**** BEGIN LOGGING AT Mon Jun 14 02:59:57 2010 Jun 14 05:22:48 gm Jun 14 05:25:14 hi likewise, all Jun 14 05:30:08 likewise: do you have hardware to test nios2+mmu? e.g. a NEEK kit Jun 14 05:56:57 greetings Jun 14 05:57:10 Im having trouble understanding the OE process Jun 14 05:57:33 Ive got a working OE now I think, I just run a bitbake virtual/kernel Jun 14 05:57:57 my goal is it to have a module compiled for my target board, so I can insmod it Jun 14 05:58:30 any pointers on how to proceed? Jun 14 05:58:56 pretty please? Jun 14 06:01:41 slackytude|evil: bitbake kernel-module-mystuff Jun 14 06:01:57 slackytude|evil: ah, you already built a full kernel, never mind Jun 14 06:02:01 and mystuff needs to be a metadata file? Jun 14 06:03:06 slackytude|evil: you ran virtual/kernel already, did the module get built? Jun 14 06:05:29 likewise, no, I was under the impression that running virtual/kernel was a requirment. It couldnt possibly build my module, as the source is nowhere near the OE environment Jun 14 06:06:25 its not a module in the linux kernel. its an external module I wrote myself, based on a skel file in the linux drivers dir Jun 14 06:06:55 I can compile it for intel's quite easily, but Im new to crosscompiling Jun 14 06:08:18 It looks like I need to generate a .bb file Jun 14 06:10:03 eFfeM_work: yes, NEEK without display Jun 14 06:10:14 eFfeM_work: I have the Cyclone III starter board Jun 14 06:10:36 eFfeM_work: http://www.altera.com/products/devkits/altera/kit-cyc3-starter.html Jun 14 06:11:31 slackytude|evil: I can show you one Jun 14 06:12:13 likewise, that would be nice, I seem to be lost Jun 14 06:14:05 slackytude|evil: http://pastebin.ca/1882692 Jun 14 06:14:24 slackytude|evil: I am assuming you already have a correct Makefile, or should I sent you one as well? Jun 14 06:15:20 I have a MakeFile for Intel Jun 14 06:15:25 It wouldnt hurt Jun 14 06:15:29 so, yes, please Jun 14 06:15:57 slackytude|evil: http://pastebin.ca/1882695 Jun 14 06:16:30 likewise: do you want to help getting Nios2+mmu into oe? I don't think we have that board here Jun 14 06:17:05 I have the toolchain for nios2+mmu building under oe, still need to iron out a few issues wrt uclibc. Jun 14 06:17:10 but this is all on our own hw Jun 14 06:17:12 likewise, thanks a lot, Ill to get on with it now Jun 14 06:18:02 likewise, the makefile uses uname -r. how does that work? wont that just use my host machine kernel? Jun 14 06:19:59 slackytude|evil: The "?=" is a conditional assignment (KERNELDIR will be already correctly set by OE for your target) Jun 14 06:20:20 slackytude|evil: so that line doesn't override the existing assignment of KERNELDIR Jun 14 06:20:25 I see Jun 14 06:20:37 thanks again Jun 14 06:21:08 eFfeM_work: sure I can spend time on that Jun 14 06:21:13 meh Jun 14 06:21:23 slackytude|evil: you're welcome Jun 14 06:22:03 likewise: i'm hoping to finish the uclibc issues later today or tomorrow, will push them to a branch and let you know Jun 14 06:22:10 thanks alot Jun 14 06:22:15 I wanted to show my development to some professor at my uni. I spent the weekend preparing a demo. I wanted to show it running on embedded systems as well. Jun 14 06:22:20 The date was tomorrow Jun 14 06:22:31 and now I get an email that it got moved to next monday Jun 14 06:22:45 slackytude|evil: more relaxed timing then Jun 14 06:22:52 indeed Jun 14 06:22:58 maybe I should sleep a bit Jun 14 06:43:32 morning Jun 14 06:50:12 03Martin Jansa  07org.openembedded.dev * rbe2bd3dbae 10openembedded.git/recipes/ (20 files in 10 dirs): Jun 14 06:50:13 recipes: PR bump after EFL SRCREV change resulting to different .so names. Jun 14 06:50:13 * mostly libelementary-ver-svn-06 -> libelementary-ver-pre-svn-06 Jun 14 06:50:13 Signed-off-by: Martin Jansa Jun 14 07:11:33 03Koen Kooi  07org.openembedded.dev * re393c44c80 10openembedded.git/recipes/mythtv/ (3 files): Jun 14 07:11:33 mythtv: add back 0.22+fixes which is still in active use Jun 14 07:11:33 * 0.23 is incompatible with 0.22 (as usual with mythtv) Jun 14 07:23:13 03Koen Kooi  07org.openembedded.dev * r38ca5cdddc 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom: pin mythtv to 0.22 Jun 14 07:23:17 