**** BEGIN LOGGING AT Thu May 17 03:00:00 2012 May 17 07:30:51 good morning May 17 08:08:18 hi all May 17 10:48:17 fray: GNUtoo : hi. I built succesfully an image with oe-core. Thanks. May 17 10:48:54 Now, if we want to build some packages which I do not find in oe-core, how should we build them? via oe-classic yet?, port from oe-classic to oe-core? May 17 11:00:12 bluelightning: please update the layers list May 17 11:00:17 bluelightning: :) May 17 11:00:37 otavio: yep, just saw your email, will do now :) May 17 11:00:43 bluelightning: we've merged meta-mozilla and meta-chromium onto meta-browser May 17 11:00:55 bluelightning: thx :) May 17 11:03:22 zrafa: try to find recipes on github e.g. May 17 11:04:53 otavio: done May 17 11:05:03 bluelightning: :) May 17 11:05:26 bluelightning: makes more sense to have a single place for browsers May 17 11:06:02 GNUtoo: :-) ericben told me you're still working on more platforms support for Chromium May 17 11:06:13 GNUtoo: anything I can help? May 17 11:07:20 otavio: indeed, good move May 17 11:12:59 Jay7: but if I find.. should I download and put somewhere and then bitbake? May 17 11:14:31 zrafa: if you find, I'm sure you will find layer :) May 17 11:14:41 so you may attach that layer and do bitbake May 17 11:14:54 or just place recipe into meta-oe, e.g. May 17 11:39:42 Jay7: no sure if I understand. Is there some example around?. Does that mean that we can not build repositories full of packages as before? (I ask this May 17 11:40:06 because oe-classic is full of errors everywhere and it stops all the time the building) May 17 11:40:16 now you can stack layers to get packages you need May 17 11:40:25 oe-core is minimal base May 17 11:40:53 e.g. we have meta-handheld on top of oe-core and meta-oe to build images for zaurus and ipaq May 17 11:45:06 Jay7: I see in layer index meta-handheld : git://git.openembedded.org/meta-handheld May 17 11:45:28 I do a git clone and then I do something like I did to build core-image with oe-core? May 17 11:45:48 kind of May 17 11:45:55 check angstrom build scripts May 17 11:46:08 there are lots of layers added May 17 12:00:32 Jay7: let me check May 17 12:00:34 angstrom May 17 13:16:52 khem, hi May 17 13:49:48 how I can add a gcc -D option if compiling with hf tune? May 17 14:06:25 only at me bitbake sometime fails and sometimes no? May 17 14:06:42 you're going to have to be a lot more specific than that May 17 14:07:16 for example if i do: May 17 14:07:19 root@ossystems:~# node May 17 14:07:20 ERROR: Binary compiled with -mfloat-abi=hard but without -DUSE_EABI_HARDFLOAT May 17 14:07:41 bitbake pcmanfm cross compilation fails May 17 14:07:43 otavio, is it chromium? May 17 14:07:53 GNUtoo-desktop: nops; nodejs May 17 14:08:00 in beagleboard May 17 14:08:05 otavio, I've the exact same error with chromium May 17 14:08:21 with armv6 May 17 14:08:26 if i do: May 17 14:08:40 bitbake -k angstrom-lxde-image May 17 14:08:47 it compile :( May 17 14:09:01 lol May 17 14:09:42 some ideas? May 17 14:10:28 silvio_: you still haven't given us anything useful May 17 14:10:32 we are not psychic May 17 14:10:42 if you're getting an error, put the exact error somewhere (ideally a pastebin) May 17 14:10:55 ok May 17 14:11:10 and the fact that -k does build stuff is irrelevant, and means *nothing* May 17 14:11:20 -k tells bitbake to continue building what it can even when encountering a failure May 17 14:11:26 it doesn't mean the failure went away May 17 14:11:39 i know is the reason that make me confused