**** BEGIN LOGGING AT Sat May 15 02:59:57 2010 May 15 03:04:05 khem, i just scp expat to another centos, untar and tar and scp back now bitbake nano works May 15 03:42:19 03Khem Raj  07org.openembedded.dev * r5203790c01 10openembedded.git/recipes/gcc/ (gcc-4.4.4.inc gcc-4.4.4/gcc-add-t-slibgcc-libgcc.patch): (log message trimmed) May 15 03:42:19 gcc-4.4.4.inc: Fix ppc issues of requiring libgcc at link time when using -Os May 15 03:42:19 * This change converts libgcc_s.so to a linker script so when -lgcc_s is May 15 03:42:19 used then it actually links to a group of libraries and one of those May 15 03:42:19 libraries is libgcc. This fixes the problem of requiring to link with May 15 03:42:19 libgcc _always_ on ppc to get the out-of-line epilogue/prologue May 15 03:42:20 functions. May 15 03:44:57 03Khem Raj  07org.openembedded.dev * rd05923bf3f 10openembedded.git/conf/machine/include/qemu.inc: May 15 03:44:57 machine/qemu.inc: Build and install qemu-native for qemu based targets. May 15 03:44:57 * Some distributions like ubuntu dont ship roms which are needed May 15 03:44:57 to boot ppc targets therefore we need to use qemu-native to run May 15 03:44:57 the images, therefore always build qemu-native for qemu based May 15 03:44:58 targets May 15 03:44:59 Signed-off-by: Khem Raj May 15 03:46:06 03Khem Raj  07org.openembedded.dev * r896f4980f7 10openembedded.git/recipes/eglibc/eglibc_2.11.bb: May 15 03:46:06 eglibc_2.11.bb: Bump PR to reflect last update. May 15 03:46:06 Signed-off-by: Khem Raj May 15 06:00:10 Hi, again May 15 06:00:41 I think I found a bug in the wxwidgets library May 15 06:02:43 I can send the log files and the full explanation by E-mail if anybody can look at it. There are at least 3 log files that needs to be reviewed, so unless you want the pastebin' post I'd rather use an E-mail May 15 06:04:38 or maybe something is wrong somewhere else.... I need someone more experienced with OE to take a look May 15 06:35:40 IgorK: please pastebin May 15 06:35:46 good morning May 15 06:36:15 mckoan: do u want to see 3 pastebin posts? May 15 06:36:22 good morning to u 2 May 15 06:39:18 Id like to post config.log, log.do_compiple and log.do_configure.... May 15 06:58:41 mckoan: I decided to put 3 different posts May 15 06:58:58 Here is the config.log: http://pastebin.com/YEFhRwtu May 15 07:01:50 Here is the log.do_configure: http://pastebin.com/K9YG1C53 May 15 07:05:54 And here is the log.do_compile: http://pastebin.com/eKSZQ4yT, which incorrectly uses "-DwxUSE_GUI=0 -DwxUSE_BASE=1". May 15 07:06:19 any comments/suggestions/workarounds? tia. May 15 07:15:36 IgorK: why do you suppose that there is an error? May 15 07:17:54 mckoan: because wx is configured to use gui and not base.... May 15 07:18:19 also if u look at config.log wxUSE_GUI is set to 1 May 15 07:19:39 but compile command uses wxUSE_GUI=0, which means that my program will link to not use GUI May 15 07:32:44 IgorK: BTW wx are working perferctly May 15 07:38:19 IgorK: why are you using an old wxwidgets-2.8.9-r7/wxWidgets-2.8.9 ? May 15 07:41:05 mckian: because it's a stable version, and I use it on x86 for development May 15 07:41:35 do u agree that there is a problem? May 15 07:42:36 IgorK: not really May 15 07:43:21 IgorK: BTW I'm using wxwidgets-2.8.10-r1 and is more stable May 15 07:44:56 mcloan: do u use it along with X11/GPE and the angstrom? May 15 07:46:04 IgorK: I use it with KaeilOS and X11 (luckily no need of GPE) May 15 07:46:28 * mckoan mantains http://www.kaeilos.com/ May 15 07:46:35 besides, Im using wxGTK-2.8.10 original that is copied to the OE tree... May 15 07:47:07 mckoan: so you use X11 implementation? May 15 07:50:03 IgorK: no, I use libwx_gtk and GTK+ May 15 07:50:21 X11 is the graphic engine May 15 07:51:42 mckoan: yes, but there is wxGTK nad wxX11 May 15 07:54:22 well I compiled x11-gpe-image and on top wxwidgets with my program May 15 07:54:57 unfortunately, since wx is compiled without GUI support it is crashing... ;-) May 15 07:55:04 sorry :( May 15 07:59:01 and Im using Angstrom kernel May 15 08:04:49 mckoan: I will be back in a couple of hoursit's 1 in the morning here ;-) May 15 08:05:13 look forward to talk to u in the morning May 15 08:09:01 morning May 15 08:09:34 I would like to have some informations, I'm a noob in OE, and have some problems with it. May 15 08:10:17 I use for the moment buildroot, and I try to use OE to make the same filesystem May 15 08:11:13 but (for example), I can't see how I can say : I would like Qt, but without Database and XML part May 15 08:11:29 for the moment, OE build all the Qt software lib May 15 08:20:32 03Koen Kooi  07org.openembedded.dev * r84de51631e 10openembedded.git/recipes/git/ (files/git-less-hardlinks.diff git_1.7.0.2.bb): May 15 08:20:32 git 1.7.0.2: fix hardlinking problems as reported by Eric Brombaugh May 15 08:20:32 Using softlinks