**** BEGIN LOGGING AT Sat Feb 16 02:59:57 2008 Feb 16 03:06:18 03pfalcon 07org.oe.dev * rab8ea819... 10/ (1 packages/tasks/task-devimage.bb): task-devimage: Stuff kernel-image inside, for kexec testing. Feb 16 03:06:28 03pfalcon 07org.oe.dev * r6453c193... 10/ (5 files in 3 dirs): Feb 16 03:06:28 initramfs-module-bootmenu, initramfs-module-kexecboot: Fixes for kexecing. Feb 16 03:06:28 * Be sure to pass original command line contents to new kernel. Feb 16 03:06:28 * Add protection against infinite recursive kexecing. Feb 16 03:06:34 03pfalcon 07org.oe.angstrom-2007.12-stable * rae6e8845... 10/ (1 BACKPORTS.txt): BACKPORTS.txt: Note pending linux-handhelds-2.6 backports. Feb 16 04:01:56 ant_: ping Feb 16 07:03:44 Does OE presuppose a dynamically linked /bin/sh on the build machine? Feb 16 07:36:00 Laibsch: morning Feb 16 07:36:44 hvontres|home: Good morning Feb 16 07:36:48 How are you doing? Feb 16 07:37:00 Shouldn't you go to bed? Feb 16 07:40:35 Laibsch: hehe.. just watching the tonight show. Feb 16 07:40:44 Aha Feb 16 07:40:45 Plus I have a three day weekend Feb 16 07:40:53 I hope it is entertaining Feb 16 07:40:59 Oh, that is nice Feb 16 07:41:16 I hope you join our "push'em"-weekend, then ;-) Feb 16 07:44:38 Laibsch: maybe. I need to replace part of my fence... Oh the joys of owning a home :) Feb 16 08:46:21 gm Feb 16 08:47:42 hvontres|home: pong Feb 16 08:53:26 ant_: nevermind... I didn't realize the patch you talked about earlier was already in .dev. Resume now works on poodle :) Feb 16 09:03:40 hvontres|home: the Z seems to resume. I kept the logs. Feb 16 09:07:51 hvontres|home: resume logs : http://pastebin.com/m76391262 Feb 16 09:14:45 ant_: interesting. I don't get the warnings on resume Feb 16 09:16:16 ant_: time to go to bed :) Feb 16 09:18:08 03koen 07org.oe.dev * r88db3959... 10/ (1 conf/machine/poodle.conf conf/machine/tosa.conf): poodle,tosa: unbreak PACKAGE_ARCH Feb 16 09:19:03 hvontres|work: poor novel Tom Sawyer (painting fences)...sleep good! Feb 16 09:56:15 hi all, is there a way how to get from OE if it build oabi or eabi? Feb 16 10:17:03 Magon_: It should tell you in the first few lines Feb 16 10:17:23 Laibsch: first line of what? Feb 16 10:18:18 look at the output of "bitbake blah" Feb 16 10:18:26 Those first few lines Feb 16 10:18:39 Laibsch: ok i have it Feb 16 10:18:55 TARGET_ARCH = "arm" <- that is eabi right? Feb 16 10:19:10 TARGET_OS = "linux-gnueabi" Feb 16 10:19:18 ok Feb 16 10:22:35 Laibsch: thx i probably have problem with that..it say when executed Illegal instruction Feb 16 10:22:55 it= builded binary Feb 16 10:25:04 TARGET_ARCH = "arm-oabi" Feb 16 10:25:04 TARGET_OS = "linux" Feb 16 10:25:16 this is going to be oabi binaries..right? Feb 16 10:55:12 howdee doodly doo Feb 16 10:55:43 XorA: hey, you are supposed to be back on monday Feb 16 10:56:00 zecke: I dont work until Monday :-) Feb 16 10:56:38 XorA: then I don't talk about your work until monday but will just put an more evil grin on my face Feb 16 10:57:16 zecke: its ok, you can say its all crap now :-), Ill just ignore the insults until monday :-D Feb 16 10:59:09 what is the right magic combination of TARGET_OS and TARGET_ARCH for arm oabi? Feb 16 12:39:57 morning boys and girls Feb 16 12:50:10 hi mickeyl Feb 16 12:50:16 hi mickey|flu, health to thee Feb 16 12:50:18 hi woglinde Feb 16 12:50:20 hi ph5 Feb 16 12:50:34 hi fooks Feb 16 12:50:38 blessings to mickey|flu Feb 16 12:52:55 thanks guys Feb 16 12:53:01 working on recovering... Feb 16 12:57:03 * * OE Bug 3781 has been RESOLVED (FIXED) by Feb 16 12:57:05 * *  Feb 16 12:57:07 * * http://bugs.openembedded.org/show_bug.cgi?id=3781 Feb 16 12:57:39 yo mickey|flu Feb 16 12:58:04 * * OE Bug 2452 has been RESOLVED (FIXED) by Feb 16 12:58:05 * * New recipe - ifplugd 0.28 Feb 16 12:58:07 * * http://bugs.openembedded.org/show_bug.cgi?id=2452 Feb 16 12:58:15 * * OE Bug 3778 has been RESOLVED (FIXED) by Feb 16 12:58:17 * *  Feb 16 