**** BEGIN LOGGING AT Mon Jul 14 02:59:56 2008 Jul 14 03:15:07 * * OE Bug 4426 has been RESOLVED (FIXED) by mwester(AT)mwester.net Jul 14 03:15:08 * * Security Vulnerabilities in Bind {8,9} Jul 14 03:15:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4426 Jul 14 03:15:39 03mwester 07org.oe.dev * r5be04634... 10/ (6 files in 3 dirs): bind - upgrade to 9.3.5-P1 in response to security vulnerabilities. Jul 14 04:05:10 finally that one finished with another failure...now I am trying mach:x86, distro:angstrom-2008.1, bitbake console-image...Does this sound reasonable? Jul 14 04:24:43 KernelKlick: depends on what you wanna do. glibc just failed for my x86 build Jul 14 05:05:50 that one failed too Jul 14 05:08:08 dcordes and all: Darn, that one failed too. glibc-intermediate do_compile. Does anyone have a working x86 bootstrap or console image? Jul 14 05:09:24 dcordes: i tried a uclibc earlier...it failed too Jul 14 05:09:53 I think it was minimal-uclibc...if I remember right Jul 14 05:12:10 KernelKlick: http://bugs.openembedded.net/show_bug.cgi?id=3284 Jul 14 05:12:33 Laibsch: yes, it is still a problem (oebug 3284) Jul 14 05:16:10 dcordes: do we add that to the local.conf TARGET_ARCH="i386"? or somewhere else? Jul 14 05:16:16 KernelKlick: what sort of machine are you building for? Jul 14 05:16:31 KernelKlick: I just defined machine x86 Jul 14 05:16:53 its a 486 with no math coproc Jul 14 05:16:58 the according .conf file will care about the rest Jul 14 05:17:42 i tried to set MACHINE=x86 before and it fails at the do.rootfs stage Jul 14 05:18:22 sorry...for the bitbake bootstrap-image Jul 14 05:18:42 what distro are you using? Jul 14 05:24:33 dcordes: I am not worried about the math coproc at this point. I just want to get an i386 image that will work on one of the VM (qemu, virtualbox) for some testing. You said you got MACHINE:x86 to work? Which distro did you use? Jul 14 05:32:48 KernelKlick: no I didn't say that. I said the build failed on glibc-intermedeate Jul 14 05:34:34 oh okay....np Jul 14 06:05:52 Hello, could someone point me in the direction of the kernel patches? I'd like to try to apply them natively rather than through OE, since I'm short on disk space right now Jul 14 06:28:24 radiochickenwax: maybe they are in org.oe.dev/packages/linux/your-kernel/ Jul 14 06:42:25 dcordes: yeah I was thinking that too, but I was wondering if they were stored online somewhere accessible Jul 14 06:46:58 morning Jul 14 06:47:46 dcodes: how do I access "org.oe.dev" is what I'm wondering Jul 14 06:48:21 do I *have* to go through monotone? Jul 14 06:52:04 KernelKlick : Use angstrom stable it works Jul 14 06:52:47 03tmbinc 07org.oe.dreambox * r2ad6039d... 10/ (1 packages/gstreamer/gst-plugin-dvbmediasink.bb): gst-plugin-dvbstreamer: update to support more video formats Jul 14 06:53:00 03tmbinc 07org.oe.dreambox * r34c41dae... 10/ (3 files in 3 dirs): tuxbox-tuxtxt-32bpp: add support for dm8000 rc Jul 14 06:53:04 03tmbinc 07org.oe.dreambox * r49173621... 10/ (3 files in 3 dirs): enigma2-defaultservices: add 'arte hd' to defaults Jul 14 06:53:04 radiochickenwax : yes. you need monotone Jul 14 06:53:08 03tmbinc 07org.oe.dreambox * r567003f1... 10/ (1 packages/images/dreambox-image.bb): dreambox-image: add networkwizard,mediascanner,graphmultiepg,hotplug,defaultservicescanner plugins and defaultservices Jul 14 06:53:12 03tmbinc 07org.oe.dreambox * rc7374691... 10/ (1 packages/wlan-rt73/wlan-rt73.bb): wlan-rt73: install kernel module into right directory Jul 14 06:53:16 03tmbinc 07org.oe.dreambox * r9c3be02e... 10/ (1 packages/tuxbox/tuxbox-common.bb): tuxbox-common: install cables.xml only when required Jul 14 06:53:20 03tmbinc 07org.oe.dreambox * r6ffcf640... 10/ (4 files in 3 dirs): linux-dm800: keep big patches outside metadata Jul 14 06:53:26 03tmbinc 07org.oe.dreambox * r3bf15b64... 