has the following effects on package size: May 15 08:20:32 git: 54MB -> 3.0MB May 15 08:20:32 git-dbg: 646MB -> 141MB May 15 08:53:20 damunix: you have to customize your own qt recipe May 15 09:22:20 hi, can I use the file:// scheme to clone a git repository from local filesystem in a _git.bb recipe? I tried with SRC_URI="file:///path/to/clone;protocol=git;branch=mybranch" but it does not seem to work. Note that cloning from command line works fine with the file:// scheme May 15 09:23:39 03Koen Kooi  07org.openembedded.dev * r4154c5f1b9 10openembedded.git/recipes/sed/sed_4.1.2.bb: sed 4.1.2: add missing 'inherit gettext' May 15 09:25:55 ao2: no May 15 09:25:59 ao2: the 'scheme May 15 09:26:28 ao2: ' indicates which fetcher to use. So you will always need to use git://, now the question is if protocol=file is known by the fetcher May 15 09:27:30 zecke, so it would be SRC_URI="git:///path/to/clone;protocol=file" if that was supported May 15 09:28:04 ao2: yes May 15 09:28:38 zecke, let's try, and thanks May 15 09:29:57 ao2: it is supported. May 15 09:30:20 cool May 15 09:31:05 way handier than appending patches to test stuff before submitting upstream May 15 09:50:27 hi all May 15 09:50:36 hi mdavison May 15 09:51:27 hi eFfeM May 15 09:51:44 im sort of new to the whole OE thing May 15 09:51:52 i've got one problem though May 15 09:52:31 is there any way you can debug the bitbake recipes? May 15 09:53:41 mdavison: bitbake -DDD recipename May 15 09:54:27 yea that shows the dump.. but well the problem for me is that i've come from more of a development background May 15 09:54:36 worked with C# C++ etc. May 15 09:54:49 i'm more used to the breakpoints and setpping in and stepping out May 15 09:54:57 is there a way to do that? May 15 09:55:07 i know pydev can help with python debugging May 15 09:55:17 and gbd for bash debugging May 15 09:55:31 but seems like bitbake recipes are not exactly bash are they? May 15 09:58:08 bbs May 15 09:58:54 mdavison: bb recipes are python, no you can't debug them in such way (AFAIK) May 15 10:08:23 have a nice weekend May 15 11:01:35 03Henning Heinold  07org.openembedded.dev * r0f801ce2b1 10openembedded.git/recipes/fastjar/ (6 files): May 15 11:01:36 fastjar: switch to BBCLASSEXTEND and clean up May 15 11:01:36 * remove older versions May 15 11:01:36 * we don't need the empty dev package, because there are no libs May 15 11:01:36 * bump INC_PR to get native and normal package to the same revision May 15 11:08:01 When doing: "ffmpeg -f video4linux2 -r 30 -s 320x240 -i /dev/video0 -vcodec libx264 -vpre ultrafast sample.mpg" I got the following error: "video4linux2 ... (VIDIOC_STREAMON) Cannont allocate memory" is there some way to free up some memory thta needs video4linux2? is it RAM or Disk space affair? I do have Beagle Board with 256 Memory... May 15 11:08:15 How can I use swap in Angstrong? May 15 11:11:11 hi,I've the g_realloc issue: http://pastebin.com/Y1LkXuMM May 15 11:11:44 cleaning gtk+-native,bitbaking glib-2.0-native and recompiling gtk+-native doesn't fix it May 15 11:12:17 I tried multiple time...still not compiling May 15 11:57:29 GNUtoo|oeee hm please look with nm or objdump if g_realloc_n is in /home/embedded/oetmp_htcdream/sysroots/i686-linux/usr/lib/libglib-2.0.so May 15 11:58:44 and please post the output of /home/embedded/oetmp_htcdream/work/i686-linux/gtk+-native-2.20.0-r8.1/gtk+-2.20.0/modules/input/im-xim.la May 15 12:13:19 ok thanks May 15 12:21:44 GNUtoo|oeee, do you know how to use a toolchain with a recent gcc in OE ? May 15 12:22:30 GNUtoo|oeee, I read in some places that it's better to use gcc-4.4.x for cortex-a8 May 15 12:22:46 GNUtoo|oeee, and OE uses arm-angstrom-linux-gnueabi-gcc-4.3.3 May 15 12:25:25 dcordes ?????? May 15 12:25:37 I dont understand your question May 15 12:30:00 woglinde, I would like to use cross a toolchain that uses gcc-4.4.x May 15 12:37:09 back May 15 12:37:14 sorry I was eating May 15 12:37:47 dcordes, PREFERED_VERSION_gcc-cross = "" May 15 12:38:05 hi mickeyl May 15 12:38:14 GNUtoo|oeee, awesome thanks May 15 12:42:34 oh okay May 15 12:42:35 yes May 15 12:42:40 set in local.conf May 15 12:46:28 bitbaking devshell for the verifiction of libglib May 15 12:50:32 000465e0 t IA__g_realloc_n May 15 12:50:33 000465e0 T g_realloc_n May 15 12:52:30 hm thiss looks good May 15 12:52:52 what is your parallel_makes option? May 15 12:53:29 none May 15 12:53:36 hm May 15 12:53:39 that is why I'm lost May 15 12:53:57 I bitbaken glib-2.0 and glib-2.0-native May 15 12:54:00 before May 15 12:54:03 and no parallel make May 15 12:54:47 hm and when you try to compile gtk+ now? May 15 12:55:00 http://pastebin.com/BYz4VhCQ May 15 12:55:18 gtk+ or gtk+-native? May 15 12:57:04 gtk+-native May 15 12:57:25 ok May 15 12:57:57 ok I do that: May 15 12:57:58 bitbake -c clean gtk+-native && bitbake gtk+-native May 15 