**** BEGIN LOGGING AT Tue Jul 06 02:59:58 2010 **** BEGIN LOGGING AT Tue Jul 06 08:26:14 2010 Jul 06 08:42:29 morning Jul 06 08:44:06 ka6sox: I know mysql if it is still needed Jul 06 08:46:00 Jay7, how about replication? Jul 06 08:46:02 :D Jul 06 08:46:22 ka6sox: that is worst but I can look :) Jul 06 08:47:08 good morning Jul 06 08:47:28 morning mckoan Jul 06 08:48:05 Jay7, i've been working on this for 16+hrs and am getting pretty groggy... Jul 06 08:48:26 Jay7, can you please verify that git and bugs are both working? Jul 06 08:50:57 moment Jul 06 08:51:55 git pull is downloading changes Jul 06 08:52:04 I can login to bugzilla Jul 06 08:52:32 Jay7, excellent..can I catch you in about 8 hours or so? Jul 06 08:52:49 I'd really appreciate it as I can't hardly keep my eyes open. Jul 06 08:52:56 yes I should be here Jul 06 08:53:00 ping'me :) Jul 06 08:53:09 thanks! Jul 06 08:53:13 catch you laters. Jul 06 08:53:18 ok Jul 06 11:53:19 hi, seems like i'm missing a '/' in glibc_2.9.bb Jul 06 11:53:54 in the SRC_URI, file://etc/ld.so.conf should be file:///etc/ld.so.conf Jul 06 11:54:21 if the slash is missing, bitbake tries to install the file to / Jul 06 11:55:08 when there is an absolute path in the arguments of os.path.join, the last absolute path is used and everything before that is thrown away Jul 06 13:25:26 whats the best recipe to create a cross sdk from ? Jul 06 13:28:26 You are a brave man, Crofton Jul 06 13:28:34 heh Jul 06 13:28:42 * mwester is frightened of toolchains. Jul 06 13:29:30 yeah, but the problem needs to be dealt with ... Jul 06 14:19:04 * Jay7 is starting to look for some work or startup in Europe with relocation.. Jul 06 14:23:55 03Sebastian Krzyszkowiak  07org.openembedded.dev * rfec235b699 10openembedded.git/recipes/tasks/task-shr-feed.bb: Jul 06 14:23:55 task-shr-feed: add xf86-video-fbdev Jul 06 14:23:55 Signed-off-by: Klaus Kurzmann Jul 06 14:23:56 03Sebastian Krzyszkowiak  07org.openembedded.dev * rfd41e21bf7 10openembedded.git/recipes/blipomoko/blipomoko_git.bb: Jul 06 14:23:56 blipomoko: bump new version Jul 06 14:37:54 Hi all. Jul 06 14:50:02 Angelox_123: hello Jul 06 14:50:17 Hello Marex. Jul 06 15:11:19 Hi guys, I'm sorry if someone answered me earlier but I didn't catch it so I'm gonna ask again : I'm not able to make my webcam use uvcview driver instead of the gpsca driver. The webcam uses this driver on my PC, so I know the cam has support. Can anyone help me Jul 06 15:28:54 screwgoth: Take out the uvcview driver? Poke udev a bit? Jul 06 15:30:57 no no ... I want uvcvideo driver .... Jul 06 15:31:00 don't want gspca Jul 06 15:31:11 Take out gspca, poke udev :) Jul 06 15:31:34 :-) gotcha ... lemme try Jul 06 15:33:18 * Tartarus pokes libxml-namespacesupport-perl to see if he can make it not fill up his drive in do_configure Jul 06 15:37:04 blindvt: the .config is here www.uclibc.org/~kraj/x86.config Jul 06 15:40:18 * mwester suggests using a VERY sharp stick when poking at perl cross-compilation stuff. :( Jul 06 15:41:13 * khem suggestes a chain saw Jul 06 15:49:22 Yeah, crap Jul 06 15:49:33 Looks like we need to fake a cpan config Jul 06 16:05:47 hi Jul 06 16:05:54 where can I get help about mtdutils? Jul 06 16:06:15 I'll try in the openmoko channel Jul 06 16:06:44 #mtd? Jul 06 16:07:07 thanks a lot Jul 06 16:07:13 what is a problem? Jul 06 16:07:33 flash_info /dev/mtd3 => Device /dev/mtd3 has 0 erase regions Jul 06 16:07:44 #mtd doesn't exist btw Jul 06 16:07:52 cat /proc/mtd Jul 06 16:07:52 it's on oftc, not freenode Jul 06 16:08:01 ah ok Jul 06 16:08:03 see the mtd webpages Jul 06 16:08:26 http://pastebin.com/3CLVjhVh Jul 06 16:08:28 ok thanks Jul 06 16:08:38 hm any idea why glibc-initial_2.9 is trying to unpack ld.so.conf to / ? i'm getting failures at do_populate_sysroot Jul 06 16:08:41 http://pastebin.com/cNue4zhF Jul 06 16:08:46 i'm behind the times on oe-dev staging now Jul 06 16:09:30 GNUtoo|laptop: mtd3 starts at same place as mtd4 Jul 06 16:09:35 jconnolly: some guy posted on the mailing list about that. sounds like its SRC_URI is wrong. Jul 06 16:09:36 ah ouch Jul 06 16:09:41 no.. size it is Jul 06 16:10:04 ah ok Jul 06 16:10:23 ah ok pb_ thx, will look Jul 06 16:10:33 it's the same for all mtd btw Jul 06 16:10:41 is it normal to have 0 erase regions? Jul 06 16:10:49 I don't think so Jul 06 16:13:08 jconnolly, file://etc/ld.so.conf \ Jul 06 16:13:11 strange no? Jul 06 16:13:36 ah no Jul 06 16:13:40 there is an /etc dir Jul 06 16:13:43 GNUtoo|laptop: yes, i gave it file:///etc/ld.so.conf and rebuilding now Jul 06 16:13:45 Something related to how the base uri stuff changed? Jul 06 16:13:52 it fixed it Jul 06 16:13:58 ok Jul 06 16:14:09 trying to build from oe-dev now Jul 06 16:14:12 ok Jul 06 16:14:17 too many sugary goodies that i want in oe-dev Jul 06 16:14:26 i.e. modemmanager Jul 06 16:14:41 hi hrw Jul 06 16:14:47 hi jconnolly Jul 06 16:15:04 jconnolly: how goes bug2? I heard rumours that you have ubercool debugplug now Jul 06 16:15:58 very cool. waiting on production-ready hw, yes TheMammal did some work to breakout all samtec pins and do usb passthrough, also new docks will be slick looking Jul 06 16:16:10 things are moving quickly now Jul 06 16:16:30 when can i expect ubuntu on bug2? Jul 06 16:16:31 ;D Jul 06 16:16:58 jconnolly: with non-ubuntu kernel and bug2 I can run it now if you provide hw Jul 06 16:17:07 jconnolly: iirc you have working videomodule now Jul 06 16:17:45 yes, it's excellent. soon enough i'll use bug2 as a workstation i think Jul 06 16:19:07 jconnolly: give 512MB by default Jul 06 16:20:10 what we give now: vram=24M omapfb.vram=0:0M,1:24M,2:0M omapdss.def_disp=dvi omapfb.mode=dvi:1280x1024MR-32@57 Jul 06 16:23:00 jconnolly, will it have 256M or 512M? Jul 06 16:23:06 512 Jul 06 16:23:09 wow!!!! Jul 06 16:23:17 jconnolly: nice Jul 06 16:23:19 yes. let me verify but i believe that was decided Jul 06 16:23:26 if I get one I'll use it as workstation too Jul 06 16:23:50 specially that I know that there is a rootkit in the BIOS of my laptop Jul 06 16:23:52 jconnolly: moving to 37xx with 1GHz too? :D Jul 06 16:23:56 s/that/because Jul 06 16:24:05 jconnolly: or is it too much? :D Jul 06 16:24:07 we're evaluating it now hrw Jul 06 16:24:27 jconnolly: I remember notes on bug2 schematics "this is not needed with 37xx" etc Jul 06 16:24:44 curious, whose 37xx? Jul 06 16:24:51 MMatt Peddicord to Dave, Matt, eng show details 11:01 AM (1 hour ago) Could it be this easy? Switch out some resistors and we can go from 3530 to the 37xx series processors? Jul 06 16:25:02 literally 1 hour ago question was asked hrw Jul 06 16:25:04 ;D Jul 06 16:25:17 Tartarus: TI OMAP Jul 06 16:26:30 Tartarus: 34/35xx were 720MHz max, 36/37xx are 1GHz max now also have faster gpu/dsp Jul 06 16:30:00 ah, k Jul 06 16:30:07 worked on a different 37xx at some point Jul 06 16:32:40 beagleboard xm uses 37xx Jul 06 16:33:33 hi Jul 06 16:33:50 how to configure kdrive to run 18bpp with this options::: Jul 06 16:33:51 ?? Jul 06 16:33:52 -screen WIDTH[/WIDTHMM]xHEIGHT[/HEIGHTMM][@ROTATION][X][Y][xDEPTH/BPP{,DEPTH/BPP}[xFREQ]] Jul 06 16:34:00 i don`t understand... Jul 06 16:34:08 Xfbdev i use... Jul 06 16:34:21 -screen 640x480@18-60? Jul 06 16:34:37 what is 18-60 (60) ? Jul 06 16:34:42 -screen 640x480x18-60 rather Jul 06 16:34:53 i will test. Jul 06 16:35:07 -screen 640x480x18-60 = 640x480 res with 18 bpp at 60 Hz Jul 06 16:35:11 hum... Jul 06 16:35:18 i use Motorola A1200... Jul 06 16:35:31 and screen is 320x240 Jul 06 16:35:38 i dont know the frequency Jul 06 16:35:44 so 240x320x18-60? Jul 06 16:35:51 testing... Jul 06 16:36:45 Xfbdev -screen 240x320x18-60 Jul 06 16:36:46 Fatal server error: Jul 06 16:36:46 no screens found Jul 06 16:36:57 Angelox_123, what fbset gives? Jul 06 16:37:00 hi btw Jul 06 16:37:08 mode "240x320-57" Jul 06 16:37:08 # D: 5.200 MHz, H: 18.571 kHz, V: 56.793 Hz Jul 06 16:37:08 geometry 240 320 240 640 16 Jul 06 16:37:08 timings 192308 20 10 3 2 10 2 Jul 06 16:37:08 accel true Jul 06 16:37:09 rgba 6/12,6/6,6/0,0/0 Jul 06 16:37:11 endmode Jul 06 16:37:26 this is my fbset.. Jul 06 16:39:13 hi? Jul 06 16:39:20 16bit Jul 06 16:39:37 i need 18bit.. Jul 06 16:39:47 18bpp Jul 06 16:39:56 I've just moved to a 64bit build system and am now seeing an issue with a recipe installing its libs into /usr/lib64 which causes the packager to miss the files as $libdir=/usr/lib - is this a common issue? Jul 06 16:50:01 tharvey: never seen that Jul 06 16:50:16 what recipe? Jul 06 16:50:16 hello? Jul 06 16:59:08 Angelox_123: It sounds like you need to go grab a datasheet for your screen so you can figure out the timings and such Jul 06 17:05:33 Tartarus, olsrd recipe is doing this Jul 06 17:05:52 so far its the only one I find if I do a find