12:58:19 * * http://bugs.openembedded.org/show_bug.cgi?id=3778 Feb 16 13:00:09 welcome back XorA Feb 16 13:00:36 mickey|flu: get well before leffe session :-)( Feb 16 13:00:37 thanks to your patches w/ gst-pulse 0.9.7 and PA 0.9.9 we have ~45% CPU load playing mp3 Feb 16 13:00:54 mickey|flu: cool, more ammo to hit lennert with :-) Feb 16 13:00:56 XorA: (leffe) no kidding, i'm happy i got it this week and not next week Feb 16 13:01:02 righto Feb 16 13:01:18 mickey|flu: you can be the evangalist of no busy looping Feb 16 13:02:48 yeah, I'm curious what he will say about that Feb 16 13:03:07 the 500µsec was just a guess or did you try other values as well? Feb 16 13:03:12 * XorA bans pulseaudio for speading up global warming Feb 16 13:03:25 mickey|flu: I worked down from 100ms Feb 16 13:03:34 *nod* Feb 16 13:04:47 500usec was first value that didnt cause skips in audio Feb 16 13:04:56 so we have the TLS problem, the priorized mutexes, and the busylooping Feb 16 13:05:03 three things to take up with him Feb 16 13:05:11 Ill bet you can work out with maths what the value should be Feb 16 13:05:21 03mickeyl 07org.oe.dev * rfa4b0c91... 10/ (6 files in 4 dirs): Feb 16 13:05:21 busybox 1.9.1 one defconfig to rule them all, add patch for long options. fixes #3781 -- thanks Feb 16 13:05:21 Andrew Paulsen! Feb 16 13:05:26 03mickeyl 07org.oe.dev * r7e284f07... 10/ (4 files in 3 dirs): ifplugd 0.25 remove nylonisms and bashisms, remove 0.20, closes #1587 Feb 16 13:05:33 certainly we need different values based on the system performance Feb 16 13:07:19 it looks like pulse acheives its low latency by dealing with audio as a stream of bytes. But for performance you need to deal with audio as a stream of blocks, I think embedded pulse needs a block version which isnt as good as the desktop version for latency but pulls down the cpu load Feb 16 13:08:15 gm Feb 16 13:11:46 morning Crofton Feb 16 13:11:55 XorA: i see. well, bummer, that's a fundamentally different approach Feb 16 13:12:17 mickey|flu: the underlying hardware works in DMA transfer sized blocks Feb 16 13:14:37 mickey|flu: I guess we shall just have to balance the two, all the usleep adds to pulse-gst is more empty space in buffer per loop Feb 16 13:14:47 mickey|flu: effectively emulating a block mode Feb 16 13:15:52 *nod* Feb 16 13:22:52 mickey|flu what ?!?!?! flu ... no no we need you at FOSDEM !!! Feb 16 13:24:12 gremlin[it]: sure, that's why i'm having it now and not next week ;) Feb 16 13:24:22 i'll surely be recovered by then Feb 16 13:25:01 we will inject him with leffe until he is a) better, b) undead Feb 16 13:26:10 ahh ok ;) Feb 16 13:27:49 hehe Feb 16 13:27:56 i need to get my act together and provide some PR material Feb 16 13:28:02 i guess no one else is likely to do it Feb 16 13:28:03 right? Feb 16 13:28:05 *sigh* Feb 16 13:28:51 seem you are right ... Feb 16 13:28:59 what kind of material ? Feb 16 13:31:54 minimum is an A5 flyer that sums up the most important points of "What can OE do for you" Feb 16 13:32:05 and a fat ass eyecatcher poster Feb 16 13:32:19 graeme had a nice proposal Feb 16 13:32:25 i'll see whether i can do something based on that Feb 16 13:32:32 will reserve tomorrow to do it Feb 16 13:40:35 i'm not good to do such things ... Feb 16 13:41:48 if i'm able (and i'll spend all next nights for that) the neuros will work with OE, so i need a television or a LCD with composite input ... Feb 16 13:43:43 zecke pasted "Qtopia configure hacks" at http://paste.lisp.org/display/55998 Feb 16 13:57:41 03mickeyl 07org.oe.dev * r8379319d... 10/ (1 packages/gpsd/gpsd.inc packages/python/python_2.5.1.bb): gpsd: remove python-ncurses dependency from gspd, it's actually python-curses we depend on Feb 16 14:11:04 * * OE Bug 3792 has been created by  Feb 16 14:11:06 * * midori-0.0.15-r0-do_compile Feb 16 14:11:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3792 Feb 16 14:15:04 * * OE Bug 3793 has been created