**** BEGIN LOGGING AT Mon Oct 17 02:59:57 2011 Oct 17 04:25:55 Has anyone run into ckermit error: ?SET SPEED has no effect without prior SET LINE ? Oct 17 06:46:12 good morning Oct 17 07:44:25 Is there a way to let bitbake detect when a recipe has been changed so that it will automatically rebuild that recipe when the image is built? C Oct 17 07:45:02 JaMa|Off, you here? Oct 17 07:57:53 ka6sox: yup Oct 17 08:01:49 JaMa|Off: hey Oct 17 08:06:12 JaMa|Off: wrt X we are not too far http://paste.debian.net/137190/ Oct 17 08:07:27 JaMa|Off: this was removing meta-oe's xserver-nodm-init and xserver-common Oct 17 08:07:33 bluelightning: hi Oct 17 08:07:43 hi ant_work Oct 17 08:08:13 morning all Oct 17 08:19:59 Good morning. I've been recently switching from oe-classic to oe-core. Nice stuff. In oe-classic I was creating a distribution of my own, based on a overlay setup using the already existing Angstrom configs. Oct 17 08:22:52 This was all working fine, until I found out I was working on an outdated branch of oe, pretty stupid. However, now I'd like to `port' my work to the oe-core way but I can't really find any documentation on creating my own distro. I guess the best way would be to use the setup-scripts of Angstrom as a start? Oct 17 08:23:16 Thanks for the help last week btw, I really appreciate it! Oct 17 08:25:25 gm Oct 17 08:39:02 mornin Oct 17 10:22:47 Could someone look at my question in the backlog? (at 10:15h) Thanks! Oct 17 12:50:45 khem: I saw a build failure in gcc http://paste.debian.net/137338/ Oct 17 13:32:31 hi all Oct 17 13:32:47 i got some guidance on initramfs Oct 17 13:32:49 need* Oct 17 13:33:55 i know i need .cpio.gz to pass into my kernel this can be done using compile virtual/kernel tweaking some values in make ARCH=arm menuconfig Oct 17 13:34:21 but i have some troubles making .cpio.gz Oct 17 13:34:32 i followed the tutorial at http://www.3open.org/d/oe/build_kernel_with_inline_initramfs Oct 17 13:34:43 instead of mtd-utils Oct 17 13:34:51 i build helloworld instead Oct 17 13:35:17 so instead of IMAGE_INSTALL += "mtd-utils i type IMAGE_INSTALL += "helloworld Oct 17 13:35:26 is what i am doing correct? Oct 17 13:39:43 you have just to add INITRAMFS_IMAGE = "initramfs-helloworld-image" in the kernel recipe Oct 17 13:40:26 initramfs-helloworld-image <--- is there such recipe? Oct 17 13:40:33 i have to create one right? Oct 17 13:41:00 yes, following the example you gave Oct 17 13:41:06 and where is the kernel recipe? Oct 17 13:41:20 isnt it at make ARCH=arm menuconfig Oct 17 13:41:24 at general Oct 17 13:41:33 under sources right? Oct 17 13:41:35 what is the virtual/kernel you're building? Oct 17 13:41:51 i used bitbake virtual/kernel Oct 17 13:41:55 to build my kernel first Oct 17 13:42:05 for sarge-at91 Oct 17 13:42:06 whic kernel was built? Oct 17 13:42:32 linux ? linux-yocto? linux-omap? ?? Oct 17 13:42:44 arm Oct 17 13:42:52 right Oct 17 13:42:53 ? Oct 17 13:43:30 you are using th e'old' openembedded-classic, isn't? Oct 17 13:43:44 what u mean old? Oct 17 13:44:01 then just add it to recipes/linux/linux.inc Oct 17 13:44:28 and be sure CONFIG_INITRAMFS_SOURCE="initramfs.cpio.gz" Oct 17 13:45:22 err i am lost Oct 17 13:45:32 khem: btw: there is ICE in samba build with latest gcc-4.6, have we talked about this one already, iirc we did.. http://paste.pocoo.org/show/493821/ Oct 17 13:47:32 