**** BEGIN LOGGING AT Wed Apr 10 02:59:58 2013 Apr 10 07:00:36 Good morning all Apr 10 07:00:54 please can any one tell me how to install ipk in target Apr 10 07:01:06 bcz opkg is not fcomming in my root file system Apr 10 07:30:56 morning all Apr 10 07:35:25 kota down load the ipk Apr 10 07:35:33 ipk install ipk-file Apr 10 07:47:27 good morning Apr 10 08:04:55 good morning Apr 10 08:05:53 hi silvio__, woglinde, mckoan Apr 10 08:13:07 is it work for cross compiled root file system also Apr 10 08:13:22 ipk binary is not part of oe core? Apr 10 08:13:40 or do we need to build ipk package if it there Apr 10 08:13:55 i want to keep part of rootfs Apr 10 08:20:47 hi apelete Apr 10 08:37:19 hi apelete Apr 10 08:37:55 kota you did not tell us what you want to achieve, I guessed you mean running ipkg on target Apr 10 08:59:41 apelete: what's the situation with kernel on nanonote? How big is the patchset? Apr 10 09:00:36 morning all Apr 10 09:01:23 gm bluelightning Apr 10 09:05:49 otavio, you here? had a question about OpenGL ES, all I get is a white screen. I suspect my Xorg configuration is wrong (it does complain about AIGLX problems) - I wonder if I am doing something wrong Apr 10 09:08:17 hi ant_work Apr 10 09:19:02 gm bluelightning Apr 10 09:19:19 hi woglinde Apr 10 09:26:03 hi ant_work Apr 10 09:26:10 hi apelete Apr 10 09:26:19 hi bluelightning Apr 10 09:26:58 sorry ant_work for not replying the other night, I was off Apr 10 09:29:09 ant_work: currently with jlime the nanonote is running on a 2.6.36 kernel Apr 10 09:29:46 but qi-hardware have a much more recent version (3.x I think) Apr 10 09:29:55 apelete: yes http://en.qi-hardware.com/wiki/Ben_NanoNote/Kernel Apr 10 09:31:03 ant_work: it may be possible to run a vanilla kernel: it depends on whether the nanonote patches have been pushed upstream or not Apr 10 09:31:44 ant_work: I will investigate this once I finish porting jlime to oe-core Apr 10 09:32:49 I'm pretty sure you could move to linux-yocto_3.2. Just need to check the mkimage part Apr 10 09:33:06 I'm planning to update the bsp, eg. u-boot and kernel once we have a working jlime distro on oe-core Apr 10 09:34:36 but I'm not there yet, please bear with me until then :) Apr 10 09:34:51 ;) Apr 10 09:35:19 ant_work: been working on jlime on my free time, so it's not going as fast as I would like Apr 10 09:35:42 I have no idea about jlime but I think it would be easier to start with distro-less defaults and a core-image-base to check the kernel is working Apr 10 09:36:39 distro customization would follow later imho Apr 10 09:37:04 ant_work: yes, that's possible, maybe I will try that (to see if vanilla yocto kernel can boot on the nn) Apr 10 09:37:59 something simple, just 3 layers Apr 10 09:38:25 what do you mean by "3 layers" ? Apr 10 09:38:26 oe-core, meta-handheld, meta-initramfs Apr 10 09:39:28 ant_work: and what would you do once you have a working kernel ? it has to boot into some userspace application, right ? Apr 10 09:40:05 (I'm just asking, trying to bounce ideas here) Apr 10 09:40:06 well, nanonote is fundamentally like a Zaurus: you'd test display, touchscreen, keymap, usb.... Apr 10 09:41:15 and let other people do whatever they want with the kernel ? Apr 10 09:41:43 what do you mean? Apr 10 09:42:33 what would you want to do once I built a yocto kernel for the nanonote ? Apr 10 09:42:49 s/want to do/want me to do/ Apr 10 09:43:00 ah, then you can forget about it, the recipe will be automatically updated Apr 10 09:43:09 zero mainainance Apr 10 09:43:15 *maintainance Apr 10 09:44:59 