for lib64 in $TMPDIR/work Jul 06 17:06:34 olsrd recipes builds plugins as .so's and installs them to /usr/lib64 for some reason Jul 06 17:07:24 is it autotools based or other? Jul 06 17:08:43 doesn't seem to use autotools Jul 06 17:10:39 looks like its just an issue in olsrd's Makefiles that I need to patch - perhaps its not a common issue Jul 06 17:13:22 Yeap Jul 06 17:14:49 there's a: '$(MAKECMD) -C lib LIBDIR=$(LIBDIR)' in its makefile that gets expanded to LIBDIR=/usr/lib64 but I can't figure out where LIBDIR gets set to that - probably need to just remove that Jul 06 17:15:23 actually probably need to do an 'export LIBDIR=${libdir}' in recipe Jul 06 17:16:02 khem, i mean these configury patches, #here Jul 06 17:16:59 khem, (not the nptl fixup that i just pushed) Jul 06 17:17:19 blindvt: ok Jul 06 17:17:29 blindvt: I havent checked my mails yet Jul 06 17:17:52 khem, if the configury series is ok, then please push them, i don't have write-access to oe Jul 06 17:18:41 blindvt: yes will do Jul 06 17:18:51 blindvt: btw. did you get my .config Jul 06 17:18:55 which fails ? Jul 06 17:20:28 hi khem Jul 06 17:20:29 khem, not yet, no Jul 06 17:22:12 blindvt: its here http://uclibc.org/~kraj/x86.config Jul 06 17:22:18 ant__: hey dude Jul 06 17:22:49 blindvt: and the fix is here http://uclibc.org/~kraj/0004-nptl-remove-generated-files-from-libpthread.patch Jul 06 17:22:50 now I'll feed family, then start battle against-kexec-tools ;) Jul 06 17:23:10 hopefully is just the missing autoconf > 2.63 patch Jul 06 17:25:08 khem, ah that one. i386 is NIH Jul 06 17:25:13 khem, use i486 Jul 06 17:25:39 khem, (or implement lowlevellock for i386) Jul 06 17:25:43 blindvt: hmm for nptl infact i386 should be not used Jul 06 17:26:05 (somehow) Jul 06 17:26:43 blindvt: but the patch to Makefile.commonarch still makes sense Jul 06 17:26:54 otherwise we end up in cyclic dep chain Jul 06 17:27:18 blindvt: btw. if you can refresh the configury patches to latest OE and resent it would help quite a bit Jul 06 17:27:30 and as you have done it you are the best man to do so Jul 06 17:27:38 khem, not sure. If you're i[456]86 then the correct impl is used. The rest is toast Jul 06 17:27:52 khem, i'll refresh them, sure Jul 06 17:27:58 blindvt: thx Jul 06 17:28:46 blindvt: however the patch to exclude generated files from targets in libpthread/nptl/sysdeps/Makefile.commonarch still makes sense Jul 06 17:28:56 I have two problems using OE. first: cp: cannot create regular file `/ld.so.conf': Permission denied Jul 06 17:29:06 Build of /home/dv/workspace/oe2/openembedded/recipes/glibc/glibc-initial_2.9.bb do_unpack failed Jul 06 17:29:20 anybody knows something about it? Jul 06 17:29:29 Dvorkin: whats your tmpdir Jul 06 17:29:38 Dvorkin: DISTRO and MACHINE ? Jul 06 17:29:51 khem, yea, but that should be filter-out gen_%.c or the like (don't know offhand if there's already a list that one can use) Jul 06 17:30:06 DISTRO="angstrom-2008.1" Jul 06 17:30:15 MACHINE = "xilinx-ml507" Jul 06 17:30:43 TMPDIR is not set Jul 06 17:31:32 at /home/dv/workspace/oe2/openembedded/recipes/glibc/glibc_2.9.bb line 62 I see: file://etc/ld.so.conf \ Jul 06 17:31:36 blindvt: yes although its only .c files that make sense for CSRC list Jul 06 17:31:51 Dvorkin: how old is your snapshot Jul 06 17:32:20 It was updated this dinner. Jul 06 17:32:30 about 5 hours ago Jul 06 17:33:16 Dvorkin: hmm ok does it show 5ac52371b09a25518b60e23e80c8c812cdea6e36 in git log Jul 06 17:33:49 Dvorkin: btw what version of bitbake are you using ? Jul 06 17:33:56 khem: what file do I have to check for log? Jul 06 17:34:18 Dvorkin: just run git log on top dir and inspect the SHAs Jul 06 17:34:22 or rather search Jul 06 17:34:35 khem: 1.8.19 Jul 06 17:36:34 khem: is this bitbake version OK? Jul 06 17:37:42 Dvorkin: you should use git master Jul 06 17:38:27 Dvorkin: or backport http://cgit.openembedded.org/cgit.cgi/bitbake/commit/?id=ca257adc587bb0937ea76d8b32b654fdbf4192b8 to your version of bitbake Jul 06 17:40:38 khem: I've got bitbake such a way: clone_or_pull "git://git.openembedded.org/bitbake.git" "bitbake" Jul 06 17:40:58 khem: and checkout_maybe_branch "bitbake" "1.10" "origin/1.10" Jul 06 17:41:56 Dvorkin: yes just do git clone git://git.openembedded.org/bitbake Jul 06 17:42:01 and start using it Jul 06 17:42:38 khem: in short I just used this script to get everything: http://github.com/aalonso/oe-utils/blob/master/oe-repos-update Jul 06 17:45:28 khem: strange thing that I already had bitbake that worked for me, but created a new instance in new directory and tryed to do everything from scratch. suddenly Bitbake asked me to install CHRPATH. why? may be it is the source of the problem? Jul 06 17:46:00 Dvorkin: OE needs it Jul 06 17:46:29 khem: I did not had it before! in my first installation Jul 06 17:46:54 Dvorkin: dont worry its a new thing needed for encoding correct rpaths Jul 06 17:49:45 khem: aha! how much time this requirement was done? 3 weeks ago I've got xilinx-ml507 uImage that worked for my embedded system and then I updated bitbake and got different binary that won't start. but as many times as I started bitbake and updated it each da it not asked for chrpath... can it be the source of the problem in generating wrong binary for my devkit? Jul 06 17:51:36 older than 3 weeks Jul 06 17:51:47 but not older than 3 months I think Jul 06 17:52:08 it does not affect kernel its for userspace only Jul 06 17:52:50 khem: It asked for chrpath only when I started to do everything from scratch... such a strange thing! So my question is "if I meet it's requirements, if my version of bitbake is up-to-date and I'm using right angstrom git tree, what can be wrong with this ld.so.conf?" Jul 06 17:55:22 Dvorkin: there is a change in bitbake which was committed to bitbake couple of days back which needs corresponding change in OE which was committed last night Jul 06 17:55:37 bitbake change is only available on master thus far Jul 06 17:56:07 so if you use git master of bitbake and git org.openembedded.dev it should be good Jul 06 17:56:21 khem: seems I'm using wrong git tree of bitbake... (origin/1.10) ? Jul 06 17:56:33 but older version of bitbake on latest OE .dev could trigger it Jul 06 17:56:44 Dvorkin: yeah Jul 06 17:56:47 use master Jul 06 17:57:03 this change should go into the 1.10 branch too but I dont know when Jul 06 18:03:55 Jay7, I'll be in the office in about an hour...hopefully catch up with you then? Jul 06 18:04:08 ka6sox: ok Jul 06 18:04:30 khem: I think the OE change should be reverted Jul 06 18:04:57 it brakes on all bitbake version except git master and you can not expect that all people are using this version Jul 06 18:04:58 cbrake, problem was write caching BBU for the HD controller... Jul 06 18:05:01 * Jay7 is fighting with ThinkPad's tablet and xorg 1.8 Jul 06 18:06:03 ensc|w: I would approach is differently and fix bitbake Jul 06 18:06:32 ensc|w: for stable I would not suggest it Jul 06 18:09:04 khem: yes; bitbake shall be fixed. But do you really want to enforce the usage of a development version of bitbake just to build OE? Jul 06 18:09:48 bitbake 1.8 is still the official latest stable bitbake version; and the next stable one (1.10) is also broken Jul 06 18:10:36 khem: sorry for stupid question... how can I get master? I just cloned git://git.openembedded.org/bitbake.git and run: $ git branch. And see: * master. Is it enoghth ? Jul 06 18:10:45 we should request RP to port that fix to 1.8 and 1.10 branch Jul 06 18:10:49 where can I find howto/guidelines on submitting patches for OE? Jul 06 18:11:01 Dvorkin: yes its ok Jul 06 18:11:43 khem: thank you so much! Jul 06 18:12:50 tharvey: http://wiki.openembedded.net/index.php/Commit_Policy Jul 06 18:15:08 khem, thanks - and is there an appropriate howto for posting to maillist using output of 'git diff' or otherwise? Jul 06 18:17:37 tharvey: git format-patch and git send-email Jul 06 18:17:40 is preferred Jul 06 18:17:50 ok, thanks Jul 06 18:24:59 * khem wants resizable qemu image Jul 06 18:25:01 khem, should i send the refreshed configury changes as new thread or as a follow-up to the old one for reference? Jul 06 18:25:22 khem, i'd follow-up, not sure Jul 06 18:26:19 blindvt: you can send [PATCH][v2] Jul 06 18:26:35 khem, k. thanks Jul 06 18:27:00 git format-patch --subject-prefix=PATCH][v2 Jul 06 18:28:11 ugh. remember carmelo's instructions that tricked me into implementing some automatic subject-prefix handling? ;) Jul 06 18:28:55 good afternoon. I see that the module I want is being built, and is showing up in the TMPDIR/deploy/images/.../modules-blah.tgz... it seems that tere must be some way to specify which you want in the jffs2 image, but I can't seem to find that