ksupar: your tutorial is ok, I'm just telling you that you can embed an image just declaring INITRAMFS_IMAGE and without doing copy by hand Oct 17 13:48:09 so the tutorial is correct right? as in the steps that i mention? Oct 17 13:49:22 khem: looks like we did .. but it's not fixed in meta-oe Oct 17 13:49:25 khem: http://git.openembedded.org/openembedded/commit/recipes/samba/samba_3.2.15.bb?id=4ba7aa07c0dcd28f94515ff9927e2a04403fcf15 Oct 17 13:50:58 khem: but upstream bug says resolved :/ Oct 17 13:57:55 erm ant the variable INITRAMFS_IMAGE is at kernel.bbclass? Oct 17 13:59:24 is evaluated there Oct 17 14:00:09 add itthere if you're lazy ;) Oct 17 14:00:24 so at kernel.bbclass i add my initramfs-helloworld-image at INITRAMFS_IMAGE at kernel.bbclass? Oct 17 14:01:05 #30 INITRAMFS_IMAGE ?= "" Oct 17 14:02:07 yup correct i found it thanks Oct 17 14:02:16 sorry ant for troubling Oct 17 14:02:25 u help alot Oct 17 14:02:27 thanks Oct 17 14:02:31 np, I should write an item on wiki Oct 17 14:02:34 :/ Oct 17 14:02:55 i will mention your name on my report :D Oct 17 14:03:09 what do you add to your image? Oct 17 14:03:18 static binaries? Oct 17 14:03:26 yea for now Oct 17 14:03:33 my end product is a web server Oct 17 14:03:45 but i still trying to get my hello world working :( Oct 17 14:04:08 heh..keep on Oct 17 14:05:01 recipes/udev/udev_162.bb <--- the source seems to be down Oct 17 14:14:21 yay i manage to create cpio.gz Oct 17 14:14:33 now for the final compile with my kernel Oct 17 14:14:33 hope it works Oct 17 14:14:35 *pray Oct 17 14:19:25 ksupar: if you have a real bootloader you can supply the cpio as initrd= parameter Oct 17 14:19:47 better: you can stack more initramfs one over the other Oct 17 14:20:05 it's interesting stuff Oct 17 14:20:18 real bootloader? Oct 17 14:20:20 u mean uboot? Oct 17 14:20:27 yes, can cope Oct 17 14:20:59 it will be easier for you to debug Oct 17 14:21:33 you'll embed the final image Oct 17 14:22:41 btw ant Oct 17 14:22:45 i am curious Oct 17 14:22:50 for example Oct 17 14:22:56 i load my kernel at 0x20008000 Oct 17 14:23:04 but when it uncompress Oct 17 14:23:12 it overwrite my rfs Oct 17 14:23:47 wait with u-boot I found this http://lists.denx.de/pipermail/u-boot/2011-July/096096.html Oct 17 14:23:54 i am curious how to know how much is the size after it uncompress? Oct 17 14:24:00 it looks like it expect an address Oct 17 14:24:45 so that method works too? Oct 17 14:24:46 about your question there is a 'multiplier' hardcoded in kernel, thus there is a max size for th einitramfs iirc Oct 17 14:25:06 i go try now Oct 17 14:37:42 the file should be cpio.gz or .cpio? Oct 17 15:41:51 has anyone seen a bunch of errors like this: rm: cannot remove `Python-2.6.6/Lib/test/decimaltestdata': Directory not empty Oct 17 15:42:03 this happens during rm_work Oct 17 15:44:40 cbrake: I havent seen this error thus far Oct 17 15:45:10 khem: very strange -- almost every package is giving me this, and then the next time I run bitbake it works fine Oct 17 15:45:21 khem: trying to build a meta-micro image Oct 17 15:46:00 i have similar problem also Oct 17 15:46:08 bitbake XXX fail Oct 17 15:46:15 bitbake XXX again works Oct 17 15:47:49 cbrake: oe-core default testing doesnt use rm_work by default thats one reason I think rm_work becomes a baddy Oct 17 15:48:40 khem: yeah, I disabled it and