**** BEGIN LOGGING AT Tue Apr 19 02:59:57 2011 Apr 19 03:44:36 Is there something a COMPATIBLE_ARCH? Apr 19 03:44:45 I know that there is COMPATIBLE_MACHINE Apr 19 03:45:17 but what if there is software that depends on Cortex A8? Apr 19 05:01:49 SidH_: COMPATIBLE_SOC Apr 19 05:05:41 early JaMa Apr 19 05:09:52 moin :) Apr 19 05:10:08 7am here already.. time to go to work.. Apr 19 05:11:18 wow..time flies when you are having "fun"...I'm still working Apr 19 05:13:49 JaMa: Thanks Apr 19 06:28:00 hey guys I have a newb bitbake problem: I have a recipe that is failing do_fetch by saying "cannot check archive integrity" Apr 19 06:28:18 I can extract it manualy just fine Apr 19 06:28:25 what check is bitbake doing? Apr 19 06:38:00 Nedlinpopo: it checks md5sum/sha1sums of the recipes (AFAIK) Apr 19 06:38:23 Nedlinpopo: or not the recipe but the tarballs (or whatever) mentioned in it recipe Apr 19 06:38:42 yeah, I tried adding that and it stopped complaining about the missing checksums but still said integrity check failed Apr 19 06:38:47 Nedlinpopo: the bitbake output should tell you which recipe is failing Apr 19 06:38:48 :/ Apr 19 06:39:09 Nedlinpopo: yeah, you should not remove the checksums ... Apr 19 06:39:12 it's the libdc1394 recipe from pixhawk Apr 19 06:39:22 i tried it with and without the checksums Apr 19 06:39:28 Nedlinpopo: if there are no checksums it will tell you the one it found out Apr 19 06:39:40 that's the one I added Apr 19 06:39:46 but it still failed to unpack Apr 19 06:39:53 Nedlinpopo: there are two checksums per file Apr 19 06:40:00 a sha1sum and a md5sum Apr 19 06:40:04 do you need both? Apr 19 06:40:08 make sure you have both Apr 19 06:40:09 yes Apr 19 06:40:15 oh, weird. Apr 19 06:40:30 however, think of what it means that your checksums dont match Apr 19 06:40:36 that's probably what went wrong, I only added the sha256sum Apr 19 06:40:49 they never existed in the first place Apr 19 06:40:52 Nedlinpopo: somehow your tarball is not the same one that the one found out who contributed the recipe Apr 19 06:41:06 i did download it on a second computer and verify the checksum is the same in both places Apr 19 06:41:28 the contributed recipe *never* had one, I'm the first to add it. Apr 19 06:41:41 Nedlinpopo: ok,that explains it Apr 19 06:42:11 Nedlinpopo: you are required to have both checksums in it.otherwise bitbake will complain. thats it Apr 19 06:42:34 * rschus wonders whether the message bitbake gives about that is not clear enough Apr 19 06:42:34 sure thing Apr 19 06:43:16 weirdly if you have one checksum, but not both, the missing checsum error message is not displayed, hence hte confusion Apr 19 06:43:30 ok, got it Apr 19 06:43:55 what is the point of having both checksums? Apr 19 06:44:04 the sha256 is much stronger Apr 19 06:44:48 Nedlinpopo: somehow I guess it is historic Apr 19 06:45:13 strange Apr 19 06:45:38 Nedlinpopo: some projects use sha256 others md5 on their website/release files that you can use to countercheck the values you have Apr 19 06:47:04 I'm aware of that. I'm just confused as to why bitbake would check *both*, since most repositories only publish one Apr 19 06:48:22 and since sha256 is cryptographically *much* stronger, I would be much mor einclined to trust it when ever I had both Apr 19 08:33:52 Hi guys Apr 19 08:34:51 About dbus support...which recipe should be in charge of creating the startup script for the initial runlevels in yocto? Apr 19 08:35:12 In my image I have /etc/init.d/dbus-1 Apr 19 08:35:35 but no one is starting it at boot Apr 19 08:36:17 I have to start it manually Apr 19 08:44:37 mrAlmond, oe-core/meta/recipes-core/dbus/dbus.inc sets them; so dbus-1.4.1 should be fine Apr 19 08:49:13 mlip: thank you for the suggestion Apr 19 08:49:37 At the moment I'm using yocto 1.0 Apr 19 08:54:07 mlip : I'm looking at files under meta/recipes-core/dbus/ but there are no explicit calls to "ln" to make links Apr 19 08:56:11 grep "INIT" meta/recipes-core/dbus/dbus.inc Apr 19 08:56:19 INITSCRIPT_NAME = "dbus-1" Apr 19 08:56:19 INITSCRIPT_PARAMS = "start 02 5 3 2 . stop 20 0 1 6 ." Apr 19 08:56:21 grep update-rc.d Apr 19 08:56:42 ook Apr 19 08:56:45 understood Apr 19 08:56:47 tnx! Apr 19 08:57:50 ok..probably the problem is that I'm using dbus_1.2.24 from meta-openembedded Apr 19 