**** BEGIN LOGGING AT Mon May 30 02:59:59 2011 May 30 07:21:49 hi all May 30 07:22:39 pinchartl, I got the same problem. Not only using opkg but also on dropbear settings, and more. May 30 07:39:15 good morning May 30 07:41:48 good morning May 30 07:42:12 hi, I'm facing to a problem building qt4e-base-image with angstrom-2010.x on a Debian lenny host, is it normal? http://pastebin.com/6SnFSYs6 May 30 07:42:37 I suspect a Debian lenny issue with new OE stuff May 30 08:18:21 Hi everyone May 30 08:19:25 I'm using dbus 1.4.0 from the meta-openembedded layer....is it possible to compile it without X11 support? I need to use it with qt-embedded. May 30 08:20:01 mrAlmond: well, compile and packaging are two things. is libdbus.so linking to X11? May 30 08:20:24 zecke : My problem is with dbus-launch that links X11 May 30 08:20:29 cfrabel: and do you have any solution to that problem ? :-) May 30 08:20:34 jama ping May 30 08:21:03 oe-core and meta-oe should have package seperation May 30 08:21:12 zecke : In one of my previous build I've compiled the xorg-xserver but now I'm not using it in the latest images that uses qt-embedded May 30 08:21:23 so yes you have to compile dbus-x11 because its in the dbus recipe May 30 08:21:37 but it will not end up in your image May 30 08:22:13 indeed...in my image there is no dbus-launch but I can find all the other binaries dbus-monitor, dbus-daemon etc etc... May 30 08:22:26 but I think I need also dbus-launch May 30 08:22:53 mrAlmond than look at the recipe in which package its packed May 30 08:22:56 gm mickeyl May 30 08:25:07 woglinde : How can I find that? May 30 08:25:50 mrAlmond -> < woglinde> mrAlmond than look at the recipe in which package its packed May 30 08:27:23 mrAlmond there are less, vi, even with eclipse you can look at the recipe May 30 08:31:03 woglinde : Yes I know...I'm using vi because I'm working on a remote server...but I just wanted to know the variable name that describes the package name May 30 08:31:15 o.O May 30 08:31:26 go start with the oe-manual again May 30 08:31:47 pinchartl, I modified /etc/init/d/dropbear script May 30 08:32:14 pinchartl, i'm looking for the modifications I made... May 30 08:32:27 cfrabel: my target system won't have dropbear, so that's not a real issue May 30 08:33:29 I also got a motd error May 30 08:34:05 I wonder who needs motd in embedded device May 30 08:34:22 woglinde, so, i put it off! May 30 08:35:27 pinchartl, I the very first boot, the system creates some links, on an ro system, it fails, so my first boot is rw May 30 08:35:40 At the very first... May 30 08:48:08 woglinde : ok...so let's start with the manual again, I've read it two months ago and now I've forgot a lot of things... May 30 08:48:36 hint May 30 08:48:39 FILES_ May 30 09:06:05 hi ant May 30 09:07:04 hello woglinde May 30 09:11:54 woglinde: p o n g May 30 09:19:22 JaMa|Off: hi there May 30 09:19:26 hi May 30 09:21:26 JaMa|Off: have you seen how koen changed volatiles and fstab (base-files)? That would permit the use of udev devices-cache May 30 09:21:30 (once root is rw) May 30 09:22:22 err.. I mean having root ro...then the remount should work May 30 09:23:02 how did you solve in shr for your devices? May 30 09:23:19 using devtmpfs May 30 09:23:39 and lot more directories on volatile May 30 09:35:58 jama can you update navit? May 30 09:36:02 *g* May 30 09:36:06 or give it a try May 30 09:36:22 some memleaks and segfaults are fixed May 30 09:37:31 woglinde: ok May 30 09:37:51 btw it was updated few days ago by Stanislav May 30 14:15:03 bluelightning: hi there May 30 14:15:17 hi anarsoul May 30 14:15:18 dind May 30 14:15:20 argh May 30 14:15:27 bluelightning: did you get 2.6.39 kernel booting on your h1940? May 30 14:15:33 didn't get very far with it I'm afraid May 30 14:15:59 I attempted to build u-boot from source but something was wrong with the hacked up OE recipe I used and I didn't investigate it further May 30 14:16:25 I can provide you a u-boot binary May 30 14:16:34 that would be great, thanks :) May 30 14:17:19 I