03Koen Kooi  07org.openembedded.dev * r809fb1a21a 10openembedded.git/recipes/mythtv/ (3 files): mythtv 0.23: make it atually fetch for the 0.23 branch instead of the 0.22 branch Jun 14 07:24:58 03Koen Kooi  07org.openembedded.dev * r5974c72fcf 10openembedded.git/recipes/opencv/ (opencv-samples_svn.bb opencv_svn.bb): opencv, -samples: bump SRCREV Jun 14 07:26:21 morning Jun 14 07:27:08 hi hrw Jun 14 07:57:25 gm hrw, woglinde Jun 14 07:57:33 hi likewise Jun 14 07:57:34 gm all Jun 14 09:12:50 03Koen Kooi  07org.openembedded.dev * r98a2f13212 10openembedded.git/recipes/ffmpeg/ (3 files): omapfbplay: bump PR Jun 14 09:15:20 03Martin Jansa  07org.openembedded.dev * r60b6d7c45e 10openembedded.git/recipes/xorg-app/ (3 files): Jun 14 09:15:21 xkbcomp-native: use xorg-app-common.inc Jun 14 09:15:21 Signed-off-by: Martin Jansa Jun 14 09:15:22 03Martin Jansa  07org.openembedded.dev * r9a4db5e566 10openembedded.git/recipes/xorg-app/mkbold-mkitalic-native_0.11.bb: Jun 14 09:15:22 mkbold-mkitalic-native: use xorg-app-common.inc Jun 14 09:15:22 Signed-off-by: Martin Jansa Jun 14 09:15:23 03Martin Jansa  07org.openembedded.dev * rc9d795c245 10openembedded.git/recipes/xorg-app/ (xterm_207.bb xterm_251.bb): Jun 14 09:15:23 xterm: use xorg-app-common.inc Jun 14 09:15:23 Signed-off-by: Martin Jansa Jun 14 09:15:24 03Martin Jansa  07org.openembedded.dev * raef543fe45 10openembedded.git/recipes/xorg-app/ (4 files): Jun 14 09:15:24 mkfontscale: always include non-native recipe and use xorg-app-common.inc Jun 14 09:15:25 Signed-off-by: Martin Jansa Jun 14 09:15:25 03Martin Jansa  07org.openembedded.dev * r8da289827b 10openembedded.git/recipes/xorg-app/ (3 files): Jun 14 09:15:26 mkfontdir: always include non-native recipe and use xorg-app-common.inc Jun 14 09:15:26 Signed-off-by: Martin Jansa Jun 14 09:15:28 03Martin Jansa  07org.openembedded.dev * r1624d1b331 10openembedded.git/recipes/xorg-driver/ (274 files): Jun 14 09:15:28 xorg-driver: use INC_PR Jun 14 09:15:28 Signed-off-by: Martin Jansa Jun 14 09:15:30 03Martin Jansa  07org.openembedded.dev * r46291053cc 10openembedded.git/recipes/ (174 files in 105 dirs): Jun 14 09:15:30 recipes: bump PR/INC_PR for packages changed in RDEPENDS/RRECOMMENDS/RSUGGESTS/RPROVIDES/RCONFLICTS/RREPLACES cleanup Jun 14 09:18:22 ping jama Jun 14 09:18:53 pong Jun 14 09:19:38 jama could you please try latest navit-svn with qt_qpainter and svg's only Jun 14 09:19:44 on freerunner Jun 14 09:20:31 woglinde: what do I need to enable qt_qpainter? Jun 14 09:20:42 in navit.xml Jun 14 09:20:58 change gtk-drawing-area to qt_qpainter Jun 14 09:21:23 ah qt_qpainter: no (configure parameter) nevermind... Jun 14 09:21:30 remove all png's and xpm's from share/navit/xpm Jun 14 09:21:35 you have to build it Jun 14 09:21:55 I meant latest svn from navit Jun 14 09:21:58 not oe Jun 14 09:22:00 *g* Jun 14 09:22:43 why not oe? Jun 14 09:22:51 sigh Jun 14 09:22:56 I'm building it with AUTOREV here.. Jun 14 09:23:04 yeah thats what I meant Jun 14 09:23:06 sorry Jun 14 09:23:10 bump the svnrev Jun 14 09:23:17 and qt-painter was disabled in navit.inc Jun 14 09:23:31 that's why it's not packaged in separate package yet Jun 14 09:23:42 yeah Jun 14 09:23:47 its only test Jun 14 09:24:09 cp15 told gtk svg support via librsvg is hell slow on free runner Jun 14 09:24:21 now I like to know if it is the same with qt Jun 14 09:27:36 jo stefan Jun 14 09:27:44 hi woglinde Jun 14 09:35:29 hi cbrake Jun 14 09:43:28 morning Jun 14 09:43:36 hi jay7 Jun 14 09:54:47 03Koen Kooi  07org.openembedded.dev * r275524cb94 10openembedded.git/recipes/ffmpeg/omapfbplay_git.bb: omapfbplay: update to latest git Jun 14 10:57:06 03Martin Jansa  07org.openembedded.dev * rc193935bb0 10openembedded.git/recipes/navit/navit-icons_svn.bb: Jun 14 10:57:06 navit-icons: bump SRCREV Jun 14 10:57:06 Signed-off-by: Martin Jansa Jun 14 10:57:08 03Martin Jansa  07org.openembedded.dev * r6339640198 10openembedded.git/recipes/efl1/elementary_svn.bb: Jun 14 10:57:09 elementary: try to fix upgrade path from older -themes and -configs subpackage Jun 14 10:57:09 Signed-off-by: Martin Jansa Jun 14 11:12:41 jama any news? Jun 14 11:13:36 gm cbrake Jun 14 11:19:51 hi gnutoo