09:18:31 btw...now I've set PREFERRED_VERSION_dbus = "1.4.1" in my local.conf to choose that version of dbus Apr 19 09:18:43 and I've compiled it by issuing : bitbake dbus Apr 19 09:18:52 and it correctly compiled the 1.4.1 version Apr 19 09:19:05 but when I go with the image compilation Apr 19 09:19:10 it gives me : Apr 19 09:19:42 Cannot satisfy the following dependencies for task-poky-apps-console: * dbus * Apr 19 09:20:01 like dbus was not available Apr 19 09:24:56 he stefan Apr 19 09:26:31 hi woglinde Apr 19 09:26:44 woglinde: great results in openjdk testing with your patches Apr 19 09:26:52 woha Apr 19 09:26:53 fine Apr 19 09:26:56 woglinde: 8 out of 8 worked without problems Apr 19 09:27:00 cool Apr 19 09:27:10 2 failed due to other problems unrelated to openjdk Apr 19 09:27:15 I just started another run Apr 19 09:27:18 *g* Apr 19 09:27:24 good work :) Apr 19 09:27:47 if nothing else pops up I will prepare a pull requets with these commist for 2011.03 later this week Apr 19 09:30:34 someone saw Robert recently? Apr 19 09:32:30 hrw schuster? Apr 19 09:33:28 yes Apr 19 09:33:38 woglinde: was this for members? http://lists.linuxtogo.org/pipermail/openembedded-members/2011-April/000714.html Apr 19 09:33:58 hm funny seems florians mail for linuxtag got lost somewhere Apr 19 09:34:06 and arrived today Apr 19 09:34:37 ups Apr 19 09:34:41 what the hell Apr 19 09:34:43 damn Apr 19 09:34:52 I catched the wrong list Apr 19 09:34:59 thanks jama Apr 19 09:35:13 yw Apr 19 09:36:35 hello Apr 19 09:36:45 anyway I hit segfault while compiling the orc examples Apr 19 09:36:54 but from the changelog that is fixed too Apr 19 09:40:14 I m building a files for a 'lsmipsel' machine..... but now I m getting error while compiling kernel ..... this are the errors.... http://pastebin.mozilla.org/1207498 Apr 19 09:40:46 can anyone help me.... Apr 19 09:40:49 http://pastebin.mozilla.org/1207498 Apr 19 09:40:55 failuire is pretty clear nor? Apr 19 09:40:57 sed: can't read /home/prashant/Downloads/oebase/build/tmp/work/lsmipsel-angstrom-linux/linux-nios2-2.6.34-r0/defconfig: No such file or directory Apr 19 09:41:09 provide a kernel config for your machine Apr 19 09:45:27 morning woglinde Apr 19 09:45:35 ka6sox go to sleep Apr 19 09:45:37 *sigh* Apr 19 09:45:58 woglinde, don't I wash Apr 19 09:46:01 er with Apr 19 09:46:03 bleh Apr 19 09:46:05 wish Apr 19 09:46:17 * woglinde breaks the heart seeing you all days working that hard Apr 19 09:50:37 hi bluelightning Apr 19 09:50:45 hi woglinde, all Apr 19 09:52:10 woglinde, if I want to bypass kernel compilation... will it be possible? and how? Apr 19 09:53:20 sanket look for ASUME_PROVIDED Apr 19 09:53:27 in local.conf Apr 19 09:53:41 but dont know if it will work for kernels Apr 19 10:01:27 where can I find recipe for dbus-x11 ? Apr 19 10:02:17 with find? Apr 19 10:02:42 or grep Apr 19 10:02:53 I've tried Apr 19 10:03:14 but the only reference is in meta-openembedded in gconf : RDEPENDS_${PN} += "dbus-x11" Apr 19 10:03:37 but there is no dbus*x11*.bb or something similar Apr 19 10:05:17 dbus.inc defines it Apr 19 10:05:27 PACKAGES =+ "${PN}-lib ${PN}-x11" Apr 19 10:06:04 ouch...I didn't grep for *x11* in dbus Apr 19 10:06:06 my fault Apr 19 10:11:42 ok the problem was that using dbus 1.4.1 there was no dbus-x11 package defined for it Apr 19 11:08:47 re mickey Apr 19 11:08:48 hi obi Apr 19 11:09:02 hi woglinde! Apr 19 11:13:21 woglinde: after debug orc is up to you now ;) I'm not so keen with recipes building a compiler building things... Apr 19 11:14:59 ant didnt you see my mail= Apr 19 11:15:05 new version is out Apr 19 11:18:17 he he 2011-01-09 Apr 19 11:18:19 David SchleefRevert "orc.h: remove stdint.h from includes" Apr 19 11:18:44 I don't have time to understand what's happened there :) Apr 19 11:20:33 orc-stdint.h is still empty..they should have changed the code to avoid stdint types Apr 19 11:21:07 look at the latest version Apr 19 11:21:18 ah, yes, your mail is in oe-members :) Apr 19 11:21:51 no Apr 19 11:21:56 send it to -devel too Apr 19 11:22:08 I hope at least Apr 19 11:22:15 yup Apr 19 11:22:27 yes, I'm catching up with thousand emails... Apr 19 11:22:42 why the hell xbmc is over 100mb? Apr 19 11:25:22 debug build? :) Apr 19 11:25:59 no the source package Apr 19 11:26:24 ah Apr 19 11:35:24 