things are working much better -- good thing I have a 2TB scratch drive :-) Oct 17 15:48:55 ksupar: you describe something very generic that could happen for many reasons. Unless you specify clearly whats thats causing it, its hard to say Oct 17 15:49:18 cbrake: yes and builds will be faster too Oct 17 15:49:24 ok i show u later Oct 17 15:49:28 i building some stuff now Oct 17 15:49:52 cbrake: but rm_work is handy on some boxes where my HD is only 50G free Oct 17 15:50:24 khem: definitely Oct 17 15:50:40 cbrake: I think its something we should discuss. weather we want to promote it with oe-core Oct 17 15:50:43 NOTE: Handling BitBake files: / (6562/7200) [91 %]ERROR: 'virtual:native:/home/lkp/oe/openembedded/recipes/glib-2.0/glib-2.0_git.bb' while parsing /home/lkp/oe/openembedded/recipes/glib-2.0/glib-2.0_git.bb ERROR: 'virtual:native:/home/lkp/oe/openembedded/recipes/glib-2.0/glib-2.0_git.bb' while parsing /home/lkp/oe/openembedded/recipes/glib-2.0/glib-2.0_git.bb NOTE: Handling BitBake files: \ (7199/7200) [99 %]Command execution failed: Tra Oct 17 15:50:58 File "/home/lkp/oe/bitbake/lib/bb/command.py", line 85, in runAsyncCommand self.cooker.updateCache() File "/home/lkp/oe/bitbake/lib/bb/cooker.py", line 829, in updateCache if not self.parser.parse_next(): File "/home/lkp/oe/bitbake/lib/bb/cooker.py", line 983, in parse_next def parse_next(self): File "/home/lkp/oe/bitbake/lib/bb/cooker.py", line 1022, in parse_next raise ParsingErrorsFound ParsingErrorsFound Oct 17 15:51:28 ~pastebin Oct 17 15:51:29 [~pastebin] A "pastebin" is a web-based service where you should paste anything over 3 lines so you don't flood the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://bin.cakephp.org/ , http://asterisk.pastey.net/ , or install pastebinit with yum or aptitude. Oct 17 15:51:45 ok sorry Oct 17 15:52:57 you have to use -DDD option and see whats going on why it fails to parse is not clear from the paste Oct 17 15:59:08 otavio: re. the gcc build error you are seeing its due to flex-native not being available Oct 17 15:59:28 can you make sure if dependency on flex-native exist ? Oct 17 16:03:52 khem: i think it was an autobuilder issue; we will test it once again, from scratch, and then I confirm the issue Oct 17 16:04:02 khem: for now, assume it was our fault Oct 17 16:04:10 JaMa|Off: re. the problem you reported. We do not have that patch yet in our gcc 4.6 tree. So I will prepare a patch if you can test Oct 17 16:04:24 otavio: ok Oct 17 16:04:29 mario-goulart: please remove tmp, sstage and pseudodone on our autobuilder to check if it works Oct 17 16:08:09 otavio: ok Oct 17 16:12:48 hi all regarding the helloworld image which format will it produce? Oct 17 16:13:02 i mean the default Oct 17 16:17:24 kaylessa: format of what ? Oct 17 16:17:54 the rfs format that it will be produce Oct 17 16:18:24 depends on what you selected or your distro selected for you Oct 17 16:19:37 its IMAGE_FSTYPES that determines it Oct 17 16:19:37 i have my kernel and my helloworld image Oct 17 16:19:48 but i dunno how to load my helloworld image Oct 17 16:19:53 i load my kernel at 0x20008000 Oct 17 16:19:59 it manage to bootm Oct 17 16:20:03 but cant find a rfs Oct 17 16:20:22 error "root=" Oct 17 16:20:35 hmmm so whats your kernel cmdline Oct 17 16:20:43 i know i can setenv bootargs root= Oct 17 16:20:51 but i dunno what to input in