by  Feb 16 14:15:06 * * midori-0.0.17-r0-do_configure Feb 16 14:15:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3793 Feb 16 14:19:02 there is an how-to about configure OE with an external toolchain ? Feb 16 14:20:29 yes Feb 16 14:20:35 its in the wiki Feb 16 14:24:10 thanks Feb 16 14:42:04 * * OE Bug 3794 has been created by  Feb 16 14:42:06 * * minimal-openmoko-image-1.0-r0-do_rootfs Feb 16 14:42:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3794 Feb 16 14:43:04 * * OE Bug 3795 has been created by  Feb 16 14:43:06 * * libmokogsmd2-0.1.0+svnr3663-r2-do_compile Feb 16 14:43:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3795 Feb 16 15:06:08 03likewise 07org.oe.dev * r5596670e... 10/ (5 files in 4 dirs): linux-(rt-)2.6.24: Fix clock resolution to nanosecond. Feb 16 15:06:13 03pfalcon 07org.oe.dev * r1821f6ea... 10/ (1 conf/machine/h3900.conf packages/tasks/task-base.bb): h3900.conf: Ship ipaq-boot-params Feb 16 15:06:23 03pfalcon 07org.oe.angstrom-2007.12-stable * ra05b9437... 10/ (27 files in 27 dirs): (log message trimmed) Feb 16 15:06:23 Merge from .dev: Feb 16 15:06:23 0d376af69d870f4d7e004a35b9ce9adbb7bb1600 linux-handhelds-2.6 2.6.21-hh20: Enable CONFIG_ATAGS_PROC consistently. (Paul, Koen) Feb 16 15:06:23 5e355cb98cf1742f5a79dfe4777b01a20d63e782 linux-handhelds-2.6 2.6.21-hh20: Add support for passing kexec command line. (Paul, Koen) Feb 16 15:06:25 da903ce92b8f5d3cb42f9ffd3e15c6b4b8563dc4 linux-handhelds-2.6 2.6.21-hh20: Enable NBD and CONFIGFS as modules. (Paul, Koen) Feb 16 15:06:28 1e069dfafed0f72201d0b9415a94299ff9638e96 linux-handhelds-2.6 2.6.21-hh20: Use noop iosched by default, make other modular. (Paul, Koen) Feb 16 15:06:31 eeedd7c12d801248e236e657abdb9f0938952b09 linux-handhelds-2.6 2.6.21-hh20: Enable few more FSes and introspection options. (Paul, Koen) Feb 16 15:06:35 03pfalcon 07org.oe.angstrom-2007.12-stable * r5d40b8d7... 10/ (1 BACKPORTS.txt): BACKPORTS.txt: Move linux-handhelds-2.6 fixes to appplied. Feb 16 15:07:01 woglinde i didn't find you remember the title of the page ? Feb 16 15:07:04 * * OE Bug 3796 has been created by  Feb 16 15:07:06 * * openmoko-feedreader2-0.0.1+svnr3645-r1-do_compile Feb 16 15:07:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3796 Feb 16 15:07:16 * * OE Bug 3797 has been created by  Feb 16 15:07:18 * * openmoko-browser2-0.0.1+svnr3646-r1-do_compile Feb 16 15:07:20 * * http://bugs.openembedded.net/show_bug.cgi?id=3797 Feb 16 15:58:38 Hello Feb 16 15:58:57 I just updated OE, and I'm having a problem building python 2.5.1 Feb 16 15:59:19 Specifically, it fails during do_install with this: http://pastebin.com/m2c39b844 Feb 16 16:00:27 I'm trying to build ml8, last version I built was ml5 Feb 16 16:00:39 I didn't update OE in between. Feb 16 16:02:03 Any ideas or help? :) Feb 16 16:06:04 03thebohemian 07org.oe.dev * rc3619a54... 10/ (4 files in 2 dirs): Feb 16 16:06:04 classpath: Feb 16 16:06:04 - make classpath-tools and classpath-examples rdepend on a runtime Feb 16 16:06:04 - make classpath-examples rdepend on classpath-awt since it contains AWT + Swing demos Feb 16 16:06:04 - adjust revisions Feb 16 16:17:16 hi, i have a small problem...how to tel angstrom to build linux and not linux-gnueabi when building on ARM? setting ARM_ABI="oabi" is not working as it probably should Feb 16 16:39:04 * * OE Bug 3798 has been created by  Feb 16 16:39:06 * * gumstix-kernel-2.6.21-r1-do_unpack Feb 16 16:39:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3798 Feb 16 16:50:59 hi all, can I include a new architecture to the linux kernel (such as arm or sparc, ...) and then add it to openembedded ? Feb 16 16:58:04 * * OE Bug 3787 has been marked as DUPLICATE of bug 3134 by Feb 16 16:58:06 * * apmd-3.2.2-r10-do_compile Feb 16 16:58:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3787 Feb 16 17:00:23 how can I get bitbake to rerun a kernel build after modifying sources? Feb 16 17:05:05 blak3: bitbake -c rebuild package ? Feb 16 17:05:38 03likewise 07org.oe.dev * r5c877070... 