documentation. Jul 06 18:29:05 How does one go about specifying which modules are in the jffs2 image? Jul 06 18:32:41 03Khem Raj  07org.openembedded.dev * ra4e69e7b7a 10openembedded.git/recipes/uclibc/ (3 files in 3 dirs): Jul 06 18:32:41 uclibc-git: Use i686 for qemux86 instead of i386 to accomodate nptl Jul 06 18:32:41 * NPTL supports i486 onwards so choose a different default for qemu. Jul 06 18:32:41 * Update SRCREV Jul 06 18:32:41 * Add UCLIBC_HAS_LONG_DOUBLE_MATH=y to uClibc.distro Jul 06 18:32:53 03Graham Gower  07org.openembedded.dev * rd6f20cbf1b 10openembedded.git/recipes/gpsd/ (gpsd-2.94/configure-libusb.patch gpsd_2.94.bb): Jul 06 18:32:54 gpsd_2.94.bb: fix CROSS COMPILE badness Jul 06 18:32:54 configure.ac shouldn't check the host system for includes/libraries, Jul 06 18:32:54 use pkg-config to find libusb instead. Jul 06 18:32:54 Also, add libusb1 to DEPENDS. Jul 06 18:35:24 tzanger, you have to add the module to the image you are building Jul 06 18:35:55 there may be a convenient var you can append to though used in image.bbclass Jul 06 18:36:10 tzanger: MACHINE_EXTRA_RRECOMMENDS += "kernel-module-x kernel-module-y ...." Jul 06 18:36:21 ya... thats the one - thanks khem :) Jul 06 18:37:43 khem: so it's not enough that the defconfig has the proper configuration? The .ko exists in the modules-blah.tgz, but not hin the JFFS2 Jul 06 18:38:30 tzanger, no, that just makes the module get built and packaged, but does not affect whats included in images Jul 06 18:38:59 and the kernel-module-x -- that would find and make sure x.ko got inserted into the root fs? no path from /lib/modules/[kernelver] required? Jul 06 18:39:14 khem, btw.. you said you had more machines, IIRC. Care to add qemux{86_64,ppc{32,64},m68k,sh3} ? For sh it would be nice to also cover EB variants Jul 06 18:39:32 s/x{/{x/ Jul 06 18:40:00 tzanger, right - the package should already exist if you enabled the module in the defconfig - the package dictates where the file goes, now what is needed is for you to tell your image to include that package Jul 06 18:40:18 khem, and for arm EB variants would also be handy Jul 06 18:41:53 khem, oh and i'd vote for toolchain configury --with-cpu --with-tune to generate proper code per default, while we're at it Jul 06 18:42:07 ahhh okay I'm specifying kernel module "packages" Jul 06 18:42:12 trying to get my head wrapped around this :-) Jul 06 18:43:24 ok I see it now in TMPDIR/deploy/glibc/ipk Jul 06 18:43:27 an image recipe produces a filesystem image that you can install onto your disk/flash media, whereas a typical recipe produces a package akin to an rpm or deb package Jul 06 18:43:33 making some sense now, trying to figure out where the one is already being added Jul 06 18:44:15 khem, --with-arch/--with-cpu/--with-tune is especially important to properly compile-test coldfire. I admit that coldfire is a bit hairy (from what i remember) but you just won't catch it properly with a normal m68k toolchain, from my experience Jul 06 18:44:22 note there are many images pre-defined - if you just want to add a package or two to be predefined you can use the MACHINE_EXTRA_* vars used in task-base.bb, if your needing something more unique you may consider creating your own image Jul 06 18:44:44 tharvey: I see now, it looks like task_base is where the one that IS being included is being included from Jul 06 18:45:10 tzanger, look at task-base.bb - you'll see some foo there that matches things in MACHINE_FEATURES with things in DISTRO_FEATURES to try to automagically install things that are needed Jul 06 18:45:28 I will, thanks! Jul 06 18:45:51 tzanger, right, and you'll notice that images (ie console-image.bb) uses task-base* Jul 06 18:46:50 tzanger, you may also be interested in module_autoload_* - you can define those (ie in kernel recipes, or machine conf) to autoload a module that is already installed Jul 06 18:49:01 tharvey: ahh, that would be handy too Jul 06 18:49:20 now shouldn't I be doing this in the machine conf file instead of in the console-image recipe? Jul 06 18:49:21 soccerfans: nl leads with 1-0 against uruguay Jul 06 18:50:27 eFfeM: best goal of tournament IMO :) Jul 06 18:50:42 yeah, that one was quite nice :) Jul 06 18:52:14 * khem predicts Netherlands Vs. Germany Jul 06 18:52:38 haha Jul 06 18:52:39 me too :) Jul 06 18:52:45 would be quite interesting Jul 06 18:53:03 * BlindMan looking for the OT-lartmaster Jul 06 18:53:32 Uruguay is hardling in posession Jul 06 18:53:44 hardly Jul 06 18:53:50 :q Jul 06 18:58:50 Bah, broken nose or gtfu Jul 06 18:59:11 thats soccer for you Jul 06 18:59:59 huh, yellow for 'em both Jul 06 19:00:01 hmm, so that didn't owrk :-) Jul 06 19:00:16 (sides) Jul 06 19:00:56 it says that the console image Im building has no buidlable providers, and goes on to say that kernel-module-rt2500usb is unbuildable Jul 06 19:01:21 tmp/deploy shows a kernel-module-rt2500usb ipk, so it did know how to build it Jul 06 19:01:33 tzanger: MACHINE and DISTRO ? Jul 06 19:01:42 And is that ipk for the machine you're building for? Jul 06 19:02:48 MACHINE is a custom one based off of at91sam9g45ekes (I just removed the GUI stuff) Jul 06 19:02:52 DISTRO... checking Jul 06 19:03:20 angstrom-2008.1 Jul 06 19:03:43 the image built before I tried to add the module, so it can't be something too horrible, I'm just not sure what :-) Jul 06 19:03:50 OK Jul 06 19:03:57 do a 'git diff' and pastebin your changes Jul 06 19:03:59 ~pastebin Jul 06 19:04:00 [~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. Jul 06 19:04:24 And if you didn't modify any existing files, pastebin the machine.conf Jul 06 19:04:34 tzanger: did you use MACHINE_EXTRA_RRECOMMENDS Jul 06 19:04:34 I didn't change anything in the oe git tree Jul 06 19:05:21 k, so pastebin your new files please :) Jul 06 19:05:33 khem: no, I tried to do it via changing the recipe Jul 06 19:05:44 ${@base_contains('COMBINED_FEATURES', 'usbhost', 'kernel-module-rt2500usb', '',d)} \ Jul 06 19:05:57 that was how the image-base did it for the module that was in there already Jul 06 19:06:00 tzanger, you can just define MACHINE_EXTRA_* in your conf files... such as local.conf and not have to touch the oe tree Jul 06 19:06:02 so I tried changing the kernel-module Jul 06 19:06:34 (well not changing, but adding that line to the DEPENDS = task-base-extended for the image recipe) Jul 06 19:07:24 http://pastebin.ca/1895633 Jul 06 19:07:29 that is the image recipe that I changed Jul 06 19:07:32 diff coming in a moment Jul 06 19:08:17 http://pastebin.ca/1895634 Jul 06 19:08:22 that's the diff from the one that DID build Jul 06 19:08:45 er wait, I don't think that pasted right Jul 06 19:09:01 Didn't, heh Jul 06 19:09:07 http://pastebin.ca/1895635 Jul 06 19:09:10 that's the diff, sorry Jul 06 19:09:34 Yeah, OK Jul 06 19:09:37 That explains it then Jul 06 19:09:58 DEPENDS means "I can find a recipe that gives me this" Jul 06 19:10:07 RDEPENDS means "I can find a package file that gives me this" Jul 06 19:10:14 usually external modules Jul 06 19:10:15 And you've got package names in a recipe names place Jul 06 19:10:20 aha Jul 06 19:10:25 So that's why you get the error you do Jul 06 19:10:33 so I should add an RDEPENDS for task-base-extended instead Jul 06 19:10:49 1-1 Jul 06 19:10:52 yeah Jul 06 19:10:54 and :( Jul 06 19:11:23 keeper could have stopped it Jul 06 19:11:54 I would say a mistake from goalie Jul 06 19:13:57 yup Jul 06 19:16:50 thats not offside Jul 06 19:17:02 nope Jul 06 19:17:19 linesmen commit so many mistakes this WC Jul 06 19:17:40 yeah Jul 06 19:17:58 Can we get some form of on/offside technology please? :| Jul 06 19:18:08 Few cameras and it should be solvable Jul 06 19:18:58 not as long as they have board members that were born before cameras existed Jul 06 19:19:23 Yeah Jul 06 19:20:08 but I think these kind of controversies make the game popular too Jul 06 19:20:13 good or bad Jul 06 19:33:59 khem, pb_, sorry, i had to resend a revised and condensed v3 version. apologies (mixing and matching 3 different boxes, won't happen in the future) Jul 06 19:36:03 problem: Jul 06 19:36:05 | make[2]: Entering directory `/home/dv/workspace/oe2/build-xilinx-ml507/conf/tmp/work/i686-linux/libx11-native-1_1.3.4-r8.0/libX11-1.3.4/specs/libX11' Jul 06 19:36:06 | GEN libX11.ps Jul 06 19:36:25 how can I get what exactly "GEN" means ? Jul 06 19:36:55 03Stefan Schmidt  07org.openembedded.dev * r1bf8f09f29 10openembedded.git/recipes/alsa/ (alsa-state.bb alsa-state/bug20/asound.state): alsa-state: Fix for muted speaker in bug20 state file. Bump PR. Jul 06 19:38:01 DvDima: Add V=1 to the do_compile make stuff Jul 06 19:39:25 DvDima, GENerate. as Tartarus said, pass V=1 to that make invocation Jul 06 19:41:08 can I do