jama and I hit a slow angstroem mirror again Apr 19 11:38:19 I did? Apr 19 11:38:42 *g* Apr 19 11:38:50 imagine the : Apr 19 11:39:09 lol Apr 19 11:53:16 JaMa, woglinde Apr 19 11:53:19 : Apr 19 11:53:34 what do you think about how modules are loaded in oe-dev? Apr 19 11:54:04 modutils-initscripts vs. udev vs. mdev vs. tmpfs ... Apr 19 11:54:36 *devtmpfs Apr 19 11:55:19 *g* Apr 19 11:56:05 (please note the colon) Apr 19 11:56:17 woglinde, what do you mean by a "slow angstrom mirror"? Apr 19 11:56:41 ka6sox download is slow here at my side Apr 19 11:56:58 *sigh* Apr 19 11:57:00 woglinde woglinde 117058259 19. Apr 13:54 /devel/archive/git_xbmc.git.sourceforge.net.gitroot.xbmc.xbmc_e2ab481ebe964321c358ab9d6402088c714adcbe.tar.gz Apr 19 11:57:13 what the hell they have in there sourcetree Apr 19 11:57:24 its huge Apr 19 11:57:26 okay... Apr 19 11:57:40 this is the xbmc one? Apr 19 11:58:08 yes Apr 19 12:00:39 thats @ OSUOSL. Apr 19 12:02:58 ka6sox did you sleep some hours? Apr 19 12:08:22 JaMa: I guess we should revisit the combinations of IMAGE_DEV_MANAGER and IMAGE_INITSCRIPTS Apr 19 12:09:09 ant_work: why? what's the problem? Apr 19 12:09:24 uh xbmc is finally there Apr 19 12:09:39 it can happen that you don't want udev and you just need modutils-initscripts Apr 19 12:10:03 or, if you have both like today's default, their runlevel must match Apr 19 12:10:11 * JaMa is using devtmpfs+shr-initscript Apr 19 12:10:12 I mean startlevel Apr 19 12:10:23 now udev starts before modutils.sh Apr 19 12:10:44 and at S03 like sysfs... Apr 19 12:11:00 I know parallelism here is a good thing but there is some order to respect... Apr 19 12:12:06 apart this, why don't we use modules-init-tools only? Apr 19 12:15:26 sorry got to finish something at work. so I'm not following very closely .. :/ Apr 19 12:15:58 ant_work ask on oe-dev and oe-core Apr 19 12:16:13 so now lets see orc Apr 19 12:16:26 woglinde, I can sleep when I'm dead. Apr 19 12:16:40 and as I thought Apr 19 12:16:45 .13 works without problems Apr 19 12:16:52 will commit now Apr 19 12:18:44 woglinde: what was pulling it in? bluez? Apr 19 12:19:07 it = schroeder Apr 19 12:22:25 xbmc build Apr 19 12:27:09 03Henning Heinold  07org.openembedded.dev * rf3dcc0656e 10openembedded.git/recipes/orc/ (orc_0.4.12.bb orc_0.4.13.bb): orc: add new version 0.4.13 and remove non-working 0.4.12 Apr 19 13:10:22 args damn it Apr 19 13:10:28 orc still sgefaults here Apr 19 13:11:22 orc-native Apr 19 13:11:24 at least Apr 19 13:34:14 woglinde: I see the commit in question orcc: remove stdint typenames from programs Apr 19 13:45:36 This is the second time I've encountered problems with the waf build system Apr 19 13:45:54 It just can't seem to ignore host libs Apr 19 13:47:21 And even worse, it doesn't halt on some errors. Reading a config.log file I see 'Build failed: -> task failed (err #1)' Apr 19 13:47:57 It ignores that error and continue to try and build against libraries that clearly isn't working Apr 19 13:48:21 I'll see if I can produce a patch to get around this problem Apr 19 13:48:32 when do you use NATIVE_INSTALL_WORKS = "1 Apr 19 13:49:18 Crofton|work: when you do not need install-native? Apr 19 13:51:39 Crofton|work: there is a strange case (mtd-utils) where oe-core removed it but oe-dev still need it or the install of the native recipe will fail... Apr 19 13:53:30 "Drop variables that do nothing", Signed-off-by: Richard Purdie -NATIVE_INSTALL_WORKS = "1" Apr 19 13:54:05 heh Apr 19 13:54:05 lot of subtle adjustments to do... Apr 19 13:54:08 so confused :) Apr 19 13:54:33 That's far from the only difference between oe-core and oe-dev Apr 19 13:54:38 do we have same bitbake at least? Apr 19 13:54:54 if I run a native recipe and the image dir in tmp/work is sane, I do not need it Apr 19 13:56:56 ant_work: bitbake master works fine in both, though there are still some small changes/fixes in the poky bitbake which haven't been synced over Apr 19 14:12:38 woglinde, the angstrom mirror is painfully slow for some people... I see 15kb/s when using the mirror... however, in turn, when i go and comment out the mirrors, the git protocol gives me 800kb/s directly. Apr 19 14:15:19 dandel yes I know Apr 19 14:18:14 kergoth: ok, thx Apr 19 14:18:34 btw, I've seen your comments about the 'issue' layers vs. .inc files