10/ (1 packages/lighttpd/lighttpd/src-server.c.patch): lighttpd: annotate upstreamed patch. Feb 16 17:05:44 03pfalcon 07org.oe.dev * r98f85bb9... 10/ (1 packages/linux/linux-handhelds-2.6.inc): linux-handhelds-2.6: More elaboration to HaRET's sample startup.txt. Feb 16 17:05:49 03mickeyl 07org.oe.dev * rb168672c... 10/ (13 files in 4 dirs): Feb 16 17:05:49 (python-)efl cvs update and catch up with upstream version changes, add missing dependency in Feb 16 17:05:49 python-ecore Feb 16 17:05:53 03mickeyl 07org.oe.dev * r28880033... 10/ (1 packages/tasks/task-openmoko-python-devel.bb): task-openmoko-python-devel: catch up with renaming python-efl to task-python-efl Feb 16 17:05:57 03pfalcon 07org.oe.angstrom-2007.12-stable * rf7a8b720... 10/ (1 BACKPORTS.txt): Feb 16 17:05:57 BACKPORTS.txt: Record: Feb 16 17:05:57 242687582dbdcc1fdca2fc9af1f32a2b5c591f3c bitbake.conf, image.bbclass: Add and use USERDISTRO var to preserve original DISTRO value. (PaulS) Feb 16 17:05:59 465f9c73626153a7e195a5b8c08ae66282c3722f bitbake.conf: Set default IMAGE_FSTYPES to "tar.gz". (PaulS) Feb 16 17:06:02 03pfalcon 07org.oe.angstrom-2007.12-stable * rf089a35f... 10/ (1 packages/linux/linux-handhelds-2.6.inc): Feb 16 17:06:05 applied changes from eeedd7c12d801248e236e657abdb9f0938952b09 Feb 16 17:06:07 through 9a8a930122d29384549135ea3382884efdb4000e Feb 16 17:06:09 linux-handhelds-2.6: Record the fact that flash booting should be done with root=mtdN (PaulS, Koen) Feb 16 17:06:13 03pfalcon 07org.oe.angstrom-2007.12-stable * r63973bab... 10/ (1 packages/linux/linux-handhelds-2.6.inc): Feb 16 17:06:14 thanks magon Feb 16 17:06:16 applied changes from 1821f6eadefc14ad7ce51d5ab026da139608f557 Feb 16 17:06:18 through 98f85bb96baab42bb0d213a35c2cd9f1c06a3f47 Feb 16 17:06:19 linux-handhelds-2.6: More elaboration to HaRET's sample startup.txt. (PaulS, Koen) Feb 16 17:06:21 03pfalcon 07org.oe.angstrom-2007.12-stable * r340f3a0f... 10/ (1 BACKPORTS.txt): BACKPORTS.txt: Couple of missed patches from previous patchset. Feb 16 17:29:30 pls help noob question: when I do a "bitbake -c rebuild virtual/kernel" it unpacks the sources again, which is not what I want Feb 16 17:29:54 can I get it just to rebuild the kernel without unpacking sources again rolling over my change? Feb 16 17:45:15 blak3: look at directory stamps/your_machine-angstrom-linux Feb 16 17:45:45 sort order build-time Feb 16 17:46:39 thank you Feb 16 17:47:01 delete all files from linux-your_kernel_version.do_compile Feb 16 17:47:29 I may have found what I need with interactive mode ... unpack, patch manually, then build Feb 16 17:47:40 I don't see a "stamp" dir Feb 16 17:48:41 oh stamps Feb 16 17:48:51 sorry ../tmp/stamps/... Feb 16 17:49:06 my bad you said stamps Feb 16 17:49:42 excellent, thanks much 8) Feb 16 17:49:54 no problem Feb 16 18:17:35 Hi all, I need to have a linux kernel for a new processor architecture (I mean by a new processor architecture a one like arm or sparc or ...) which is not included in the linux kernel sources. I thought I could have linux on it by developing a new arch for the kernel but I don't know how or who can help. could you please guide me through this. Feb 16 18:19:34 Aha. Found it. A statically linked tar and coreutils on the host will cause fakeroot to generate a cooked filesystem image. Feb 16 18:19:47 Do you have a compiler and toolchain for this architecture? Feb 16 18:19:48 s/host/buildhost/ Feb 16 18:31:05 mwester: no I don't, but it's so simple architecture Feb 16 18:33:48 mwester: I think I could develop an assembler and compiler for this arch Feb 16 18:34:32 mohamed: well, then that's where you start. Until you have a toolchain (preferably a gnu toolchain) that can generate code for the target, you can't do anything. Feb 16 18:35:59 mwester: then I could use oe to build a distro for the new arch :) ? Feb 16 18:39:28 