May 17 14:11:55 if you knew, you wouldn't be confused May 17 14:12:03 * kergoth` gets food May 17 14:12:17 ? means? May 17 14:12:32 GNUtoo-desktop: any idea how I set it depending on tune? May 17 14:12:51 feed me please May 17 14:13:04 :) May 17 14:13:31 otavio, what do you mean? May 17 14:16:18 now i have similar with gobjectintrospection May 17 14:16:21 http://paste.ubuntu.com/992418/ May 17 14:17:20 GNUtoo-desktop: i am unsure if it is a nodejs compilation problem or a oe issue May 17 14:18:34 otavio, what's your arch? May 17 14:18:51 armv7a-vfp-neon-oe-linux-gnueabi May 17 14:19:03 otavio, ah, I saw the issue on armv6 May 17 14:19:12 the eukrea-cpuimx35 May 17 14:19:46 adn this is the error that i got with "bitbake pcmanfm": May 17 14:19:52 http://paste.ubuntu.com/992423/ May 17 14:20:08 someone could show me a way to understand more pls May 17 14:21:29 silvio_, I guess you need a native xml-purge May 17 14:21:39 because it tries to execute the arm version on x86 May 17 14:21:50 or something like that May 17 14:21:56 also check permissions of xml-purge May 17 14:23:18 GNUtoo-desktop: your failure is at v8 lib I think May 17 14:23:20 mine too May 17 14:24:03 otavio, yes indeed May 17 14:24:29 GNUtoo-desktop: we ought to package v8 and use it in nodejs and chromium May 17 14:24:37 yes May 17 14:24:49 GNUtoo-desktop: so we avoid this code duplication May 17 14:25:00 GNUtoo-desktop: debian has a nice patch for a issue in v8 too May 17 14:25:00 and fixes duplications May 17 14:25:05 ok May 17 14:25:31 GNUtoo-desktop: can you look at it? May 17 14:27:29 not right now May 17 14:27:38 maybe next week May 17 14:32:16 otavio, basically I work part time for eukrea May 17 14:34:51 it fails again May 17 14:42:16 i see May 17 14:52:46 silvio_, maybe you could try to fix it May 17 14:52:55 first run the file program May 17 14:53:05 on the binary that cannot be executed May 17 14:53:16 that will print the architecture of the binary May 17 14:53:33 i m doing... May 17 14:53:39 ... or better i m try May 17 14:57:45 GNUtoo-desktop: http://anonscm.debian.org/gitweb/?p=collab-maint/libv8.git;a=blob;f=debian/rules;h=7a4fa206b8f5149f92302f05a8c907978427de74;hb=HEAD#l47 May 17 15:01:13 khem, Agate is out of HD space :D May 17 15:08:01 -Dv8_use_arm_eabi_hardfloat=false May 17 15:08:05 thanks May 17 15:32:48 hhumm May 17 15:32:50 #!/bin/sh May 17 15:32:52 arm-oe-linux-gnueabi-gcc -march=armv7-a -fno-tree-vectorize -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8 -mfloat-abi=softfp --sysroot=/home/otavio/hacking/ossystems/embedded-linux/tmp-eglibc-eglibc/sysroots/beagleboard $@ May 17 15:32:58 it passes the soft May 17 15:33:05 dunnow why it try to use hard May 17 15:42:16 i m in a cul de sac May 17 15:42:29 moin May 17 16:04:23 -rwxrwxr-x 1 silvio silvio 51718 mag 17 16:33 xml-purge May 17 16:08:36 silvio_, file ./xml-purge May 17 16:14:06 now i did the same commandfirst time fail then works!!! May 17 16:15:46 i m lost May 17 16:16:33 ? May 17 16:17:18 silvio_: since we can't see your screen, so are we May 17 16:17:33 http://paste.ubuntu.com/992622/ May 17 16:19:43 $ bitbake -k angstrom-lxde-image 2>&1 | tee $OEHOME/build.log May 17 16:33:43 mimimosa May 17 16:36:42 bb May 17 17:37:21 GNUtoo-desktop: did you find where to set the -Dv8_use_arm_eabi_hardfloat=false? May 17 17:59:58 otavio, I didn't look yet May 17 18:00:33 but I guess an CFLAGS += or something similar could work May 17 18:01:00 if it's not covered by the v8 build system or by chromium gyp build system