Apr 19 14:20:46 hi kergoth btw. Apr 19 14:23:09 also, i talked to the xbmc devs to try to figure out a good fix for the x86_64 to i686 cross compile... This issue was noted, however none of the fixes that where done have improved the problem. Apr 19 14:23:29 dandel hm Apr 19 14:23:38 dandel why the hell xbmc source code is over 100mb? Apr 19 14:23:52 git repository is over 100mb Apr 19 14:24:00 dandel and again I told you the fix yesterday Apr 19 14:24:07 remove all old libtool m4 macros Apr 19 14:24:21 than systroot is used the right way Apr 19 14:24:27 i'm trying to update the bitbake file to do that before the configure step. Apr 19 14:24:39 I tried it today Apr 19 14:24:46 but orc suckz me no out Apr 19 14:25:47 my fully built tree is 2gb locally. Apr 19 14:28:58 woglinde, this might help a bit... here's the bakefile i am working on: http://pastebin.com/by9eDgYz Apr 19 14:30:46 this grabs the sources form the current repository for xbmc 10.1. (instead of the 10.05 from july 2010) Apr 19 14:30:57 hm Apr 19 14:32:33 also, the steps for this bakefile i found that are reliable to get this to build is to bake nano and xfce46-image, and then try to build xbmc. Apr 19 14:33:29 *i found that the reliable method to make xbmc is to bake nano and xfce image Apr 19 14:33:41 hm Apr 19 14:33:50 no remove of the libtool m4 macros Apr 19 14:34:41 i'm trying to make the bitbake file do that before the bootstrap command is issue. Apr 19 14:39:04 I am away for a while Apr 19 15:01:47 Second patch ever submitted =) Apr 19 15:02:47 It's excilerating to feel that you actually contribute to something =) Apr 19 15:03:35 s/exhilerating/exhilarating/ Apr 19 15:04:39 * Tartarus kicks mdadm Apr 19 15:05:06 In patchwork, should I mark my patch as anything other than new, or is that up to someone else? Apr 19 15:07:36 leave it as is, its up to whomever reviews/applies it to accept it or reject it now Apr 19 15:08:18 Okay Apr 19 15:08:51 Now it's time to split. Got a movie to catch Apr 19 15:21:58 03Tom Rini  07master * rad388e9967 10openembedded.git/recipes/mdadm/ (mdadm-3.1.2/remove-werror.patch mdadm.inc mdadm_3.1.2.bb): (log message trimmed) Apr 19 15:21:58 mdadm: Update 3.1.2 to obey CFLAGS/LDFLAGS/etc Apr 19 15:21:58 The Makefile in mdadm 3.1.2 doesn't obey what we're telling it. Apr 19 15:21:58 We first need to shove the extra parts of ${CC} into the ${CXFLAGS} Apr 19 15:21:58 variable and then our normal CFLAGS. We need to do something similar Apr 19 15:21:58 for LDFLAGS. Finally, we need to drop -Werror to get it building Apr 19 15:21:59 again. Apr 19 17:10:30 anyone building gstreamer-ti with 2011.03? I'm running into this: http://thread.gmane.org/gmane.comp.handhelds.openembedded/43375 - looks like libtool isn't configured properly Apr 19 17:46:56 hello all, I have a customer running a custom C app on an OE-based system (custom distro based on angstrom 2008.1) that is occasionally getting a dynamic linker error with their application (Inconsistency detected by ld.so: dl-lookup.c...) Apr 19 17:47:01 full error is here http://pastebin.com/PG6brhKe Apr 19 17:47:16 glibc 2.9 Apr 19 17:47:57 my question is: is it possible for this to be caused by their application corrupting memory used by the linker? Apr 19 17:48:29 we've determined it's not hardware-related since it can be reproduced on multiple machines Apr 19 17:48:58 machine type is at91sam9g20 Apr 19 18:17:36 hi, i have latest version of oe and the latest bitbake version, when parsind .bb files i get this error: MalformedUrl: ${E_SVN}/trunk/BINDINGS/python;module=python-edje;proto=http Apr 19 18:17:45 anybody familiar with that one? Apr 19 18:19:05 what bitbake version is that? Apr 19 18:19:20 also, where did you clone the OE repo from? Apr 19 18:21:30 note that it's showing an unexpanded variable in there, that's why it's malformed. E_SVN isn't defined Apr 19 18:22:00 * kergoth thinks that behavior is idiotic, especially given the ${} syntax being the same between bitbake and shell, would be better off erroring Apr 19 18:22:03 heh Apr 19 18:22:23 Nedlinpopo BitBake Build Tool Core version 1.12.0, bitbake version 1.12.0 Apr 19 18:23:40 kergoth: I've been noticing that some mirrors of sourceforge don't have the same checksums occasionally Apr 19 18:23:48 has anyone else seen this? Apr 19 18:23:56 