mwester: I mean the linux kernel will be equipped with the new arch, then I can specify the target arch in the conf file and start to build for it ? Feb 16 18:46:51 mwester: can I tailor gnu toolchain to work for the new arch? Feb 16 19:22:04 * * OE Bug 3799 has been created by  Feb 16 19:22:06 * * uclibc-0.9.29-r9-do_compile Feb 16 19:22:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3799 Feb 16 20:01:40 hi Feb 16 20:24:12 is here someone who understands bitbake? i might found a bug, but dont know exactly and dont know how to ask google if it wasnt found earlier Feb 16 20:24:36 my problem is that i added PACKAGE_ARCH = "arm-oabi" in my local.conf Feb 16 20:25:52 all building went smoothly but at the end a cannot build rootfs Feb 16 20:26:43 dunno about oabi but you might try replacing "=" with "+=" Feb 16 20:27:22 my problem is that ipkg.conf is configured wrongly and i dont know how to presduade it to right setting Feb 16 20:27:50 sorry, dunno offhand. Feb 16 20:28:18 it doesnt mention deploy/glibc/ipk/arm-oabi just deploy/glibc/ipk Feb 16 20:28:26 and all and machine name Feb 16 20:28:34 but not arm-oabi where most of the files reside Feb 16 20:30:09 Magon: what makes you think that you can add random settings to local.conf and expect any sane result? Feb 16 20:30:30 psokolovsky: that is my problem..i dont know where it was supposed to go :-) Feb 16 20:30:54 psokolovsky: that is why i am asking...it all went just fine but failed here an i thougth that it might be the problem Feb 16 20:31:23 problem is that i was unable to force angstorm to build arm-oabi instead of eabi Feb 16 20:31:23 Magon: to build Angstrom, you need just to select it as a distro and set MACHINE. doing any other changes is not really supported. Feb 16 20:31:41 psokolovsky: that want build oabi Feb 16 20:32:30 psokolovsky: there is ARM_ABI variable but it seems to have no efect Feb 16 20:33:17 it always include angstrom-glibc.inc and there selects TARGET_OS = "linux-gnueabi" Feb 16 20:33:30 Magon: again, you set DISTRO and MACHINE, that's all. But to get idea how it works, have a look at angstrom-2007.1.conf:215 Feb 16 20:34:31 psokolovsky: what is supposed to be there..i might have old version of OE it seems Feb 16 20:34:57 ARM_ABI ?= "${@['','oabi'][bb.data.getVar('MACHINE',d) in ['collie','h3600', 'h3800', 'simpad', 'htcwallaby']]}" Feb 16 20:35:14 yes..that exact line was my problem for most of today Feb 16 20:35:19 So, it selects needed mode based on machine. All machines for oabi are strongarm. Feb 16 20:36:12 my problem is that i have machine which is not there..i have zipit Z2 and it is not in OE yet but i have working machine file Feb 16 20:37:59 seems like i need to add my machine to the machine list there and in angstrom-glibc.inc Feb 16 20:38:18 Magon: well, obvious solution is to add you machine to that list! but again, oabi support is intended for SA devices. Feb 16 20:38:22 Magon: sadly, there is another such list in conf/distro Feb 16 20:39:02 it seems like the list are in more places..that is why i added it hardcoded to local.conf Feb 16 20:39:19 psokolovsky: problem is i have kernel i cannot reflash and it supporst just oabi Feb 16 20:40:14 03likewise 07org.oe.dev * rc9c9f335... 10/ (1 classes/image.bbclass): image.bbclass: Innocent typo fix. Feb 16 20:40:33 cannot that list be a variable, and so i can add my machine to it from my local.conf ? Feb 16 20:42:06 Magon: you need to change the lists. To progress. You can cleanup and make sure there is only one place where this list is, send a patch and then set this list in your local.conf Feb 16 20:42:36 where to send patch? Feb 16 20:42:52 the mailinglist, bugtracker :) Feb 16 20:42:56 i will probably do that patch...if there is chance it will be acceptet Feb 16 20:43:02 d Feb 16 20:43:20 Magon: sure, it is long overdue but OABI is legacy so the need to fix it was not too high Feb 16 20:44:01 i am not on mailing list..but