Jun 14 11:24:01 hi pwgen Jun 14 11:24:31 *G* hi . i was just thinking on hmm is jdk 6 allready there ... Jun 14 11:25:33 in the jalimo overlay Jun 14 11:25:39 but I will check it in soon Jun 14 11:25:41 maybee today Jun 14 11:26:08 i am trying to get systemd running .. Jun 14 11:26:29 he Jun 14 11:26:30 good Jun 14 11:26:38 that was on my tasks too Jun 14 11:26:39 it looked impressive . Jun 14 11:26:53 besides its from lennart *g* Jun 14 11:27:13 .. task 889 of 4547 . i think it will need some time to build all their dependencies .. Jun 14 11:27:18 jepp Jun 14 11:27:27 yeah you need glib Jun 14 11:27:33 and couple of other stuff Jun 14 11:27:47 and udev and posix caps and and and ... Jun 14 11:27:55 did you attend his talk? Jun 14 11:28:04 jepp and i was impressed Jun 14 11:28:56 i got upstart running on my c-3200 but using the old sysv initstuff was not a good approach Jun 14 11:29:09 hm debian packages are in experimental Jun 14 11:29:13 somehow Jun 14 11:29:19 i had to rewrite the complete sysv init to benefit from upstart Jun 14 11:29:32 hm this should be gone with luci Jun 14 11:29:32 d Jun 14 11:30:03 oe upstart support isn't that good either Jun 14 11:30:07 :( Jun 14 11:30:56 yea. i had an 5.3 running. but didnt test the latest one . but lets systemd give a chance Jun 14 11:32:27 I hope you can integrate it nicely Jun 14 11:32:42 *FG* netwalker with ubuntu and gxine was too slow to play dvb-t stream . Angstrom + mplayer can do this .... Jun 14 11:33:06 hm Jun 14 11:33:22 maybee xine has not such a good neon support Jun 14 11:33:27 like ffpmeg Jun 14 11:33:32 ups ffmpeg Jun 14 11:34:33 the ubunti cam with totem as default media player, but this was ugly slow. so xine was the only alternative mediaplayer ubuntu offers Jun 14 11:35:22 woglinde: not yet.. rebuilding from scratch over weekend and I didn't have qt built Jun 14 11:35:35 jama okay Jun 14 11:35:38 I will wait Jun 14 11:35:51 do we build qt-svg support in oe? Jun 14 11:35:58 woglinde: and I don't have fr with me.. so I can only build it now.. but test on device will be tonight first Jun 14 11:36:08 woglinde, hi Jun 14 11:36:30 y Packaged contents of libqtsvg4 into /OE/tmpdir-dev-shr/deploy/ipk/armv4t/libqtsvg4_4.6.2+4.7.0-beta1-r17.1.5_armv4t.ipk Jun 14 11:36:55 okay Jun 14 11:36:58 fine Jun 14 11:48:49 When i tried to bitbake recipes/e2fsprogs/e2fsprogs_1.41.9.bb, I got this error in compilation "unix_io.o: file not recognized: File truncated" and exit by errormake[2]: *** [libext2fs.so.2.4] Error 1 Jun 14 11:49:16 soccer now Jun 14 11:49:29 ananth strange Jun 14 11:49:36 are you using oe stable or dev? Jun 14 11:49:50 gumstix stable one Jun 14 11:49:57 org.openembedded Jun 14 11:50:34 When i saw the release notes of e2fsprogs they prefer to say like this in configure BUILD_CFLAGS and BUILD_LDFLAGS instead of CFLAGS and LDFLAGS Jun 14 11:50:39 http://www.btree.org/linux/sources/fs/e2fsprogs/RELEASE-NOTES Jun 14 11:51:06 anath sorry ask gumstick Jun 14 11:51:12 dont know what the did to oe Jun 14 11:51:14 So how to modify the configure parameter in recipes/e2fsprogs/e2fsprogs_1.41.9.bb ? Jun 14 11:52:01 ok woglinde. In oe did you came across this problem Jun 14 11:53:15 no Jun 14 11:54:10 ok thanks Jun 14 11:57:59 woglinde any clues about the error and how this error can be corrected ? Jun 14 12:00:06 sorry can only speculate this file is either not compiled for the right architecture or dont compiled at all Jun 14 12:00:12 look into the logs Jun 14 12:00:17 and the file it sle Jun 14 12:00:19 self Jun 14 12:00:28 it has to be a binary file Jun 14 12:03:52 ok Jun 14 12:21:28 Hi woglinde in e2fsprogs lib/ext2fs/ unix_io.o,unlink.o,valid_blk.o and version.o are the files with empty zero length. So i get this compiling issues. Did you have any idea how to make this right Jun 14 12:26:00 no Jun 14 12:26:05 maybee you have borken ram Jun 14 12:26:13 try to compile it again Jun 14 12:26:23 run bitbake -c clean e2fsprogs Jun 14 12:26:31 and bitbake e2fsprogs Jun 14 12:45:07 woglinde: systemd needs udev_154 .. Jun 14 12:45:28 hm ayes Jun 14 12:45:35 but 154 is oe Jun 14 12:45:46 only def_pref = _1 Jun 14 12:45:50 ups -1 Jun 14 12:45:56 pwgen 1:0 btw. Jun 