Oct 17 16:21:22 where is your rfs Oct 17 16:21:33 is it on a partition ? or nfs Oct 17 16:21:44 erm i got no idea Oct 17 16:21:47 trying to learn now Oct 17 16:21:56 my helloworldimage end with .tar.gz Oct 17 16:22:05 how do i make it work as rfs Oct 17 16:22:59 ok then you can either write it to a flash partition on your board or you can untar it into a nfs location that the board can see and boot from Oct 17 16:23:19 and there are endless documents on google on how to boot from nfs Oct 17 16:23:28 i know about nfs Oct 17 16:23:28 but Oct 17 16:23:40 i cant use it Oct 17 16:23:53 due to some design specification Oct 17 16:23:57 ok Oct 17 16:24:14 i have a sample file Oct 17 16:24:15 then your board should have some storage attached or SDCARD Oct 17 16:24:30 nope dont have Oct 17 16:24:30 only 32mb of sdram and 2mb of flash Oct 17 16:24:33 thats all i have Oct 17 16:24:58 2MB of flash ? Oct 17 16:25:18 yup Oct 17 16:25:30 it could be challanging to fit in linux there Oct 17 16:25:48 but you can try to strip down kernel and use a mimimal initramfs Oct 17 16:26:01 yea i figuring out initramfs now Oct 17 16:26:10 i followed this tutorial Oct 17 16:26:14 but doesnt seems working Oct 17 16:26:21 http://www.3open.org/d/oe/build_kernel_with_inline_initramfs Oct 17 16:26:39 i changed IMAGE_INSTALL += "mtd-utils" --> IMAGE_INSTALL += "helloworld" Oct 17 16:27:03 the kernel didnt get compressed with my initramfs-helloworld Oct 17 16:27:51 i am trying to load in Sdram first before trying to figure out how to squeeze into 2mb flash Oct 17 16:29:49 gm Oct 17 16:43:35 I am unable to fetch util-linux-ng-2.18 to build a devimage-image. http://pastebin.com/p7mwZ9Jv Though I have the tarball. Is it possible to place the file in the sources dir and force bitbake to look there before attempting to fetch? Oct 17 16:57:38 yea can Oct 17 16:57:45 thats what i always do when i unable to fetch Oct 17 16:57:57 change the source to a working one Oct 17 17:01:51 kaylessa: do you change the recipe? Oct 17 17:02:01 ya i change it Oct 17 17:02:25 change to a working source Oct 17 17:10:57 Does anyone know why and/or how long mirrors.openembedded.org will be down? Oct 17 17:13:22 khem: seems it was our fault indeed; it built fine. Oct 17 17:13:32 is someone seeing build failures to openssl ? Oct 17 17:13:33 oe-core Oct 17 17:14:37 in fact, meta-oe Oct 17 17:17:42 khem: you seem to have change the recipe latelly; can you try to build it from scratch removing sstate? Oct 17 17:17:49 khem: it seems to fail in do_install Oct 17 17:18:10 * cbrake likes! -- micro image with mtd-utils is only 1MB :-) Oct 17 17:19:43 unfortunately, it is not working yet: /bin/sh: symbol '__exidx_end': can't resolve symbol Oct 17 17:20:25 cbrake: uclibc? Oct 17 17:20:29 otavio: yes Oct 17 17:20:35 cbrake: awesome Oct 17 17:21:08 * otavio does another openssl build to check if it keeps failing Oct 17 17:21:45 pb_: have you tested micro recently? Oct 17 17:22:07 pb_: or would you happen to have the git hash for the oe-core version you last tested it with? Oct 17 17:25:27 when I use my sdk as an external toolchain, bitbake tells me that "multiple providers are available for runtime locale-base-en-us (eglibc, glibc)" Oct 17 17:25:46 then it tries to build eglibc Oct 17 17:26:14 How can I figure out where it is deciding this? Oct 17 17:30:52 it seems to be a parallel