10/ (41 files in 22 dirs): add dm8000 support Jul 14 07:02:45 steliosk why? you could use git Jul 14 07:02:56 it's not official, but you have the repository Jul 14 07:03:29 until we switch 'officially' mtn is the way to go.... Jul 14 07:03:45 ok Jul 14 07:26:48 03koen 07org.oe.dev * r45d7f296... 10/ (1 packages/ffmpeg/ffmpeg_git.bb): ffmpeg git: bump SRCREV Jul 14 08:28:36 morning Jul 14 08:55:15 good morning Jul 14 10:17:47 Jul 14 10:21:49 03koen 07org.oe.dev * raa0f5212... 10/ (14 files in 5 dirs): qmake2: add sdk version Jul 14 10:22:10 03koen 07org.oe.dev * r614ea89d... 10/ (4 files in 2 dirs): qmake2 sdk: fix packaging Jul 14 10:22:30 03koen 07org.oe.dev * rf879a25c... 10/ (1 packages/meta/meta-toolchain-neuros.bb): meta-toolchain neuros: add qmake2-sdk Jul 14 10:22:50 03koen 07org.oe.dev * r43ceaaee... 10/ (7 files in 3 dirs): uicmoc4: add sdk version, fix build with gcc 4.3.x Jul 14 10:23:10 03koen 07org.oe.dev * rdff33d19... 10/ (1 packages/meta/meta-toolchain-neuros.bb): meta-toolchain neuros: also ship uicmoc4-sdk Jul 14 11:28:57 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:28:57 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:28:57 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:29:20 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:29:25 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:29:35 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:30:12 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 11:31:05 florian: something wrong with openembedded-issues, is not up to date Jul 14 11:33:07 obviously... Jul 14 11:34:48 florian: question: is http://bugs.linuxtogo.org/ the official gpe bugtracker? Jul 14 11:36:43 ant|work: yes Jul 14 11:37:21 if yes, why don't move to bugs.openembedded.net/org? Jul 14 11:37:38 (more visibility) Jul 14 11:39:28 florian: only 95 open bugs there? You lucky guy! Now I see why... Jul 14 11:40:02 ant|work: having one bugtracker for ditributions, build system and applications is a major pain. we had very bad experience with that using the same bugtracker than Familiar. Jul 14 11:41:00 then I cannot argue against Jul 14 11:45:16 florian: befor pestering your shiny bugtracker, I want be sure the issues are 100% GPE-related and not OE/Angstrom legacy Jul 14 12:04:57 florian: What problems were those? No doubt, there is a price for consolidation, there certainly is a gain, too. Jul 14 12:09:41 urgh, a search for 'gpe' finds out 109 bugs in gugs.openembedded.net Jul 14 12:16:33 03mickeyl 07org.oe.dev * rd39fcbb2... 10/ (8 files in 3 dirs): Jul 14 12:16:33 linux-openmoko-devel: clean up a bit, compute defconfig for om-gta01 and om-gta02 out of one defconfig, Jul 14 12:16:33 enable IKCONFIG in defconfig, disable APM_POWER in defconfig. Jul 14 12:16:33 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:16:35 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:18:09 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:18:09 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:18:09 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:18:09 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:18:14 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:19:13 florian: some messages sent from the bugtracker seems to have been skipped in the openembedded-issues ML Jul 14 12:19:32 queue flushed ? Jul 14 12:19:53 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:22:01 mickeyl just cant get his changes right first time :-D Jul 14 12:22:34 03mickeyl 07org.oe.dev * r93d087e4... 10/ (1 packages/linux/linux-openezx-devel_svn.bb): linux-openezx-devel bump to 2.6.25 Jul 14 12:22:56 ahh, is Mickeyl then... ;-) Jul 14 12:23:06 this CIA is bothering Jul 14 12:23:50 anyway time to find new accomodation Jul 14 12:34:30 03mickeyl 07org.oe.dev * rd39fcbb2... 