**** BEGIN LOGGING AT Tue Aug 21 02:59:56 2007 Aug 21 06:55:07 can i have a look at .bb recipes online? Aug 21 07:09:59 readon: http://www.openembedded.org/filebrowser/org.openembedded.dev Aug 21 07:15:49 ] Aug 21 07:30:42 03pH5 07org.oe.dev * r4000d75b... 10/ (1 packages/gsm/libgsmd_svn.bb): libgsmd: remove band-aid, gsmd.h / linux_list.h issue fixed upstream Aug 21 08:00:52 03pH5 07org.oe.dev * re649ed66... 10/ (3 files in 2 dirs): matchbox-keyboard{,-inputmethod}: fix RPROVIDES Aug 21 08:08:03 morning all Aug 21 08:08:43 moin Aug 21 08:10:43 thanks Aug 21 08:27:55 Hi Aug 21 08:28:05 morning folks Aug 21 09:08:53 good morning all Aug 21 09:28:07 morning koen Aug 21 09:28:57 hey mickeyl Aug 21 09:38:09 hi koen, mickeyl Aug 21 09:40:07 hey RP Aug 21 09:45:15 good morning Aug 21 09:53:35 hey florian Aug 21 09:55:32 mickey|taxes: oh... enjoy ;) Aug 21 09:55:41 *sigh* Aug 21 09:55:45 bb later tonite Aug 21 09:55:47 (hopefully) Aug 21 09:55:59 03koen 07org.oe.dev * r0ff00843... 10/ (1 packages/glibc/glibc_2.6.bb): glibc-2.6: remove ppc-nofpu patched, as they seem to have been applied Aug 21 09:56:00 03koen 07org.oe.dev * r228a0376... 10/ (1 packages/openmoko2/libmokojournal2_svn.bb): libmokojounal2: inherit lib_package to fix packaging Aug 21 10:22:11 * RP has spotted a significant problem with the current location of pkgdata. It contains machine specific data yet its common accross all machines :/ Aug 21 10:22:55 right Aug 21 10:23:16 that's why I build a recipe for all machines whenever possible Aug 21 10:23:16 Does anyone not use multimachine anymore? Aug 21 10:23:50 * koen has no idea Aug 21 10:24:10 I wondering if anyone would object to merging it into the standard bitbake.conf? Aug 21 10:24:53 koen: If you build a kernel with some modules enabled, then try and repackage a kernel which doesn't have those modules it will crash Aug 21 10:42:41 hi Aug 21 10:43:07 which variable helds the download file of SRC_URI? Aug 21 10:44:38 downloaded? Aug 21 10:44:51 kxml2-2.3.0.jar Aug 21 10:44:53 bb.fetch.localpath returns the localpath Aug 21 10:44:54 as exmaple Aug 21 10:45:10 http://foo.de/kxml2-2.3.0.jar Aug 21 10:45:17 *sigh* Aug 21 10:45:35 I only want to know which variable helds kxml2-2.3.0.jar Aug 21 10:46:02 or is there none? Aug 21 10:46:03 haeh? Aug 21 10:46:17 brb Aug 21 11:02:57 * * OE Bug 2812 has been created by autobuild(AT)openembedded.org Aug 21 11:03:00 * * libmokogsmd2-0.1.0+svnr2703-r1-do_configure Aug 21 11:03:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2812 Aug 21 11:08:24 MACHINE_EXTRA_RECOMMENDS adds module to the image, correct? Aug 21 11:08:44 What's the line that makes them get installed at boot? Aug 21 11:09:00 it adds RRECOMMENDS to task-base Aug 21 11:09:09 so if your image doesn't use task-base... Aug 21 11:09:36 angstrom uses task-base Aug 21 11:10:08 somewhere I though I had a line that installed usb on OSK at boot Aug 21 11:10:10 I tihkn Aug 21 11:11:47 koen, Apparently some dude is arranging for me to get some gumstix Aug 21 11:13:25 the updated gumstix kernel should be using our generic kernel Aug 21 11:13:49 1 - we should build working stuff Aug 21 11:14:02 2 - As devs have time we should migrate Aug 21 11:14:02 everytime a linux- kernel gets added god kills 5 kittens Aug 21 11:14:10 I know Aug 21 11:14:46 It is important to have goals and work toward them, without driving unpaid devs insane Aug 21 11:15:00 * Crofton|home is in thesis finishing mode and nearing insanity Aug 21 11:15:39 I did fix uboot-tools to only make mkimage and actually work Aug 21 11:15:58 * * OE Bug 2812 has been RESOLVED (FIXED) by Aug 21 11:16:00 * * libmokogsmd2-0.1.0+svnr2703-r1-do_configure Aug 21 11:16:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2812 Aug 21 11:17:48 koen, which blackfin board do you have? Aug 21 11:18:58 * * OE Bug 2813 has been created by jbs(AT)bandos.homelinux.com Aug 21 11:19:00 * * libsdl-net fails sanity checks in configure Aug 21 11:19:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2813 Aug 21 11:22:02 Crofton|home: adzs-bf538f, but that doesn't seem work have a working kernel port (yet) Aug 21 11:22:24 538, we may have one I can get access to Aug 21 11:22:28 the bf537-stamp is the reference linux portboard Aug 21 11:22:42 but they professor that has it is annoying Aug 21 11:23:02 he'd probably want a 20 page report from me, just for looking at the board Aug 21 11:41:24 crofton hehe Aug 21 11:41:39 ship it to some one else and let write the rport Aug 21 11:41:43 Crofton|home: http://www.flickr.com/photos/koenkooi/1192511235/ Aug 21 11:42:27 heh Aug 21 11:42:45 koen for what are the golden connectors on the right side Aug 21 11:43:00 interesting, I think that is a different one Aug 21 11:43:11 woglinde: to but it in a rack iirc Aug 21 11:47:03 Crofton|home: the BOM nor schematics mention the fpga and second cpu... Aug 21 11:47:14 weird Aug 21 11:54:05 hm Aug 21 11:54:19 git and quilt installing some files at staging Aug 21 11:54:26 dbus-1 too Aug 21 12:00:29 Has anyone tried to build gcc-sdk lately ? does it build/work ? Aug 21 12:15:05 I'm trying to find out if the Freescale MITX ppc board is supported in OE. Is it? Aug 21 12:17:00 ~lart git perl install Aug 21 12:17:00 * ibot beats git perl install over the head with a microkernel Aug 21 12:23:04 micropal : as far as i know no Aug 21 12:26:46 steliosk: I'm doing a gcc 4.2.1/glibc 2.6 build for efika now Aug 21 12:27:44 koen : cool ! Aug 21 12:29:27 koen : angstrom-minimal-image for efika fails Aug 21 12:29:30 http://rafb.net/p/k7hCDS66.html Aug 21 12:30:06 steliosk: that looks like a problem with your filesystem Aug 21 12:30:19 | makedevs: /storage/angstrom/tmp/libc/rootfs/bin/tinylogin: file can not be created! Aug 21 12:32:36 there is no rootfs/bin dir Aug 21 12:32:43 that's really weird Aug 21 12:51:57 * * OE Bug 2800 has been RESOLVED (FIXED) by Aug 21 12:51:59 * * angstrom-x11-image-1.0-r11-do_rootfs Aug 21 12:52:01 * * http://bugs.openembedded.org/show_bug.cgi?id=2800 Aug 21 13:33:59 <__law__> hi any ideas why bitbake doesnt fetch from this git repo? http://vivien.chappelier.free.fr/typhoon/download.html Aug 21 13:34:15 <__law__> SRC_URI = "git://manwe.homelinux.org/~glaurung/linux-typhoon.git;protocol=git Aug 21 13:34:27 <__law__> is my SRC_URI Aug 21 13:35:46 <__law__> git-clone http://manwe.homelinux.org/~glaurung/linux-typhoon.git/ test works fine.. on my host Aug 21 13:36:12 __law__ what's the error you get ? Aug 21 13:44:55 Crofton : Just pushed the xilinx patches (kernel, bbclass etc) for the ml403. It should be pretty similar for your board Aug 21 13:45:05 <__law__> steliosk, http://pastebin.ca/665619 it looks like there is a connection timeout... Aug 21 13:47:08 cool Aug 21 13:47:24 must finish thesising b4 I get distracted by all the other cool stuff Aug 21 13:48:45 __law__ are you sure the three has git support > maybe it has only http. change protocol=http and try again Aug 21 13:49:34 koen, be nice to the steve guy, he is arranging for our gumstix Aug 21 13:49:49 so we can actually test our meta-data on hw Aug 21 13:50:02 yes, *our* metadata Aug 21 13:50:09 hehe Aug 21 13:50:30 which likely has bitrot because no one tests on gumstix ..... Aug 21 13:50:33 him adding random crap does not make debuggin easier for us Aug 21 13:50:51 him getting us boards does Aug 21 13:51:08 * Crofton|home crosses fingers and hopes they really appear, not like TI slackers Aug 21 13:51:22 <__law__> steliosk, http://pastebin.ca/665628 here is my new error Aug 21 13:51:32 He is trying a lot on his own Aug 21 13:53:14 Crofotn :reminds me of a sign i had seen sometime ago . "Don't mind of the dog beware of the owner" hehe Aug 21 13:54:14 do the kernel module ipks end up in the machine directory in ipk's? Aug 21 13:54:24 03ifaistos 07org.oe.dev * r0dea00d5... 10/ (3 files in 2 dirs): Aug 21 13:54:24 conf/machine/include/tune-ppc440e.conf and tune-ppc603e : Add -mhard-float gcc switch as Aug 21 13:54:24 these cores have an fpu Aug 21 13:54:29 03ifaistos 07org.oe.dev * r3758bac2... 10/ (1 classes/xilinx-bsp.bbclass): classes/xilinx-bsp.bbclass : Add a class to handle xilinx kernel requirments Aug 21 13:54:34 03ifaistos 07org.oe.dev * r89033af2... 10/ (4 files in 3 dirs): Aug 21 13:54:34 packages/linux/linux-ml403-mvista-2.6.x_git.bb : Add kernel for the xilinx ml403 board based Aug 21 13:54:34 on mvista git tree Aug 21 13:54:40 03ifaistos 07org.oe.dev * reabb862a... 10/ (4 files in 3 dirs): Aug 21 13:54:40 packages/linux/linux-ml403-slab-2.6.x_git.bb : Add kernel recipe for xilinx ml403 board Aug 21 13:54:40 using Secretlab's git tree Aug 21 13:54:45 03ifaistos 07org.oe.dev * r5ee4afc8... 10/ (1 conf/machine/xilinx-ml403.conf): Aug 21 13:54:45 conf/machine/xilinx-ml403.conf : Set default kernel for the xilinx ml403 the one build Aug 21 13:54:45 from the slab fit tree Aug 21 13:55:09 __law__ did you clean the package build before trying again ? Aug 21 13:57:06 koen : Did gcc 2.6 build ok ? Aug 21 14:02:33 steliosk: it's building X at the moment Aug 21 14:04:55 <__law__> steliosk, yes Aug 21 14:08:08 koen, do you know what Steve's gumstix problem is? Aug 21 14:08:26 I started helping him over the weekend, when it would not build images Aug 21 14:10:34 his kernel recipes sucks Aug 21 14:10:58 due to messing with PV it won't do dependencie resolving Aug 21 14:11:02 ok Aug 21 14:11:14 (among other problems) Aug 21 14:11:19 * koen mentions linux.inc Aug 21 14:11:26 hopefully, we really do get some boards out of this and can do some testing Aug 21 14:11:51 my problem is that I do not have time to go through linux.inc atm Aug 21 14:12:04 koen: thanks for fixing libmokojournal2 earlier today, this and the matchbox RPROVIDES drove me crazy yesterday evening Aug 21 14:12:20 and it sound like this guy as actually promoting OE as a solution to gumstix Aug 21 14:12:49 __law__ It looks like you have CVS_TARBALL_STASH defined and bitbake looks at that and fails as it can't find the file Aug 21 14:13:53 <__law__> steliosk, ok i´ll take a look, but now i have to leave bbl Aug 21 14:14:02 Crofton|home : Does the gumstix builds ? haven't try to build it for sometime now Aug 21 14:14:26 apaprently we spit out images now, but they have issues Aug 21 14:14:34 