apelete: bluelightning: and we'll finally put under /archive toe obsolete 2.6x kernels Apr 10 09:46:19 ant_work: I get your point, it's quite reasonable actually Apr 10 09:46:42 ant_work: are you personnally using a nn ? Apr 10 09:46:58 no :) Apr 10 09:47:10 just curious because it's a mips Apr 10 09:47:50 and we'd like to have a kexecboot-image for a mips device (promised to ELC) Apr 10 09:48:18 once you'll get your image working I'll bother you with kexec on mips :p Apr 10 09:49:10 ant_work: I see what you're getting at, I just didn't thought of just working on the bsp for the device Apr 10 09:49:18 see, I was almost getting one nanonote but then I fatally found an older armv4 (collie) Apr 10 09:49:35 I'm wasting my free time with it now... Apr 10 09:50:08 well, not totally wasted :) Apr 10 09:50:10 hopefully Apr 10 09:50:17 ant_work: maybe I will shift priorities and focus on delivering a kernel for the nn, it seems like a good idea Apr 10 09:50:53 ant_work: what's kexecboot-image ? what's ELC ? Apr 10 09:51:27 ELC = Embedded Linux Conference Apr 10 09:52:35 apelete: today it's Linux Foundation nowadays Apr 10 09:52:37 http://elinux.org/CEWG_Open_Project_Proposal_2012 Apr 10 09:54:11 ok Apr 10 09:57:27 ant_work: so that's why you would like a kexecboot-image. Apr 10 09:57:47 just google'd kexecboot to see what it is Apr 10 09:58:13 ant_work: how is it different from u-boot ? Apr 10 09:58:49 is a second-stage bootloader atm, it's a kernel Apr 10 09:59:03 (even though a kernel could bot on bare metal...) Apr 10 10:00:03 besides, you could just test kexec on the device from shell. I fear it's still broken on mips :/ Apr 10 10:01:10 but do you still need u-boot on the device to run kexecboot ? Apr 10 10:01:30 u-boot or another bootloader Apr 10 10:02:11 but... http://elinux.org/images/7/7b/LinuxWithoutABootLoader.pdf Apr 10 10:03:46 so the answer is: "Does Linux need a boot loader? Apr 10 10:03:46 Yes (currently :-)" Apr 10 10:05:03 IIRC Linux used to be able to boot on bare metal but that functionality got removed years ago because it was unmaintained, everyone was using a bootloader Apr 10 10:06:13 ant_work bluelightning: what's the goal of having kexecboot on the device if we already have u-boot ? Apr 10 10:06:25 booting multiple images ? Apr 10 10:06:36 apelete: yep, being able to select an image from a nice graphical menu for one :) Apr 10 10:07:02 including booting easily from removable media Apr 10 10:07:26 okay, and I guess that is only possible with a stage2 bootloader, which kexecboot is Apr 10 10:08:26 bluelightning ant_work: it would be like having grub, but for embedded devices, right ? Apr 10 10:08:31 remember u-boot needs a driver to access any device Apr 10 10:09:00 it's size is unjustified, thus people have been pushing the idea of linux as bootloader Apr 10 10:09:39 it's size is unjustified Apr 10 10:09:47 talking about u-boot's size ? Apr 10 10:10:14 yes, and note you need a specific u-boot for each machine Apr 10 10:10:23 yes Apr 10 10:10:43 and cross th efingers on every kernel update... Apr 10 10:11:02 (a.k.a. beagleboard / u-boot odissey) Apr 10 10:11:44 yeah, I'm aware of that problem (already failed at booting a random kernel with a random u-boot on the nn) Apr 10 10:13:10 ant_work: anyway, that's interesting. I will try to build a yocto kernel for the nn and test it Apr 10 10:13:45 in theory you'd just copy the defconfig+patches of 2.6.36 and list them in SRC_URI Apr 10 10:13:50 see Zaurus example Apr 10 10:14:08 this is somehow 'abusing' of the Yocto kernel-tools.. :) Apr 