10/ (8 files in 3 dirs): Jul 14 12:34:30 linux-openmoko-devel: clean up a bit, compute defconfig for om-gta01 and om-gta02 out of one defconfig, Jul 14 12:34:30 enable IKCONFIG in defconfig, disable APM_POWER in defconfig. Jul 14 13:55:10 woglinde: Hiya Jul 14 13:55:16 hi Jul 14 13:55:29 Any luck with the ftdi package? Jul 14 13:55:31 * piroko winces Jul 14 13:55:31 I am still working on it Jul 14 13:55:34 lol Jul 14 13:55:50 It's a pain isn't it? Jul 14 13:55:54 hm Jul 14 13:56:12 In have to understand how libtool2 sets the flags Jul 14 13:56:21 thats what I am in now Jul 14 13:56:51 Gotcha Jul 14 14:00:22 woglinde: did you yet stumble in "error: m4_require: circular dependency of AC_LANG_COMPILER(C++)" ? Jul 14 14:00:54 (as in bug 4427 & 4391) Jul 14 14:01:24 libtool/autotools issue? Jul 14 14:01:53 woglinde: hey Jul 14 14:09:59 woglinde: I ask 'cause I found this: Jul 14 14:10:02 OK, that's a trivial bug (libtool 2 related). Jul 14 14:10:09 configure.ac needs a AC_PROG_CXX line. Jul 14 14:10:14 " Jul 14 14:25:55 This is just sad Jul 14 14:26:12 I'm installing an ubuntu virtual machine just so I can get past that glibc error in arch Jul 14 15:48:34 morning Jul 14 15:50:17 hi kergoth`work Jul 14 15:56:21 hi kergoth Jul 14 15:56:41 you know what would be nice. screen that can split out each window in the screen session into a seperate terminal or terminal tab Jul 14 15:56:48 or a way to save/load a urxvtd session Jul 14 15:57:37 somewhere, the original author of screen just screamed, and doesn't know why Jul 14 15:57:51 hahah Jul 14 15:58:14 g'day kergoth`work Jul 14 15:58:16 i'm using a tiled wm, i like spawning 2-4 terminals + a gvim for each project/bug i'm working on.. and itd be nice to persist that session Jul 14 16:36:14 ~oebug 4428 Jul 14 16:37:06 * * OE Bug 4428 has been created by  Jul 14 16:37:08 * * gcc-cross-sdk 4.3.1 fails to compile Jul 14 16:37:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4428 Jul 14 16:52:19 regarding http://bugs.openembedded.net/show_bug.cgi?id=4428 where do I put the -march=i486?? Jul 14 16:57:31 where does OE decide what to do with the ARCH setting? If I knew where to look I could test the arch setting and then add the appropriate flag to the compiling command, does that sound reasonable, or should this fix be done another way? Jul 14 17:00:13 or maybe a better question is what type of pc104 SOC does openembedded work well with under $400...Maybe we should forget about running embedded linux on the x86 platform, noone seems to be doing it now... Jul 14 17:00:18 :) Jul 14 17:00:58 KernelKlick, -march=i486 would typically be setup in an include for compiler tuning Jul 14 17:01:27 KernelKlick: lots of folks have used OE for targeting x86 platforms in the past Jul 14 17:02:37 KernelKlick, specifically the -march=i486 would go in 'TARGET_CC_ARCH' which is typically set in conf/machine/include/*.inc files which are included from machine conf files to 'tune' for a specific architecture Jul 14 17:04:17 I have run into issues with some of the x86 based machines - specifically the x86-prescott config seems to have an issue as it includes i686 in PACKAGE_EXTRA_ARCHS even though TARGET_ARCH=i686 Jul 14 17:04:40 so just because a machine exists don't assume it will work - I've also found that the lite5200 won't build b/c of misconfigurations Jul 14 17:05:02 I use the existing machine confs as examples and starting points... Jul 14 17:20:01 any issue running multiple instances of bitbake on same working directory concurrently? they probably would step on each other's cache right? Jul 14 17:23:23 i don't see the point to run multiple instance of bitbake.... Jul 14 17:25:29 I'm working on multiple machines at once... while one is building I want to go tweak and rebuild another... Jul 14 17:26:05 I recall the PC guys saying something like 'well never need more than 640KB' :P Jul 14 17:27:11 hehe! Jul 14 17:27:34 or 32-bits for addressing physical ram, more recently. Jul 14 17:27:43 yup... Jul 14 17:28:45 doing a parallel building of the same target should be faster anyway Jul 14 17:29:52 diego__, your missing my point... one build is already running then I decide to make some changes to something that affects a different machine... Jul 14 17:30:22 I'm fairly certain the answer is that the two will step and each other... Jul 14 17:30:25 how do you