13:17:19 still May 15 13:17:24 /im-xim.la does not export GTK+ IM module API: /home/embedded/oetmp_htcdream/work/i686-linux/gtk+-native-2.20.0-r8.1/gtk+-2.20.0/modules/input/.libs/im-xim.so: undefined symbol: g_realloc_n May 15 13:17:30 mmm May 15 13:30:11 lol just tried to build gcc-gross May 15 13:31:00 mornign woglinde May 15 13:31:19 woglinde, do I have to set gcc-cross-initial and gcc-cross-intermediate to the same version as gcc-cross ? May 15 13:48:45 dcordes yes May 15 13:49:05 dcordes it's not 100% nesccary but better May 15 13:52:56 next cleanup May 15 13:53:06 03Henning Heinold  07org.openembedded.dev * r3f052215b0 10openembedded.git/recipes/jamvm/jamvm-initial_1.5.0.bb: jamvm-initial: use new staging system May 15 14:07:41 dev built failed : http://pastebin.com/WH1Sy9FC "Can't find recent OpenSSL libcrypto" May 15 14:07:45 anyone seen this? May 15 14:10:24 try bitbake openssl May 15 14:10:26 before May 15 14:13:01 woglinde, what should I try then? May 15 14:13:45 woglinde, trying...made sure libssl-dev is on lucid May 15 14:15:14 gntuoo sorry dont know May 15 14:17:17 openssl_0.9.8m.bb failed May 15 14:18:03 i did 'bitbake -c clean openssl && bitbake openssl', thought it's built as i failed on openssh which should be _after_ openssl May 15 14:26:30 http://pastebin.com/yAHjJxZA openssl fail on dev May 15 14:26:55 bitbake -c clean May 15 14:27:08 bitbake -c clean openssl May 15 14:33:16 ok np and thanks a lot May 15 14:46:32 woglinde, thanks. baking May 15 14:57:15 same error after bitbake -c clean && bitbake -c clean openssl && bitbake openssl May 15 14:57:34 libcrypto.so: file not recognized: file truncated May 15 14:59:40 are you using ccache? May 15 14:59:49 make the cache clean May 15 15:00:24 yes I am May 15 15:00:25 how? May 15 15:00:38 GNUtoo I meant xxiao May 15 15:00:42 ok May 15 15:05:37 woglinde, thanks. did ccache -C now retrying May 15 15:18:53 woglinde_, sigh, same error May 15 16:16:44 GNUtoo|oeee: can you see wich deps have not been built? May 15 16:16:58 how? May 15 16:17:08 look in workdir May 15 16:17:15 bitbake package-index? May 15 16:17:42 no, in /home/embedded/oetmp_htcdream/work/i686-linux May 15 16:17:57 ok May 15 16:17:58 see e.g atk-native, pango-native May 15 16:18:19 jo ant May 15 16:18:27 hello May 15 16:18:56 xxiao ubuntu lucid? May 15 16:19:05 hm but that shouldnt matter May 15 16:19:08 its arm May 15 16:19:24 hm maybee same problem as gnutoo May 15 16:19:30 ok atk-native and pango native have log.do_sysroot_stage May 15 16:19:42 hm no May 15 16:19:45 hm hm May 15 16:19:46 they should only have /tmp and the logs May 15 16:19:51 I've gentoo btw,old gentoo May 15 16:20:01 Gentoo 32bit here May 15 16:20:45 image pango-1.24.4 staging-pkg sysroot-destdir temp May 15 16:20:58 32bit also May 15 16:21:09 is not finished then May 15 16:21:17 ah ok May 15 16:21:21 bitbake package-index? May 15 16:21:22 should only have /temp inside May 15 16:21:33 I do not have rm_work May 15 16:21:38 each package should clean workkdir with rm_work May 15 16:21:41 ahhh May 15 16:21:45 soory :) May 15 16:21:54 I keep sources for more debugability May 15 16:21:59 and also for gdb May 15 16:22:16 ok, for mer is immediately visible if has completed May 15 16:22:18 xxiao maybee its easier when you build from scratch May 15 16:22:29 xxaio or remove the whole ccache May 15 16:22:31 dir May 15 16:22:49 GNUtoo|oeee: and I was building in tmpfs..ram is not unlimited May 15 16:23:22 ok May 15 16:23:35 woglinde_, that's what i'm doing, still wip May 15 16:23:41 personally I just rebuit from scratch May 15 16:23:43 GNUtoo|oeee: while you're there, would you check for duplicated tasks (e.g. fetch) May 15 16:23:51 ok May 15 16:24:21 I always have two run.base_do_fetch (same creation time) May 15 16:24:39 really oe and openwrt are the two projects pushing me to buy the highest end pc :) May 15 16:24:40 and two run.do_fetch May 15 16:24:45 it seem that I have some duplicated May 15 16:24:51 these are equals May 15 16:24:57 need build them fast, core2duo is still too slow with -j4 May 15 16:24:58 zlib-native: May 15 16:25:32 run.staging_helper.3753 run.staging_helper.3753 May 15 16:25:35 GNUtoo|oeee: but..run.do_staging_helper differs each other May 15 16:25:49 ah? I'll look May 15 16:26:26 possibly one is for install and one for packaged staging May 15 16:27:17 ok May 15 16:27:35 indeed May 15 16:27:38 ok apart that no May 15 16:27:45 no duplicated tasks May 15 16:27:50 yep, one is for task-setscene and the other is for task-package_stage May 15 16:28:06 (still talking about run.staging_helper May 15 16:28:09 ) May 15 16:28:24 ok May 15 16:28:28 ant__ its maybee the same problem with gtk+native builds gtk+ May 15 16:30:32 woglinde_: if you look at it, gtk+è-native does not depend directly on glib-2.0-native (iirc ;) May 15 16:31:36 well, the sources do depend, the recipe