i will try bugtracker Feb 16 20:44:13 Magon: no, that list cannot be set in local.conf. the distro controls it. Feb 16 20:44:29 Magon: don't forget to poke us on irc as well Feb 16 20:45:04 psokolovsky: why cannot i add items to that list form local.conf? Feb 16 20:46:03 Magon: because you should not add *anything* via local.conf at all (for angstrom) Feb 16 20:46:50 psokolovsky: ok..so i have to add it directly to angstorm-2008.1.conf Feb 16 20:47:09 yep Feb 16 20:47:14 ... Feb 16 21:17:04 * * OE Bug 3800 has been created by  Feb 16 21:17:06 * * uclibc-0.9.29-r9-do_install Feb 16 21:17:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3800 Feb 16 21:18:42 psokolovsky: Have you tried the alternative suggestion I mentioned for the initramfs? Feb 16 21:19:25 RP: Hi! nope, mostly waiting for proposed suggestions for packaging issue. Feb 16 21:19:59 psokolovsky: Which packaging issue? Feb 16 21:20:43 RP: see my response on ML. one of the reason kernel is built is to get kernel-image package with it ;-) Feb 16 21:22:18 hi all, I have a new simple processor architecture which I want to have a linux kernel for, someone advised me to use gnu toolchain to generate code for the target machine. does any one know how to tailor the toolchain for the new arch? or what else should I do to have linux on that new arch? Feb 16 21:22:44 psokolovsky: Sorry, I'd missed that mail. We don't and never will support mixing two different libcs like that. You'll have to run two different commands to generate a uclibc based kernel and then a glibc based rootfs Feb 16 21:22:58 psokolovsky: I really don't see that as a problem though Feb 16 21:23:14 RP: pity Feb 16 21:23:22 RP: what about packaging then? Feb 16 21:23:30 psokolovsky: What about it? Feb 16 21:23:45 RP: how to write package for kernel with initramfs? Feb 16 21:23:54 psokolovsky: With my proposal, it should "just work" Feb 16 21:24:00 mohamed: do you have compiler for that arch? Feb 16 21:24:14 psokolovsky: I just said that if it didn't, I would look into why. I think it will Feb 16 21:24:37 RP: how it can, if you add that task after do_install? or as separate task altogether? Feb 16 21:25:02 Magon: no I don't, but I think I could develop a compiler for that new arch. Feb 16 21:25:11 psokolovsky: Can you explain exactly what you think the problem is? Feb 16 21:25:28 psokolovsky: What is it you want to package? Feb 16 21:25:34 mohamed: so first you need to develop compiler...probably just tweak gcc Feb 16 21:25:57 RP: your (essentially my) task just deploys kernel, that's all. how it will go thru do_package_write and friends? Feb 16 21:26:14 Magon: have you got any idea about how to tweak gcc? Feb 16 21:26:27 RP: I want to package new zImage-* with linked in initramfs into kernel-image package. Feb 16 21:26:46 Magon: and what the copiler should generate what and use which prog. language? Feb 16 21:27:17 psokolovsky: Your solution does that? Feb 16 21:28:27 RP: my workaround recipes to overcome bitbake deficiencies do overcome that issue too. however, I personally much more dissatisfied with that than recursive invocations ;-) Feb 16 21:28:36 mohamed: ok..gcc can compile C/C++ and few more, it is written in C i suppose but i am not shure...you need to look at gcc sources and make it compile code do your architecture Feb 16 21:28:54 mohamed: but that is not exactly easy and i dont know how to do that Feb 16 21:29:22 OT: anyone knows how to create EFI partitions? Feb 16 21:29:31 psokolovsky: quit with the "deficiency" crap please. I'm trying to hold a sensible conversation and I can do without that. Feb 16 21:29:48 RP: what I do is to write package in separate recipe from deploy area, using that PKGPV magic. That's not that bad ;-), but caveat: that causes overwriting of the old kernel-image package with new content. that's what a bit ;-\ Feb 16 21:30:02 RP: ok Feb 16 21:30:37 RP: and I hope you have better solution ;-) Feb 16 21:30:48 