14 12:46:57 grmpf. dvb-s card has crashed again .. too windy here .. Jun 14 12:48:41 what the hack i libcgroup ?? Jun 14 12:48:53 hm? Jun 14 12:50:00 systemd wants to find headers for libcgroup .. "checking for CGROUP.." failed Jun 14 12:50:27 hm Jun 14 12:50:31 lets google Jun 14 12:50:56 here we go Jun 14 12:50:57 http://libcg.sourceforge.net/ Jun 14 12:51:02 debian has something Jun 14 12:51:09 seems you need to make an other recpie Jun 14 12:53:12 the last udev i can see is 151. i copy/renamed it to 154 and it compiled ( dunno if its working ) Jun 14 12:53:22 hm Jun 14 12:53:24 okay Jun 14 12:53:42 okay I thought we had 154 alreay Jun 14 12:53:54 the latest is now 157 Jun 14 12:54:33 libcgroup is autotools based Jun 14 12:54:42 so it should be easy to make a recipe Jun 14 12:55:06 i am on it .. Jun 14 12:58:25 03Martin Jansa  07org.openembedded.dev * rf6cd1f9f32 10openembedded.git/recipes/xorg-data/xorg-data-common.inc: Jun 14 12:58:25 xorg-data: convert to new staging Jun 14 12:58:25 Signed-off-by: Martin Jansa Jun 14 12:58:32 03Martin Jansa  07org.openembedded.dev * r68641750dd 10openembedded.git/recipes/navit/ (navit.inc navit_svn.bb): Jun 14 12:58:32 navit: move packaging from svn version to .inc file, add qt-painter package, enable qt Jun 14 12:58:32 * for qml you need qt-4.7.0-beta1 or newer Jun 14 12:58:32 Signed-off-by: Martin Jansa Jun 14 12:58:32 03Martin Jansa  07org.openembedded.dev * rb661aa5c43 10openembedded.git/ (3 files in 2 dirs): Jun 14 12:58:32 navit: remove old versions to simplify .inc file Jun 14 12:58:33 Signed-off-by: Martin Jansa Jun 14 13:03:51 woglinde, khem as you are the uclibc maintainers: i'm trying to build a system with uclibc, but when buildign gcc-cross building libstdc++v3 fails because of a configure problem" undefined reference to __uClibc_main (which in turn causes later tests to fail), I found this symbol in tmp/sysroots/nios2-linux-uclibc/usr/lib/libc.a but apparently that file is not linked in. Any idea? Jun 14 13:04:26 should this lib be elsewher (e.g in cross) ? Jun 14 13:06:51 effem uh never saw this error Jun 14 13:07:04 hm Jun 14 13:07:12 effem I suggest try the _git version of uclibx Jun 14 13:07:22 remove def_pref = "-1" Jun 14 13:07:35 here is teh configure command and the gcc spec: http://www.pastebin.ca/1882864 Jun 14 13:07:45 and set it in preffered_version for uclibc an uclibc-initial to git Jun 14 13:07:51 in local.conf Jun 14 13:08:02 ok will try taht Jun 14 13:09:07 btw TARGET_OS is set to linux-uclibc Jun 14 13:09:54 and the version i was building is 0.9.;30.3 Jun 14 13:10:43 hms what was the command to rebuild the ipkg index at the buildhost? Jun 14 13:11:00 (:-(( 2:0 .. Jun 14 13:11:23 yes Jun 14 13:11:32 only 5 minutes lef Jun 14 13:11:36 pwgen :-) Jun 14 13:13:14 ... i lived a long time near to the neatherlands/german border. so i would loved it when the danish had beaten the netherlands. but . Jun 14 13:13:25 morning Jun 14 13:13:44 ah found it Jun 14 13:13:48 hi chouimat Jun 14 13:13:55 hi chouimat Jun 14 13:14:24 * chouimat is happy that contract from hell is over ... Jun 14 13:15:52 hehe Jun 14 13:15:53 gratz Jun 14 13:17:14 woglinde: now I just need to cut my drinking (side effect of the contract) Jun 14 13:17:17 now i have a libcgroup recipe, that compiles, but the pkg-config is still complaining . Jun 14 13:18:00 pwgen hm why? Jun 14 13:18:11 autotools should configure it right Jun 14 13:18:22 otherwise use inherit pkgconfig too Jun 14 13:20:16 http://pastebin.com/MAmSv81S Jun 14 13:37:58 pwgen still there? Jun 14 13:46:41 woglinde: adding pkgconfig helps. now its stuck on missing vala ... lets see. Jun 14 13:47:20 hehe yes Jun 14 13:47:21 vala Jun 14 13:47:22 sorry Jun 14 13:47:30 could have tell you before Jun 14 13:47:42 luckly mickeyl helds vala in shape in oe Jun 14 13:48:24 i try building some higher level image . and check the deps with an clean build later Jun 14 14:35:31 woglinde: for what target do you typically do your uclibc build? suspecting my problem might not be a uclibc problem bu ta toolchain problem Jun 14 14:35:39 beagleboard Jun 14 14:35:41 or simpad Jun 14 14:35:45 would like to build a known good setup and