not :D May 15 16:31:54 I added the depend locally May 15 16:32:05 I tried too, no help :/ May 15 16:32:10 when I had the issue May 15 16:32:11 ok May 15 16:32:31 but I was with 4 threads May 15 16:33:04 when I added glib-2.0-native to depends of the native recipe May 15 16:34:27 I've a pentium M 2GHz May 15 16:34:34 so only default settings May 15 16:35:24 It really sounds like a bitbake bug May 15 16:35:31 Do a -e and see what's listed in DEPENDS May 15 16:35:38 It's stuff that you said was still building May 15 16:36:59 ok May 15 16:40:16 woglinde_, with a from-scratch build openssl is built without errors, i did not have libssl-dev on lucid first and the only way to fix that is to build oe from scratch. thanks May 15 16:45:21 xxiao??????? May 15 16:45:40 xxiao if you nedd libssl-dev, then oe has failure May 15 16:45:49 we normaly dont need host stuff May 15 16:45:55 http://pastebin.com/jACVCXrp May 15 16:46:00 besides the common tools like find or grep May 15 16:57:23 woglinde_: btw someone should add that we need chrpath May 15 16:57:36 since last changes May 15 16:58:32 bbl May 15 16:59:25 woglinde_, it seems i do need it May 15 17:00:41 i'm build base-image now, see if it can go through, originally it failed at openssh->openssl->libssl-dev May 15 17:06:12 anyone here know what the recommended way to capture an image from a webcam is? May 15 17:06:18 this is for image proicessing May 15 17:06:23 the current method I am looking at is cvcam May 15 17:07:24 for instance, is there a way to capture a frame reight on the dsp instead of donig it on the gpp and sending it over? May 15 17:08:15 aditya_111 hm what camera you have? May 15 17:08:30 I have a lifecam cinema from MS May 15 17:08:47 it works fine with mplayer so atleast I know the current kernel/uvc can access the came fine May 15 17:08:49 cam* May 15 17:09:11 using v4l2 as the driver May 15 17:09:17 dunno if thats useful to know May 15 17:09:34 yes May 15 17:09:50 cvcam seems nice, but I'm thinking it'd be even better if my task on the DSP was directly accessing the cam and just pumping out processed info to the GPP at a regular interval May 15 17:10:04 hm May 15 17:10:16 than you need a proxy May 15 17:10:36 which fetches stuff from cam and sends to dsp May 15 17:10:44 dsp on omap3530? May 15 17:10:49 so essentially what I am planning now yeah? May 15 17:10:57 yes, beagleboard/omap May 15 17:11:03 hehe May 15 17:11:09 like the proxy is still in gpp space right? May 15 17:11:15 or is this somethign better than just running a gpp prob May 15 17:11:17 prog* May 15 17:11:19 why? May 15 17:11:29 < doesn't know what a proxy is May 15 17:11:42 hm with gpp you mean maincpu? May 15 17:11:49 yes May 15 17:11:55 hm okay May 15 17:11:57 gpp-->the arm itself in this case i assume May 15 17:12:04 yeah May 15 17:12:11 I've just been using gpp/dsp as the terms May 15 17:12:18 is usb the interface May 15 17:12:18 cause thats how the code is organized in dsplink May 15 17:12:19 the dsp cant do it directly May 15 17:12:28 yesh its a usb webcam May 15 17:12:47 is there anything specific I can search for for these proxies? May 15 17:12:53 normally dsplink will do better if it's using camera-sensor-interface or omap's own video interface May 15 17:12:54 hm May 15 17:13:05 for usb you need use gpp's usb stack May 15 17:13:31 xxiao:are these special hardware interfaces? May 15 17:13:31 hm maybee I am not so firm with dsplink yet May 15 17:13:41 aditya yes May 15 17:13:44 aditya_111, yes May 15 17:14:14 "camera-sensor-interface" -- what they are called? May 15 17:14:38 CSI, i think omap has one, dm365 has one as well May 15 17:14:41 I dont really mind changing the hardware interface if its possible, but USB webcam is nice cause the overall platform is cheaper May 15 17:15:06 guess I should probably try just webcam->gpp->dsp May 15 17:15:12 depends on your app, usb webcam has latency issues May 15 17:15:13 and see if it will work May 15 17:15:35 yeah, I am trying to get around 10-15 frames including processing May 15 17:15:43 hmm May 15 17:15:53 so if I did do it through the usb stack May 15 17:16:03 would cvcam be my best option? May 15 17:16:07 or is there something more basic? May 15 17:16:19 I basically just need a frame dump, nothing image processing-y yet May 15 17:20:44 never used cvcam, but if you need opencv library for that...remember opencv is opimitized for x86 only May 15 17:21:31 so what should I do now after bitbake -e? May 15 17:23:07 03Khem Raj  07org.openembedded.dev * rf75f6d005a 10openembedded.git/ (4 files in 4 dirs): May 15 17:23:07 qemuppc: Add new machine qemuppc running ppc on QEMU. May 15 17:23:07 * Tested with DISTRO=minimal image=minimal-image May 15 17:23:07 * Adjust run-qemu.sh to use native-qemu and fix May 15 17:23:07 variables needed for powerpc run. May 15 17:23:07 Signed-off-by: Khem Raj May 15 17:26:58 