Magon: well anyway I got the drift of what you said, just a little more explanation excuse me; I'll write code in c/c++ and the result will be assembly for the traget arch, isn't it? Feb 16 21:31:15 mohamed: more or less Feb 16 21:31:21 psokolovsky: The only way I can see this working is to package that file later in its own package task Feb 16 21:31:47 RP: how to make "own package task"? Feb 16 21:32:03 psokolovsky: If we need to change the package classes so we can rerun a second sweep of packaging so be it Feb 16 21:32:15 RP: ... and apparently, for the case initramfs is to be used, kernel-image should not be written initially too Feb 16 21:32:15 mohamed: i am not shure how this exactly work...but generaly compiler is something you put sources(ea C) and it output binary Feb 16 21:32:42 psokolovsky: The recipe shouldn't building that then ;-) Feb 16 21:32:59 Magon: I'll try to work in that way; thank you for your time :) Feb 16 21:33:36 RP: yep, all that talk about do_p_w_ipk being a child subtask of d_p_w was towards that. I failed to figure out how to write out a "new iteration" of packages after initial d_p_w was done. Feb 16 21:34:03 psokolovsky: The honest answer is I don't know how to do that Feb 16 21:34:16 RP: well, please have a look at kernel-initramfs.bb if possible, to see how I did, and see how bad it is. Feb 16 21:34:36 psokolovsky: I've looked and recursive bitbake is not an option Feb 16 21:34:56 RP: please look how package is written out. Feb 16 21:34:56 psokolovsky: You are not the one who will have to support the resulting mess... Feb 16 21:35:54 psokolovsky: You mean that you just limit PACKAGES and FILES? Feb 16 21:36:41 RP: not just. there's whole morbid machinery. want me to comment it line by line? Feb 16 21:37:29 psokolovsky: No, I understand what it does (graping the right PV/PR etc.) Feb 16 21:38:50 RP: first of all it builds initramfs, and ask *that* task, running in recursive bitbake, to delivery it to *our* location (it doesn't fish in deploy dir) Feb 16 21:39:04 psokolovsky: You could write a python function that changes PACKAGES and FILES in the data dictonary, then calls do_package, do_package_write* etc Feb 16 21:39:29 RP: then, it calls task to link in that initramfs into kernel, and again, asks the task to delivery resulting zImage to our location. Feb 16 21:39:30 psokolovsky: yep, I saw the DEPLOY_TO stuff Feb 16 21:39:57 RP: then it just indeeds greps for its PV/PR and just "installs" zImage for do_package Feb 16 21:40:30 RP: as you see, the very first problem is thar it's unclear what do_package_write* to call! Feb 16 21:40:40 psokolovsky: I admire the ingenuity in the approach, its great. Its not something we can maintain or allow though :( Feb 16 21:41:30 RP: yep, I exactly made an aim to do it no matter what ;-). well, any better ways to do that?.. Feb 16 21:41:45 psokolovsky: "no matter what" is not acceptable Feb 16 21:45:30 brb Feb 16 21:56:00 have anyone tryied to build arm oabi recently? it seems like my changes wasnt that fatal as i thougth..there is another problem probably Feb 16 21:58:12 hi everybody Feb 16 21:58:31 thesing: hi Feb 16 22:08:34 Nice Feb 16 22:40:24 howdy all Feb 16 22:40:35 hi soopaman Feb 16 22:41:04 i've finally managed to build a few OE images on my machine, but don't seem to be able to get them to boot in qemu Feb 16 22:41:37 nor am i sure what the "proper" way of burning it cf to test on the device Feb 16 22:41:57 any ideas? (i've google and keep getting stuck on an unmountable root device) Feb 16 23:11:24 Soopaman: what sort of images did you make? Feb 16 23:11:59 opie, x11 and console Feb 16 23:12:42 I mean, what filesystem type did you use? Feb 16 23:13:24 tar and jfffs2 Feb 16 23:16:23 okay - there's your problem - neither one of those would work for qemu or compact flashes Feb 16 23:16:55 you should build a ext2 image instead Feb 16 23:17:39 so replace "jffs2 tar.gz" with "ext2" in my