it from command-line of bitbake? Jul 06 19:41:10 khem, oh, any help to differenciate mips ABIs is very welcome. I've switched the default back to O32 but that is not acceptable in the long run Jul 06 19:41:24 DvDima, IIRC no. Jul 06 19:42:46 DvDima: try to use devshell Jul 06 19:45:55 03Frans Meulenbroeks  07org.openembedded.dev * r9101746564 10openembedded.git/recipes/tgt/tgt_1.0.6.bb: Jul 06 19:45:55 tgt: fixed checksums for 1.0.6, apparently they didn't make it into the recipe Jul 06 19:45:55 Signed-off-by: Frans Meulenbroeks Jul 06 19:46:11 03Frans Meulenbroeks  07org.openembedded.dev * r77c8a823d4 10openembedded.git/recipes/gcc/ (4 files in 2 dirs): (log message trimmed) Jul 06 19:46:12 gcc 4.1.2: added support for nios2 Jul 06 19:46:12 patches are lifted from the windriver toolchain Jul 06 19:46:12 (see http://www.nioswiki.com/Build_the_gcc4_toolchain) Jul 06 19:46:12 also fixed a few issues for use on 64 bit systems. Jul 06 19:46:19 03Frans Meulenbroeks  07org.openembedded.dev * rad29e9bc60 10openembedded.git/: Merge branch 'org.openembedded.dev' of git.openembedded.org:openembedded into org.openembedded.dev Jul 06 19:46:19 03Frans Meulenbroeks  07org.openembedded.dev * rbc4d128812 10openembedded.git/recipes/linux/ (linux-nios2-2.6.34/defconfig linux-nios2_2.6.34.bb): Jul 06 19:46:20 linux-nios2_2.6.34: created Jul 06 19:46:20 This version is from http://sopc.et.ntust.edu.tw/?p=linux-2.6.git;a=shortlog;h=refs/heads/nios2mmu Jul 06 19:46:21 but as this git is very slow, Axon Digital Design kindly donated space Jul 06 19:46:31 to host a mirrorred tarball. The recipe has the original src as comment Jul 06 19:46:31 Signed-off-by: Frans Meulenbroeks Jul 06 19:46:31 03Frans Meulenbroeks  07org.openembedded.dev * rb1e37cc9eb 10openembedded.git/recipes/glibc/ (5 files in 2 dirs): (log message trimmed) Jul 06 19:46:31 glibc 2.5: added support for nios2 Jul 06 19:46:31 added support for nios2. Jul 06 19:47:33 as this is untested on other machines it has DEFAULT_PREFERENCE = "-1" Jul 06 19:47:33 for all other machines than nios2 Jul 06 19:47:33 Signed-off-by: Frans Meulenbroeks Jul 06 19:48:52 cia is getting confused by my commits Jul 06 19:49:05 it missed some lines Jul 06 19:49:37 It only does the first few Jul 06 19:50:15 Tartarus: it missed ac416f9de36ff85598072b746f58b05665e6181b but the DEFAULT_PREFERENCE line is from that commit not from the glibc commit Jul 06 19:51:27 k Jul 06 19:51:42 anybody using MACHINE=xilinx-ml507 ? Jul 06 19:51:57 gm Jul 06 19:52:23 khem did you by any chance look at the include headers issue we discussed a while ago Jul 06 19:52:58 likewise: hi; i just committed the nios2 patches (for nios2+mmu) only need to cleanup the machine conf a little bit Jul 06 19:53:45 eFfeM: excellent spending of precious time :-) Jul 06 19:54:43 eFfeM: could you test on a standard dev board already? Jul 06 19:54:49 likewise: :-) took quite some time to iron out the bugs that didn't allow it to work on 64 bit Jul 06 19:55:31 likewise: tested it on the baseboard (which as is told to me is the same as the neek, actually I used the neek sof) Jul 06 19:56:29 likewise: http://www.nioswiki.com/@api/deki/files/487/=cycloneIII_embedded_evaluation_kit_standard.sof Jul 06 19:57:02 from http://www.nioswiki.com/Try_out_without_compilation Jul 06 19:57:46 likewise only thing I am not sure of is whether or not to pin the gcc version in nios2.inc Jul 06 19:57:54 (and binutils etc) Jul 06 19:57:57 2-1 Jul 06 19:58:28 hrm, much going on Jul 06 19:58:33 (football wise) Jul 06 19:59:08 family is watching behind my back, I hear the audio and watch the replays :-) Jul 06 19:59:12 hehe Jul 06 19:59:24 I just use the commentators pitch :) Jul 06 20:01:31 kergoth, do you by chance know a way out of the python2 vs. python3 jungle with current bitbake master, btw? I've just pushed the trivial print vs. print() hunks but the rest is something i don't see how to deal with in an un-intrusive way Jul 06 20:02:56 1-3 (had the order wrong earlier) a man can't even go to the restroom for a minute :-) Jul 06 20:03:00 NL leads Jul 06 20:03:29 kergoth, just curious since i already apted python2 just for you^Wbitbake but it would be nice not to have to download that old interpreter just for bitbake, mid-term Jul 06 20:03:46 eFfeM: hehe, definately not! there is a 15 minute break, use it! Jul 06 20:04:19 BlindMan: I don't care too much about soccer, that's why I am chatting here Jul 06 20:04:41 jo Jul 06 20:04:49 hi there, wog Jul 06 20:05:38 nice game Jul 06 20:05:55 woglinde: my tip, .de vs. .nl :) Jul 06 20:06:13 hehe Jul 06 20:06:15 (khem's as well) Jul 06 20:08:15 woglinde: actually I am in favour of germany the whole tournament ... something must be wrong with me! ;) Jul 06 20:08:57 haha Jul 06 20:09:09 we have the best team since 1974 Jul 06 20:09:41 woglinde: granted, and they are so young and so cool! Jul 06 20:09:57 quite fond of them, cu rrently Jul 06 20:10:20 yeah only against serbia Jul 06 20:11:12 i am wondering, they must have shot the most goals during this tournament Jul 06 20:11:35 the preussian germans! ;) Jul 06 20:12:57 Dutch will win the WC be is DE or ES Jul 06 20:13:10 it seems to me Jul 06 20:13:10 khem no Jul 06 20:13:12 khem: hehe, NEVER if germany is in the finals :) Jul 06 20:13:41 woglinde: they seems to be very united Jul 06 20:14:03 we will see tomorrow Jul 06 20:14:07 hehe Jul 06 20:14:07 I was looking at their preformances they are yet to draw a game Jul 06 20:14:16 very consistent Jul 06 20:14:20 woglinde: prolly half germany will watch it :) Jul 06 20:14:31 uhh Jul 06 20:14:33 robben Jul 06 20:14:36 robben robben Jul 06 20:14:43 hehe Jul 06 20:14:49 2 chances that my son would have converted Jul 06 20:14:56 don't spoil it, my streams lacks several seconds! :( Jul 06 20:14:59 err, lags Jul 06 20:15:00 hmm, interesting Jul 06 20:15:07 I did not change anything pertaining to the CPU name Jul 06 20:15:20 khem: is he a pro kicker? :) Jul 06 20:15:23 #oe-football :) Jul 06 20:15:27 but the kernel won't build, ld saying "no machine record defined" Jul 06 20:15:37 Jay7: ack :) Jul 06 20:15:41 yes all crockery is half broken in the kitchen Jul 06 20:16:07 who else is coming? :) Jul 06 20:16:49 woglinde: we still want revenge for 1974 Jul 06 20:17:22 Jay7: as you are quite right, see it from this angle, it's all over on sunday! :) Jul 06 20:18:39 effem no chance Jul 06 20:18:50 they (nl) got sooo many chances and don't turn them into goals Jul 06 20:18:54 we know all moves from robben Jul 06 20:18:56 was the same in the previous match Jul 06 20:18:59 * Jay7 is designing new kexecboot gui.. Jul 06 20:19:03 woglin(.de) Ve. eFfeM(.nl) Jul 06 20:19:19 Jay7: hrm, whats kexecboot? Jul 06 20:19:22 and we have mueller Jul 06 20:19:26 khem ;-) Jul 06 20:19:27 *g* Jul 06 20:19:33 woglinde: 8-) Jul 06 20:19:44 woglinde: but can't play tomorrow Jul 06 20:19:45 goal Jul 06 20:19:46 uh Jul 06 20:19:58 woglinde: i said, don't spoil! ;) Jul 06 20:20:09 BlindMan: LAB successor Jul 06 20:20:09 oh well thats just to make figures look nice Jul 06 20:20:15 it doent matter now Jul 06 20:20:26 Jay7: ok, what's LAB? :) Jul 06 20:20:30 2-3 big defense flaw i'd say Jul 06 20:20:35 linux as a bootloader :) Jul 06 20:20:46 khem: hey, there lots of incidences where two goals within a minute where shot :) Jul 06 20:20:53 what score is it now? Jul 06 20:20:56 Jay7: ic Jul 06 20:20:58 Jay7: 3-2 for .nl Jul 06 20:20:59 2:3 Jul 06 20:21:06 hehe... Jul 06 20:21:09 uru shot a goal Jul 06 20:21:14 BlindMan: I have crystal ball I can see into future Jul 06 20:21:24 khem: haha Jul 06 20:21:24 last throw in Jul 06 20:22:04 end Jul 06 20:22:26 hehe Jul 06 20:22:29 hm Jul 06 20:22:41 heh that was yellow card Jul 06 20:22:45 crazy Jul 06 20:22:54 yes Jul 06 20:22:56 no just a yellow card Jul 06 20:22:59 didnt see why Jul 06 20:23:00 you? Jul 06 20:23:12 uh Jul 06 20:23:15 time wasting Jul 06 20:23:25 now Jul 06 20:23:26 really Jul 06 20:23:37 that was close Jul 06 20:23:47 finally Jul 06 20:23:52 hehe Jul 06 20:23:52 sometimes i don't understand the referees Jul 06 20:23:54 hm uruguay makes argentina 2006 Jul 06 20:23:55 heh but dutch have collected too many yellow cards for their next match Jul 06 20:23:59 now get back to bug fixing :-) Jul 06 20:23:59 eFfeM: nobody does Jul 06 20:24:16 likewise: I have a week off :) Jul 06 20:24:24 khem: there are now penalities for having tow yellow cards Jul 06 20:24:28 khem: that's even better :-0 Jul 06 20:24:30 my company has compulsory shut down Jul 06 20:24:33 khem: they were neutrulized after quarter finals Jul 06 20:24:47 khem: because of world cup? :) Jul 06 20:25:01 heh no the employees wanted it Jul 06 20:25:07 around july 4th Jul 06 20:25:16 kids have summer