compare Jun 14 14:35:49 ok will buidl for beagle Jun 14 14:35:53 effem look at the config which is generate Jun 14 14:35:54 d Jun 14 14:36:04 maybe it isnt good for your target Jun 14 14:36:41 the problem seems to be that a configure test fails Jun 14 14:36:45 config for gcc ? Jun 14 14:36:53 or uclibc? Jun 14 14:37:15 uclibc Jun 14 14:37:23 which gcc version? Jun 14 14:37:26 do you use Jun 14 14:37:37 4.1.2 :-( Jun 14 14:37:40 the 4.3 angstroem uses worked for me Jun 14 14:37:41 oh Jun 14 14:38:34 i'm trying to get uclibc work for nios2 and the nios2 is for 4.1.2, i could port to 4.2 but not 4.4 (didn't try 4.3) Jun 14 14:42:51 woglinde: will try bb and compare things, will get back on this tomorrow (time to go home :-) ) Jun 14 15:33:15 Hello Jun 14 15:33:32 I'm looking for a .bb file for alsaplayer ? Can anyone help me ? Jun 14 15:37:27 fjanot, isn't that part of alsautils? Jun 14 15:37:46 what are the differences between alsaplayer and aplay? Jun 14 15:39:30 alsautils contains aplay not alsaplayer Jun 14 15:40:21 can aplay play .mp3 file with libmad ? Jun 14 15:42:54 no Jun 14 15:42:58 but with mplayer yes Jun 14 15:43:04 but libmad is not good enough Jun 14 15:43:10 use ffmpeg's libmp3 implementation Jun 14 15:43:16 it's more performant nowadays Jun 14 15:43:59 there was a debian bugreport with diego something(a mplayer developer) about it Jun 14 15:44:06 in fact I have a small application which used alsaplayer as a daemon and I would like to port it to a beagleboard Jun 14 15:44:54 http://www.mail-archive.com/debian-bugs-dist@lists.debian.org/msg688202.html Jun 14 15:45:00 ah ok Jun 14 15:45:03 then I'll look Jun 14 15:50:27 fjanot, I think you'll need to make a recipe for alsaplayer Jun 14 15:50:33 I've grepped the recipe dir Jun 14 15:50:45 grep -i player -r * | grep -i alsa Jun 14 15:50:56 and it found no alsa player Jun 14 15:51:09 we dont have it Jun 14 15:51:20 gstreamer xine and mplayer is enough Jun 14 15:51:41 fjanot, maybe you could be interested in mpd Jun 14 15:51:53 he had a custom application for alsaplayer Jun 14 15:51:56 that is the issue Jun 14 15:53:55 I just see that mplayer has a slave mode, so I can control it with my application Jun 14 15:54:13 I will try to solve my problem using this way Jun 14 15:54:24 tahnk you very much for your help Jun 14 15:54:44 it has a slave mode Jun 14 15:55:13 http://www.mplayerhq.hu/DOCS/tech/slave.txt Jun 14 15:55:41 thank you Jun 14 16:29:24 if "make -e" is run on a Makefile, which calls $(MAKE), will the "-e" (environment variables get priority) be inherited by $(MAKE) ? Jun 14 16:40:00 woglinde: systemd now fails because of problems linking libc and socket funtion "accept4" . any idea ? Jun 14 16:43:20 hm Jun 14 16:43:28 pastebin complete error? Jun 14 16:44:13 http://pastebin.org/330751 Jun 14 16:45:57 pwgen its easy to fix Jun 14 16:46:01 its linking time Jun 14 16:46:14 try to add -lc before -lgcc in the makefile Jun 14 17:02:00 woglinde: its aufocon/make created, how ca i add this on confuration time ? Jun 14 17:07:43 in the Makefile.am Jun 14 17:08:04 please pastebin it for test-engine Jun 14 17:08:20 normaly linker adds -lc itself Jun 14 17:08:28 but sometimes not Jun 14 17:08:41 depends on what is used gcc/g++ as linker too Jun 14 17:11:17 http://pastebin.org/330820 Jun 14 17:14:50 http://pastebin.org/330833 Jun 14 17:34:19 hmm dosnt help . Jun 14 17:34:49 enough for now , let have a look at it next day. Jun 14 17:34:52 CU Jun 14 17:34:53 hms Jun 14 17:34:56 args Jun 14 17:35:01 you maybee can mail it to me Jun 14 17:35:04 please Jun 14 17:35:12 heinold@inf.fu-berlin.de Jun 14 17:35:14 thanks Jun 14 17:42:40 woglinde: email is on the way Jun 14 17:42:43 cu Jun 14 18:31:02 likewise: since you would have already read them into the make env on top level because you used -e they should be passed automatically to sub-makes, iow submake would not get -e but it will get the environment which it otherwise would have processed with -e Jun 14 18:33:32 hmmm so another weird case of libtool la files I am seeing. The recipe is libsdl, its having libdir=/usr/lib in libSDL.la as a result libtool happily expands it to -L/usr/lib -rpath /usr/lib on linker