know that not affect the current building? Jul 14 17:30:51 because I'm being careful :P Jul 14 17:31:20 ah ok, i suppose you don't build gdb... :D Jul 14 17:31:58 I'm mainly plain around with images right now... and not building the same images per machine Jul 14 17:34:01 i think is possible to run more bitbake instances if you duplicate configurations, but i'm not sure is whorth the pain Jul 14 17:34:41 I think I agree with you Jul 14 17:35:36 is anyone using 'devshell'? Jul 14 17:38:37 uhm, what is it a devshell? Jul 14 17:38:54 bitbake task ('bitbake -c devshell') Jul 14 17:40:24 no, i've played with bitbake -i time ago, but i stopped when someone had pointed out it was broken Jul 14 17:40:52 devshell is described here http://www.openembedded.org/repo/org.openembedded.documentation/usermanual/html/index.html (I think - I can't seem to load that page right now) Jul 14 17:41:09 I ask because I don't understand how its very useful the way its defined... Jul 14 17:41:30 im reading this: http://wiki.openembedded.net/index.php/Interactive_debugging Jul 14 17:42:39 I was more interested in something that let you go into a working directory and tweak the configuration - ie, busybox config file, kernel make menuconfig etc - I believe that's what devshell was created for but it drops you in a shell 'prior' to do_patch and do_configure which I don't understand Jul 14 17:43:32 it makes sense, probably you may need to "debug" also configure or patch Jul 14 17:44:01 whee, i'm awake and actually somewhat energic and motivated for the first time in a -long- time Jul 14 17:48:56 perhaps the concept with devshell is to first build a target (so that its patched and configured) and 'then' do a devshell to it - I just don't understand why the devshell task isn't inserted after do_configure Jul 14 18:21:33 tharvey: do you have a working x86 bootstrap-image or whatever? I can get bitbake nano to work, helloword, and task-base. I just cant get bootstrap or console to work Jul 14 18:23:31 jo kernelklick Jul 14 18:27:42 jo woglinde, did you sleep well? Jul 14 18:28:05 btw I have no idea what jo means...sorry Jul 14 18:28:23 KernelKlick: jo means "yo" Jul 14 18:28:47 ah...thanks. Jul 14 18:28:54 KernelKlick: I'm sure woglinde did not get any sleep at all as he's been working really hard on finding a way around libftdi's rpath problems. :D Jul 14 18:29:47 thought it might mean "jump off" or "jerk off" or something since i bothered him all yesterday and he couldnt work on rpath problems :) Jul 14 18:30:13 KernelKlick, sorry I don't use bootstrap... just kernel and bzImage Jul 14 18:35:08 tharvey: are those bitbake recipes too? bzimage, i thought was kernel compiling thing... Jul 14 18:36:41 so you build your x86 by hand...compile your own kernel bzimage...do you then use Openembedded to add packages or what? Jul 14 18:36:55 KernelKlick, well... every 'target' is from a recipe. The bzimage is from kernel recipe which is overloaded or virtual, the image I'm using is base-image - sorry I mispoke when I said 'kernel and bzImage' Jul 14 18:37:46 I created my own machine for an x86 ETX board... I started by trying x86-prescott but it won't build without a tweak or two then I realized its probably not the right machine for me anyway Jul 14 18:38:05 LOL...thats where I am Jul 14 18:38:27 I'm unaware of what existing x86 based machines in OE tree actually build, but I can tell you what you need to do to x86-prescott to get it to build Jul 14 18:40:10 well actually it sounds like I need to reproduce your solution. I will eventually have to build my own kernel because the SOC i am using does not have a math coprocessor even though it is a 486...go figure. Jul 14 18:41:06 yeah I dont see the kernel recipe in packages/images Jul 14 18:41:37 thats where you would find 'images' which would mean 'root filesystem images' Jul 14 18:41:48 kernel recipes are in packages/linux Jul 14 18:42:16 * broonie is amused to see someone surprised by a 486SX :) Jul 14 18:42:36 