vacation Jul 06 20:25:20 khem the yellow cards do not really matter as the cards for earlier rounds have been scrapped Jul 06 20:25:28 yes Jul 06 20:25:34 it does not matter too much if you start with 0 or 1 yellow card in the finals Jul 06 20:25:41 because ballack missed final 2002 Jul 06 20:25:42 but how about the one they picked today Jul 06 20:25:56 doesn't matter for the finals :) Jul 06 20:25:58 khem only 2 gets you out Jul 06 20:26:05 *g* Jul 06 20:26:10 oh in final all slate is clear Jul 06 20:26:11 because the seconde yellow Jul 06 20:26:14 khem some players have 1 but a 2nd one in a different match does not get you out immediately Jul 06 20:26:18 is yellow-red Jul 06 20:26:38 oh 2 yellow means you can play this game but you are out for next Jul 06 20:26:51 where as red means you are out now and also for next game Jul 06 20:26:54 hmm cool Jul 06 20:27:11 or if it was terrible foul you are out for some other games a swell Jul 06 20:27:13 likewise: dont you like soccer Jul 06 20:28:04 prince of .nl was there watching the game too and if he notices that you did not watch it he will raise your taxes Jul 06 20:28:38 lol I was just creating a tax spreadsheet to see if I can pay my bills :-) Jul 06 20:29:05 khem: I like socces, it's on, but I (too) only watch when the pitch of the commenter reaches 10 kHz or higher. Jul 06 20:29:17 s@ces@cer@\ Jul 06 20:29:32 likewise: hehe then you havent watched the mexican goal TV Jul 06 20:29:41 because 10khz means normal Jul 06 20:29:43 haha 6 players from bundesliga are now in the final Jul 06 20:30:34 woglinde: bundesliga will be broadcasted on ESPN3 here I am looking forward to it Jul 06 20:30:50 and also la liga and serie A Jul 06 20:32:48 * khem hates flash Jul 06 20:33:02 when will people switch to html5 Jul 06 20:33:47 when it standardizes on a good video codec? :-) Jul 06 20:33:54 when there is a free codec everybody is willing to use? Jul 06 20:35:34 vp8? Jul 06 20:36:20 woglinde: i want to stress the everybody again :) Jul 06 20:37:53 actually my son had to watch over internet, he said that there was enough lag so the noise from outside would tell him when they would score before the goal was visible over ethernet Jul 06 20:38:13 eFfeM: same here ... sometimes a bit weird Jul 06 20:38:29 effem we had the same here via dvb-t in 2006 Jul 06 20:39:34 03Frans Meulenbroeks  07org.openembedded.dev * r4d6a63850b 10openembedded.git/conf/machine/ (include/nios2.inc neek.conf): Jul 06 20:39:34 neek: board added; first nios2 board Jul 06 20:39:34 see comments in nios2.inc and neek.conf Jul 06 20:39:34 this is still somewhat work in progress and Jul 06 20:39:34 could use some work Jul 06 20:39:43 likewise: there's the neek conf file, feel free to give it a try and report back any issues to me Jul 06 20:43:31 03Frans Meulenbroeks  07org.openembedded.dev * r6112efc57e 10openembedded.git/MAINTAINERS: Jul 06 20:43:31 updated my entry (added nios2) Jul 06 20:43:31 Signed-off-by: Frans Meulenbroeks Jul 06 20:43:38 hmm Jul 06 20:43:59 tharvey: my changes didn't seem to make anything move in the jffs2 image Jul 06 20:44:19 tzanger, what were your changes? Jul 06 20:44:36 (I can't get to pastbin.ca from my current connection) Jul 06 20:44:53 tharvey: just the RDEPENDS to include a couple extra kernel modules in the jffs2 image Jul 06 20:44:56 did you make sure to rebuild your image? Jul 06 20:45:11 and did you make sure to bump the PR or clean whatever task/recipe you made your change to? Jul 06 20:45:13 tharvey: yes; I even blew away the TMPDIR/deploy/glibc/at91sam9gekes directory Jul 06 20:45:18 tharvey: no... Jul 06 20:45:31 I didn't know that was a step :-) Jul 06 20:45:40 don't just blow away the dir, there are some stamps kept around - make sure you bitbake -cclean Jul 06 20:45:41 I figured the cache was rebuilding so it saw the changes Jul 06 20:45:53 no..... that would be a huge misconception heh Jul 06 20:46:18 apparently so :) Jul 06 20:46:28 I also found an interesting bug, but I am thinking it's more of an at91 thing Jul 06 20:46:32 recipe related dependencies are handled by PR vars pretty much, so you need to either know exactly what your doing and what to clean/rebuild, or just blinding bump the PR's of the recipes your tweaking each time you bitbake Jul 06 20:46:38 I created a custom machine file based on one of the existing machine conf files Jul 06 20:47:13 take care changing things in machine conf as well... ie if you change something that affects compiler flags then you effectively need to clean like 'everything' Jul 06 20:47:18 did my changes, built... linux fails to link with "no machine record defined". If I rename the original (what I based mine) machine.conf and rename mine to the original filename, modify local.conf... it builds great Jul 06 20:47:19 I know, I'm so stupid, but I can't see how to enable extended logging to see what's wrong with do_compile Jul 06 20:47:40 if ever in doubt, rm -rf $TMPDIR and build clean but a little bit of care and you can avoid having to do that Jul 06 20:47:58 tharvey: well that -cclean is a godsend to know, thank you for that tip Jul 06 20:48:10 OE is a rather large system ot try and grapple with all at once Jul 06 20:48:12 tzanger, that sounds like it could be what I just mentioned - something you changed that needs a clean or PR bump Jul 06 20:48:36 OE has a large learning curve but if/when you get through it you'll find its extremely flexible compared to other linux build systems Jul 06 20:48:45 tharvey: ok, I'll give that a go too afterward Jul 06 20:48:52 buildroot was giving me the exact same error too Jul 06 20:49:35 if you change a kernel config or patch or anything in the kernel recipe make sure to clean your kernel as well (bitbake -c clean virtual/kernel) Jul 06 20:52:09 tharvey: nope, bitbake -cclean didn't seem to help it at all Jul 06 20:52:43 (I'm checking this by looking at the rootfs .tar.bz2 it's generating along with the jffs2 image) Jul 06 21:00:06 tzanger, cleaning the kernel will only rebuild it and kernel modules, it will not for example rebuild your rootfs - you would have to clean that or bump PR's for that. It's even a bit more convoluted as your image will refer to task-base-* and that is what would really need to be rebuilt Jul 06 21:00:37 again if you know exactly what your doing you can clean/rebuild - but seeing as your new to OE I would recommend always bumping appropriate PR's and letting bitbake do the right thing Jul 06 21:01:13 tharvey: RDEPENDS = "task-base-extended ${ANGSTROM_EXTRA_INSTALL} ${@base_contains('COMBINED_FEATURES', 'usbhost', 'kernel-module-rt2500usb'', '',d)}" <-- is that enough to add on a module if the combined features of the machine+distro list "usbhost" ? Jul 06 21:01:58 ack... make sure thats a .= for starters Jul 06 21:02:46 no, it isn't... I just riffed off of the DEPENDS = line above it (that was already there in the machine's conf file) Jul 06 21:03:07 IMAGE_INSTALL is similar... = not .= Jul 06 21:03:28 just make sure your not clobbering an already existing variable when what you really want to do is append to it Jul 06 21:03:37 I see += elsewhere in the file (for ANGSTROM_EXTRA_INSTALL and ROOTFS_POSTPROCESS_COMMAND Jul 06 21:03:55 make sure your appending in the correct way to put whitespaces around it where necessary - if in doubt add whitespaces around it Jul 06 21:04:27 take a look at http://bitbake.berlios.de/manual/ch02.html#id868864 Jul 06 21:04:33 *nods* do spaces/tabs matter? Jul 06 21:05:01 multiple whitespace does not, but you have to make sure you have whitespace between fields Jul 06 21:05:29 ok, but it's not like python where spaces/tabs are an essential part of the description :-) Jul 06 21:05:41 .