03Khem Raj  07org.openembedded.dev * r30063031f3 10openembedded.git/recipes/vlc/x264_r2245.bb: May 15 17:26:58 x264_r2245.bb: Set X264_ECFLAGS="-fPIC" for mips mips May 15 17:26:58 Signed-off-by: Khem Raj May 15 17:27:00 taditya_111 hm maybee you can look at gstreamer stuff May 15 17:27:12 and building a pipe for your stuff May 15 17:28:06 pipeline even May 15 17:28:21 gstreamer modules for accessing usb cam should be there May 15 17:31:33 hmm May 15 17:31:59 xxiao: they have a version for oe so was hoping that'd be enough May 15 17:32:20 woglinde_: I always thought gstreamer was like.. a media player May 15 17:32:34 look at gstream plugin v4l2src May 15 17:32:52 aditya_111, TI is behind gstream, better leverage it May 15 17:33:22 it's not just a media player, a multimedia framework that can do v4l2 May 15 17:42:24 aditya_111,what do you need for gstreamer May 15 17:42:41 aditya_111, I got it working on my bug devices and streamed v4l2 over wifi May 15 17:43:14 GNUtoo|oeee: I am trying to get quick access to images from the webcam May 15 17:43:21 ah ok May 15 17:43:25 so I can perform some iamge processing for tracking etc for a small robot May 15 17:43:31 ok May 15 17:43:44 why not using screenshot from mplayer then May 15 17:43:48 maybe it would be easier May 15 17:43:50 so what you are saying is that I can bug you later if I get stuck :P May 15 17:44:18 I dont really want to have to run mplayer along with my app if I dont have to do that May 15 17:44:26 aditya_111, also but what I'm saying is that I made this wiki page: http://bugcommunity.com/wiki/index.php/Video May 15 17:45:01 aditya_111, ah? May 15 17:45:04 mmm May 15 17:45:06 let me think May 15 17:45:15 hmm May 15 17:45:29 those calls look like they are meant to save to a file/network stream May 15 17:45:35 yes May 15 17:45:39 is there any way to directly put stuff into memory? May 15 17:45:43 mmm May 15 17:45:53 like a gstreamer_capture(camera_num) May 15 17:45:57 maybe ask in #gstreamer May 15 17:45:57 something along those lines May 15 17:46:00 ok May 15 17:46:22 because I know how to save to file using gst-launch May 15 17:46:32 but I didn't write a program that use their api May 15 17:59:05 hi just to check, do we have access with floss to the video dsp of the S3C6410 soc? May 15 18:00:36 I... have no idea lol May 15 18:11:57 ok May 15 18:12:02 asking in #mer May 15 19:56:17 Hi, guys May 15 19:56:37 hi igork May 15 19:56:41 After much struggle, I found the porblem with my program crashing May 15 19:56:48 ah May 15 19:57:08 the problem is that for some reason wxwidgets compiled in non-GUI mode May 15 19:57:16 lol May 15 19:57:17 okay May 15 19:57:46 even though configure correctly identifies the GUI presence and defines allneccessary variables May 15 19:58:00 still Makefile generated is wrong May 15 19:58:43 I can re-post the pastebin' log if someone will wants to take a look May 15 19:58:55 s/log/logs May 15 19:59:30 explicitely asking to compile a GUI library didn't help at all May 15 19:59:58 any ideas/suggestions/workarounds? May 15 20:00:44 I am using the Angstrom kernel along with X11 and GPE environment May 15 20:01:11 no May 15 20:01:19 seems an wxwidgets problem May 15 20:01:37 arm-angstrom-linux-gnueabi-g++ (GCC) 4.2.4 May 15 20:02:56 well one question: who is responcible for writing Makefile after running configure? May 15 20:09:51 khem: Good job tracking down a fix for ppc stuff, doing a loop build now to confirm it everywhere May 15 20:10:14 since I have a failing one for everything, should be semi quick (everything is there except qemuppc and calamari, my usual e500v2 target) May 15 20:15:11 ericben, hi welcome aboard May 15 21:12:18 no news on g_realloc? May 15 21:12:45 no May 15 21:12:49 ok May 15 21:12:55 should I write a mail about it? May 15 21:13:19 because every other person has the issue with multiple threads etc... May 15 21:13:26 and can pass the issue bitbaking glib-2.0 May 15 21:14:03 GNUtoo: Try bitbake 1.8.x and see if that's still working May 15 21:14:13 ok May 15 21:14:16 This really does feel like a bitbake issue, since DEPENDS looks right May 15 21:14:58 I've 1.8 May 15 21:15:10 Oh, it was someone else that was using 1.10? May 15 21:15:23 I'm at d99de1c3f35a01f6d06b2965da823ae3cd16d0eb May 15 21:15:38 is 1.10 risky ? should I try it? May 15 21:15:50 Is that hash 1.8 or 1.10? May 15 21:16:11 1.8 May 15 21:16:39 k May 15 21:16:46 Try like 1.8.18 release or something then May 15 21:19:57 I'm at 1.18 May 15 21:20:07 I'm at lastest commit in bitbake 1.8 May 15 21:20:20 that is one commit after 1.18 release May 15 21:20:26 which is " Increment version post release" May 15 21:23:00 k May 15 21:23:07 Then, yeah, not sure what's going on May 15 21:23:18 But, like I said, bitbake -e has the DEPENDS right May 15 21:23:23 And they aren't being obeyd May 15 21:24:02 ok,thanks a lot May 15 