 09:29:02 hrw: no, not sure sorry Oct 03 09:34:34 np Oct 03 09:50:29 hi bluelightning Oct 03 09:54:47 During building of my image I had this error: "opkg_install_cmd: Cannot install package foopackage" Oct 03 09:55:01 in the log I only find this warning Oct 03 09:55:14 WARNING: could not open /home/oe-core/build/out-eglibc/work/colibri_t20-angstrom-linux-gnueabi/btw-image-V2.0-r3/T20_LinuxImageV2.0/rootfs/lib/modules/3.1.10/modules.order: No such file or directory Oct 03 09:55:15 WARNING: could not open /home/oe-core/build/out-eglibc/work/colibri_t20-angstrom-linux-gnueabi/btw-image-V2.0-r3/T20_LinuxImageV2.0/rootfs/lib/modules/3.1.10/modules.builtin: No such file or directory Oct 03 09:55:15 marking as unpacked only, configuration required on target. Oct 03 09:55:20 SOme suggest? Oct 03 10:25:06 morning all Oct 03 10:52:07 hi pb_ Oct 03 10:55:31 hi bluelightning Oct 03 10:56:59 pb_: hi, can you add that .text relocation description to qa check header? easier to find it there then on ML :) Oct 03 10:59:21 yup, will do Oct 03 12:48:42 I made my package, then add it to a new image, then I make some change to it to deploy a new version and when I try to build image aI get this: Oct 03 12:48:50 | * opkg_install_cmd: Cannot install package mypackage Oct 03 12:55:21 Someone have any idea? http://paste.ubuntu.com/1257897/ Oct 03 12:57:38 | * opkg_install_cmd: Cannot install package qtcardiopad. Oct 03 12:57:38 | * opkg_install_cmd: Cannot install package qdbusnetworkmanager. Oct 03 12:57:56 silvio:these lines are why your do_rootfs stage fails Oct 03 12:58:05 check why these to packages fail to install Oct 03 12:58:19 *two Oct 03 12:58:42 Unknown package 'qtcardiopad'. Oct 03 12:59:17 yes, both were not found Oct 03 12:59:46 silvio:first, try to build these two packages individually Oct 03 13:00:41 dromede, I did: Oct 03 13:01:06 NOTE: Running task 3440 of 3440 (ID: 12, /home/oe-core/build/../stuff/meta-btw/recipes-btw/qtcardiopad/qtcardiopad.bb, do_rm_work_all) Oct 03 13:01:06 NOTE: package qtcardiopad-1.0-r0: task do_rm_work_all: Started Oct 03 13:01:06 NOTE: package qtcardiopad-1.0-r0: task do_rm_work_all: Succeeded Oct 03 13:01:06 NOTE: Tasks Summary: Attempted 3440 tasks of which 3439 didn't need to be rerun and all succeeded. Oct 03 13:01:35 silvio_: have you checked if packages-split actually have some files in qtcardiopad ? Oct 03 13:02:28 JaMa: opkg will fail like this if a packages is empty? Oct 03 13:03:19 if it's empty then it's not created Oct 03 13:03:30 ah, riht Oct 03 13:03:33 *right Oct 03 13:03:56 and bitbake should complain about it (unless you specify ALLOW_EMPTY) but that imho doesn't work now Oct 03 13:03:56 JaMa, dromede, maybe there is a regression problem in the .pro file... Oct 03 13:04:19 and also it can be renamed when debian.bbclass is used Oct 03 13:04:30 so actuall package name could be libqtcardiopad now Oct 03 13:10:25 I have yhe package split for qdbusnetworkmanager Oct 03 13:10:25 I have the package split for qdbusnetworkmanager, so I dont understand errot Oct 03 13:10:25 but for qtcardiopad I need some adjustment Oct 03 13:13:54 it could be also built for architecture not enabled in your opkg.conf Oct 03 13:14:42 but I don change nothing about it Oct 03 13:15:02 before this package went compile Oct 03 13:15:14 before this package wad compiled and installe Oct 03 13:15:56 opkg arch priority was changed in latest oe-core Oct 03 13:16:00 if you track master Oct 03 13:17:25 so where I can check? Oct 03 13:43:21 jackmitchell: I sent the patches I had ready Oct 03 13:45:15 otavio: looking good! Oct 03 13:45:41 jackmitchell: have time to look at ffmpeg with me? Oct 03 13:46:32 otavio: I can be around, but I literally just copied the libav package, renamed it to ffmpeg and added the extra binaries Oct 03 13:47:02 otavio: did you see my intitial dirty ffmpeg commit yesterday? Oct 03 13:47:11 jackmitchell: yes Oct 03 13:47:20 jackmitchell: as you said, dirty heheh Oct 03 13:47:53 otavio: indeed, the first issue with it I believe it attempts to double strip the binaries Oct 03 13:48:21 jackmitchell: I think we could use dynamic packages for it Oct 03 13:48:33 jackmitchell: so it reduces the size of package a lot Oct 03 13:48:43 jackmitchell: the recipe size Oct 03 13:49:07 otavio: that sounds good, it's a bit of a monster at the moment Oct 03 14:00:16 otavio: is there an recipe that uses dynamic packages which I can look at as an example of what you mean? Oct 03 14:00:31 jackmitchell: grep PACKAGE_DYN Oct 03 14:00:59 or DYNAMIC_PACK I don't know now Oct 03 14:06:48 jackmitchell: why did you move to ffmpeg source? Oct 03 14:06:52 so PACKAGE_DYNAMIC is basically a mixture of a package-* and a python function to collate and name all the packages automatically? Oct 03 14:07:04 jackmitchell: and did not update libav? Oct 03 14:07:07 jackmitchell: yes Oct 03 14:07:42 otavio: libav doesn't have the required binaries to build xbmc anymore Oct 03 14:08:22 for PACKAGES_DYNAMIC please refer to: https://wiki.yoctoproject.org/wiki/Handling_optional_module_packaging Oct 03 14:08:29 hopefully hitting the manual soon Oct 03 14:08:42 otavio: specifically libpostproc Oct 03 14:08:47 jackmitchell: and ffmpeg is api/abi compatible? Oct 03 14:09:06 otavio: libav is a fork of ffmpeg Oct 03 14:09:46 otavio: I believe Koen was using libav instead of ffmpeg due to better support for neon instructions in libav Oct 03 14:10:18 jackmitchell: and has it improve? Oct 03 14:10:51 otavio: possibly in the 1.0 release, I don't really keep up with it, seeing as though libav can no longer build xbmc it is a moot point Oct 03 14:12:02 otavio: it wouldn't surprise me if someone was working on a patch to remove the libpostproc dependancy from xbmc though Oct 03 14:12:34 bluelightning: thanks, will take a quick gander Oct 03 14:12:58 jackmitchell: or we can package libpostproc only Oct 03 14:13:47 ok, "found" why gcc did not built for me. I removed most of OE patches Oct 03 14:13:56 once I added them back it builds Oct 03 14:14:16 otavio: yes, I didn't think about mixing the two together! Oct 03 14:14:46 otavio: if libav and ffmpeg could be convert to DYNAMIC then we would have an elegant solution Oct 03 14:16:06 How to clean completely a package, also git download? Oct 03 14:16:22 -c cleanall Oct 03 14:17:01 JaMa, thanks Oct 03 14:17:19 jackmitchell: yes Oct 03 14:17:34 jackmitchell: i will check xbmc need of postproc now Oct 03 14:19:48 jackmitchell: libav has postproc but it is disabled by default Oct 03 14:20:02 jackmitchell: so we can work on libav and don't move to ffmpeg Oct 03 14:20:38 jackmitchell: last release is 0.8.3 Oct 03 14:20:38 otavio: latest libav removes it from tree Oct 03 14:20:50 http://git.libav.org/?p=libav.git;a=tree Oct 03 14:22:05 otavio: sorry, in libav master it is removed from tree Oct 03 14:22:11 jackmitchell: http://git.libav.org/?p=libav.git;a=commit;h=b315042c8ce984bec431c5965120853a843cbfa5 Oct 03 14:22:21 jackmitchell: and moved to an external git Oct 03 14:23:21 otavio: ah, ok - so it could be packaged as a completely seperate library? Oct 03 14:23:29 jackmitchell: yes Oct 03 14:23:33 jackmitchell: makes sense Oct 03 14:23:48 that does make things a bit nicer - I missed that when I was going through this Oct 03 14:24:00 jackmitchell: since are you looking at libav I will look at libpostproc Oct 03 14:24:04 jackmitchell: set? Oct 03 14:26:31 otavio: ok Oct 03 14:34:14 can anyone see what i'm doing wrong here: http://pastebin.com/4c4ATzzu Oct 03 14:34:33 I am recieving NOTE: Error expanding variable populate_packages | ETA: --:--:-- Oct 03 14:35:43 wrong indentation Oct 03 14:37:26 jackmitchell: I just checked and 0.8.3 has postproc Oct 03 14:37:44 jackmitchell: it has been removed at tree after 0.8 Oct 03 14:37:59 jackmitchell: http://anonscm.debian.org/gitweb/?p=pkg-multimedia/libav.git;a=blob;f=debian/confflags;h=c31f88d496920929fa2133384387a06bf6c49bf6;hb=HEAD#l140 Oct 03 14:38:25 otavio: yes, removed in master Oct 03 14:38:44 jackmitchell: but we can use 0.8.3 Oct 03 14:40:41 jackmitchell: Debian and Ubuntu are using it so I think we can stick at 0.8.3 for now Oct 03 14:41:38 Where/how do I set what MACHINE_ARCH_ I'm going to use for a kernel build? Oct 03 14:41:49 oh sorry Oct 03 14:41:54 didn't read the last part in the title Oct 03 14:43:55 otavio: ok Oct 