10 10:14:51 let's just say it's for the greater good ;) Apr 10 10:15:15 once you'll understand it, you'll see that you can magically define subsets of .config fragments. I'm doing that with 3.8 Apr 10 10:16:15 ok will try to build a newer kernel, and keep you posted. Apr 10 10:16:22 great Apr 10 11:44:50 where is datadir, libdir, bindir etc. set? Seems to get the full path to sysroot on mac, so I'm thinking of fixing it. Can't seems to find where they are originally set however Apr 10 11:45:39 martiert: bitbake.conf Apr 10 11:45:57 specifically meta/conf/bitbake.conf Apr 10 11:47:44 thanks bluelightning Apr 10 11:50:11 martiert: I guess even more specifically you'd want to be looking at the _native overridden values we're setting Apr 10 11:53:45 bluelightning: thinking of the native/cross/nativesdk bbclass as it says on the top of bitbake.conf? Apr 10 13:16:34 Hello everybody. I have some packages which are not present in the generated OpenEmbedded IPK repository.. Am I missing something ? Apr 10 13:23:30 you me Angstrom repo? Apr 10 13:23:35 you mean ... Apr 10 13:55:47 morning all Apr 10 13:58:36 hi pb Apr 10 13:58:47 pb its afternoon here ;) Apr 10 13:59:00 oh yes Apr 10 14:03:54 woglinde: it's always morning in UGT timezone Apr 10 14:04:40 jama yes Apr 10 14:04:41 http://www.total-knowledge.com/~ilya/mips/ugt.html Apr 10 14:04:52 hence the smiley Apr 10 14:05:53 hi JaMa woglinde Apr 10 14:06:02 hi denisATeukrea Apr 10 14:08:04 btw I saw you had some commit to repair the wizard Apr 10 14:08:18 I hiu pb Apr 10 14:08:29 *hi pb Apr 10 14:09:49 hi denis Apr 10 14:10:06 I guess that firefox and ogg are still broken though Apr 10 14:10:08 hi Apr 10 14:15:06 till later Apr 10 14:16:44 ok Apr 10 14:20:59 denisATeukrea: which one? Apr 10 14:21:27 and I still don't know about firefox/ogg Apr 10 14:22:25 in jansa/test I found a commit that bump some PR to make the wizard work again Apr 10 14:22:27 ok Apr 10 14:23:39 http://git.shr-project.org/git/?p=meta-smartphone.git;a=commit;h=cfe8ee6b5418f51a4ee80791c73900b2d5af5950 Apr 10 14:23:46 it was needed only for e-wm upgrade to 0.17.2 Apr 10 14:23:51 yes Apr 10 14:24:08 with lastest image I got some issue with wizard not running automatically Apr 10 14:24:21 built from shr branches? Apr 10 14:24:31 no, the staging official images Apr 10 14:24:44 then this commit is not related to that Apr 10 14:24:48 ok Apr 10 14:24:52 it's only in shr branch Apr 10 14:24:59 ok Apr 10 14:25:02 not in danny which is used for official staging images Apr 10 14:25:09 ah ok Apr 10 14:25:10 yes Apr 10 14:25:16 I wondered if it fixed it Apr 10 14:25:26 but as you said the problem is not even in danny branches Apr 10 14:25:34 can you check if shr-wizard is in wrong directory? Apr 10 14:25:40 but I don't expect so Apr 10 14:25:43 I'll do after work Apr 10 14:25:47 ping me Apr 10 14:26:18 I'll check buildhistory, mmt Apr 10 14:26:23 ok Apr 10 14:26:34 I checked like yesterday morning Apr 10 14:26:57 om-gta04? Apr 10 14:27:01 yes Apr 10 14:27:30 manually it started though Apr 10 14:27:38 but it didn't made my desktop profile change Apr 10 14:27:56 looks correct, all modules in linux-gnueabi-arm-ver-pre-svn-08 Apr 10 14:28:05 yes it does run manually Apr 10 14:28:23 like DISPLAY=:0 /.../shr-wizard.py Apr 10 15:34:48 re Apr 10 15:37:19 hi florian Apr 10 15:38:37 hey pb_ Apr 10 15:53:32 how can i find out what variables are availabe in the various "conf" files? Apr 10 15:54:08 You don't want to know. there are hundreds, and not all are variables you should ever be setting ,many are used