Crofton|home: the only big problem I can thnk of is 2.6.15 not supporting EABI] Aug 21 14:14:53 he is trying to update the recipe to 22 Aug 21 14:15:07 is someone aware of a bug building GTK ? (package gtk+-2.10.14-r5: task do_compile: failed) compiling gtkcombobox.c... Aug 21 14:15:29 anyway, my point here, is we have a chance to work though the gumstix issues Aug 21 14:15:34 Crofton|home: make him post his recipe and defconfig and I'll add it to linux.inc Aug 21 14:15:41 as we have an interested party testing :) Aug 21 14:15:45 ok Aug 21 14:16:03 we'll need some help liekly Aug 21 14:16:23 and when I get past thesis edits, I'll look at moving linux-omap.inc that way Aug 21 14:16:42 arm machines only need a kernel and the right jffs2 params to work Aug 21 14:16:45 I am still annoyed at myself for getting sucked into the gumstix vortex Aug 21 14:16:50 angstrom takes care of all other details Aug 21 14:17:21 well, the gumstix guys are aware of OE, and there is interest moving from buildroot to OE Aug 21 14:17:29 I would like to see that happen. Aug 21 14:17:49 Every idiot in my particular SDR field likes to demo stuff on gumstix Aug 21 14:18:13 some guys at a workshop put one on a table, with no cables, and said our stuff runs on that Aug 21 14:18:33 they were idiots, but people where impressed by a plastic box on a table :) Aug 21 14:19:18 Crofton|home : "wow" factor :) Aug 21 14:19:29 steliosk, exactly Aug 21 14:20:08 so basically, I am getting sucked into the gumstix universe from two directions Aug 21 14:20:36 which creates more OE "wins" Aug 21 14:21:24 Crofton|home : Does that mean you will not be using your fpga board soon ? wink wink Aug 21 14:21:59 Crofton|home: demo your stuff on a neo :) Aug 21 14:22:41 03koen 07org.oe.dev * re5d7ecec... 10/ (1 packages/geoclue/geoclue_git.bb): geoclue: enable system bus and gpsd Aug 21 14:22:47 steliosk: how powerfull is the fpga on this one http://www.flickr.com/photo_zoom.gne?id=1192511235&size=o ? Aug 21 14:24:45 steliosk, after thesis :) Aug 21 14:25:08 koen : hehe, its small and has a nice screen also ;) plus can you place a call with a gumstix ? :) Aug 21 14:25:30 koen, neo is also post thesis Aug 21 14:25:47 * steliosk looks at Crofton's todo list Aug 21 14:25:48 * Crofton|home must focus Aug 21 14:26:14 I'll ask Steve to pastebin the kernel file and the defconfig Aug 21 14:28:15 morning Aug 21 14:29:03 koen : The Spartan 3 XC3S250 is on the low end of a rather low cost series Aug 21 14:36:42 03mickeyl 07org.oe.dev * rd8e1a34f... 10/ (1 classes/distutils.bbclass): Aug 21 14:36:42 classes/distutils.bbclass: remove easy-install.pth, since this ends up in every package and Python 2.5 seems to be able to import .egg packages even Aug 21 14:36:42 without it Aug 21 14:36:53 03mickeyl 07org.oe.dev * r3289b5a8... 10/ (5 files in 2 dirs): python-[ecore|edje|emotion|evas]: bump PR as per recent distutils.bbclass change Aug 21 14:49:58 * * OE Bug 2814 has been created by autobuild(AT)openembedded.org Aug 21 14:50:00 * * webkit-gtk-0.0+svnr25169-r25169-r1-do_install Aug 21 14:50:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2814 Aug 21 14:56:58 * * OE Bug 2815 has been created by jorik(AT)kippendief.biz Aug 21 14:57:00 * * net-snmp fails to compile on slugos/arm Aug 21 14:57:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2815 Aug 21 15:00:13 * oxo sometimes wonders if there is any point in submitting