broonie: kids :) Jul 14 18:42:47 im not sufprised by the 486...but linux sure is Jul 14 18:42:53 surprised Jul 14 18:43:27 KernelKlick: Hehe. Nah. I'm just grateful that he's even looking at it :) Jul 14 18:43:28 all the linux distros including 386 compats die when they cant find the coprocessor Jul 14 18:43:29 KernelKlick, make a conf for your specific motherboard, assuming its not there already in conf/machine/.conf Jul 14 18:43:33 Linux is 386-era, when the only way to get a FPU was with a fancy extra chip :) Jul 14 18:43:52 yep Jul 14 18:44:40 KernelKlick, use one of the other x86 based board as an example... perhaps based off something else i486 like godelx.conf Jul 14 18:44:47 geodelx.conf that is... Jul 14 18:44:58 or wrap.conf... Jul 14 18:45:17 tharvey: okay...I will try Jul 14 18:46:18 KernelKlick, you'll want to customize PREFERRED_PROVIDER_virtual/kernel - probably set it to just 'linux' unless you need specific patches for your system, in which case you'll want to create your own kernel recipe for it Jul 14 18:47:12 can the recipe turn on FPU emulation? Jul 14 18:47:19 the kernel one...sorry Jul 14 18:47:29 if you can find out the proper gcc params for your FPU situation I 'think' you would put them in a 'TARGET_CC_ARCH' var Jul 14 18:48:00 anything that you would do in linux .config you would need to create your own kernel recipe for (or add your defconfig to an existing one) Jul 14 18:50:38 KernelKlick, if you specify linux-x86 for your PREFERRED_PROVIDER_virtual/kernel there is a packages/linux/linux-x86_2.6.20.bb recipe that may be appropriate but if you need your own defconfig you would edit that recipe, and place your defconfig in its subdir Jul 14 18:50:58 (I'm still learning OE by the way, so if I'm saying anything wrong hopefully someone else will jump in) Jul 14 19:14:30  tharvey: which distro should I favor? No gui, just busybox, ftp, ssh, nfs, samba, usb hotmount, watchdog. That should be about it. No webservers, databases, email...no frills Jul 14 19:14:59 i know we want to be able to plug and mount usb flash drives Jul 14 19:20:40 KernelKlick, I've been working with angstrom-2008 - it seems thats what most OE based projects either use or base their own distro on Jul 14 19:22:19 okay..thanks Jul 14 19:26:51 KernelKlick whats in the image defines only the image Jul 14 19:27:24 distro is for setting which libc is used or which compiler and sane versions of packages Jul 14 19:28:44 KernelKlick so likely you should look at the images.bb and find a suitable one or define your own image and task Jul 14 19:32:27 distro also defines packing options such as packager, feeds, etc Jul 14 19:32:43 sure Jul 14 19:32:58 okay....speaking of...where to find list of possible machine_features Jul 14 19:33:13 hm Jul 14 19:33:25 unfornatly there os none Jul 14 19:33:29 is Jul 14 19:33:34 lol Jul 14 19:33:40 just my luck Jul 14 19:33:40 *g* Jul 14 19:33:51 yeah lack of documentation Jul 14 19:34:12 look in the task packaes and compare them with the machine confs Jul 14 19:34:54 i was afraid of that....if I leave it blank, does it inherit it from somewhere? Jul 14 19:35:28 KernelKlick, have you read poky's manual? by far the best documentation for OE Jul 14 19:35:49 it has a pretty good list but the best thing to do is look at task-base.bb where the features are 'used' Jul 14 19:36:12 okay Jul 14 19:38:47 start small... just use kernel26 until you have a kernel and root filesystem booting Jul 14 19:43:49 Anyone having problems with the OE website? Jul 14 19:44:07 mewyn: Better question: anyone NOT having problems with the OE website? ;) Jul 14 19:44:32 lol Jul 14 19:44:41 Well, when you phrase it like that ;) Jul 14 19:44:58 ya... website seems to be down Jul 14 19:47:15 in the task-base there is a uboot...is that saying uboot already exists and we want to tie in or is it adding uboot to image...does it use grub or lilo if not chosen? I wouldn't know...haven't had any of these work yet... ;) Jul 14 19:50:32 KernelKlick