internally Apr 10 15:54:22 best to read the yocto documentation, they cover the *important* variables there Apr 10 15:54:25 rather than all :) Apr 10 15:59:05 waynr: yes, short answer, check the Yocto Project Reference Manual, there is an appendix listing the important variables Apr 10 16:01:30 okay i was curious for example because LAYERDIR did not expand and I want some subdirectory of the current git repository so i can initialize a git.Repo object (from GitPython) to acquire repo version information Apr 10 16:02:00 but i am currently trying to accomplish the same thing from within a recipe so maybe LAYERDIR will expand there Apr 10 16:02:32 waynr: LAYERDIR's value exists while layer.conf is parsed, and it's set to that layer Apr 10 16:02:35 thats it Apr 10 16:02:37 you can't expect it anywhere else Apr 10 16:02:59 okay Apr 10 16:03:28 hmm i think i remember a variable that points to the current recipe's directory Apr 10 16:03:35 FILE_DIRNAME and THISDIR Apr 10 16:03:43 see also ${@os.path.dirname(FILE)} Apr 10 16:03:46 as FILE is the recipe filename Apr 10 16:03:52 (technically the currently parsed file, actually) Apr 10 16:04:15 you'll have to use := if you reference them, to force immediate expansion Apr 10 16:04:29 cool thanks kergoth, that's even more useful :D Apr 10 16:05:38 moin Apr 10 16:27:44 this is resulting in TypeError: object of type 'ExpansionError' has no len() Apr 10 16:27:45 http://paste.debian.net/248710/ Apr 10 16:29:41 waynr: was it you I was discussing this same error with a few weeks ago? Apr 10 16:30:10 in any case that is an error in the error handling code Apr 10 16:30:13 yes, i am currently going through what i recall of the steps we went through Apr 10 16:30:19 which is masking the real error Apr 10 16:30:42 can you remind me which exact bitbake version you are using? we ought to just get this fixed... Apr 10 16:31:14 1.17.1 Apr 10 16:31:53 on yocto branch 1.4_M4.final Apr 10 16:32:01 erh tag* Apr 10 16:33:12 i am looking at bitbake/lib/bb/cooker.py line 1703 Apr 10 16:33:28 it looks like there should be an except block for "ExpansionError" Apr 10 16:34:27 er, nevermind i see bb.data_smart.ExpansionError Apr 10 16:35:26 but there is a bb.data_smart.ExpansionError and a bb.pysh.sherrors.ExpansionError Apr 10 16:36:26 BTW: is anyone still seeing bitbake hangs with bitbake/master? Apr 10 16:37:37 JaMa: I don't think I ever saw them myself, and not seeing them now... Apr 10 16:42:29 waynr: the temporary fix would be to make sure it's not calling len() on something that's an exception Apr 10 16:46:15 yeah i just put a try-except block around the len() call so i print the result, figured out the problem Apr 10 16:47:16 waynr: what was triggering the real error? Apr 10 16:51:29 i forgot to remove the "${@get_layer_version()}" assignment from my layer conf file which is where i originally wrote the function Apr 10 16:52:30 ah ok Apr 10 16:52:58 and after that there was a typo in the function itself which led to another uncaught exception Apr 10 16:54:23 wooo it builds now to see what my /etc/issue looks like Apr 10 16:57:00 waynr: hmm, I suspect the error handling issue may be fixed; if I copy and paste your function into base-files it gives me the proper error during parse Apr 10 16:57:07 fixed in master, I mean Apr 10 16:58:01 okay that's one less thing to worry about Apr 10 16:58:52 waynr: confirmed, with 1.4_M4 and that change I get the error you mentioned earlier Apr 10 16:58:59 indeed, phew :) Apr 10 17:03:16 bluelightning: just out of curiosity, is your nickname a reference