bugs to OE Aug 21 15:02:15 <_law_> bb Aug 21 15:03:52 I mean, of the bugs that I have submitted, it does not even look like someone actually does something with it.. Aug 21 15:04:43 what machine? Aug 21 15:05:03 what do you mean? Aug 21 15:05:43 Jin^eLD: Poky has a fix for the apt-native issue you mentioned (not building apt-ftparchive) Aug 21 15:07:13 hrw|gone, ping Aug 21 15:07:32 RP: ah, cool Aug 21 15:08:28 RP: I will get back from holidays on the 25., then I will try it again Aug 21 15:23:55 koen, the gumstix kernel is at http://www.sakoman.net/oe/ :) Aug 21 16:13:24 koen, ping Aug 21 16:15:24 who are the linux.inc gurus? Aug 21 16:37:33 Crofton: don't forget to tell steve to use DISTRO=angstrom-2007.1 Aug 21 16:38:59 I think I already did :) Aug 21 16:39:29 in linux.inc, it is using the "manual" method of uImage production Aug 21 16:39:40 03koen 07org.oe.dev * r46c332d7... 10/ (47 files in 4 dirs): linux 2.6.21: add gumstix support Aug 21 16:39:44 03koen 07org.oe.dev * rf80dfa84... 10/ (1 conf/machine/gumstix.conf): gumstix: update machine file for new kernel Aug 21 16:39:48 and I think it hard coes the entry points Aug 21 16:39:57 which is bad Aug 21 16:40:12 Crofton: it hardcodes the entry point that are in linux/arch/arm/Makefile Aug 21 16:40:33 Crofton: what do you think about UBOOT_ENTRYPOINT = 0xfoo in machine.confs? Aug 21 16:40:40 hmm Aug 21 16:40:43 not sure Aug 21 16:40:52 even what that is for :) Aug 21 16:42:03 his latest email from the gumstix guy said "probably send" Aug 21 16:42:10 he is going to stay after them Aug 21 16:42:12 sent? Aug 21 16:42:21 also, our TI intern returned Aug 21 16:42:36 as is the "will probably send us some" Aug 21 16:42:44 still promising Aug 21 16:43:14 The TI intern suggests our group at VT apply under some kind of donation program for the HW Aug 21 16:43:21 not quite as good as it going to OE Aug 21 16:43:51 but we should be able to justify the need, and we have some relation with TI we can use Aug 21 16:49:45 koen, I dropped him a note asking him to test. I am off to the coffee shop with chapter 3, and no computer! Aug 21 16:50:06 bye crofton Aug 21 16:50:12 (not an Amsterdam coffee shop ) Aug 21 16:50:16 bye woglinde Aug 21 16:50:17 hehe Aug 21 16:55:23 hm Aug 21 16:55:44 I puzzeld why at the native recipes --prefix=/devel/arm/oetmp2/staging is set Aug 21 17:16:21 * koen spots a few missing ABIs in http://wiki.openmoko.org/wiki/Binary_compatibility Aug 21 17:36:12 hi Aug 21 17:53:43 koen, ping Aug 21 17:58:53 hi khem Aug 21 18:00:00 koen, gumstix failed on missing uboot-mkimage Aug 21 18:00:18 he woglinde Aug 21 18:02:35 crofton, what distro are you building for when you are getting the missing uboot-mkimage? Aug 21 18:02:45 angstrom Aug 21 18:02:49 working on fix now Aug 21 18:02:52 interm Aug 21 18:02:58 ok, i just wanted to confirm Aug 21 18:03:14 just bitbake u-boot-mkimage-openmoko-native Aug 21 18:03:27 that will build uboot-mkimage Aug 21 18:03:38 I have some complaints with linux.inc :) Aug 21 18:27:53 03koen 07org.oe.dev * rd57c8ad6... 