May 17 18:16:30 GNUtoo-desktop: http://paste.debian.net/169306/ May 17 18:16:36 GNUtoo-desktop: didn't help May 17 18:17:33 can you also look in bitbake -e May 17 18:17:40 to see if CFLAGS are the same? May 17 18:17:52 s/are the same/are for target/ May 17 18:18:05 usually they are I think May 17 18:18:25 or also look C++ flags May 17 18:18:50 (CXXFLAGS if I remember well) May 17 18:21:29 khem: did you ever build v8? May 17 18:22:06 I am having problem to build nodejs and GNUtoo-desktop for chromiun May 17 18:22:19 root@ossystems:~# node May 17 18:22:19 ERROR: Binary compiled with -mfloat-abi=hard but without -DUSE_EABI_HARDFLOAT May 17 18:29:47 otavio: hmmm how is your toolchain confugured May 17 18:29:57 beagleboard May 17 18:29:58 oe-core May 17 18:30:12 otavio: and did u explicitly enable hf ? May 17 18:30:17 no May 17 18:30:29 khem: in fact: May 17 18:30:54 (devel)~/hacking/ossystems/embedded-linux% cat tmp-eglibc-eglibc/work/armv7a-vfp-neon-oe-linux-gnueabi/nodejs-0.6.18-r0/g++ May 17 18:30:57 #!/bin/sh May 17 18:31:00 arm-oe-linux-gnueabi-g++ -march=armv7-a -fno-tree-vectorize -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8 --sysroot=/home/otavio/hacking/ossystems/embedded-linux/tmp-eglibc-eglibc/sysroots/beagleboard $@ May 17 18:31:10 oh dont paste that into channel May 17 18:31:19 oh that was small May 17 18:31:25 so it is softfp May 17 18:31:36 lol May 17 18:31:47 :) May 17 18:32:18 ok and where can I look at node recipe May 17 18:32:26 and actually in node build tree May 17 18:32:27 meta-oe May 17 18:32:38 devtools May 17 18:32:42 nodejs May 17 18:32:42 I would be curious to see how it infers hf May 17 18:32:49 me too May 17 18:33:10 be prepared to fight with the build system May 17 18:36:35 otavio: I see export CC=gcc in do_compile May 17 18:36:39 I wonder why is that May 17 18:36:49 scons May 17 18:36:56 it creates a script May 17 18:37:03 and it just calls make and oe_runmake May 17 18:37:14 and not oe_runmake May 17 18:37:33 at any rate can you post the do_compile log somewhere May 17 18:38:07 I just tried to pass softfp in CXXFLAGS May 17 18:38:16 let it finish to build May 17 18:38:27 do_configure_virtclass-native () { sed -i -e s:\'/usr/lib:\'${STAGING_LIBDIR}:g wscript sed -i -e s:\'/usr/local/lib:\'${STAGING_LIBDIR}:g wscript May 17 18:38:30 ./configure --prefix=${prefix} --without-snapshot May 17 18:38:32 thats wrong too May 17 18:38:34 } May 17 18:38:46 it will do damage when you rerun do_configure task May 17 18:38:46 ? May 17 18:39:23 and you will get paths like ${STAGING_LIBDIR}/${STAGING_LIBDIR} May 17 18:39:31 and it will keep appending May 17 18:39:42 as many times as you do configure May 17 18:39:45 not a nice thing May 17 18:39:53 yes May 17 18:39:55 but thats besides are point May 17 18:40:10 but i did cleansstate all tries May 17 18:40:34 yes these problems are not related to the issue you are seeing May 17 18:40:36 but they are problems May 17 18:41:17 the nodejs in fact is in really bad packaging quality May 17 18:42:46 otavio: compiler should default to soft-fp May 17 18:42:54 yep May 17 18:43:10 I think nodejs build system is injecting float-abi=hard somewhere May 17 18:43:22 let me see if I have the sources to look at it somewhere May 17 18:44:08 i will be right back May 17 19:03:47 somehow scons is not getting to understand armeabi:softfp May 17 19:03:57 otavio: I have not much idea how scons work May 17 19:16:06 I downloaded meta-openembedded, how do I use it ? :) readme is not useful for that. Am I