am interesting to know about how is going to be handled video drivers (to Xorg) in oe-core? May 30 14:26:00 bluelightning: sent May 30 14:26:31 bluelightning: boot u-boot via haret, then u-boot takes uImage from first partition on card (it should be vfat) May 30 14:28:17 usb should work in u-boot, so you can get u-boot console on ttyACM0 May 30 14:42:49 03Koen Kooi  07org.openembedded.dev * r21145b253f 10openembedded.git/recipes/udev/udev_171.bb: May 30 14:42:49 udev: add 171 May 30 14:42:49 As the recipe says, machine specific rules should be in their own recipe May 30 14:42:49 Signed-off-by: Koen Kooi May 30 14:42:50 03Koen Kooi  07org.openembedded.dev * recea326c7b 10openembedded.git/recipes/udev/udev_git.bb: May 30 14:42:51 udev git: bump SRCREV past 171 May 30 14:42:51 Signed-off-by: Koen Kooi May 30 14:43:01 03Koen Kooi  07org.openembedded.dev * r600ce4b489 10openembedded.git/conf/distro/include/angstrom-2010-preferred-versions.inc: May 30 14:43:01 angstrom: prefer udev 171 May 30 14:43:01 Signed-off-by: Koen Kooi May 30 14:57:47 Hello, I just resent the pending patches we have in our local tree for regular oe (dev) and these are rebased against today's tip. May 30 14:58:06 It would be nice if someone could review and push the interesting ones from these May 30 14:58:38 otavio I think you can push them yourself after review May 30 14:58:50 koen somehow dont care this much May 30 14:58:59 woglinde: I don't have the ssh key setup here May 30 14:59:11 hm okay May 30 15:00:36 woglinde: since oe-core is pull model based I won't care too much about setting it up May 30 15:04:48 hm understand May 30 15:06:15 otavio hm the firefox update May 30 15:06:18 its still old May 30 15:06:58 woglinde: it is what it had on oe dev at the time May 30 15:09:03 otavio: I'm going to retest that ICE in xserver with xephyr.. May 30 15:11:48 JaMa|Off: nice May 30 15:14:17 JaMa|Off: this would be nice May 30 15:14:37 JaMa|Off: I still need to figure why xephyr has some bugs with evdev but this is for later May 30 15:17:45 otavio: still emits ICE.. May 30 15:18:09 will try with gcc-4.6 May 30 15:18:17 :-) May 30 15:29:17 hi gnutoo May 30 15:29:43 hi May 30 15:30:18 otavio: enabled in meta-oe :) May 30 15:30:58 hi May 30 15:31:13 how can I build a toolchain with oe? May 30 15:31:45 i've been using codesourcery's (yes - foolish me) May 30 15:32:00 but I'ld like to build it myself May 30 15:32:30 follow the setup guide May 30 15:32:37 and do bitbake gcc-cross May 30 15:32:40 thats it May 30 15:33:19 toolchain will be under youroetmp/sysroot/host-arch/usr/targetarch/bin May 30 15:33:42 tdebrouw, also learn about devshell May 30 15:34:44 k May 30 15:34:47 thx May 30 15:37:05 tdebrouw, basically if you do that: May 30 15:37:09 bitbake devshell May 30 15:37:18 yeah May 30 15:37:25 that will give you a script in: May 30 15:37:46 ..../deploy/addons/arm-oe-linux-gnueabi-shr-nokia900-devshell in my case May 30 15:37:55 source it May 30 15:38:01 and you have all in PATH May 30 15:38:06 plus some helper functions May 30 15:38:11 such as oe_runconf May 30 15:38:13 oe_runmake May 30 15:38:15 etc... May 30 15:38:52 k May 30 15:38:53 thx May 30 15:39:17 this info isn't on the wiki - just noticed that May 30 15:50:25 When buying a SD card for the BeagleBoard, should I go for the high quality cards or is the cheapest sufficient? May 30 15:52:24 Oops, I thought I posted this question on the BB channel May 30 15:52:53 hi khem May 30 15:52:58 But in general, what are peoples' experience of SD cards? May 30 15:52:59 risca depends May 30 15:53:10 but most cards work on bb May 30 15:53:18 is it xm or older? May 30 15:53:35 Older May 30 15:53:43 woglinde: I've checked navit repo and there is only 11 commits after our SRCREV, it that what you wanted to test? May 30 15:54:06 uh? May 30 15:54:22 the last pull request was for meta was 395x May 30 15:55:07 otherwise I havent ask May 30 15:55:36 woglinde, I've had 2 cards where unpacking the Angstrom-image was fine, but once I tried booting it starts spitting out read errors