install problem Oct 17 17:30:54 humm Oct 17 17:31:41 parallel install has been enabled in oe-core Oct 17 17:31:48 it might not work for everything Oct 17 17:31:56 search ml to disable it Oct 17 17:34:03 khem: right; it seems openssl fails sometimes Oct 17 17:34:12 khem: I built it again and it worked Oct 17 17:34:19 khem: but seems fragile Oct 17 19:11:00 hi all, anyone awake ? Oct 17 19:18:25 hmm, thats odd, glib-2.0-native failure Oct 17 19:20:45 * cbrake tries a elibc micro build for size comparison ... Oct 17 19:20:52 *eglibc Oct 17 19:22:23 anyone used to kexecboot ? Oct 17 19:27:56 lautriv, yes I tried Oct 17 19:28:02 it was with oe.dev tough Oct 17 19:28:51 khem: yec, I can test samba for you Oct 17 19:28:52 yes Oct 17 19:29:11 JaMa|Off: ok. I will see to it tonight Oct 17 19:29:19 and you might see a patch on ml soon Oct 17 19:29:23 may be tomorrow Oct 17 19:30:21 no problem.. I've forced arm mode in it again to resolve it temporary Oct 17 19:33:41 GNUtoo, for a long time i had a running kexec (internal, SD and CF) unfortunately the CF died and this thingie collected dust. now while i'll put it back to work, it seems i miss something. Found some statements in the net this kexecboot can be tuned by an additional boot.cfg but whatever i poke in, no reaction..........maybe the file had another name ? ( AFAIR i was fine without any configfile before) Oct 17 19:35:32 | ./.libs/libglib-2.0.so: undefined reference to `qsort_r' Oct 17 19:35:38 anyone run into this one before? Oct 17 19:36:42 lautriv, basically it works like that Oct 17 19:36:45 kergoth, yes morphis Oct 17 19:36:58 it scans the filesystem for a boot.cfg Oct 17 19:37:04 *filesystems Oct 17 19:37:19 not all filesystems are supported Oct 17 19:37:25 for instance ubifs is not supported Oct 17 19:37:55 http://kexecboot.org/documentation/how_to_write_config for the config file format Oct 17 19:46:38 kergoth: you need to remove the qsort-patch for the native version of the recipe Oct 17 19:46:55 kergoth: and add it again for cross version until this is fixed id HEAD Oct 17 19:46:59 s/id/in/ Oct 17 19:48:12 GNUtoo, what could be the reason for "unknown blockdevice (0,0)" if i have all neccessary drivers/fs build-in . AFAIK unknown block comes only if there is no driver for the device. Oct 17 19:48:37 bbl Oct 17 20:02:38 morphis: ah, thanks Oct 17 20:03:50 micro image with eglibc+mtd-utils is 2MB, vs 1MB for uclibc -- still not bad Oct 17 21:12:45 pb_: ping Oct 17 21:45:26 Yay for the OE patch on pwclient being like #5 on google for how to use pwclient Oct 17 21:45:31 and the only one with .pwclientrc Oct 17 21:45:37 s/OE patch/OE wiki page/ Oct 17 21:46:28 what is pwclient at all? Oct 17 21:46:52 ah patchwork client Oct 17 21:46:53 okay Oct 17 23:43:39 I am trying to start with openembedded but I am facing this problem http://wklej.org/hash/9db843b275d/ Oct 17 23:43:50 any ideas? :/ it's just spewing these errors Oct 17 23:44:03 I have set up openembedded according to the "Getting started" guide Oct 17 23:47:38 RushPL: looks like too new bitbake for too old oe metadata Oct 17 23:48:15 actually probably the other way around .. I had bitbake 1.8 ... it was mentioned as the nwest in the guide but now I see it is the version from 2009 Oct 17 23:49:22 JaMa|Off: cause I got the metadata from git Oct 17 23:50:24 okay now it fails even worse :) Oct 17 23:50:56 