03 14:45:18 JaMa: what do we have these days, 3 space python, tab? Oct 03 14:46:10 jackmitchell: mix, but for python 4 spaces Oct 03 14:48:32 ok, I now have 4 spaces and still the same problem: http://pastebin.com/deMMFZwt Oct 03 14:49:14 jackmitchell: "libav lib %s' Oct 03 14:50:14 JaMa: well spotted! The bash syntax in the do_install messes up my coloured syntax, it's all bright blue after that line ;( Oct 03 15:12:20 otavio: I have started on the libav recipe but I need to rebuild for a new architecture as it won't build for the raspberry pi due to some unsupported ASM Oct 03 15:13:07 otavio: it won't finish building by the time I will be going home so I will have to come back to it on Tuesday as I am on holiday in Belgium (yay Beer!!) till then :D Oct 03 15:13:31 otavio: I'll push what I have so far to my branch if you want to take a look Oct 03 15:16:06 otavio: https://github.com/CommunistCode/meta-oe/commits/xbmc Oct 03 15:17:49 Ok; thx Oct 03 15:38:16 I get a weird error after an update Oct 03 15:38:39 ERROR: No recipes available for: pulseaudio_1.1.bbappend Oct 03 15:39:00 remove it from the layer Oct 03 15:39:05 oh wait Oct 03 15:39:13 oe core is at 2.0 or so Oct 03 15:39:15 got it, pulseaudio recipe was update to 2.1 Oct 03 15:40:27 Is there a way to ignore a specific append from a layer? Or would using a BBMASK be the best approach Oct 03 15:42:20 BBMASk yes Oct 03 15:42:31 or deleting it Oct 03 18:45:54 evening all Oct 03 18:57:02 Hi pb_ Oct 03 18:57:13 pb_: you mentioned some problem with textrels on mips Oct 03 18:57:21 I did not read the email in detail Oct 03 18:57:41 if you can cook up a small testcase I can take a look Oct 03 19:14:21 morning Oct 03 19:16:34 khem: I'll have a go. iirc, if you compile pretty much anything for mips you will see it. Oct 03 19:17:16 pb_: hmm Oct 03 19:17:22 ok Oct 03 19:17:23 Please excues my ignorance, but I'm trying to get a development environment under CentOS 6.3 to compile to an anstrom distribution (kernel 2.6.11) for an armv5tel chipset. I've followed the first 3 steps here: http://www.openembedded.org/wiki/Getting_Started_with_OE-Core but I'm unclear how to proceed Oct 03 19:17:27 I dont do much mips these days Oct 03 19:17:39 Should I change the conf/local.conf and add in 'armv5tel' for the machine name? Oct 03 19:17:51 but I will try to build an image and see Oct 03 19:18:05 skorket: armv5tel is not a machine Oct 03 19:18:09 unless you defined yours Oct 03 19:18:19 which means you should put it in conf/machine Oct 03 19:18:26 as a global machine conf file Oct 03 19:19:00 I do not see a machine conf file... Oct 03 19:19:24 and just so I understand, a "machine" is the whole environment whereas the armv5tel is a chipset? Oct 03 19:20:31 armv5tel is an ISA Oct 03 19:23:11 So what is the proper question? "How do I set up an environment for an ARM target under kernel 2.6.11"? Oct 03 19:24:59 Well, I suspect that the first step is probably going to be to get a newer kernel. Oct 03 19:25:45 I guess first step is to dust off this board :) Oct 03 19:27:22 I can run whatever, but the target is 2.6.11 Oct 03 19:28:40 re Oct 03 19:29:10 Whoops, and I mispoke, it's Arcom, not Angstrom Oct 03 19:32:50 Even pointers to what I should be doing or other resources would be much appreciated Oct 03 19:33:47 I'm not sure you will be able to get much of a modern oe-core system to run under 2.6.11. If I remember right that kernel is missing several important system calls. Oct 03 19:34:12 But, as far as general documentation goes, I think the yoctoproject wiki is probably the place to start. Oct 03 19:34:28 You can set up an oe-core build for qemuarm which is pretty much a generic armv5te. Oct 03 19:35:05 pb_ what you would recommend to get a development environment for that system going? Oct 03 19:41:59 oh my god 2.6.11 Oct 03 20:09:41 eglibc we have configured with minimal 2.6.16 Oct 03 20:30:32 skorket: you would start by defining a machine probably (in conf/include/machine), taking an example from the existing machines already there Oct 03 20:31:19 depending on what you are using for a bootloader, i would then try to get u-boot compiled and try to boot it