= is concatenation without spaces, += is with, good ot know Jul 06 21:06:15 is there a way to print out what these variables expand to once all the conf files are parsed? Jul 06 21:06:39 bitbake -e will show you global env Jul 06 21:06:46 bitbake -e will show you env for that recipe Jul 06 21:08:31 whoa mama that's a lot of data Jul 06 21:09:14 grep is your friend Jul 06 21:09:28 well I'm curious too, :-) Jul 06 21:09:42 piping it to a file to look at it all Jul 06 21:09:50 there's like complete python programs in here Jul 06 21:10:10 recipes can contain python functions and/or sh functions - theres a good mix of both Jul 06 21:11:53 tzanger, another nice trick that may help you understand packages is that you can use 'dpkg-deb --contents' and 'dpkg-deb --info' on *.ipk files to see what they install and depend on etc Jul 06 21:12:28 hmm Jul 06 21:12:37 it looks like my module packages are making it into RDEPENDS Jul 06 21:13:24 so check your packages if you placed your module in with a specific package. Note that all the task-base-* stuff gets built as meta packages Jul 06 21:14:15 RDEPENDS="task-base-extended kernel-module-rt2500usb kernel-module-rt2x00lib kernel-module-rt2x00usb task-base-extended usbview thttpd iperf dosfstools mtd-utils initscripts udev sysvinit sysvinit-pidof tinylogin opkg-nogpg angstrom-version angstrom-feed-configs opkg-collateral" Jul 06 21:14:36 but what's interesting is that I *do not* see the kernel-module- that task-base is putting in there Jul 06 21:14:39 (by default) Jul 06 21:14:50 maybe I should be using task-base not task-base-extended Jul 06 21:16:33 I think you may be confused Jul 06 21:16:45 maybe this ins't the right way at all. I see DISTRO_EXTRA_RRECOMMENDS with kernel-module-vfat and such, but those are not in the jffs2 image either Jul 06 21:16:47 you made your edit to task-base.bb right? Jul 06 21:17:11 DISTRO_EXTRA_* is same functionality as MACHINE_EXTRA_* but its mean for distro maintainers to use Jul 06 21:17:28 tharvey: no, I made my edit to recipes/images/console-at91sam9-image.bb Jul 06 21:17:41 tharvey: I wanted to keep the git repo clean if at all possible Jul 06 21:17:56 tharvey: it's extraordinarily likely that I am confused. :-) Jul 06 21:18:09 if you modified task-base.bb then you should bump the PR defined in there and re-bitbake it, then use dpkg-deb --contents on the resulting ipk and you should see that your kernel-module-* is listed as a dependency of task-base Jul 06 21:19:08 if you want to keep the git repo clean you should add things to MACHINE_EXTRA_RRECOMMENDS in a local conf file outside the oe tree Jul 06 21:19:31 aha, let me change the var that I'm buggering with Jul 06 21:19:48 so something like MACHINE_EXTRA_RRECOMMENDS += "kernel-image-foo" ? Jul 06 21:19:53 ok I understand what you did... you copied probably console-image.bb and are adding things in there Jul 06 21:20:17 tharvey: copied the console-at91sam9-image.bb yes (it's one that the at91 people made for at91 stuff) Jul 06 21:20:31 that shouldn't affect any task-base stuff (which is where I thought you were putting it) - it should just add those into the tarball when you re-bitbake your image recipe Jul 06 21:20:52 maybe you should make sure that your kernel-module-*.ipk actually contains your kernel module? :) Jul 06 21:21:01 I'm getting an error building console-image... even after doing a - apparently a missing dir that can't be found via stat.... pastebin of error output is here: <>http://pastebin.com/q0YiVUJx> Jul 06 21:22:21 tzanger, or look at your $TMPDIR/work/$MACHINE/console-at91sam9-image*/temp/log.do_rootfs* file to see what it installed to rootfs Jul 06 21:22:49 Should I simply remove and start up again? Jul 06 21:26:06 how can I see what's commands was trying to execute before GEN libX11.ps make[2]: [libX11.ps] Error 1 (ignored) Jul 06 21:26:41 tharvey: I got it, thanks to your diety-like patience with me Jul 06 21:27:07 tharvey: there was an IMAGE_INSTALL variable being built right under what I was trying to do. I added them to that variable. Jul 06 21:27:48 Hmm Jul 06 21:28:09 So, not horrible? You can configure cpan to not use $HOME, but whatever dirs you want Jul 06 21:28:25 So next up is figuring out how to automate that (did it by hand now) Jul 06 21:29:02 Maybe not that horrible, hmm Jul 06 21:29:45 tharvey: thank you for all of your patience and help. I won't know if I did something useful until the boards arrive, but it did build Jul 06 21:29:51 How to fix this: Jul 06 21:29:55 ERROR: function do_install failed Jul 06 21:29:55 ERROR: log data follows (/home/rob/OE/build/tmp/work/overo-angstrom-linux-gnueabi/angstrom-version-1_2010.7-test-20100706-r4/temp/log.do_install.9401) Jul 06 21:29:55 | install: cannot stat `/home/rob/OE/build/tmp/work/overo-angstrom-linux-gnueabi/angstrom-version-1_2010.7-test-20100706-r4/lsb_release': No such file or directory Jul 06 21:29:56 sed in some paths, use the kernel.org mirror which should be good enough for everyone Jul 06 21:29:56 ? Jul 06 21:30:51 robtow: thining outloud, sorry Jul 06 21:33:28 Tartarus - no problem. Jul 06 21:33:44 Tartarus - any thoughts on *my* problem? ;-) Jul 06 21:34:33 should I blow away my dir? Jul 06 21:35:12 * Tartarus reads Jul 06 21:35:34 glad everything oe.org is back up Jul 06 21:35:37 Shouldn't have to blow away that much stuff Jul 06 21:35:44 rwhitby: thanks ! Jul 06 21:35:45 But there's a package that needs rebuilding I bet Jul 06 21:35:46 sec.. Jul 06 21:36:36 robtow: Bah, my tree isn't that up to date (been kicking a world -k build around for a few weeks now) Jul 06 21:36:46 robtow: git log -p, search for lsb_release Jul 06 21:36:52 That should give a hint at what to -c clean Jul 06 21:37:01 hmm Jul 06 21:38:39 heh, just found an old email (10y old!) from my brother about Canadians. The one that makes me laugh is "you understand the phrase, "Excuse me, could you please pass me a serviette? I have just spilled my poutine." Jul 06 21:58:16 GNUtoo|laptop: you said ml is preferred over bugzilla Jul 06 21:58:29 yes Jul 06 21:58:36 but I said for build bugs Jul 06 21:59:05 robtow: are you using .dev Jul 06 21:59:30 I a wmondering if I can just sand a mail "hey, matchbox-panel-2 do_fetch doesn't work. I think it's SRC_URI broken. something must be wrong with svn revision. anybody can confirm"? Jul 06 21:59:42 robtow: switch to using git master of bitbake then clean the offending recipe and rebuild Jul 06 21:59:52 khem: hi. good to see you Jul 06 22:00:19 dcordes, I think that's fine but usually you ask some people on irc to mirror the file Jul 06 22:00:32 dcordes, if the svn doesn't exist anymore Jul 06 22:01:00 GNUtoo|laptop: hey man :D I think matchbox-panel-2 is broken can you try to fetch it ? Jul 06 22:01:09 j/k Jul 06 22:01:28 dcordes, I'll try one sec Jul 06 22:03:34 thx Jul 06 22:07:42 GNUtoo|laptop: that would be nice Jul 06 22:08:20 how to regenerate glibc locales ? I had the locale generation limited in local.conf and now I have problemn distro or image want some languages I don't have. Jul 06 22:08:44 now I remevoed this from local.conf and need to reply Jul 06 22:08:46 rebuild Jul 06 22:09:25 dcordes, I've the source Jul 06 22:09:27 it seems Jul 06 22:11:26 dcordes, I mirror the source on my server and then you ask someone to mirror it? Jul 06 22:12:14 GNUtoo|laptop: you were able to pull it or have it locally already ? Jul 06 22:12:23 I have it locally Jul 06 22:12:47 dcordes, what's the pull address? Jul 06 22:13:05 ok. can you remove it and t ry to pull ? Jul 06 22:13:24 no Jul 06 22:13:30 It's precious Jul 06 22:13:38 give me the pull address Jul 06 22:13:40 if I rm it Jul 06 22:13:42 we loose it Jul 06 22:13:47 and we can't mirror it Jul 06 22:16:32 ping dcordes Jul 06 22:16:43 ok I'll look it myself Jul 06 22:16:53 but I really want to finish what I'm doing and go to sleep Jul 06 22:18:25 sorry my link is slow today Jul 06 22:18:25 dcordes, it works trough http Jul 06 22:18:36 change svn:// to http:// Jul 06 22:18:41 ah Jul 06 22:18:45 ok Jul 06 22:18:59 hi, i'm trying to build oe, but i'm getting 'cp: cannot create regular file `/ld.so.conf': Permission denied' can anyone help ? Jul 06 22:19:01 svn co http://svn.o-hand.com/repos/matchbox/trunk Jul 06 22:19:02 works Jul 06 22:19:32 cool Jul 06 22:19:33 this problem is caused by eglibc task Jul 06 22:19:56 GNUtoo|laptop: maybe you also have the answer to my glibc locale generation question above ? Jul 06 22:20:17 I'll look Jul 06 22:20:28 zenob, ask jconnolly|away Jul 06 22:20:49 dcordes, paste the exact error Jul 06 22:21:54 zenob, Jul 06 18:13:38 GNUtoo|laptop: yes, i gave it file:///etc/ld.so.conf and rebuilding now Jul 06 22:22:09 zenob, so change file://etc/ld.so in that Jul 06 22:22:44 is a problem with bitbake, see the OE mailing list Jul 06 22:23:45 zenob: See OE mailing lisrt topic "native glibc-initial-2.9 unpack error". Google for that. Jul 06 22:24:17 Yes, but why now? Was it a new bug introduced in a commit to bitbake, and thus only affects those who use the bitbake git repo? Jul 06 22:24:38 Or was it a latent bug triggered by a new commit to OE? Jul 06 22:24:51 (and thus might have an OE-only workaround for users?) Jul 06 22:25:39 didn't read the thread fully, I haven't seen the error myself yet Jul 06 22:26:24 GNUtoo|laptop: hm opkg cannot package locale-base-fr-fr iirc Jul 06 22:26:48 GNUtoo|laptop: and in local.conf I had the usual glibc generate locales limitation Jul 06 22:26:48 dcordes, ls locale* Jul 06 22:26:56 GNUtoo|laptop: there is no such recipüe Jul 06 22:27:02 can't find it Jul 06 22:27:07 that is the problem. Jul 06 22:27:56 dcordes, no look in ipk dir Jul 06 22:28:10 ls locale*.ipk Jul 06 22:38:04 mwester: it was a bug in bitbake that OE relied on Jul 06 