21:24:14 but strange thing May 15 21:24:33 even bitbaking glib-2.0-native then bitbaking gtk+-native fails May 15 21:24:53 and I rebuilt from scratch not so long ago May 15 21:24:58 because the compiler was broken May 15 21:25:06 s/compiler/toolchain May 15 21:31:39 Yeah, there's other deps for gtk+-native May 15 21:31:43 panto/cairo/atk May 15 21:31:47 pango May 15 21:32:06 ok thanks May 15 21:36:45 bye and thanks a lot May 15 21:43:52 we're trying to get a USB keyboard working on an embedded box (typical MIPS raw flash-only kinda uclinux+busybox build... May 15 21:44:51 we have enough drivers loaded so /dev/event## node that barfs out raw PC scancodes... how do we connect that to get a tty/login kinda driver? May 15 21:50:15 peerce do you have a monitor for it? May 15 21:52:19 right now we're on a serial console, we got ethernett (ssh etc). but we're just trying to get the keyboard working as an ascii input device May 15 21:53:44 peerce hm May 15 21:53:50 sorry than I cant help May 15 21:54:38 we hvae an fb device too, that eventually we'd like to hook up as an output display but right now, its just pixels and lines, not 'tty out'... May 15 21:55:53 hi May 15 21:56:02 re dcordes May 15 21:56:32 woglinde, hey. I ran into some error building with gcc-cross*_4.4.4 May 15 21:57:04 woglinde, wondering if it makes sense to solve it or file a bug May 15 21:57:13 sure May 15 21:57:18 write to ml May 15 21:57:22 or ask khem here May 15 21:57:35 khem, ping :) May 15 21:58:34 woglinde, I am wondering what selects gcc-cross version May 15 21:58:53 dcordes distro May 15 21:59:30 angstroem choose 4.3.3 May 15 21:59:34 I'm using angstrom-2008.1 May 15 21:59:46 why does it not chose a more recent version? May 15 22:00:04 ask in #angstroem May 15 22:00:05 dcordes: hi May 15 22:00:15 dcordes: what error do you see May 15 22:00:18 topic hint May 15 22:00:30 jo khem May 15 22:00:32 * khem is back from healthy lunch May 15 22:00:36 woglinde: wie gehts May 15 22:00:50 khem ganz gut May 15 22:01:36 khem, wait a sec. I will dig out the log May 15 22:01:46 * dcordes could use some unhealthy dinner May 15 22:02:00 dcordes arent you using tinderbox? May 15 22:02:17 hm no May 15 22:02:24 put May 15 22:02:25 INHERIT += "oestats-client" May 15 22:02:25 OESTATS_SERVER = "tinderbox.openembedded.net" May 15 22:02:25 OESTATS_BUILDER = "woglinde" May 15 22:02:29 woglinde: ich muss mein duetsch ueben :) May 15 22:02:41 hm OESTATS_BUILDER = "dcordes" May 15 22:02:45 into your local.conf May 15 22:02:48 ok May 15 22:02:59 and tinderbox hosting the logs for you May 15 22:03:04 when I run into the same error it will upload it? May 15 22:03:06 khem warum? May 15 22:03:07 dcordes: and which gcc version is it ? May 15 22:03:08 awesome May 15 22:03:17 khem, I tried 4.4.4 May 15 22:03:37 dcordes yes indeed it's awesome May 15 22:04:14 hm I wonder why some fixes ipkg staging May 15 22:04:21 args why/when May 15 22:04:25 woglinde: so dass ich es nicht vergessen :) May 15 22:04:38 khem sieht noch gut aus May 15 22:05:44 woglinde: ich hatte ein duetsch kollege aber er ist nicht mehr hier May 15 22:06:36 khem yes you already told me May 15 22:10:45 are there any other embeded linux channels? May 15 22:11:02 peerce: there are many depends what u need May 15 22:11:13 woglinde: I got qemuppc working finally May 15 22:11:21 woglinde: next I am trying SH4 May 15 22:12:05 trying to get usb keyboard and fb tty working on a really stripped embeeded 2.6 system, finding nothing but really old and wrong advise on the web :( May 15 22:12:23 embedded ... wtf is that ee coming from? :D May 15 22:13:40 is it bad to 'spam' tinderbox with accidentl malconfigurations etc etc? May 15 22:14:17 dcordes: should be bearable to a certain extent May 15 22:14:55 ok I will just turn it off soon as I start debugging or stuff May 15 22:21:22 this kind of stuff is really starting to annoy me. the 'official' HOWTO on linux framebuffer, http://tldp.org/HOWTO/Framebuffer-HOWTO.html ... and its 10 freekin years old and talking about 2.0 and 2.2 stuff May 15 22:21:44 peerce, I've noticed this too when I was wondering about msm_fb May 15 22:22:45 ok I started 3 runs now accidently. only one of them is valid to the bug it's supposed to report. hope this will work May 15 22:22:47 bbl May 15 22:24:16 on PC's, the keyboards and framebuffers just work, because the hal or udev or whatever understands it all. but on an embedded chip, I've never figured out how to configure, how the pieces fit together. May 15 22:27:37 ooh, we just found the option for the framebuffer console driver, so building a new kernel now May 15 22:28:07 peerce *g* May 15 22:36:41 well, gotta run, my friend is gonna keep on hacking. May 15 23:01:43 woglinde: .."LogFS had finally made it into the kernel in 2.6.34..." May 15 23:01:48 The name LogFS was chosen for several reasons: May 15 