should I get openembedded from: git clone git://git.openembedded.org/openembedded ? Oct 17 23:51:31 and if yes, what bitbake version? now with bitbake 1.12 I get an error that stops the bb files scan Oct 17 23:51:36 this is old OE-classic Oct 17 23:51:44 new is in multiple layers Oct 17 23:51:49 oe-core/meta-oe/meta-* Oct 17 23:52:06 look at ML for pointers do documentation.. Oct 17 23:52:12 * JaMa|Off going to bed Oct 17 23:52:18 gnight all Oct 18 00:29:27 NOTE: Task failed: Fetch failed: Unable to fetch URL ftp://invisible-island.net/ncurses/5.7/ncurses-5.7-20110108-patch.sh.bz2;apply=yes;name=p20110108sh from any source. Oct 18 00:29:29 help:) Oct 18 01:53:01 hi. I am trying to use icecc. I am following the instructions in http://www.openembedded.org/wiki/Speeding_up_compilation. but it is not working. Oct 18 01:53:13 could someone help me on that? Oct 18 01:54:00 the error that I get is Oct 18 01:54:02 ERROR: Unable to open conf/bitbake.conf Oct 18 01:55:16 you need a sane environment setup first Oct 18 01:56:36 khen, the env that I have is working. I use it to build my projects. Oct 18 01:56:50 khen, thanks for the fast reply! Oct 18 01:57:00 if you're getting that error, your oe setup is busted, and has nothing at all to do with icecc Oct 18 01:57:04 as khem pointed out Oct 18 01:58:33 kergoth, I see. I was thinking that it was related because I am able to build my projects without icecc. I only saw that error after adding the icecc related lines. Oct 18 01:59:03 that error means your bbpath is wrong, which means your setup is broken, icecc or no. Oct 18 01:59:17 kergoth, do you have an idea of what is wrong? Oct 18 01:59:26 sorry for the very newbie questions... Oct 18 01:59:29 I left my crystal ball at home Oct 18 01:59:41 I don't have your environment in front of me, so no, afraid not.. Oct 18 02:00:17 kergoth, hehehe. I understand. Oct 18 02:00:52 you'd need to provide a heck of a lot more information for anyone to be able to do anything more than guess at this point Oct 18 02:01:16 we don't even know if you're using oe or oe-core, much less how your exact setup is done Oct 18 02:01:41 hi anyone can explain after virtual/kernel Oct 18 02:02:07 kaylessa: you'll need to elaborate. what about it? Oct 18 02:02:09 whats the difference between the uimage in /deploy and the uimage in /arm/boot/ Oct 18 02:02:33 i am able to load the uimage in /deploy but unable to load the uimage in /arm/boot Oct 18 02:02:42 what exactly is "/arm/boot"? Oct 18 02:02:47 this path isn't familiar to me at all Oct 18 02:03:09 its inside my work/sarge-linux/linux/linux Oct 18 02:03:43 work/sarge-linux/linux/linux/arm/boot/uimage Oct 18 02:04:29 why are you using the files from workdir directly at all? Oct 18 02:05:12 i am just curious Oct 18 02:05:17 isnt it suppose to be the same file Oct 18 02:05:48 the files in deploy are populated from the source tree, yes. read the recipe and classes if you want to see exactly what it's doing Oct 18 02:07:07 so the one in deploy is the correct one? Oct 18 02:07:39 the contents of workdir are intermediate. temporary. you shouldn't rely on them Oct 18 02:07:58 another thing is Oct 18 02:08:04 at the deploy theres two uimage Oct 18 02:08:13 one is the link uimage and one is the actual uimage Oct 18 02:08:42 the link uimage isnt important right **** ENDING LOGGING AT Tue Oct 18 02:59:57 2011