yikes, that's worrisome. haven't seen that personally yet Apr 19 18:24:17 (it might be some stupid caching at my office site thouhg) Apr 19 18:25:25 I was having a hell of a time trying to get libdc1394 Apr 19 18:26:08 I would get 1 of 2 consintent checksum sets Apr 19 18:26:36 also, is there a way in a recipe to not specify a checksum without an error? Apr 19 18:35:25 kergoth: it would be great to have different syntax for bb and shell variables.. but I'm not sure how it is possible.. Apr 19 18:38:27 Nedlinpopo: yeah, there are two variables, one makes it so missing checksums are nonfatal. I don't recall the exact name offhand, though. grep for md5sum in classes/ :) Apr 19 18:40:06 any C experts have any insight into the error I mentioned earlier? http://pastebin.com/PG6brhKe Apr 19 18:40:32 just wondering if this is even possibly caused by badness in the application (memory corruption) or if it is more likely a glibc bug, etc Apr 19 18:41:12 i'd guess differences in glib versions Apr 19 18:41:34 yeah, i came across that as a possible explanation as well when googling Apr 19 18:41:35 that assert looks like it's in the shared object header/load functions Apr 19 18:41:56 they get it maybe 3 out of 200 times when running their app Apr 19 18:42:06 and it always fails when making a system() call instead a pthread Apr 19 18:42:25 can they recompile on the native system? Apr 19 18:42:30 i haven't had any issues with glibc otherwise Apr 19 18:42:54 good question, something to try Apr 19 18:43:16 i've asked if they can reproduce it outside their app and send code, but they haven't been able to (on target system) Apr 19 18:44:01 if they use autotools, make sure they're the same version as well Apr 19 18:44:22 how does ${datadir} change between native and target for BBCLASSEXTEND = native? Apr 19 18:44:48 tNedlinpopo: they *should* be..they should be building using the same OE tree to build their app as they are to produce the entire filesystem Apr 19 18:45:36 well, as long as the tree was static across al builds Apr 19 18:45:48 it sure should, but it's probably good to make sure Apr 19 18:46:52 i was thinking that if it were some incompatibility like that, it would fail consistently Apr 19 18:47:24 yeah I'd guess that too, but it sounds like it's maybe load order dependant Apr 19 18:47:44 yeah, possible, i have their LD debug traces Apr 19 18:47:51 nothing unusual in them Apr 19 18:48:28 kergoth: OE_STRICT_CHECKSUMS is the one I was looking for Apr 19 18:48:43 ah, yeah, thats the one Apr 19 18:49:12 Crofton|work: the same way all the other paths do for native vs target. native gets its prefix changed to the sysroot Apr 19 18:49:26 hmm Apr 19 18:50:14 http://pastebin.com/ZaSqVPGX Apr 19 18:50:20 I am trying to get sip going Apr 19 18:50:29 and the native side is not quite write Apr 19 18:51:08 any hints on getting this going with packaged staguing Apr 19 18:53:31 03Tom Rini  07master * rb55dcd636a 10openembedded.git/recipes/iozone3/iozone3_263.bb: Apr 19 18:53:32 iozone3: Update EXTRA_OEMAKE Apr 19 18:53:32 iozone3 knows about both a generic linux target and some architecture Apr 19 18:53:32 specific targets. We now default to building for vanilla linux and Apr 19 18:53:32 arch-specific targets when appropriate. Pointed out by Freescale. Apr 19 18:53:32 Signed-off-by: Tom Rini Apr 19 18:57:22 Nedlinpopo: thanks for the ideas Apr 19 19:09:23 how do I write do_install for the bbclassextend = native case? Apr 19 19:36:41 Crofton|work: what do you mean? Apr 19 19:36:58 if you need something other than the normal one you can do do_install_virtclass-native() { ... } Apr 19 19:37:09 well, the paths are set in configure Apr 19 19:37:16 http://pastebin.com/ZaSqVPGX Apr 19 19:50:49 re tasslehoff Apr 19 19:53:39 woglinde: not sure what re means in irc, but hi :) Apr 19 19:54:07 something like welcome back Apr 19 19:55:20 woglinde: :) Apr 19 19:59:10 Hi. When bitbake in interactive mode is sort of gone, how do I then do all that nice interactive stuff ? Apr 19 19:59:46 you don't. not until we reimplement it as a new UI, or add the functionality to an existing UI like was intended for the ncurses one Apr 19 20:01:41 anyone building gstreamer-ti? getting libtool issues for release-2011.03 Apr 19 20:01:55 tharvey hm Apr 19 20:02:23 same issue described