most embedded boards and proz dont use either grub or lilo Jul 14 19:51:19 oops...sorry....inserting foot into my mouth :P Jul 14 19:52:00 Gross Jul 14 19:52:06 * piroko uses u-boot Jul 14 19:53:15 so if I add Machine_Features=".....uboot..." it will add uboot to the image it creates? wow. that is cool Jul 14 19:54:15 re Jul 14 19:54:37 Am I SOL for grabbing the mtn file? Jul 14 19:58:13 in conf/Machine, what is FEED_ARCH? Jul 14 19:58:29 mewyn: Probably. Just keep tryin Jul 14 20:00:03 lol. As you said that, my remote machine at home kicked in and started grabbing it, but my local machine is just in a retry loop. Jul 14 20:00:51 Irony :) Jul 14 20:01:47 Makes it hard to use in my VM here. Jul 14 20:02:06 But, I'll get it eventually. :) Jul 14 20:24:36 hi pb Jul 14 20:26:52 should nfsroot work when your distro (angstrom in my case) defaults your interface to use dhcp? from what I'm seeing when init trys to configure the network interfaces it kills the rootfs before it can obtain a dhcp IP and recover Jul 14 20:31:43 tharvey, comment out the #auto eth0 in /etc/network/interfaces Jul 14 20:32:42 do other see failures building m4-native using parallel make? Jul 14 20:32:54 it fails, then I restart and the builds goes on Jul 14 20:36:12 I only have one cpu Jul 14 20:39:19 Crofton|work, that causes networking service to leave eth0 alone right? otherwise when dhcp client is started interface is temporarily unconfigured vs static assigment where the IP changes but was never unconfigured Jul 14 20:39:33 yeah Jul 14 20:39:39 thx Jul 14 20:39:51 apparently the auto thing takes the interface down before bringing it up Jul 14 20:40:02 which is bad if root is on nfs :) Jul 14 20:40:37 actually I think its a feature of dhcp client... auto is fine if you use static Jul 14 20:40:53 but dhcp client assignes 0.0.0.0 then sends out a request... Jul 14 20:45:19 i have a question at is surely dumb but i can't seem to figure out. how are modules loaded at runtime determined? I don't have anything in /etc/modutils, /etc/modules.d, /etc/modules.conf. Some kernel modules load upon boot and others don't. I'm stumped. Jul 14 20:46:19 kgilmer: I'd assume udev is doing it, unless you have them being inserted in a boot script Jul 14 20:47:51 mewyn...so there is no reference to the modules that are loading in /etc/udev Jul 14 20:48:12 should i be looking somewhere else? Jul 14 20:48:34 Well, udev likes to autoload modules if it finds a mapping. Jul 14 20:48:42 What are the modules being loaded? Jul 14 20:49:42 kgilmer, I've been wanting to ask the exact same question... Jul 14 20:50:41 well, one is isp1504_arc Jul 14 20:50:54 we have several kernel modules specific to our device Jul 14 20:51:09 i have a usb host yet I have to manually modprobe ehci-hcd - wondering why its not automatically loaded Jul 14 20:51:19 some of them are loading upon boot (ie i see them when i type lsmod), others are not listed. Jul 14 20:51:31 i can loaded them by hand with modprobe. Jul 14 20:51:46 i can't find any file that references the specfic modules that are loading. Jul 14 20:52:03 tharvey, well i hope we can find an answer :) Jul 14 20:55:23 what modules are autoloading and what image are you using? I have none on my target Jul 14 20:55:46 kgilmer: Well, udev looks in /sys to see what hardware is on the system, then looks at your modules to see what it can load up. Jul 14 20:56:27 So, it's my guess that udev is loading up those modules based on what it's finding in /sys Jul 14 20:56:32 i'm using Poky. There are about 10 modules Jul 14 20:56:34 4 are loading on boot Jul 14 20:56:41 that makes sense mewyn Jul 14 20:56:52 I also just stumbled on /etc/modutils.sh Jul 14 20:57:00 which looks like it reads /etc/modules Jul 14 20:58:06 adding a line to that file to see if boot behavior changes... Jul 14 21:00:10 no luck with /etc/modules Jul 14 21:01:08 Are you trying to not load these modules? Jul 14 21:01:31 well there are some that are being loaded and others that are