commandline Jun 14 18:33:46 now -L/usr/lib starts looking for libraries on the build machine Jun 14 18:34:50 and on ubuntu/debian it goes fine but on archlinux libgcc_s.so is installed in /usr/lib and not in /lib like debian based systems Jun 14 18:35:04 so it picks this libgcc and kaboom Jun 14 18:35:25 ~lart libtool Jun 14 18:35:26 * ibot beats libtool over the head with a microkernel Jun 14 18:35:27 khem: I am seeing a case (install stage of recipes/udev/{acl,attr}) where this is not the case. Only if I do export MAKE="make -e" in the recipe, it gets the other exported variables in the recipes. Jun 14 18:36:29 khem hi Jun 14 18:36:30 likewise: firstly its a bad idea to use -e anyway Jun 14 18:36:34 woglinde: hello Jun 14 18:36:42 khem how do I debug a illegal instruction Jun 14 18:36:53 woglinde: kernel ? Jun 14 18:37:19 likewise: did you check MAKEFLAGS Jun 14 18:37:31 khem: we use -e *all over the place* by default in OE, don't we? Jun 14 18:37:53 likewise: yes yes :) Jun 14 18:37:58 khem: :-) Jun 14 18:38:33 likewise: still that does not make it a good idea Jun 14 18:38:38 khem no runtime Jun 14 18:38:42 userspace Jun 14 18:38:59 I finally got openjdk compiled with uclibc Jun 14 18:39:14 but at runtime I get illegal instruction Jun 14 18:39:24 woglinde: you can enable use_debug=-1 on kernel cmdline Jun 14 18:39:26 woglinde: does openjdk work for you for non-uclibc? Jun 14 18:39:33 likewise yes Jun 14 18:39:36 and compile kernel with USER_DEBUG=y Jun 14 18:39:43 khem ah okay Jun 14 18:39:45 then it should dump you address and opcodes Jun 14 18:39:57 cool Jun 14 18:40:07 but I wonder where it may comes from Jun 14 18:40:11 for uclibc self Jun 14 18:40:19 woglinde: is it happening in uclibc ? Jun 14 18:40:28 or from the openjdk arm-asm Jun 14 18:40:34 khem I couldnt locate it Jun 14 18:40:38 otherwise I wouldnt ask Jun 14 18:40:50 woglinde: my first suspect would be inline assembly Jun 14 18:41:13 we compile with thumb so the inline asm has to interworking safe Jun 14 18:41:30 hm let see if beagleboard kernel as USER_DEBUG on per default Jun 14 18:41:34 people dont know to write proper inline asm let alone make them interwork safe Jun 14 18:41:48 * woglinde has to wait to catch the next bus Jun 14 18:42:00 khem 4:0 btw. Jun 14 18:42:10 woglinde: yes I watched it Jun 14 18:42:17 action packed stuff :) Jun 14 18:42:29 I was not even settled in my sofa and it was 1:0 Jun 14 18:43:04 likewise: congrats for the win today :) Jun 14 18:43:35 khem: didn't even see the match; was debugging a device driver :-) Jun 14 18:43:42 likewise: how is submake invoked in udev ? Jun 14 18:43:49 may thats where it looses Jun 14 18:44:01 khem: $(MAKE) -C subdir Jun 14 18:44:02 hm okay Jun 14 18:44:05 likewise: eh. I would have done both at same time Jun 14 18:44:06 no USER_DEBUG Jun 14 18:44:08 khem: $(MAKE) install -C subdir Jun 14 18:44:12 khem: lol Jun 14 18:44:28 likewise: my video watching is all on pc Jun 14 18:44:36 I dont even own a TV anymore :0 Jun 14 18:44:41 khem I watched via zattoo Jun 14 18:44:53 tofay Jun 14 18:44:57 today Jun 14 18:45:02 hm italy is running Jun 14 18:45:08 I hope they will lost Jun 14 18:45:15 khem: my video watching is one movie per month, on the beamer, from my MBP. Only watch telly when my girlfriend cheers "you should really really really see this!" Jun 14 18:45:47 heh that cheer means a command "watch this !!" Jun 14 18:45:54 should watch some more, it relaxes the mind a little Jun 14 18:46:12 khem: that's why I'm still unmarried :-) Jun 14 18:47:02 woglinde: cool but zattoo is not available in US :( Jun 14 18:47:07 ah damn rtl is not on zattoo Jun 14 18:47:17 I forgot it Jun 14 18:47:17 this is diverting too much :-) Don't be surprised when I check in a recipe saying $(MARRY)... Jun 14 18:47:28 likewise: heh Jun 14 18:47:43 likewise: so make invocation seems to be right hmmm Jun 14 18:48:02 woglinde: I have similar webtv here Jun 14 18:48:04 okay 0:0 Jun 14 18:48:10 will leave now Jun 14 18:48:14 till later maybee Jun 14 18:48:19 woglinde: italy will play a draw today Jun 14 18:50:34 likewise: check if before submake call MAKEFLAGS is being modified in some way Jun 14 18:51:10 likewise: some options are passed to submakes thru MAKEFLAGS I am not sure -e is one of them Jun 14 18:51:19 but like -k is passed for sure Jun 14 18:55:19 khem: will check, thanks for the hint Jun 14 18:56:49 likewise: if you are lucky -e will be passed if not then use sledgehammer and "MAKE=make -e" Jun 14 18:57:40 likewise: I wonder what udev needs from env Jun 14 18:59:59 khem: http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/udev/attr_2.4.44.bb Jun 14 19:00:42 khem: and this is the failure: http://tinderbox.openembedded.net/packages/attr/ Jun 14 19:01:49 khem: the subdirs do not get the exported variables correctly. I hate to inject sledgehammer fixes, so I try to understand what the neat solution would be. Jun 14 19:04:43 likewise: what happens if you pass -e in EXTRA_OEMAKE Jun 14 19:04:49 instead Jun 14 19:05:24 add EXTRA_OEMAKE += " -e " and retry Jun 14 19:06:00 EXTRA_OEMAKE = "-e MAKEFLAGS=" Jun 14 19:06:08 is already set in bitbake.conf ? Jun 14 19:06:10 hmmm Jun 14 19:12:43 khem: the subdir Makefiles include another file which does PKG_BIN_DIR = /usr/bin/ Jun 14 19:13:04 hmm Jun 14 19:13:15 khem: so I think even though the subdir makes do inherit the good values from their parent make, they get overridden later Jun 14 19:13:28 yes its messed somehow Jun 14 19:13:42 I think our MAKEFLAGS should also contain -e Jun 14 19:14:28 hm yes but the assumption is that toplevel does all the right things Jun 14 19:14:33 and then calls submakes Jun 14 19:16:31 khem: in this case, the include file probably messes things up in the subdir. Jun 14 19:16:55 khem: so probably I should just remove them from the include file Jun 14 19:17:42 we have 15 other recipes doing EXTRA_OEMAKE = "-e ...", such as qt. Jun 14 19:24:33 likewise: yes remove them Jun 14 19:30:43 khem: I think we strongly rely on them. I could remove them and see what happens. Jun 14 19:31:03 yeah and adding -e to MAKEFLAGS is a good idea too Jun 14 19:31:06 now that we use it Jun 14 19:41:11 how do I get the interactive bitbake console to rereade a bbclass file? Jun 14 19:47:41 nm Jun 14 20:28:33 re Jun 14 20:28:39 khem ping Jun 14 20:29:04 DEBUG_USER=y an debug.user=-1? Jun 14 20:31:24 likewise: debug_user=-1 Jun 14 20:31:39 why not debug.user? Jun 14 20:31:41 *sigh* Jun 14 20:31:48 I hate this kernel debug stuff Jun 14 21:38:09 khem ping Jun 14 21:38:10 khem ping Jun 14 21:47:30 woglinde: yes Jun 14 21:47:44 [ 197.198791] [1252] java: obsolete system call 405dad68. Jun 14 21:47:44 [ 197.204071] Code: ebfffe2f e1a0c000 e8bd0007 ef000005 (e52d0004) Jun 14 21:47:53 heh Jun 14 21:47:54 now tell me whats wrong Jun 14 21:48:05 *g* Jun 14 21:48:11 google has nada Jun 14 21:48:13 woglinde: btw java is not very EABI friendly just so you know Jun 14 21:48:36 khem glibc java is working on omap3 Jun 14 21:48:44 ok Jun 14 21:49:33 but let me build glibc again Jun 14 21:49:52 maybee it will not work with this kernel Jun 14 21:51:00 e52d0004 is saving r0 on stack Jun 14 21:51:08 and why would be fail Jun 14 21:51:10 check SP Jun 14 21:51:15 is it still ok Jun 14 21:51:18 03Yuri Bushmelev  07org.openembedded.dev * r38f8326bf5 10openembedded.git/recipes/neon/neon_0.25.5.bb: neon_0.25.5: Disable GSSAPI to remove kerberos dependency. Jun 14 21:51:27 hm intressting Jun 14 21:51:32 what is SPR value Jun 14 21:51:43 sorry there isnt anymore output Jun 14 21:51:57 and I cant install debug symbols because my mmc is to small Jun 14 21:53:13 ok Jun 14 21:56:43 woglinde: although I dont think obsolete syscall is causing it Jun 14 22:00:29 woglinde: can you paste the complete log somewhere ? Jun 14 22:01:51 * khem has switched to using archlinux these days its impressive Jun 14 22:02:07 KDE 4.5 runs like a hill goat here :) Jun 14 22:03:17 khem this was the complete loh Jun 14 22:03:20 ups log Jun 14 22:05:29 woglinde: hmmm Jun 14 22:06:37 sorry Jun 14 22:08:07 woglinde: did you compile complete system in arm mode ? Jun 14 22:09:58 yes Jun 14 22:10:31 woglinde: is the option you used user_debug=-1 ? Jun 14 22:10:44 yes Jun 14 22:10:51 I typoes it last time Jun 14 22:10:56 and when avahi segfaults there is more output Jun 14 22:11:20 woglinde: and you are using uclibc-git Jun 