here - http://thread.gmane.org/gmane.comp.handhelds.openembedded/43375 - I don't see any resolution to it Apr 19 20:03:22 hm let me think Apr 19 20:03:32 tharvey did you bild from scratch? Apr 19 20:03:45 yes Apr 19 20:04:16 seems like gstreamer didn't get staged properly? Apr 19 20:04:32 hm yes Apr 19 20:04:50 try bitbake gestreamer before Apr 19 20:05:12 last built it successfully with release-2010.12 and angstrom-2008.1, these issues are with release-2011.03 and angstrom-2010.x Apr 19 20:05:52 yes, I tried bitbaking gstreamer again - let me try cleaning it though. Will -cclean cause package to be restaged on next build? Apr 19 20:06:57 yes Apr 19 20:16:54 woglinde, same issue - and gstreamer 'appears' to be staged properly - I see libs in SYSROOTFS/$MACHINE_ARCH but compile looks like its trying to find libs in $SYSROOTFS/$MACHINE Apr 19 20:17:17 ah hm Apr 19 20:23:54 not really clear where the libtool config happens seems strange that this is the only affected recipe I've run into like this Apr 19 20:26:59 tharvey, What package is failing for you? And what error do you get? Apr 19 20:27:31 gstreamer-ti - its exactly the same issue as posted to ml here: http://thread.gmane.org/gmane.comp.handhelds.openembedded/43375 Apr 19 20:28:35 not clear what the 'warning: library was moved' error from libtool indicates, then its clear that its looking for libgstreamer in the wrong place (machine path vs arch path) Apr 19 20:32:40 Yeah, that is odd Apr 19 20:36:11 lots of changes since my last build - angstrom-2010.x uses gcc-4.5.x, gstreamer now at 0.10.32 - gstreamer-ti's log.do_configure is completely different Apr 19 20:36:42 bah... all this damn 'auto'tools stuff seems to get in the way ;) Apr 19 20:37:15 Tell me about it. I'm currently trying to solve a bug with nmap Apr 19 20:38:18 things are changing pretty rapidly - I may be completely out of date. I'm not at all clear whats been going on the the splitup of packages into various trees - don't really see anything on OE wiki explaining the new state of things Apr 19 20:40:58 tharvey, See if this thread can help Apr 19 20:40:59 http://www.mail-archive.com/angstrom-distro-devel@linuxtogo.org/msg04160.html Apr 19 20:41:36 oh thx! didn't notice a 2nd thread about it Apr 19 20:41:59 I do see where an inappropriate --with-libtool-sysroot is slipping into oe_runconf - not clear why yet Apr 19 20:42:59 tharvey, I extracted it from your first ML post. Apr 19 20:43:51 The thing was that someone had answered your post with a dead link. Just altered the link a bit and voilá Apr 19 20:43:55 aaah... yes I see the reply you found... I didn't notice it had an obfuscated path in it Apr 19 20:44:14 yup... couldn't follow the dead link Apr 19 20:44:32 lol... gotta love the workaround at the end of the thread - hack in a manual symlink Apr 19 20:45:30 Haha Yeah, I just saw and was gonna comment on that Apr 19 20:45:36 It isn't really a solution Apr 19 20:46:13 However, the thought already crossed my mind that it probably would work ^^ Apr 19 20:46:22 had already* Apr 19 20:53:15 the offending --with-libtool-sysroot is from autotools.bbclass but not clear yet how it has changed and why the issue didn't always exist Apr 19 20:53:49 tharvey: sysroot support was added to libtool in 2.4. that wasn't there until we moved to it Apr 19 20:53:55 make sure it isn't using old libtool macros or something Apr 19 20:54:50 kergoth, I'm looking at the older oe tree I was building from 2010.12 - http://cgit.openembedded.net/cgit.cgi/openembedded/tree/classes/autotools.bbclass?id=release-2010.12 Apr 19 20:55:28 oe_runconf there adds CONFIGUREOPTS which has ${@append_libtool_sysroot(d)} yet in my old builds run.do_configure there is no --with-libtool-sysroot in those args Apr 19 20:56:13 so 1) I would think --with-libtool-sysroot is just pointing to the wrong thing (STAGING_DIR_HOST machine not arch?) and 2) not clear why it wasn't previously even an are to configure Apr 19 20:56:28 tharvey wasnt khem suggesting it Apr 19 20:56:34 in the thread Apr 19 20:56:47 sorry I cannt look at it Apr 19 20:56:49 kergoth, oh I follow you - LIBTOOL_HAS_SYSROOT was not defined Apr 19 20:56:53 previously, and now it is Apr 19 20:56:59 still try to debug orc Apr 19 20:58:19 woglinde, is it a bug regarding orc-stdint.h? Apr 19 20:59:47 I had