10/ (1 packages/linux/linux.inc): Aug 21 18:27:53 linux.inc : Need uboot-mkimage to build uImage for gumstix kernel. Aug 21 18:27:53 linux.inc: require mkimage for gumstix Aug 21 18:28:16 heh Aug 21 18:28:30 Crofton: it looks like we both made the exact same change :) Aug 21 18:28:48 rofl Aug 21 18:28:57 jesus Aug 21 18:29:08 the git Makefile is nasty Aug 21 18:29:14 but anyway now it works Aug 21 18:29:25 woglinde: it's written by kernel people Aug 21 18:29:49 koen, we should switch back to ubbot-tools for mkimage Aug 21 18:31:09 I hate the dependency on a bb file with openmoko in the name for general builds Aug 21 18:31:47 hm is there a Makefile sanity checker/analyzer? Aug 21 18:32:50 I'm building something and I need to dump something into the udev rules directory, is there a variable for that? Aug 21 18:33:14 * oxo waves somewhat at koen Aug 21 18:33:23 install -d ${D}${sysconfdir}/udev/rules.d/ Aug 21 18:33:24 install -m 0644 ${WORKDIR}/50-hostap_cs.rules ${D}${sysconfdir}/udev/rules.d/50-hostap_cs.rules Aug 21 18:33:27 hey oxo Aug 21 18:33:31 koen: thanks Aug 21 18:33:53 voor jou, dat zou beter zijn Aug 21 18:33:54 time to watch dr house Aug 21 18:44:15 ~lart dbus-native Maintainer Aug 21 18:44:15 * ibot holds dbus-native Maintainer to the floor and spanks him with a cat-o-nine-tails Aug 21 18:44:29 koen, whats the tinderbox url? Aug 21 18:45:42 Crofton: openembedded.org, the 'Tinderbox' link Aug 21 18:46:00 dah Aug 21 18:46:03 just found it Aug 21 18:46:52 there does not seem to be much content? Aug 21 18:47:31 hmm tree =angstrom? Aug 21 18:48:59 yes Aug 21 18:49:21 koen could you please look at packages/dbus/dbus-native_1.0.1.bb Aug 21 18:49:22 we need a script we can run in cron to builds at night Aug 21 18:49:40 I dont get why you call oe_runmake install and autotools_stage_all Aug 21 18:51:38 woglinde: I didn't write the bb, just added the new version Aug 21 18:51:52 koen oh okay Aug 21 18:51:58 but it made no sense Aug 21 18:52:19 because autotools_stage_all will invoke oe_runmake anyway Aug 21 18:53:27 any found an error in classes/native sysconfdir and sharedstatedir isnt override in Aug 21 19:00:04 ouch that are many patches to gumstis Aug 21 19:00:06 ouch that are many patches to gumstix Aug 21 19:00:12 yes Aug 21 19:00:27 I think the gumstix dev is over worked and not getting stuff upstream .... Aug 21 19:00:38 they shpuld start tickling rmk Aug 21 19:00:49 *sigh* Aug 21 19:01:11 OE could use a kernel guy just to collect patches and help them flow upstream Aug 21 19:02:02 * woglinde wonders why gadet_patch isnt upstream Aug 21 19:03:47 args Aug 21 19:03:50 I hate this Aug 21 19:04:03 pxa2xx_udc.c and saying Intel PXA25x Aug 21 19:07:10 these are also against .21, not .22 Aug 21 19:07:26 so maybe they are going upstream :) Aug 21 19:08:05 crofton hm??? Aug 21 19:08:20 if no one sending them to rmk Aug 21 19:08:23 hello Aug 21 19:08:28 nothing is going upstream Aug 21 19:09:31 When I have some time, I should look at the gumstix list and see what they are thinking Aug 21 19:09:33 is it worth to install Linux distro on MacBook in order to work on OE? Aug 21 19:10:02 the ucb1400-sound fix seems nice Aug 21 19:10:14 I've read BuildOnMac page, and it seems like stupid idea to boot Linux on MacBook because everything is in place for Mac OSX also.. Aug 21 19:12:32 yakov: you could just try I guess :) if you can get all dependencies/needed packages then cahnces are probably good Aug 21 19:13:08 but its just a guess, I'm not rally an OE guru to give a qualified answer :P Aug 21 19:19:11 :) let's try .. Aug 21 19:23:58 * * OE Bug 2816 has been created by jonas.bonn(AT)gmail.com Aug 21 19:24:00 * * qt4-x11-free Aug 21 19:24:02 * * http://bugs.openembedded.org/show_bug.cgi?id=2816 Aug 21 19:28:34 