not. Jul 14 21:01:35 i want to load the ones that are not Jul 14 21:01:54 and i want to be able to define that behavior in a recipe or as part of my system image Jul 14 21:03:09 I'd have to bone up on my udev to figure it out with udev. Jul 14 21:05:44 well looks like modutils.sh init script calls insmod Jul 14 21:05:51 so i need to pass the full path to my module... Jul 14 21:05:58 trying again Jul 14 21:25:54 well i figured it out Jul 14 21:26:03 should have been much easier but i'm pretty jetlagged atm Jul 14 21:26:17 to get something to load on boot that isn't done with udev Jul 14 21:26:25 simply add the name of the module to /etc/modules Jul 14 21:26:38 but, if you have it in /etc/modules.d, then it will not work Jul 14 21:27:12 a fatal error will occur in modprobe but the init script will not display it Jul 14 21:27:38 so tharvey, add ehci-hcd to /etc/modules Jul 14 21:27:45 mewyn, thanks for your help :) Jul 14 21:28:43 kgilmer, I don't have an /etc/modules currently... its just a list of modname per line? Jul 14 21:29:01 yep. i didn't have it by default either. Jul 14 21:30:07 nice... worked great Jul 14 21:30:08 thanks! Jul 14 21:30:38 what parses that? is that standard or just an OE thing? Jul 14 21:31:35 and if you didn't have one, how were the 4 modules getting loaded? Jul 14 21:35:33 it's a rcS (startup) script linked from /etc/init.d Jul 14 21:35:37 modutils.sh Jul 14 21:35:46 loads /etc/modules and calls modprobe on each line Jul 14 21:36:10 not sure if it's standard or not Jul 14 21:36:32 regarding the 4 other modules, I assume mewyn is right and udev is somehow figuring out they need to be loaded. Jul 14 21:36:43 i'll have to dig into udev to figure it out Jul 14 21:36:53 does anyone know why many recipes when creating 'pkg_postinst' functions check if the env has ${D}? Jul 14 21:37:16 like we have for openssh Jul 14 21:38:50 kgilmer, ah... that make sense - similar to how openWRT does it - curious though why it doesn't instead source things from /etc/modules.d so that packages can easily add/remove files listing the modules they need Jul 14 21:40:33 tharvey, yeah...no idea. i'm just glad my lcd works now when i boot. :) Jul 14 21:43:24 heading home, gnite Jul 14 21:43:32 take care Kgilmer Jul 14 21:55:53 mickeyl: ping Jul 14 21:56:18 püng Jul 14 21:56:23 hello Jul 14 21:56:42 I committed a first try of udev_124 Jul 14 21:57:00 who is expected to commit this "core-package"? Jul 14 21:57:09 (oebug 4118) Jul 14 21:57:49 good question Jul 14 21:57:56 on c7x0 the situation is excactly as (bad) as with udev_118 Jul 14 21:58:09 so why don't move? Jul 14 21:58:34 agreed Jul 14 21:59:01 i will commit it these days with DEFAULT_PREFERENCE=-1 and give people a bit time to test Jul 14 21:59:08 then making it the new default Jul 14 21:59:26 * chouimat|away wonders if he order a freerunner Jul 14 21:59:27 yes, please test it on other devices as well Jul 14 22:27:19 mickeyl: I'll open a RFC for udev to avoid you asbestos... Jul 14 22:28:04 mickeyl, how good are the meta toolchain targets? Jul 14 22:28:04 with the latest udevsettle patch you can really shrink the udevsettle timeout to the min Jul 14 22:29:44 ant_: ok, cool. Jul 14 22:30:09 mickeyl, we are getting interest in OE from TI these days Jul 14 22:30:10 Crofton|work: i don't know the current status, last time i used them a couple of months ago, they were fine Jul 14 22:30:16 (TI) glad to hear Jul 14 22:30:19 largely due to the beagle baord work Jul 14 22:30:28 we need to _move_ now re. git Jul 14 22:30:36 this blocade is a nightmare Jul 14 22:30:37 mickeyl, make it so Jul 14 22:30:39 heh Jul 14 22:31:02 lets wait for RP to be back, then we move Jul 14 22:31:07 ok Jul 14 23:07:42 good nite Jul 14 23:16:05 * * OE Bug 3575 has been marked as DUPLICATE of bug 3489 by andrea.adami(AT)gmail.com Jul 14 23:16:07 * * gpe-bluetooth deactivation on hx4700 Jul 14 23:16:10 * * http://bugs.openembedded.net/show_bug.cgi?id=3575 **** ENDING LOGGING AT Tue Jul 15 02:59:56 2008