that earlier, but got lazy and just copied a working orc-stdint.h instead. Not much of help I guess Apr 19 21:00:42 woglinde, because of my configure problems, what is the command to just tell oe to rerun configure and compile? (still having troubles with the xbmc system) Apr 19 21:01:00 bitbake -c configure xbmc Apr 19 21:01:10 bitbake -c compile xbmc Apr 19 21:01:19 Just a wild guess =) Apr 19 21:02:38 risca no its segfault Apr 19 21:03:01 Oh, that's worse =( Apr 19 21:03:19 yes Apr 19 21:03:27 but maybee its a bit related Apr 19 21:03:57 for some type_names there is Apr 19 21:03:59 type_name = 0x2
Apr 19 21:05:09 so not understanding libtool all that well, what 'should' sysroot point to? autotools.bbclass has it pointing to STAGING_DIR_HOST and at least for gstreamer-ti (others I would assume) this needs to be STAGING_DIR_TARGET (I think) - that gets me past the libtool issue but not out of the woods yet still failing on link with a: arm-commonos-linux-gnueabi/bin/ld: .gnu.attributes not found for insert Apr 19 21:06:01 also still seeing a lot of arm-commonos-linux-gnueabi-libtool: link: warning: seems to be moved - not clear if thats an issue Apr 19 21:06:07 tharvey but anyway dont expect gstreamer-ti running against .32 version Apr 19 21:07:10 I have used it with 2.6.32-omap-psp for ages... Apr 19 21:07:20 woglinde, why do you say that? Apr 19 21:08:01 not kernel Apr 19 21:08:03 technically I use a flavor of ridgerun's branch b/c upstream gstreamer-ti at least was horrible with respect to DSP decode Apr 19 21:08:04 gstreamer version Apr 19 21:08:14 oh? Apr 19 21:08:37 or was it .30 Apr 19 21:08:41 hm Apr 19 21:08:45 * tharvey starting to regret embarking on upgrade from release-2010.12 to release-2011-03 Apr 19 21:08:46 I slightly head Apr 19 21:08:52 args heard Apr 19 21:08:57 thats not working Apr 19 21:09:02 but dont know for sure Apr 19 21:09:31 I have been using .30 in the past... and noticed that gstreamer_0.10.30.bb has been removed from release-2011-03 Apr 19 21:09:58 gotta love it when someone obsoletes something that is still needed for certain platforms Apr 19 21:15:45 I have some problems with nmap. do_configure fails complaining about syntax errors in libpcap-possiblymodified/configure. Apr 19 21:16:30 I could fix the syntax errors in libpcap-possiblymodified/configure, but when I rerun do_configure that configure script is regenerated with the same syntax error Apr 19 21:17:18 I have to do something with configure.ac, but don't really know where to begin Apr 19 21:18:52 thats the 'right' way to fix it but I've seen people work around the regenerating of the file as well Apr 19 21:19:34 but pastebin your syntax errors and your ac and maybe someone with good knowlege of autoconf can help Apr 19 21:20:21 http://pastebin.com/26V1Q9Ry <-- libpcap-possiblymodified/configure.ac Apr 19 21:22:14 http://pastebin.com/3D2bmL48 <-- libpcap-possiblymodified/configure with syntax error on line 2329 Apr 19 21:23:15 he ant Apr 19 21:23:43 removing line 2330-2331 and adding a trailing ')' to line 2329 would fix the syntax I think Apr 19 21:24:52 uhm overflow? Apr 19 21:24:58 there is no line 2330 Apr 19 21:25:16 Ah, maybe... Apr 19 21:25:18 Hm Apr 19 21:25:19 hm ah javacscrtipt Apr 19 21:25:45 I used the pastebinit script. Guess it doesn't work that well Apr 19 21:26:27 hm seems its pastebin itself which cuts the first digit Apr 19 21:26:36 The aliases save the names the user supplied, while $host etc.? Apr 19 21:26:45 hm Apr 19 21:26:48 Nope Apr 19 21:27:02 { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5 Apr 19 21:27:47 risca which line? Apr 19 21:28:04 Wait one sec. I think I actually pastebin:ed a corrected configure Apr 19 21:29:31 line 329 yes: Apr 19 21:29:43 2328 case $cross_compiling:$ac_tool_warned in Apr 19 21:31:29 hm Apr 19 21:31:32 There are 4 occurances of 'case $cross_compiling:$ac_tool_warned in'. The first one has a syntax error Apr 19 21:32:04 hm Apr 19 21:32:36 5 occurances* Apr 19 21:33:33 Is the configure script generated in the same order as the configure.ac, so to speak? Apr 19 21:34:41 http://pastebin.com/438vKCvC <-- log.do_configure Apr 19 21:39:12 woglinde: should I rebuild orc? Apr 19 21:39:27 ant I already put newest version in Apr 19 21:39:36 I am now hunting a sgefault Apr 19 21:39:40 I also