local conf? Feb 16 23:18:23 yeah, I think that will do - you might need to specify some other config options - like how big to make the filesystem, but there should be examples in the existing confs Feb 16 23:18:36 k Feb 16 23:32:02 Someone just reported on #openmoko that the build for bb-native was calling perforce for some reason. It seems to be the --get option to patch; OE should probably set PATCH_GET=0 in the environment. Feb 16 23:36:34 cesarb: please stick that in a bug Feb 16 23:46:48 XorA: I read on your website that you are a fellow D&D player. :) Feb 16 23:48:22 thesing: yup Feb 16 23:49:14 XorA: Nice that there are other people around here sharing this great hobby. Feb 16 23:49:28 thesing: waiting for 4E to be released Feb 16 23:50:02 XorA: I'm afraid of the changes they do to my realms. Feb 16 23:50:40 thesing: well as a DM I skipped 3E only done that as a player Feb 16 23:51:52 XorA: I'm a DM since 2. and nearly never played myself. (Well I played other rpgs, but not D&D :( Feb 16 23:52:14 thesing: I normally run WFRP, thats my true passion :-) Feb 16 23:53:50 XorA: thats warhammer? I never played it. How magical is it? Feb 16 23:54:21 thesing: if you like the sort of RPG where the players are fighting to survive let alone be heroes then its great Feb 16 23:54:34 thesing: if you want golden hero boys, then its not the one Feb 16 23:55:17 XorA: so its more low power? No fireballs at 5th level? Feb 16 23:56:23 thesing: yes, lower power, less magic items, and the very act of using magic can cause insanity or mutation Feb 16 23:56:49 thesing: so at highest level mages can move mountains, but risk their souls doing it :-) Feb 16 23:56:50 XorA: the problem with creating a bug report for that is that it's hearsay (sorta)... I didn't see the problem, someone reported on IRC to me Feb 16 23:58:23 XorA: Are wizards still playable? If I ever happen to be in UK you have to show me WFRP ;) Feb 17 00:00:00 thesing: yes, they are playable, its only truly bad luck that gives mutations, 1 in 100 chances, and insanity chances go down the more power gained Feb 17 00:01:14 XorA: That reminds me of some d100 tables in AD&D. Feb 17 00:02:34 thesing: basically mage power is based on a number of d10 rolled, all 1s give a chance of insanity, doubles, triples, quadruples are chances of magic backfiring Feb 17 00:02:51 thesing: btw, did you try to build kexec against klibc? Feb 17 00:03:06 thesing: one of the possible results of a backfire is mutation, so in fact its even less than 1 in 100 Feb 17 00:03:29 psokolovsky: I did for x86. But I to copy some headers around. Feb 17 00:03:46 thesing: did you intend to add that to OE? Feb 17 00:04:07 XorA: So the main game mechanic is like shadowrun? Feb 17 00:04:14 thesing: but WFRP again lacks a publisher :-( Feb 17 00:04:27 thesing: no, its d100 based apart from magic rolls Feb 17 00:05:06 psokolovsky: Yes. But I finally got my mainboard and have to clean my pc after not using it for nearly six months. Feb 17 00:05:17 ok Feb 17 00:05:26 thesing: you have repaired kit, WOOWOO Feb 17 00:06:04 XorA: I thought I would never get a working mainboard. Feb 17 00:06:26 thesing: the work of Tzeentch :-) Feb 17 00:06:40 I had to download 1.5 G debian updates. Feb 17 00:11:37 cosmicpenguin, my build all of a sudden fails with ext2 as my IMAGE_FSTYPES, any clue why this would happen? Feb 17 00:13:57 gnite all Feb 17 00:15:09 Night. Feb 17 00:22:08 Soopaman: depends on where it fails, I guess Feb 17 00:22:37 NOTE: package x11-image-1.0-r0: task do_rootfs: failed Feb 17 00:22:37 ERROR: TaskFailed event exception, aborting Feb 17 00:22:37 NOTE: package x11-image-1.0: failed Feb 17 00:22:45 oops, only meant to paste 1 line Feb 17 00:40:04 night all Feb 17 01:05:31 03dyoung 07org.oe.dev * ra534fd15... 10/ (1 packages/gcc/gcc_4.2.2.bb): Add forgotten patch to re-enable big endian builds based on gcc 4.2.2 **** ENDING LOGGING AT Sun Feb 17 02:59:57 2008