and such. May 30 15:55:58 risca on both cards? May 30 15:56:04 Yes May 30 15:56:24 hm are sure that this doesnt come from the internal nand? May 30 15:56:37 erease internal nand and boot all from sd-card May 30 15:56:43 Now, I can't unpack the image on them before getting read/write errors May 30 15:57:17 I am positive that it booted from SD card May 30 15:57:35 whats the sd-card manufactor? May 30 15:57:56 It says Zap glider on the front May 30 15:58:12 lol May 30 15:58:16 never heard May 30 15:58:18 which class? May 30 15:59:00 I don't know. Says Hi-speed certified for fast data transfer on the front of the package May 30 15:59:13 there must be somewhere a class May 30 15:59:19 otherwise its crap May 30 15:59:29 for sure May 30 15:59:52 Nope, not a word... May 30 16:00:06 hm hehe May 30 16:00:15 so better buy another one May 30 16:00:48 According to the shops' website: 11 MB/s read, 3 MB/s write May 30 16:01:50 Ah, now it managed to unpack image again ^^ May 30 16:02:33 It also takes forever to unmount the card May 30 16:03:21 maybe 5 minutes or so May 30 16:03:43 I better get a better SD card, this is driving me insane May 30 16:04:26 What class should I have, then? May 30 16:05:41 class6 or class10 May 30 16:06:01 I got 2 two 8 gig class10 from transcend they are working okay for me May 30 16:08:22 Thanks for your help, woglinde =) May 30 16:08:50 rsica but I have one noname also working May 30 16:09:03 was christmas present from my vserver provider May 30 16:09:22 Yeah, I have one of these zap cards working. May 30 16:09:44 We are a bunch of friends that all got BeagleBoards and I'm helping them setting it up May 30 16:09:54 hehe May 30 16:10:01 and than? May 30 16:10:08 So it's not really my cards ^^ May 30 16:10:37 The next step is producing a sdk for DSP developing for them =) May 30 16:10:47 uhm? May 30 16:10:57 you mean improving c6accel? May 30 16:11:09 sdk is already in angstroem May 30 16:11:12 ups oe May 30 16:11:16 No, just handing it over to them in a neat package May 30 16:11:44 so you need to tweak sdk stuff in oe May 30 16:12:16 I don't think so, but my friends don't have OE, so it's up to me to build it May 30 16:12:17 better start from the beginning with oe-core meta-oe meta-angstroem and meta-ti layer May 30 16:12:24 lol May 30 16:12:32 Hehe ^^ May 30 16:12:55 I'm the most nerdy one in the group May 30 16:14:09 I'll just package my toolchain in a nice tar package, write down some instructions and hand it over to them May 30 16:14:25 Then spend a weekend helping them set it up May 30 16:25:30 JaMa|Off: nice and thx May 30 17:22:00 re anarsoul May 30 17:22:24 erm, so I'm trying to create a uimage but need to know the entrypoint/loadaddress May 30 17:31:21 bbl May 30 18:15:45 jo crofton **** ENDING LOGGING AT Mon May 30 18:58:39 2011 **** BEGIN LOGGING AT Mon May 30 19:00:14 2011 May 30 19:00:36 cdbot2's logs are lost because of expired domain hentges.net May 30 19:00:46 keep in mind :) May 30 19:01:26 ka6sox-home: ping May 30 19:04:17 jo effem May 30 19:05:00 hi woglinde, all May 30 19:14:26 hi May 30 19:14:31 I have aproblem with identification ofa kernel module May 30 19:14:37 kernel-module-hci-usb May 30 19:14:40 yes? May 30 19:14:42 this is btusb.ko? May 30 19:14:59 opkg files kernel-module-hci-usb? May 30 19:15:12 type it without the? May 30 19:16:03 Jay7, pong May 30 19:16:13 woglinde: were? May 30 19:16:18 type were? May 30 19:16:50 ka6sox-home: may you add #kexecboot to nslu logging bot? May 30 19:17:04 Jay7, sure...gimme a few minutes May 30 19:17:07 ka6sox-home: we are using cdbot2 now but his logs are lost May 30 19:17:19 ka6sox-home: this isn't first priority task at all :) May 30 19:17:42 kk...I'm splitting 1 server into 2 now...so its a bit busy ATM. May 30 19:17:54 ka6sox-home: np, I'll remind you later then May 30 19:18:01 please do... May 30 19:18:04 ok **** ENDING LOGGING AT Mon May 30 20:11:13 2011 **** BEGIN LOGGING AT Mon May 30 20:12:15 2011 May 30 20:24:43 woglinde: SRCREV = "3915" is only in navit-icons and xpm dir has only cmake