to the PSX game Suikoden? Apr 10 17:03:30 waynr: actually, it isn't... :) Apr 10 17:04:13 sounded cool at the time (many years ago) and it kind of stuck Apr 10 17:29:23 I get on my archlinux-box a libgcc failure. Someone has this issue? -> http://pastebin.com/2hiehrwU Apr 10 17:31:43 silviof: either you have a hardware issue or this is a bug in the version of gcc in archlinux Apr 10 17:32:19 (hardware issue being bad RAM, overheating CPU or something inbetween) Apr 10 17:33:02 bluelightning: thx - i will try a chroot environment :-P Apr 10 17:33:21 silviof: chroot won't help unless you supply a different version of gcc inside the chroot... Apr 10 17:33:41 yes - i mean a debian or ubuntu installation in a chroot environment Apr 10 17:33:56 I see, I guess that may help Apr 10 17:34:18 bluelightning: thx again :-) Apr 10 17:44:18 ouch: http://lwn.net/Articles/546840/ Apr 10 17:48:14 bluelightning: damn Apr 10 17:49:12 timing is a bit unfortunate as well... "this stuff is gone... oh btw I'm going away for a week" Apr 10 17:49:38 (not that I blame the guy for taking his holiday, just bad timing) Apr 10 17:51:53 bl uha Apr 10 17:54:29 ? Apr 10 17:54:40 vfp code removed Apr 10 17:55:30 right Apr 10 17:55:33 bbl Apr 10 18:52:11 Hey guys, I'm still trying to figure out how to configure my bbappend for a kernel so that opkg does not think it can upgrade it from the distribution feeds, any suggestions? Apr 10 18:54:00 spacecolonyone arent there serval examples which takes the kernel number adds one in the bbapend? Apr 10 18:54:08 s Apr 10 18:55:07 Right, my understanding what that I set MACHINE_KERNEL_PR_append='foo' and that distinguishes it Apr 10 18:56:32 but that doesn't seem to mark my image as distinct Apr 10 18:57:46 opkg wants to take me from 3.2.28-r122afoo to 3.2.28-r122a.1 Apr 10 18:58:59 What am I missing? Apr 10 18:59:29 I cannot find recipes for the bitstream vera, dejavu etc. fonts Apr 10 19:00:10 does xorg-minimal-fonts contain them? Apr 10 19:09:31 woglinde: any further pointers? Apr 10 19:09:44 no Apr 10 19:09:48 test some schemes Apr 10 19:10:09 what exactly has 3.2.28-r122afoo Apr 10 19:10:17 you need r122b Apr 10 19:12:38 then when r122c is out it will still report as upgradable. What I'm trying to learn is how this chain of upgradability is determined. I then want to break it. Apr 10 20:22:00 hello Apr 10 21:29:06 So I'm trying to overwrite a file in the source archive for a recipe I'm making. I've tried listing it as a file in SRC_URI and also FILESPATH_prepend :=${THISDIR}/${PN} in my recipe, but neither work. If it was a git source, I would make a patch and then add the patch to SRC_URI, but that doesn't seem to work here. Suggestions or examples? Apr 10 21:36:14 spacecolonyone: "that doesn't seem to work here" why? Apr 10 21:36:48 the source isn't under version control Apr 10 21:37:27 at least that is what I'm guessing Apr 10 21:38:07 and? Apr 10 21:38:15 bitbake will unpack source archive and apply patch Apr 10 21:38:41 but I can't use git to generate my diff Apr 10 21:39:01 and as I type that I realize I could probably just use diff -u Apr 10 21:39:11 on the unversioned data Apr 10 21:39:27 yes, patches have been used for decades, long before git even existed Apr 10 21:39:30 somehow we managed Apr 10 21:39:41 * spacecolonyone is an astronomy Apr 10 21:39:47 damn autocorrect Apr 10 21:39:53 astronomer Apr 10 21:40:06 actually that was my typing, crap Apr 10 21:40:29 ok, all that said, that probably isn't the correct way to handle this Apr 10 21:40:54 diff -uNr is right way Apr 10 