on the target device Oct 03 20:33:16 gmc, could you give me a reference to a template that I might be able to use? Oct 03 20:34:05 are you using the classic oe or oe-core? Oct 03 20:34:17 oe-core Oct 03 20:34:21 ok, good :) Oct 03 20:35:01 if you check out http://www.openembedded.org/wiki/LayerIndex , theree's several device specific layers available Oct 03 20:35:42 i would pick one that is somewhat close to my target device.. lett's say meta-kirkwood.. Oct 03 20:35:58 https://github.com/kelvinlawson/meta-kirkwood/tree/master/conf/machine has some examples of machine definitions of kirkwood-based boards Oct 03 20:36:31 Here is the machine I'm running: http://www.eurotech-inc.com/single-board-computer-pxa270-pc104-titan.asp ... should I do kirkwood or something like meta-fsl-arm? Oct 03 20:39:11 hmm yes, kirkwood might serve as an example, but check the datasheet/specs of your processor to find the right tuning file Oct 03 20:39:47 actually, i see there's a tune-xscale.conf: http://cgit.openembedded.org/openembedded-core/tree/meta/conf/machine/include/tune-xscale.inc Oct 03 20:40:47 so you recommend using kirkwood as a base then overwriting this tune-xscale.conf file into it? Oct 03 20:40:53 although that's for older xscaleÅ's Oct 03 20:41:09 well, the tune-*.conf gets included from the machine definition in conf/machine/foobar.conf normally Oct 03 20:41:30 i would look at the various existing machine confs and tune confs to work out what i would need for my target board Oct 03 20:43:52 There's an arch-armv5.inc here: http://cgit.openembedded.org/openembedded-core/tree/meta/conf/machine/include/arm/arch-armv5.inc Oct 03 20:44:15 gmc, I'm sorry to do this to you. I don't want to waste your time. Is there a 'walkthrough' that I can use to try and get something running? Oct 03 20:44:25 or can you give me some broad strokes of how to set this up? Oct 03 20:50:33 skorket: well, it is all kinda complicated.. as was mentioned, you could try the qemu target first, to get a feel for the whole process Oct 03 20:50:38 kirkwood is not same as pxa270 Oct 03 20:50:57 i'm not claiming it is.. Oct 03 20:52:01 skorket: the documentation is all a bit spartan, i've found.. one thing that would have saved me a lot of time had i known, is that the openembedded.org docco is all outdated, the yoctoproject docco is more up to date (i believe the two projects have merged at some point) Oct 03 20:53:16 skorket: there's also a bunch of README's in oe-core itself that may be helpful Oct 03 20:53:35 there are layers which has closer machines Oct 03 20:53:40 to pxa270 Oct 03 20:53:47 check in layerIndex Oct 03 20:53:57 for eg http://cgit.openembedded.org/openembedded-core/tree/meta/conf/machine/include/README Oct 03 20:54:14 skorket: and in case you haven't found it yet, http://www.yoctoproject.org/documentation Oct 03 20:54:50 there's a link to a 'quick start' there, which explains how to build a working environment that boots in qemu Oct 03 20:55:40 see http://cgit.openembedded.org/meta-handheld/tree/conf/machine/hx4700.conf Oct 03 20:55:43 e.g. Oct 03 20:57:21 ok, thanks gmc and khem, I'm going through the yocto quick start now Oct 03 20:58:50 skorket: good luck :) oe can be overwhelming sometimes (in my experience), but the people here tend to be very patient and helpful, even when you do stupid things (as i have done many times :) Oct 03 20:59:19 gmc, thanks Oct 03 21:01:20 for the "MACHINE" vairable in the local.conf, I should put qemuarm? or should it be something like "arch-armv5"? Oct 03 21:04:14 machine is the BSP or board you want.. i.e. qemuarm Oct 03 21:08:33 byte stream protocol? Oct 03 22:13:41 does autoconf automatically support "configure --with-sysroot", or not? I'm seeing contradicting info Oct 03 22:43:19 hollisb, i want to say yes... Oct 03 22:44:02 mr_science: it looks like it's something handled by m4/libtool, and OE must be putting its own copy of that file in place Oct 03 22:44:26 because the m4/libtool file I see in the source tree doesn't do what it would need to Oct 03 22:44:54 i guess i would punt and ask kergoth then Oct 03 22:45:02 yeah, thanks! **** ENDING LOGGING AT Thu Oct 04 02:59:58 2012