changes since then... main navit recipe is SRCREV = "4495" **** BEGIN LOGGING AT Tue May 31 07:34:15 2011 **** BEGIN LOGGING AT Tue May 31 07:48:14 2011 May 31 10:01:01 hi woglinde_ May 31 10:03:19 jo pb May 31 10:24:22 hi all... quiet in here today May 31 10:40:17 ~seen koen May 31 10:40:18 dcordes: i haven't seen 'koen' May 31 10:43:41 dcordes ask in #beagle May 31 10:50:06 woglinde_: ok thanks May 31 10:50:53 woglinde_: cannot wait for my ac100V to arrive and run oe images on it. would be nice to compile small gnome image or so May 31 12:34:39 Hi May 31 12:36:40 hi otavio May 31 12:42:12 pb_: I guess you didn't manage to convince florian to make the checkin? May 31 12:42:19 :) May 31 12:43:49 RP: about the chicken-cross, I made the pastebin yestarday http://paste.debian.net/118456/ May 31 12:48:36 otavio: It looks like its installing into roughly the right place at least: install -m755 i586-oe-linux-chicken "/home/otavio/hacking/el/tmp-eglibc-eglibc/work/i586-oe-linux/chicken-cross-4.7.0-r0.0/image/home/otavio/hacking/el/tmp-eglibc-eglibc/sysroots/x86_64-linux/usr/bin May 31 12:49:56 otavio: What is in /home/otavio/hacking/el/tmp-eglibc-eglibc/work/i586-oe-linux/chicken-cross-4.7.0-r0.0/sysroot-destdir ? May 31 12:50:21 RP__: I pasted a just built one: http://paste.debian.net/118497/ May 31 12:50:33 * otavio checks May 31 12:51:14 (devel)~/hacking/el% ls /home/otavio/hacking/el/tmp-eglibc-eglibc/work/i586-oe-linux/chicken-cross-4.7.0-r0.0/sysroot-destdir May 31 12:51:17 home/ May 31 12:51:24 otavio: I'm thinking the problem might be its installing into [...]/sysroots/x86_64-linux/usr/bin and not sysroots/x86_64-linux/usr/bin/i586-oe-linux/ May 31 12:51:39 bindir = "${exec_prefix}/bin/${CROSS_TARGET_SYS_DIR}" May 31 12:51:55 and its only installing into ${exec_prefix}/bin for some reason May 31 12:52:12 RP__: I can move things after install May 31 12:52:23 RP__: probably a build-system issue May 31 12:52:51 otavio: I'd suggest you configure it with the correct bindir :) May 31 12:53:08 RP__: the build-system is a self-done one May 31 12:53:20 RP__: by chicken guys May 31 12:53:53 otavio: right... May 31 12:55:49 * otavio checks how hard is to fix bindir on build system May 31 12:56:42 RP__: ah, right, no. I forgot about it over the weekend. I don't think I have the right ssh key here but I will try to remember to do it tonight. May 31 12:57:48 * woglinde tries to get CPU_ALLOC_SIZE into uclibc May 31 12:58:47 heh, I wonder how many people are using uclibc on multiprocessor systems May 31 12:59:38 hm ac100 might be the first arm May 31 13:00:02 pb does the underlaying functions are defined in glibc or kernel? May 31 13:00:19 __sched_cpucount for instance May 31 13:01:26 depends what you mean by "underlying". obviously the kernel does need to expose the number of cpus at some point, but I think the majority of the api implementation is in glibc. May 31 13:01:55 I believe that this page: http://www.openembedded.org/index.php/OEandYourDistro is missing a required package for fc11 (at least) May 31 13:02:16 I ran the command as written and didn't get mercurial. May 31 13:02:59 Is that correct? If so, who do I contact about getting a wiki account? May 31 13:03:09 fc11? May 31 13:03:27 thats old May 31 13:03:32 I know :) May 31 13:03:38 It's still what I have though. May 31 13:03:41 so dont use it May 31 13:03:52 Gosh that's helpful, thanks. May 31 13:03:54 and get a recent distribution May 31 13:04:09 vm's are cheap May 31 13:04:12 and usefull May 31 13:04:22 So the change should be to remove it from the list on that page if it's no longer supported. May 31 13:04:41 pb_: thanks May 31 13:05:40 celston: it does seem to be hard/impossible to get wiki accounts at the moment. you could try asking on the mailing list. May 31 13:06:00 celston but youdont need mercurial installed May 31 13:06:01 * RP__ tried to change the wiki and couldn't as the image captcha wouldn't work in firefox May 31 13:06:13 it will be build by oe anyway May 31 13:06:13 and thats with an account :( May 31 13:06:18 