23:01:48 * JFFS2, the Journaling flash filesystem 2 is actually a log-structured filesystem. LogFS is a journaling filesystem, so its name has to hide this fact to continue a tradition. May 15 23:01:53 see the rest..lol May 15 23:02:10 http://logfs.org/logfs/ May 15 23:17:08 Hi, guys May 15 23:17:55 Is it possible to re-create a bitbake environment and run make? May 15 23:18:49 ??? May 15 23:18:51 IgorK, bitbake -c do_devshell May 15 23:21:07 s/do_// May 15 23:21:14 and INHERIT += "devshell" May 15 23:21:21 oops May 15 23:21:25 :) May 15 23:21:48 IgorK, bitbake -c devshell May 15 23:31:08 dcordes: so this command will execute make for the build the >packagename>, right? May 15 23:32:53 IgorK, http://wiki.openembedded.net/index.php/Devshell May 15 23:36:43 almost 800 tasks finished successfully with gcc-cross*_4.4.4 May 15 23:57:55 03Henning Heinold  07org.openembedded.dev * rfe9ca1480c 10openembedded.git/recipes/ecj/ecj-initial.bb: ecj-initial: convert to new staging system May 15 23:57:56 03Henning Heinold  07org.openembedded.dev * r89187f3050 10openembedded.git/recipes/xalan-j/ (xalan-j-native_2.7.1.bb xalan-j_2.7.1.bb): May 15 23:57:56 xalan: convert to new staging system May 15 23:57:56 * remove some linkingcruft which is now in jaxp recipe May 15 23:57:58 03Henning Heinold  07org.openembedded.dev * r47f8089db4 10openembedded.git/recipes/classpathx/gnujaf_1.1.1.bb: May 15 23:57:58 gnujaf: remove do_install method to fix native package May 15 23:57:58 * remove extra depenedency for fastjar-native, because May 15 23:57:58 the java classes are doing it for us May 15 23:57:58 03Henning Heinold  07org.openembedded.dev * r9f59941ae4 10openembedded.git/recipes/ant/ant-native_1.7.1.bb: May 15 23:57:58 ant-native: convert to new staging system May 15 23:57:59 * bump PR to cover same strangness May 15 23:58:00 03Henning Heinold  07org.openembedded.dev * rf458275fc3 10openembedded.git/classes/java-native.bbclass: java-native.bbclass: remove do_stage function, because of NATIVE_INSTALL_WORKS = "1" takes care May 15 23:58:01 03Henning Heinold  07org.openembedded.dev * r3da1686952 10openembedded.git/recipes/zip/ (zip-native_2.32.bb zip.inc): zip: use BBCLASSEXTEND = "native" for native recipe May 15 23:58:03 03Henning Heinold  07org.openembedded.dev * ra18801ce4f 10openembedded.git/recipes/jikes/ (jikes-native_1.21.bb jikes_1.21.bb jikes_1.22.bb): May 15 23:58:03 jikes: use BBCLASSEXTEND = "native" May 15 23:58:03 * removed native recipe May 15 23:58:03 * version 1.22 is now default May 15 23:58:04 03Henning Heinold  07org.openembedded.dev * r33d7328f3c 10openembedded.git/recipes/classpath/classpath-native.inc: classpath-native.inc: convert to new staging system May 15 23:58:04 03Henning Heinold  07org.openembedded.dev * r0664036a82 10openembedded.git/recipes/ecj/ecj-bootstrap-native.bb: ecj-bootstrap-native: convert to new staging system May 15 23:58:05 03Henning Heinold  07org.openembedded.dev * rb273ab953b 10openembedded.git/recipes/ecj/libecj-bootstrap.inc: libecj-bootstrap.inc: convert to new staging system May 16 00:13:35 http://tinderbox.openembedded.net/packages/572880/ Package util-linux-ng - 2.17 - r23.1 fails with gcc-cross_4.4.4 May 16 00:13:59 works in same env with default angstrom-2008.1 toolchain May 16 00:18:53 khem, ping May 16 00:30:59 so is the tinderbox enough or should I post it to bugzilla? May 16 00:38:48 dcordes: where is link to tinderbox log May 16 00:39:00 ah i see May 16 00:39:47 dcordes: whats your binutils version May 16 00:42:48 khem, what's the correct way to find out? May 16 00:44:54 dcordes: $OETMP/cross/arm.../bin/-ld -b May 16 00:44:56 -v May 16 00:45:27 if it is older than 2.20 then please choose a newer binutils May 16 00:46:22 GNU ld (Linux/GNU Binutils) 2.18.50.0.7.20080502 May 16 00:46:36 I told you what to do May 16 00:47:05 like you selected newer compiler select newer binutils too May 16 00:47:23 anything else to take care of? May 16 00:47:28 dcordes: command "bitbake -c devshell wxwidgets" failed May 16 00:47:35 should I rebuild entirely or are the succeeded packages fine? May 16 00:48:33 IgorK, post the error log else nobody can help May 16 00:49:39 Here it is: http://pastebin.com/x6FcbTHx May 16 00:50:07 I'm running KDE-3.5.10 on Gentoo Linux May 16 00:50:16 /home/arm/OE/temp/work/armv4t-angstrom-linux-gnueabi/wxwidgets-2.8.9-r7/temp/run.do_devshell.28358: line 440: gnome-terminal: command not found May 16 00:50:19 dcordes: just build binutils-cross May 16 00:50:32 and then clean up util-linux-ng May 16 00:51:02 it should compile all May 16 00:51:25 then if you not lucky because you mixed binutils then you have no choice but nuke oetmp and start fresh May 16 00:51:53 I will just rm -rf right away May 16 00:52:58 dcordes: thats better but you have to rebuild all thats kind of sucks May 16 00:54:29 PREFERRED_VERSION_gcc-cross = "4.4.4" May 