22:38:23 mwester: it got fixed in bitbake and thus needed the OE to be fixed Jul 06 22:38:27 otherwise OE wont work Jul 06 22:38:53 but that means that one needs patched oe and patched bitbake to go hand in hand Jul 06 22:39:04 Ah. So users who are using a static tag to fetch bitbake are screwed. Jul 06 22:39:14 mwester: sort of Jul 06 22:39:33 They must upgrade bitbake and see what breaks because of it, or ... quit using oe. Jul 06 22:39:36 Sigh. Jul 06 22:39:57 mwester: or backport the bitbake fix Jul 06 22:40:02 OE is it's own worst enemy sometimes. This is a change that should have been submitted as an RFC and coordinated. Jul 06 22:40:10 or backout the OE fix whatever fits Jul 06 22:40:17 Either that, or bitbake should be released *inside* the OE repo. Jul 06 22:40:32 yes that wud be better imo Jul 06 22:40:42 ~lart OE Jul 06 22:40:43 something i have been claiming for a while now... Jul 06 22:40:43 * ibot explains, ever so gently, that if OE doesn't give the channel more information, they can't help Jul 06 22:41:02 you all have NO idea how painful this is. Jul 06 22:42:07 mwester: We could think of making OE compatible with or without bitbake change but thats more mess Jul 06 22:42:20 yeah, yeah. Jul 06 22:42:45 1.8.18 --- is there a 1.8.x that fixes this, or must we go to 1.10.x Jul 06 22:43:13 no the fix is thus far only on master Jul 06 22:43:26 it needs to go into those branches Jul 06 22:43:35 I have requested RP to apply it Jul 06 22:44:11 khem: eh Jul 06 22:44:22 khem: you can eaisly change oe to work with bitbake with or without the change Jul 06 22:44:30 before the commit part was in host, part was in path Jul 06 22:44:35 now for file:// its all path Jul 06 22:44:45 kergoth: yeah Jul 06 22:44:50 let me see Jul 06 22:45:44 So we've introduced a change in OE that makes it incompatible with EVERY bitbake version except the development head... Jul 06 22:45:54 Allow me to express some surprise. Jul 06 22:46:06 kergoth: right so path contains all after file:// Jul 06 22:46:16 mwester: yes Jul 06 22:46:18 * kergoth should have spotted that fix didn't account for the old way when he reviewed it Jul 06 22:46:58 * mwester images automobiles in the OE world: Every automobile is rendered useless in the garage, because a change was made such that only those who have purchased brand spanking new 2010 model autos can use the highways... Jul 06 22:47:15 oh please, if its that mission critical stop blindly following .dev Jul 06 22:47:23 whine some more Jul 06 22:47:34 mwester: cars also have recalls Jul 06 22:47:43 a new stable branch is overdue anyways Jul 06 22:48:00 i guess we now found the hash for when to branch :) Jul 06 22:48:11 or you could just fix it Jul 06 22:48:56 .dev is for test pilots only (they get paid well they say). Jul 06 22:49:09 heh Jul 06 22:49:36 i don't get why there's always this drama around .dev. its master/trunk, just like every other project. we try not to break it, but shit happens Jul 06 22:49:46 we're no different than anyone else in that respect Jul 06 22:50:07 kergoth: I agree. Jul 06 22:50:24 mickeyl: the iphone proj didn't go to me. jfyi Jul 06 22:51:04 likewise_: ah, too bad. ok, perhaps next time. Jul 06 22:51:45 I must say, it's also important to keep in mind what folks are likely to have access to when using stuff Jul 06 22:51:49 kergoth: before fix host=file://etc path=ld.so.conf and after host=file:// path=/etc/ld.so.conf right ? Jul 06 22:52:07 file is the type. :// is a separator Jul 06 22:52:09 The normal rule is that if sanity.bbclass or whatever doesn't gripe, you should be good to go Jul 06 22:52:12 host will be 'etc' Jul 06 22:52:22 just os.path.join host & path if host is non-empty and type is file Jul 06 22:52:23 basically Jul 06 22:52:24 kergoth: right I see Jul 06 22:52:43 So it sounds like some bad luck happened in trying to quickly fix something related to a bitbake/master change Jul 06 22:52:56 And we should be able to resolve things so that 1.8.18 works again Jul 06 22:53:02 'cuz 1.10.x has a broken bitbake -k Jul 06 22:53:12 And that's much worse than whatever was fixed by the uri parsing change I bet Jul 06 22:53:49 the change to bitbake was just a correctness thing Jul 06 22:54:04 kergoth, don't be a dick about it -- of course dev gets screwed up. And then you FIX it, you don't piss and moan about that others point out that a mistake was made. Simple. Jul 06 22:54:18 I'm not pissing and moaning, I'm watching you bitch and moan Jul 06 22:54:24 No, you're bitching again. Jul 06 22:54:26 Tartarus: there's a problem with bitbake 1.8.18 and OE? Jul 06 22:54:45 well, yes, when i get sick of listening to the shit Jul 06 22:54:45 kergoth: Right, and in the end, everyone needs to be able to use the metadata with both 1.8.x and 1.10.x and master Jul 06 22:55:00 I'm pointing out a legitimate issue, and highlighting that it was a mistake, you are arguing that it's not a issue, and I should go away. Jul 06 22:55:00 Sigh. Fixed my problem by doing a . Jul 06 22:55:03 it's a 30 second fix Jul 06 22:55:05 mwester: no, i'm not Jul 06 22:55:08 i already said that its broken Jul 06 22:55:08 a Jul 06 22:55:12 and is a quick fix Jul 06 22:55:16 /ignore kergoth Jul 06 22:55:18 and the commit that went in was wrong Jul 06 22:55:20 kergoth: To the metadata? Then yes, it needs doing if it wasn't already :) Jul 06 22:55:25 but clearly you dont' need to bother reading Jul 06 22:55:46 So long as in the end it works for everyone, yes, that's the important part Jul 06 22:55:49 kergoth: http://pastebin.com/syC7A63J Jul 06 22:55:54 something like this ? Jul 06 22:56:10 Just the usual clash between folks working on the weekend vs folks working during the week :( Jul 06 22:56:11 afk, errand Jul 06 22:57:06 khem: looks that results in no dirname happening to the path, so the filename may stick around Jul 06 22:57:06 kergoth: I think I still need dirname Jul 06 22:57:11 :) Jul 06 22:57:13 yeah there we go Jul 06 22:57:16 "if not host" is a bit more pythonic Jul 06 22:57:19 but looks fine functionality wise Jul 06 22:57:28 er, "if host" Jul 06 22:58:04 * kergoth really doesn't see why mwester seems to think .dev is solid enough to use for something like automotive.. what a joke Jul 06 22:58:16 sorry, no master/trunk is enterprise Jul 06 23:00:30 kergoth: http://pastebin.com/vUKmAhxd Jul 06 23:00:55 yeah looks good to me Jul 06 23:01:15 kergoth: let me try it with a bitbake without the nitin fix Jul 06 23:01:20 thanks for the quick fix, sorry I didn't spot it at review Jul 06 23:01:21 * kergoth nods Jul 06 23:01:54 meanwhile I will post it for the impatients Jul 06 23:02:41 what's the problem with bitbake 1.8.18? Jul 06 23:03:35 a fix went into oe for a change that went into bitbake master, and that oe fix didn't retain handling of the previous way bitbake was working Jul 06 23:03:46 khem is working re-adding backward compatibility Jul 06 23:07:00 * Jay7 -> sleep Jul 06 23:07:15 what is the effect of the change? I'm just having an issue and was wondering if it was related to that Jul 06 23:07:29 03Andrea Adami  07org.openembedded.dev * r89ab4e3296 10openembedded.git/recipes/kexec-tools/ (3 files in 2 dirs): Jul 06 23:07:29 kexec-tools: add patch for autoconf >2.63 to version 1.101. Jul 06 23:07:29 * taken from kexec-tools-2.0.1 recipe Jul 06 23:07:29 * bump PR. Jul 06 23:08:29 * khem mistype 'buttbake' Jul 06 23:09:12 kergoth: hmm it did not work so well lemme see Jul 06 23:09:40 hmm, best add notes to see the output of decodeurl between the two Jul 06 23:09:44 ssieb: failure to unpack Jul 06 23:10:06 ah, that would be the problem I'm having :-) Jul 06 23:11:10 kergoth: yeah doing that now Jul 06 23:12:29 ssieb: you should be able to revert the oe change if you aren't on master. alternatively, khem could revert it and re-commit an improved version Jul 06 23:13:12 originally I had .dev, but didn't realize it. does the non-.dev repository have the same problem? Jul 06 23:13:46 doubt it Jul 06 23:13:51 * kergoth wanders off for a bit Jul 06 23:15:00 goes sleep a bit Jul 06 23:15:19 cya all, thanks for the fish Jul 06 23:16:38 03Andrea Adami  07org.openembedded.dev * r76e745add3 10openembedded.git/recipes/kexec-tools/kexec-tools-klibc-static_1.101.bb: Jul 06 23:16:38 kexec-tools-klibc-static: follow PR bump of main 1.101 recipe. Jul 06 23:16:38 * (added patch for autoconf). Jul 06 23:17:15 khem: this is the output of the problem that I'm having: Jul 06 23:17:27 NOTE: Unpacking openembedded/recipes/glibc/glibc-2.4/etc/ld.so.conf to zipitz2-tmp/work/armv5te-angstrom-linux-gnueabi/glibc-initial-2.9-r36.3/ Jul 06 23:17:28 cp: cannot create regular file `/ld.so.conf': Permission denied Jul 06 23:17:35 is that the issue you're working on? Jul 06 23:18:36 