missing some doc? May 17 19:18:56 downloading one layer by itself isn't going to do anything useful May 17 19:19:10 khem: yep; this was my conclusion too May 17 19:19:13 kergoth`: I used oe-core yesterday May 17 19:19:22 khem: but i couldn't figure how to fix it May 17 19:19:52 kergoth`: to build an image core. And now I would like to use meta-openembedded to build opencv.. but no sure how to integreate meta-oe with previous building May 17 19:20:25 what is "previous building"? May 17 19:20:32 insufficient information May 17 19:20:38 i left my crystal ball at home May 17 19:22:57 kergoth`: "previous building" : I git cloned oe-core, did oe-init-build-env, modified local.conf for my needs, and then bitbake core image minimal May 17 19:23:22 kergoth`: and I got that image built for my arch (mips) May 17 19:24:00 kergoth`: now I would like to build more stuff, like opencv. I found the recipe at meta-openembedded May 17 19:24:56 okay, now you add the path to meta-openembedded/meta-oe to the BBLAYERS variable in oe-core/build/conf/bblayers.conf May 17 19:28:50 kergoth`: great, thanks. It is working... May 17 19:29:43 khem: no idea how to debug it? May 17 19:33:25 zrafa: np May 17 19:39:59 khem: it seems it is not the building system the problem May 17 19:40:01 but runtime May 17 19:41:49 otavio: yes it shows up at runtime but its build time that induces the problem May 17 19:42:38 khem: gcc is never called with hard May 17 19:46:54 otavio: hmm can you show me output of cross-gcc -dD -E -xc /dev/null please May 17 19:48:13 i am doing a new build and will post the log May 17 19:48:23 Dont need the log May 17 19:48:34 just output of above cmd for now May 17 19:50:13 khem: http://paste.debian.net/169326/ May 17 19:50:47 __SOFTFP__ __VFP_FP__ May 17 19:51:10 https://github.com/joyent/node/blob/master/deps/v8/src/platform-linux.cc#L170 May 17 19:51:15 I think I see the problem May 17 19:51:51 yes thats the offending code May 17 19:53:06 do you see the problem? i don't May 17 19:53:34 otavio: yes I see :) May 17 19:53:45 I am doing something wait a se May 17 19:53:47 c May 17 19:53:53 you have a good eye! May 17 19:53:56 tell me :) May 17 19:55:22 try this patch http://paste.ubuntu.com/992975/ May 17 19:57:38 khem: can you paste in debian or other pastebin service? i don't have an account there May 17 19:57:53 you need account ? May 17 19:58:15 to download it you do May 17 20:02:48 otavio: http://fpaste.org/SIBN/ May 17 20:04:25 testing May 17 20:04:59 ok May 17 20:11:05 khem: building May 17 20:11:20 khem: can you tell me what is the bugz? May 17 20:14:14 they are using wrong ABI define May 17 20:14:20 !@@##$##% May 17 20:17:29 WORKS! May 17 20:17:31 yey! May 17 20:18:58 cool enjoy May 17 20:19:12 and post it to nodejs bugger and attribute it to me with my signoff May 17 20:19:28 so if they have questions they can ask me May 17 20:19:33 do you mind to write a patch header? May 17 20:19:44 so i can put it in the patch May 17 20:19:47 here it goes May 17 20:21:19 gcc has a builtin define to denote hard abi when in use, e.g. when using -mfloat-abi=hard it will define __ARM_PCS_VFP to 1 and therefore we should check that to determine which calling convention is in use and not __VFP_FP__ which merely indicates presence of VFP unit May 17 20:42:30 great. we just lost bldg power May 17 20:42:32 \ May 17 21:02:11 mr_science: just need a bigger UPS ;) May 17 22:31:03 otavio, hi May 18 00:24:46 Gaston|Home: hi May 18 00:24:52 grr **** ENDING LOGGING AT Fri May 18 02:59:58 2012