14 22:11:29 or psplash Jun 14 22:11:32 khem sure Jun 14 22:11:41 woglinde: ok Jun 14 22:11:42 otherwise openjdk wouldnt be build Jun 14 22:11:48 bescause of thread stuff Jun 14 22:11:59 can you boot your box from nfs ? Jun 14 22:12:09 because it will be lot more helpful Jun 14 22:12:12 to have symbols Jun 14 22:12:13 should be possible Jun 14 22:12:16 hm right Jun 14 22:12:30 if you have symbols its straight forward Jun 14 22:13:01 * khem thinks we should use patchelf instead of chrpath Jun 14 22:24:09 * Tartarus thinks we should just link correctly the first time Jun 14 22:24:36 It's really not that hard Jun 14 22:24:51 Tartarus: with libtool its not that easy Jun 14 22:25:32 Tartarus: libtool has no idea of runtime location and staging concept Jun 14 22:39:54 No, it's pretty easy Jun 14 22:40:08 Mentor does it Jun 14 22:40:13 I believe mvista does it Jun 14 22:40:50 It's just that gcc/binutils/gdb require a little extra thought to make it happen, due to their re-running configure Jun 14 22:42:59 Well with sysroot its better Jun 14 22:43:14 Nope, that's not it Jun 14 22:43:30 We're talking linking cross gcc & co with $ORIGIN Jun 14 22:43:59 using $ORIGIN is also not 100% right Jun 14 22:45:12 Well, the path we chrpath in Jun 14 22:46:50 we need to use ORIGIN with libtool too Jun 14 22:47:01 Yes Jun 14 22:47:04 That's easy Jun 14 22:47:27 The only "hard" part is gcc/binutils/gdb and their re-running of configure & such Jun 14 22:47:32 but libtool would have to know where the program will be installed Jun 14 22:47:34 That means really odd quoting, or just a regex Jun 14 22:47:36 does it have that info Jun 14 22:47:41 Easy Jun 14 22:48:27 export RELATIVERPATH = "${@base_path_relative(d.getVar('bindir', 1), d.getVar('STAGING_LIBDIR_NATIVE', 1))}" Jun 14 22:48:37 how about soft links Jun 14 22:48:38 That gets you, roughly 99% of the world right Jun 14 22:48:54 khem hms usb nfsroot dont works Jun 14 22:49:02 What about soft links? Jun 14 22:50:08 like /lib -> /usr/lib Jun 14 22:50:13 oh.. it is dawning here.. time to sleep Jun 14 22:50:19 some libs reach out that way Jun 14 22:50:21 khem, what about? Jun 14 22:50:28 nite jay7 Jun 14 22:50:32 will relative path still find Jun 14 22:50:37 those cases Jun 14 22:50:46 khem, you mean DISTRO=micro? Jun 14 22:50:47 Or what? Jun 14 22:50:51 yes Jun 14 22:50:55 Yes Jun 14 22:51:40 We don't hard-code $ORIGIN/../usr/lib, we evaluate the variable that tell us the layout Jun 14 22:51:45 just like the code does does :) Jun 14 22:51:50 ok Jun 14 22:51:53 and we just link in what we would be chrpath'ing Jun 14 22:51:57 yeah I guess libdir Jun 14 22:52:19 No, what I pasted works for micro I bet Jun 14 22:52:20 woglinde: why it doesnt work did you enable it in kernel Jun 14 22:52:28 relative of bindir to staging libdir Jun 14 22:52:31 yes it will Jun 14 22:52:35 I'll hopefully have the time to post a branch rfc'ing sometime Jun 14 22:52:44 khem no nfsroot over usb segfaults the kernel at boottime Jun 14 22:53:00 Tartarus: that wud be cool Jun 14 22:53:14 post something Jun 14 22:53:23 heh Jun 14 22:53:32 new rpath '$ORIGIN/../../../../../../../../../..' too large; maximum length 8 Jun 14 22:53:33 kergoth posted the bulk of it a year ago almost Jun 14 22:53:37 I am annoyed by them Jun 14 22:53:39 pre-leaving MV :) Jun 14 22:54:00 brbish Jun 14 22:56:10 http://patchwork.openembedded.org/patch/972/ Jun 14 22:56:16 why was this not pushed ? Jun 14 23:01:49 It's not a complete solution Jun 14 23:02:06 It's also not quite right anymore, heh Jun 14 23:14:09 Tartarus: yeah see that Jun 14 23:14:33 Tartarus: I will look for something if you or kergoth will post Jun 14 23:15:18 I'll push a branch sometime "soon" I hope Jun 15 00:19:44 woglinde: may me gdbserver is an option Jun 15 00:20:20 khem I will buy a bigger sd-card Jun 15 00:20:38 yeah something like 2G wil be ok :) Jun 15 00:26:37 woglinde: hey! Jun 15 00:26:46 woglinde: why are you awake and not doing your thesis? Jun 15 00:26:47 zecke: hello Jun 15 00:27:12 he is doing part of praktikum :) Jun 15 00:30:58 now he is hiding. :) Jun 15 00:33:07 Iam waiting Jun 15 00:41:04 mwester: howdy **** ENDING LOGGING AT Tue Jun 15 02:59:57 2010