khem: I could fix the klibc version but the dietlibc fails badly with | make: *** [bin-i386/elftrunc] Segmentation fault Jul 06 23:19:09 hmm where is woglinde? Jul 06 23:20:18 khem: if that was not enough, second job got: | make: *** [bin-i386/dnsd] Segmentation fault Jul 06 23:20:53 does it ring any (libc) bell?bell Jul 06 23:30:05 ssieb: yes you have the same problem as everybody else Jul 06 23:30:25 ssieb: I posted a fix couple of moments ago to ml please test that Jul 06 23:30:32 and report if that works for you. Jul 06 23:30:41 oh, ok. I just reverted the commit, but I'll try that instead Jul 06 23:30:57 ssieb: thx takes the v2 patch Jul 06 23:32:05 khem: posted where? Jul 06 23:32:11 to mailing list Jul 06 23:34:11 ssieb: http://article.gmane.org/gmane.comp.handhelds.openembedded/34123 Jul 06 23:34:28 thanks, I did find it :-) Jul 06 23:36:31 can somebody help me with the locales problem described above ? Jul 06 23:39:11 dcordes: what problem Jul 06 23:41:16 khem: I can't finish the angstrom-gnome-image cause it can't package locale-base-fr-fr Jul 06 23:41:24 khem: klibc fix for i386 is upstream :) Jul 06 23:41:57 khem: I suspect it is because I had only english locale enabled in local.conf Jul 06 23:43:52 ant__: which one ? Jul 06 23:44:04 dcordes: hmm Jul 06 23:44:17 khem: bitbake fix looks good, thank you Jul 06 23:44:29 +KLIBCARCHINCFLAGS = -I$(KLIBCKERNELOBJ)/arch/x86/include Jul 06 23:44:34 ssieb: ok then add your tested by to ML Jul 06 23:44:47 * khem would like to share the blame this time Jul 06 23:44:50 heh Jul 06 23:44:56 ant__: ah cool Jul 06 23:44:57 I'm not actually on the list though... Jul 06 23:45:13 ssieb: you can still reply from gmane Jul 06 23:45:22 oh, didn't know that. I'll try Jul 06 23:47:15 ant__: oh did you post it for them ? Jul 06 23:47:37 well, not exactly, I've confirmed it works Jul 06 23:47:46 their is strangely similar Jul 06 23:48:08 http://www.zytor.com/pipermail/klibc/2010-July/thread.html Jul 06 23:50:41 yeah someone might have copied our stuff Jul 06 23:50:43 its ok Jul 06 23:51:05 important is upstream improves Jul 06 23:57:04 right Jul 07 00:02:55 good night all Jul 07 00:06:12 Anyone else noticed that linux' IO scheduler sucks quite badly in the early stages of a build (BB_NUMBER_THREADS >1) - when some of the larger tarballs like glibc get extracted? Jul 07 00:06:51 i'm using an i7 and interactivity is really poor during this part... Jul 07 00:07:39 grg: hmm Jul 07 00:09:37 i suppose i should build on a different disk to my root... Jul 07 00:10:35 khem: no idea ? Jul 07 00:11:41 grg: balance out make threads and bb threads Jul 07 00:12:29 khem, i have 8 cores (according to /proc/cpuinfo) and am using PARALLEL_MAKE="-j 6", BB_NUMBER_THREADS=3. Jul 07 00:12:44 PARALLEL_MAKE="j16" Jul 07 00:13:11 grg yeah 8 cores can run 16 make jobs Jul 07 00:13:35 khem, i know, but i often want to do other stuff with the cpus :) Jul 07 00:13:47 grg: in winter it will help :) Jul 07 00:13:52 I just did (based on a post from sakoman) - and it fails with "nothing provides"; the recipes/x-load dir has only x-load-omap3pandora_git.bb as a bb file. Jul 07 00:14:03 khem, it is winter here! Jul 07 00:14:13 How do I build the MLO file for omap? Jul 07 00:16:07 sakoman says here: , but that doesn't work for MLO, but his directions for u-boot do work. Jul 07 00:17:46 robtow: those might be for some other repo based on OE Jul 07 00:18:07 and recipes might have not been contributed Jul 07 00:22:00 khem. Hmmm. Jul 07 00:22:34 ...trying Jul 07 00:23:30 Aha! :-) Jul 07 00:26:57 How do i know which kernel openembedded is using for my image? angstrom-2008.1.conf specifies kernel26, but im not sure where i would find out which kernel it is actually using? Jul 07 00:32:46 bkinman - look in you /build/tmp/deploy/glibc/images/ dir; the kernel version gets put in the file name for the kernel.... for example: Jul 07 00:33:27 robtow: but where is it actually selected at? Jul 07 00:33:39 /home/rob/OE/build/tmp/deploy/glibc/images/overo Jul 07 00:33:39 uImage-2.6.29-r80-pm2+gitre63cf0710a4fb639d91d3e8b05aa485fbfa381b3-overo.bin Jul 07 00:33:39 uImage-2.6.32-r80+gitra6bad4464f985fdd3bed72e1b82dcbfc004d7869-overo.bin Jul 07 00:33:39 uImage-overo.bin Jul 07 00:34:04 You see here I have built two kertnels - one is the "current" one, the other is the power management kernel. Jul 07 00:39:29 khem, I bumped to -j12 and got a build failure in gcc-cross-initial: http://pastebin.com/XKYZrJuG simply starting the build again made it succeed. I have no idea how to track down such parallel build failures... Jul 07 01:16:47 grg: hmmm seems libiberty is issue Jul 07 01:17:33 khem, i got the same failure with gcc-cross, so its at least partially reproducable Jul 07 01:17:52 grg: seems a gcc bug to me Jul 07 01:18:11 grg: try same with say gcc 4.5 Jul 07 01:18:18 and see if you still see it Jul 07 01:18:43 i've been doing 4.4.4 builds for a week or more without problem... (at -j6) Jul 07 01:18:45 I have debugged such issues in past its not easy Jul 07 01:18:51 khem: can you check matchbox-panel-2 ? Jul 07 01:19:13 khem, testing 4.5 is on my todo list... Jul 07 01:19:13 and change are generally dependencies on stamp files Jul 07 01:19:14 in gcc Jul 07 01:24:42 what's wrong about following SRC_URI ? Jul 07 01:24:52 SRC_URI = "http://svn.o-hand.com/repos/matchbox/trunk;module=${PN};proto=http \ Jul 07 01:24:53 ? Jul 07 01:31:58 dcordes: you mean, why doesn Jul 07 01:32:05 dcordes: you mean, why doesn't it use svn to check it out Jul 07 01:33:27 dcordes: you want to write it as svn://svn..;proto=http Jul 07 01:33:51 zecke: no, I mean what's wrong about the SRC_URI line Jul 07 01:33:55 zecke: because it does not parse Jul 07 01:34:42 dcordes: with that context it looks fine... unless you do not have a '"' on the last line Jul 07 01:34:54 zecke: there are other things following Jul 07 01:35:10 zecke: the thing is in HEAD it is svn: Jul 07 01:35:19 zecke: but that does not fetch Jul 07 01:35:34 zecke: GNUtoo recommended to replace it with http Jul 07 01:35:38 dcordes: well, it is supposed to be svn://, as this tells bitbake to use the svn fetcher Jul 07 01:35:52 $ svn co svn://svn.o-hand.com/repos/matchbox/trunk Jul 07 01:35:52 svn: Kann nicht mit Host »svn.o-hand.com« verbinden: Connection timed out Jul 07 01:37:54 dcordes: I am not following, you do have a proto=http, so it should have done svn co http://svn.. Jul 07 01:38:22 dcordes: e.g. if you take a look at bitbake/lib/bb/fetch/svn.py and search for proto, it exists :) Jul 07 01:39:54 zecke: :) Jul 07 01:39:58 zecke: can you suggest a patch ? Jul 07 01:40:37 dcordes: I am confused. So first at all using SRC_URI = "http://" is wrong. Jul 07 01:40:55 dcordes: second, which recipe are you trying to fetch? matchbox-???.bb? Jul 07 01:52:29 grg: 4.5 works ok on mips (be) Jul 07 01:53:05 I have tried it here on qemu Jul 07 01:53:37 there is one problem Jul 07 01:53:45 khem, cool. I'm sure there will be some obscure problems that fall out with a fully running system though... Jul 07 01:53:50 while compiling native gcc Jul 07 01:54:11 grg: could be Jul 07 01:54:17 but i doubt Jul 07 01:54:20 zecke: matchbox-panel-2_svn.bb Jul 07 01:57:44 dcordes: trying it now Jul 07 01:58:29 zecke: thx Jul 07 01:59:06 dcordes: bad news, it works here http://paste.lisp.org/display/112211 Jul 07 02:01:58 I will install the oe subversion in my host Jul 07 02:02:57 dcordes: no, the interesting bit is where/how the proto=http part is dropped Jul 07 02:04:03 zecke: proto=http is dropped ? Jul 07 02:15:17 dcordes: well, if it tries to use the svn protocol, then something has ignored proto=http Jul 07 02:15:40 dcordes: anyway, I have not seen a full log, so your issue might be something completely different. :) Jul 07 02:16:00 dcordes: maybe you should go to bed, and try again with a fresh mind? Jul 07 02:28:58 is cross compile badness an OE thing, or does it come from autotools, or gcc, or something else? Jul 07 02:30:33 i would like to add /usr/X11R6 to the badness Jul 07 02:32:09 grg: we patch gcc to add it. zecke did the patch, iirc Jul 07 02:32:25 kergoth, ah. thanks Jul 07 02:35:17 hmmm... i'll have to rediff patches for each of the different versions :( Jul 07 02:37:02 grg: the patch is quite evil, it would be cool if it could get a blacklist via the cli Jul 07 02:46:42 aggh... gnu's 3 space indentation... my eyes are bleeding! Jul 07 02:47:55 zecke: hm bad idea. I never have fresh mind. Jul 07 02:48:08 zecke: if you like to take a look here is the output http://pastebin.ca/1895817 Jul 07 02:59:24 khem, I have your NSLU2 ready to ship; when you get a moment, send me an address for shipping privately. Thanks! **** ENDING LOGGING AT Wed Jul 07 02:59:56 2010