21:41:28 nm that last comment, I'm confusing myself Apr 10 21:41:54 was thinking the patch would be applied after configure, but that makes no sense Apr 10 22:29:36 thanks guys, got that working. Apr 10 22:30:49 Now back to a previous issue: I'd tried changing PR for my kernel again to prevent opkg update from reporting it as upgradable. Now I get the following error: Package version for package kernel-modules went backwards which would break package feeds from (0:3.2.28-r122ac to 0:3.2.28-ca) Apr 10 22:34:06 Is the only way to prevent opkg from reporting updates are available to not use a bbappend for my kernel and copy/rename the recipe? Apr 10 22:45:14 i am currently trying to create a qt4 image based on a headless image that i have so i "include emac-image.bb" but the rootfs_ipk_do_rootfs_append function which works for that image does not seem to get called for the new image Apr 10 22:46:59 actually there are many things which do not actually seem to get included from the base image Apr 10 22:47:10 wonder what i am doing wrong here Apr 10 22:51:10 ah, i see here that is not the correct way to extend the functionality of the image: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#usingpoky-extend-customimage Apr 10 22:52:31 maybe using an append on rootfs_ipk_do_rootfs is not the right way to accomplish this goal Apr 10 22:54:25 waynr: no, that wouldn't be the right way Apr 10 22:54:42 waynr: btw you should almost always use "require" and not "include" Apr 10 22:54:59 since "include" will continue without any error if the file can't be found Apr 10 22:57:21 ParseError: Could not include required file emac-image.bb Apr 10 22:57:57 hmm doesn't say it can't find the file just that it can't include (which could mean it can't be found) Apr 10 22:59:08 it does mean that Apr 10 22:59:43 if the file is not in the same directory you will need to use the full path from the base of the layer to the file Apr 10 23:00:19 (since it traverses BBPATH) Apr 10 23:02:54 cool, that works thanks again bluelightning Apr 10 23:03:00 np Apr 10 23:40:41 where is the proper place to override PREFERRED_PROVIDER_virtual/kernel for a particular machine? Apr 10 23:54:00 is it acceptable that a native tool built by OE contains internal absolute paths to the native sysroot? Apr 10 23:55:36 dv_: the paths get mangled in and out Apr 10 23:55:49 for instance, lets say that qmake hardcodes the path to the .pri files Apr 10 23:55:51 its not ideal, but it does a find / sed on the files Apr 10 23:55:58 and qmake is a binary Apr 10 23:56:13 for binaries, its a definite problem, we usually work around it by creating wrappers that set env vars and the like Apr 10 23:56:19 see uses of create_wrapper in the metadata Apr 10 23:56:50 have there been instances where no env vars were usable? for example, because the tool does not even look at env vars? Apr 10 23:56:55 I guess in that case only a patch helps Apr 10 23:57:29 yep, patch it to either be fully relocatable itself, or to obey vars or some other mechanism Apr 10 23:57:34 if there's a cmdline arg that could be handled too Apr 10 23:57:43 ah, thats the term - relocatable Apr 10 23:58:06 it'd be ideal if everything was relocatable, there have been efforts in that direction in the past, by projects like autopackage Apr 10 23:58:17 binreloc can be helpful if you want to move in that direction, particularly for libraries Apr 10 23:58:27 well, the qt developers dont seem to care about this :/ Apr 10 23:59:22 yeah, lots of folks don't, sadly **** ENDING LOGGING AT Thu Apr 11 02:59:59 2013