16 00:54:29 PREFERRED_VERSION_gcc-cross-initial = "4.4.4" May 16 00:54:29 PREFERRED_VERSION_gcc-cross-intermediate = "4.4.4" May 16 00:54:29 PREFERRED_VERSION_binutils-cross = "2.20" May 16 00:54:59 dcordes: good. You can even chose 2.20.1 May 16 00:55:27 dcordes: if you want you can also choose DISTRO=minimal and you dont need all that May 16 00:55:34 its default for minimal distro May 16 00:55:50 oh. any downsides vs angstrom-2008.1 ? May 16 00:56:02 what is your end goal May 16 00:56:24 make the machine work I'm building for. htcleo May 16 00:56:50 minimal should work fine May 16 00:59:04 require conf/distro/include/sane-toolchain.inc May 16 00:59:36 # Some systems need a special gcc version May 16 00:59:40 PREFERRED_GCC_VERSION_armv7a ?= "4.4.1" May 16 00:59:49 ugh May 16 01:00:00 that is my problem May 16 01:00:05 why I want new toolchain. May 16 01:00:09 hmmm ok you can bump that up May 16 01:00:19 for now May 16 01:00:25 and if it works then report back May 16 01:00:30 and we can bump it up May 16 01:00:34 how about sticking with angstrom if I need to set it manually anyway? May 16 01:00:43 your choice May 16 01:00:44 dcordes: how do I work around that problem? May 16 01:01:04 IgorK, install gnome-terminal :D or find a way to set proper terminal you have installed May 16 01:01:41 khem, ok then I stick with angstrom. but is there anything you can think of I have to add additional to the 4 lines I pasted? May 16 01:02:06 inside my gentoo environment? May 16 01:02:10 yes May 16 01:02:41 dcordes: as I said it depends on your needs May 16 01:02:55 if you want more packages then you need to customize May 16 01:03:08 OE is a mothership it does not hold you back May 16 01:04:23 fine I will just try with gcc-cross*_4.4.4 and binutils-cross_2.20.1 May 16 01:04:41 ok May 16 01:04:49 03Khem Raj  07org.openembedded.dev * rfbc6721327 10openembedded.git/ (4 files in 4 dirs): May 16 01:04:49 machine/qemush4: Add new machine for SH4 on qemu May 16 01:04:49 * Make improvements to run-qemu script. May 16 01:04:49 Signed-off-by: Khem Raj May 16 01:05:42 03Henning Heinold  07org.openembedded.dev * r8c094783fb 10openembedded.git/recipes/giflib/giflib_4.1.6.bb: giflib: add dependency to libsm from xorg-libs May 16 01:07:10 * dcordes lights candle for tmpdir May 16 01:09:03 Is there any way to run this under KDE? May 16 01:09:25 khem, thanks a lot for the assistance. I hope I can help making the qsd machines work May 16 01:09:52 IgorK: set TERMCMD = ${XTERM_TERMCMD} in your local.conf May 16 01:10:06 dcordes: oh sure. OE just works May 16 01:10:21 its fastest when it comes to new machine bringup May 16 01:10:23 :) May 16 01:10:30 you just have to get kernel booting May 16 01:10:31 yes I love it May 16 01:10:41 everything else will fall into places May 16 01:11:31 IgorK: and also INHERIT += "devshell" if you havent done already May 16 01:14:52 khem: "TERMCMD = ${XTERM_TERMCMD}", right? no set? May 16 01:17:41 yeah May 16 01:17:59 khem: btw, I think the ppc405 override to 4.4.1 is wrong, i'll post something soon May 16 01:18:06 almost done w/ my loop, stuff is otherwise looking decent May 16 01:18:32 Tartarus: hey I committed a fix to straighten out -Os May 16 01:18:36 in gcc 4.4 May 16 01:18:43 yeah, saw, testing it everywhere now May 16 01:18:46 thanks & good job May 16 01:18:50 it builds all hunky dory May 16 01:19:04 but but it does not boot the userspace :( May 16 01:19:07 what'd you test? May 16 01:19:08 on my qemu system May 16 01:19:19 I committed a machine for ppc on qemu :) May 16 01:19:24 i saw May 16 01:19:27 it looks mostly right :) May 16 01:19:42 can you try to boot a real ppc target ? May 16 01:19:47 been a while since i cared about pmac and if qemu emulated one well :) May 16 01:20:12 yeah I committed sh4 as well but it needs qemu 12.4 + sh fixes May 16 01:20:13 Depends on how busy we are w/ testing product on it :) May 16 01:20:16 which I have locally May 16 01:20:21 I am going to propose those May 16 01:20:45 people like me who dont have much hardware are served well. May 16 01:20:55 now I can test almost all what I wanted May 16 01:20:58 if it works ;) May 16 01:21:18 afk a while May 16 01:21:19 it boots without the Os patch and if I set FULL_OPTS to -O2 May 16 01:22:09 Tartarus: Let me know if you are able to boot a ppc device with this Os fix May 16 01:26:57 wondering if anybody else on here has a qualcomm QSD8* based device. like e.g. htc passion (nexus one), htc leo (hd2), toshiba tusnagi tg01 or htc incredible (verizon droid incredible) May 16 01:27:34 good nite May 16 01:59:51 03Khem Raj  07org.openembedded.dev * r2f2a374e3a 10openembedded.git/recipes/linux/ (linux-2.6.33/qemumips/defconfig linux_2.6.33.bb): May 16 01:59:51 linux-2.6.33: Make 2.6.33 default kernel for qemumips. May 16 01:59:51 Signed-off-by: Khem Raj **** ENDING LOGGING AT Sun May 16 02:59:57 2010