see that there is a considerably large patch (autofoo.patch, 800 lines) that is messing with several configure.ac Apr 19 21:39:45 yes, but you have issues I read Apr 19 21:42:23 risca oha 800 lines? Apr 19 21:42:43 What do you mean? Apr 19 21:42:53 that big Apr 19 21:43:00 sounds something wrong Apr 19 21:43:06 does it change configure? Apr 19 21:43:10 827 to be exact Apr 19 21:43:13 than its wrong Apr 19 21:43:31 It is in recipes/nmap/nmap-3.81 Apr 19 21:43:38 lets see Apr 19 21:43:39 autofoo.patch Apr 19 21:44:28 ah okay its patchin acinclude Apr 19 21:44:53 Not only acinclude.m4 Apr 19 21:45:04 It patches several files Apr 19 21:46:13 hm yes Apr 19 21:46:15 woglinde: where do you have segfault? building orc-native? here both build on Gentoo 32bit Apr 19 21:46:22 but I am to tired to look at Apr 19 21:46:30 ant which compiler? Apr 19 21:46:50 gcc version 4.4.5 (Gentoo 4.4.5 p1.2, pie-0.4.5) Apr 19 21:48:16 let me see that strcmp code... Apr 19 21:50:24 okay gcc 4.5 and gcc 4.6 here Apr 19 21:50:36 it was in orcc.c, isn't? Apr 19 21:50:54 yes Apr 19 21:50:59 yes, one of the latest commits added +static const char *orcify_typename (const char *s) Apr 19 21:51:06 but it looks the parser is doing something wrong Apr 19 21:51:21 I am investigate it now Apr 19 21:51:42 example3orc.orc Apr 19 21:51:55 .dest 4 d1 is parsed right Apr 19 21:51:58 but after it Apr 19 21:52:01 all is wrong Apr 19 21:53:30 do_configure: http://paste.debian.net/114529/ Apr 19 21:54:13 argh_do_compile exceeds lenght.. Apr 19 21:54:27 ant whats wrong wit do_configure? Apr 19 21:54:38 nothing, just for you to compare Apr 19 21:56:10 this is compile line of orcc.c Apr 19 21:56:12 http://paste.debian.net/114530/ Apr 19 21:58:11 no it isnt compile of orcc.c Apr 19 21:58:16 its execution of orcc Apr 19 21:58:43 wait, I just did grep Apr 19 22:11:10 hm Apr 19 22:11:13 time for bed Apr 19 22:11:17 something odd Apr 19 22:11:26 strdup is going wrong Apr 19 22:11:31 or is the other side Apr 19 22:11:43 where mem is poiting to wrong stuff Apr 19 22:11:51 good nite Apr 19 22:12:06 schleef should do a better job at ther parser Apr 19 22:12:11 code looks awfull Apr 19 22:12:21 why he didnt use lexx Apr 19 22:14:01 anyone familiar with koen's patch http://cgit.openembedded.org/cgit.cgi/openembedded/commit/recipes/ti/gstreamer-ti.inc?h=2011.03-maintenance&id=2bb663abd461b04663b0459f54a133dc844b4eb4 that appends '-T' to linker cmd's for newer toolchains? Not understanding how that works or how to apply it to different Makefiles Apr 19 22:14:26 curious where the -T option to gcc is documented Apr 19 22:23:56 This nmap syntax error bug is driving me crazy. Can't seem to resolve it Apr 19 22:54:54 03Andrea Adami  07org.openembedded.dev * r30e0900daa 10openembedded.git/recipes/udev/ (31 files in 31 dirs): (log message trimmed) Apr 19 22:54:54 udev: regroup overrides for Zaurus devices Apr 19 22:54:54 * device cache still disabled by default Apr 19 22:54:54 * extend mount.blacklist to all insertable media Apr 19 22:54:54 * add /dev/mmcblk* and /dev/sd* Apr 19 22:54:54 * let /dev/hd* for compatibility with deprecated IDE driver Apr 19 22:54:54 * bump INC_PR Apr 19 23:15:49 03Tom Rini  07master * rfca5ec0478 10openembedded.git/recipes/inetutils/inetutils_1.8.bb: Apr 19 23:15:49 inetutils: Add missing inherit gettext Apr 19 23:15:49 Signed-off-by: Tom Rini Apr 19 23:16:55 03Andrea Adami  07org.openembedded.dev * r169a2a9334 10openembedded.git/recipes/linux/linux-kexecboot_2.6.38.bb: Apr 19 23:16:55 linux-kexecboot: upgrade from 2.6.38.2 to 2.6.38.3 Apr 19 23:16:55 Signed-off-by: Andrea Adami Apr 19 23:17:06 03Andrea Adami  07org.openembedded.dev * rb0d4dbd1b5 10openembedded.git/recipes/linux/linux_2.6.38.bb: Apr 19 23:17:06 linux: upgrade from 2.6.38.2 to 2.6.38.3 Apr 19 23:17:06 Signed-off-by: Andrea Adami Apr 19 23:32:57 Nailed it! Fixed my nmap errors! Apr 19 23:33:04 * risca does a little victory dance Apr 19 23:33:56 I notified the oe-dev ML of my findings Apr 20 02:08:42 hey, if i have a recipe in my user collection, and there is a newer version in the oe git tree, will bitbake use the newer one or wil it strictly prefer mine? Apr 20 02:09:57 ah right, BBFILE_PRIORITY Apr 20 02:43:57 * kergoth_ chases down some rogue dependencies **** ENDING LOGGING AT Wed Apr 20 02:59:58 2011