03woglinde2 07org.oe.dev * r6243293d... 10/ (1 packages/kxml2 packages/kxml2/kxml2_2.3.0.bb): kxml2: adding recipe for kxml2 Aug 21 19:36:32 what was the problem with do_rootfs failing trying to create tinylogin? Aug 21 19:36:45 does something need cleaning after pulling? Aug 21 19:44:01 Crofton: what other cool sdr stuff can we include in an sdr image besides gnuradio? Aug 21 19:44:14 I was thinking about gnuplot and stuff like that Aug 21 19:47:54 none of the machines I use have displays :) Aug 21 19:48:25 when we have some OSSIE tarballs, we can add those Aug 21 19:50:32 One topic for OEDEM shoudl be how can we have a "stable" OE for people Aug 21 19:50:50 but that takes the fun out of it. :) Aug 21 19:50:58 hm I think it was last oedem topic too Aug 21 19:51:03 and the pain of answering questions Aug 21 19:51:16 from people who just need something that works Aug 21 19:51:30 oe works Aug 21 19:51:34 if didnt Aug 21 19:51:36 fix it ;) Aug 21 19:51:39 but things break Aug 21 19:51:41 *duck and hide* Aug 21 19:51:53 personally, I am with woglinde :) Aug 21 19:52:03 hasnt the stability always just been a matter of manpower? do we have enough people to maintain a stable branch? i doubt it, half the bb files in oe today arent maintained as well as they should be as is Aug 21 19:52:10 heh Aug 21 19:52:17 right Aug 21 19:52:40 this conversation exactly mirrors a situation we have at the uni also Aug 21 19:53:05 I updated my org.embedded.dev and rebuilt for the gumstix and then loaded the jffs2 anstrom-minimalist-image and 2.6.21 kernel, it uncompresses and then says Starting kernel ... and then nothing Aug 21 19:53:13 but it seems like there is a subset of bb files and machines that work really well Aug 21 19:53:20 heh Aug 21 19:53:36 I am betting the linux.inc file is making brken gumstix kernels Aug 21 19:53:38 :) Aug 21 19:53:50 I think the entry points are wrong Aug 21 19:54:02 asmola, how good are you at kernel stuff? Aug 21 19:54:53 I know nothing :( Aug 21 19:54:58 koen, linux.inc hard codes the -a and -e parameters to mkiamge Aug 21 19:56:50 it appears to have produced two kernel images, one I am guessing include uboot?? Aug 21 19:57:05 the may be soft linked Aug 21 19:57:14 I used the one without uboot. I fear that if I try the other one and try to write a new uboot and something doesn't go write that I'll be SOL ( I don't have JTAG) Aug 21 19:57:37 heh Aug 21 19:58:15 I am looking for the u-boot docs Aug 21 19:58:41 I am pretty sure linux.inc is not doing the kernel entry points correctly Aug 21 19:59:47 when it boots it is printing out an entry point of 20008000 Aug 21 20:00:04 ok Aug 21 20:00:11 with the same load address Aug 21 20:00:24 -a is load address and -e is entry point Aug 21 20:01:15 * Crofton reads linux.inc Aug 21 20:01:49 03woglinde2 07org.oe.dev * r343be325... 10/ (4 files in 2 dirs): Aug 21 20:01:49 classpath-minimal: rework recpies and using share/java Aug 21 20:01:49 * remove classpath-minimal.inc, 2 files are enough Aug 21 20:01:49 * use ${DATADIR}/java/classpath-minimal for installing glibj.zip Aug 21 20:02:38 hmm, need to read kernel bbclass Aug 21 20:03:20 bother, I do not have gumstick built anywhere to look Aug 21 20:04:03 asmola, I need to see what comes from linux.inc Aug 21 20:04:17 what would you like me to send? Aug 21 20:04:22 you should not need new u-boot Aug 21 20:04:38 give me some time, I will try a build at home Aug 21 20:04:47 work machine is building from scartch Aug 21 20:04:56 i didn't think that i would, i figured that i'd be able to load the kernel and rootfs and be good, i am using uboot 1.2 Aug 21 20:05:29 the entry point stuff can be tricky Aug 21 20:05:40 I had some issues when I started with the OSK Aug 21 20:07:22 koen, ping Aug 21 20:07:47 Do you want me to send over my linux.inc? Aug 21 20:08:10 linux.inc does not set KERNEL_IMAGE_TYPE for gumstix Aug 21 20:08:29 I have the same one, as long as you have pulled recently nd not edited :) Aug 21 20:09:02 When Uboot tries to boot it is showing an Image Type: ARM Linux Kernel Image (gzip compressed) Aug 21 20:09:23 yeah I just pulled it a few hours ago Aug 21 20:09:47 what entry point does it give? Aug 21 20:09:49 and then it appears to successfully uncompress the kernel image and then it goes kaput Aug 21 20:09:55 Entry Point: 20008000 Aug 21 20:10:03 Load Address: 2000800 Aug 21 20:10:12 I think they should be the same Aug 21 20:10:48 Crofton: gumstix.conf sets the imagetype Aug 21 20:11:02 (see marcins rfc on the ml) Aug 21 20:11:34 h Aug 21 20:11:35 ah Aug 21 20:12:03 why does linux.inc do the mkiamge stuff in the do_deploy method? Aug 21 20:13:01 it seems like it will install the kernel from the build dir, then create another uImage and .... Aug 21 20:13:28 and overwrite the one from the kernel build tree Aug 21 20:13:46 asmola, has the entry points set by linux.inc do_deploy Aug 21 20:14:04 when the gumstix source gets built, I will check what is in the source tree Aug 21 20:14:26 Crofton: http://rafb.net/p/nZcaW070.html ? Aug 21 20:14:51 UBOOT_ENTRYPOINT = "a0008000", that is Aug 21 20:15:14 I prefer we just use the uImage from the kernel build whenever possible Aug 21 20:15:39 I bet deleteing the if test .... run mkimage .. fi fixes the problem Aug 21 20:15:46 at least for OSK and gumstix Aug 21 20:16:03 the entry point wil very from machine to machine Aug 21 20:16:18 make uImage for the kernel does the right thing Aug 21 20:17:35 also, I would guess for some freak cases, load address != entry point Aug 21 20:19:55 http://rafb.net/p/lWxdEW96.html Aug 21 20:19:57 forgive my ignorance here but if that file is setting it to a0008000 why is mine 20008000 or is that the problem? Aug 21 20:21:31 crap Aug 21 20:21:39 I think I crashed our svn repo Aug 21 20:21:48 uh oh Aug 21 20:22:35 I need to stop doing oe builds on the repositoy/webserver machine .... Aug 21 20:23:04 alright kernel is arrving at hoem Aug 21 20:28:05 stupid router Aug 21 20:29:01 koen, ping Aug 21 20:29:51 given that it looks like the load/entry are supposed to be a* vs 2* i just hexedited the kernel and i am pushing it back down, we'll see what happens Aug 21 20:30:09 you are brave :) Aug 21 20:30:27 where did you find the entry points? Aug 21 20:30:50 it looks like they are right at the beginning Aug 21 20:31:12 you are comparing a bootable image Aug 21 20:31:43 I have a strong urge to go break koen's TV :) Aug 21 20:31:48 03koen 07org.oe.dev * r486b9834... 10/ (1 conf/machine/gumstix.conf packages/linux/linux.inc): gumstix: make uboot entry point overridable per machine Aug 21 20:33:44 i just did a quick search on gumstix user request and it looks like their kernel entry point is a* Aug 21 20:34:03 i'll know if this was a good move or if i am an idiot in a second Aug 21 20:34:15 most likely the latter Aug 21 20:34:29