You don't. But if you don't have it only the native tools get built. May 31 13:06:18 RP__: doh May 31 13:06:38 ? May 31 13:07:33 I didn't have mercurial, and 'bitbake core-image-minimal' just built the host tools and then stopped. May 31 13:08:36 core-image-minimal isnt in my actual copy of oe .dev May 31 13:08:40 or master May 31 13:08:57 it's in oe-core May 31 13:09:05 *g* May 31 13:09:15 I'm using oe-core May 31 13:09:21 afaik, notwithstanding the fact that sanity.bbclass checks for it, you shouldn't need mercurial since all the mercurial-requiring recipes are meant to depend on hg-native May 31 13:09:28 not my fault if oe-core doesnt have the maercurial native May 31 13:09:49 [chris@desktop cora_oe]$ find ./ -name mercurial\*.bb May 31 13:09:50 :) May 31 13:10:13 So does mercurial native recipe need to be in oe-core too then? May 31 13:10:24 oh yeah, so it doesn't. May 31 13:10:32 in that case, I guess sanity.bbclass is correct in requiring it. May 31 13:13:21 damn it where is the webbrowser for glibc May 31 13:14:16 Is there anything I can do to fix this then? Given that it caught me out, I guess others might find it too. May 31 13:14:32 compil mercurial yourself May 31 13:14:39 or find a rpm May 31 13:14:54 or use an actual distro which has mercurial rpm out of the box May 31 13:15:23 celston: depends what you mean by "fix" it. any solution which involves editing the wiki would appear to be a non-starter, unfortunately. May 31 13:15:50 if you wanted to mount an argument for having oe-core include mercurial-native (and remove the requirement from sanity.bbclass) then the oe-core mailing list would be the place to do it. May 31 13:17:14 pb_ I'll post to the mailing list, thanks. May 31 13:18:46 ok, good luck May 31 13:31:18 RP__: ok I got it building and it has installed the binaries as 'x86_64-linux/usr/bin/i586-oe-linux/i586-oe-linux-csc May 31 13:31:39 RP__: the problem now is that it seems libraries wheren't copied May 31 13:32:38 RP__: on package 'image' dir it has it complete May 31 13:35:59 RP__: got it May 31 13:36:03 RP__: same for libdir May 31 13:54:11 I'm not 100% sure but looks like that building on Debian Lenny x86_64 raises a problem with gcc-4.3. Does OE still work with Lenny or is required an upgrade to Squeeze? May 31 13:54:52 afaik, it will work fine with lenny. May 31 13:55:19 pb_: I'm struggling since last week May 31 13:55:27 I don't actually have any lenny machines left to test on but it did use to work until quite recently and I can't think of any reason why it should have stopped. May 31 13:55:35 pb_: tried kaeilos, angstrom: no way May 31 13:56:21 pb_: http://pastebin.com/pPyaDXvt May 31 13:56:44 why do you think that's related to lenny? May 31 13:57:28 pb_: beause its old gcc May 31 13:58:05 so you think your host gcc is miscompiling ld? May 31 13:58:18 I guess that's possible, but it seems unlikely unless you've actually debugged the crash. May 31 13:58:41 seems more likely that it's just a regular bug in ld. May 31 13:58:45 hm I am believe its a problem with armv5 May 31 13:58:52 because on armv7 its build May 31 13:59:34 is something happened bettween OE March version and last week version May 31 14:00:11 again!!!!!!!!!!!!!!!!!!!!!! May 31 14:00:26 march is qt 4.7.1 or so and latest is 4.7.3 May 31 14:00:34 there could be lot of changes in webkit May 31 14:00:44 woglinde: yes of course May 31 14:01:04 so its not I hade qt 4.7.3 than and now its not running anymore May 31 14:01:15 woglinde: BTW this error with gcc seems not related to qt May 31 14:01:38 its in qtwebkit code May 31 14:01:51 so choose yourself it is qt or not May 31 14:03:54 woglinde: considering that I'm probably the only one building on Debian Lenny x86_64 I'd prefer to comply with the most May 31 14:05:01 mckoan yes and I consider you are the only one compiling qt for armv5 May 31 14:07:56 woglinde: probably I've missed something recently in AT91 machines May 31 14:09:45 mckoan did you try to compile for beagle? May 31 14:10:02 if it fails too May 31 14:10:20 woglinde: actually I don't care about beagle, but I could try May 31 14:10:21 we have some more knowledge May 31 14:10:31 woglinde: I see May 31 14:11:14 in that case is OE ignoring armv5 boards? May 31 14:14:41 not as a matter of policy, but someone who cares about armv5 needs to make sure that they go on working. May 31 14:15:25 OE is not a conscious entity May 31 14:15:57 Not yet, anyway. May 31 14:16:00 mckoan dont think so, but most people now have arvm7a stuff May 31 14:16:04 more a distributed intelligence, an some parts are less aactive than others May 31 14:16:06 so its better tested May 31 14:17:08 but there are a plenty of boards and people still using armv5 ! May 31 14:17:25 not TI branded of course ;-) May 31 14:17:32 sure, I use armv5 myself May 31 14:17:50 pb did you try to compile qt 4.7.3? May 31 14:18:05 but I don't happen to care much about qt, at least on that platform, so I haven't tested the exact combo you are trying to use. May 31 14:18:36 to be honest, though, rather than speculating about what might or might not have caused this bug to surface, I think you would be better off just trying a newer linker and see if that fixes it. May 31 14:19:27 irrespective of whether it's newly introduced in qt 4.7.3 or not, the crash is almost certainly not qt's fault and qt is not going to be the place to install a fix. May 31 14:20:06 * mckoan is concerned that his worst fears are materializing May 31 14:20:08 similarly, it isn't obvious to me that knowing whether it works on armv7 or not is really going to help in finding a solution. May 31 14:20:11 till later May 31 14:20:27 I hope I can catch my son before the thunderstorm begins May 31 14:20:35 its getting dark already May 31 14:20:45 that sounds ominous. is someone throwing him off a tall building? May 31 14:21:00 woglinde: rush! :-) May 31 14:25:48 mckoan: what are the "worst fears" of which you speak? May 31 14:32:14 pb_: I fear that some BIG players can drive OE to their interest by reducing the number of machines and fatures May 31 14:32:33 it's already happening May 31 14:32:53 that is a concern that I share, but I don't think there is any evidence that it is already happening May 31 14:33:28 pb_: I migrated to OE because it was the most open and comprehensive May 31 14:33:51 yeah, and I think that is still true May 31 14:34:17 pb_: now it seems is becoming a commodity for larger projects May 31 14:35:01 maybe it's time to take another leap May 31 14:36:31 mckoan: some devs have already expressed any further interest for oe-classic (how they called it) May 31 14:36:40 others are silent May 31 14:36:44 BTW for now my main problem is to have a working system on ARMv5 May 31 14:37:41 which machines and features do you feel have been driven off by the big players? May 31 14:38:12 mckoan: we (zaurus hobbists) have strongarm/armv4 and armv5 to support May 31 14:38:15 pb_: should be easy to guess May 31 14:38:57 ant_work: have you recently built successfully? May 31 14:39:10 yes, apart udev / tmp issues May 31 14:39:24 but any real qt image, just opie May 31 14:40:26 any serious issue compiling, most problems arise testing runtime on real hw... May 31 14:41:24 guys, is there a way to use external (codesourcery) toolchain with oe? May 31 14:41:40 I certainly haven't noticed any thing or any one being driven off lately and there doesn't seem to be any evidence that (for example) oe-core is turning into an intel-centric zone. May 31 14:42:40 if you're talking about armv5 then I think you are blowing your current problem out of proportion. what you have encountered is a linker bug, not some conspiracy. May 31 14:43:20 oe has never provided a guarantee that either: (a) the tree will be free of bugs on any given day, or (b) someone else will automatically fix the bugs for you. May 31 14:44:15 as I mentioned earlier, the most obvious way to proceed with a linker bug is to firstly try a newer/better/different linker to see if that solves the problem, and if it doesn't to either debug the crash or send a bug report to the maintainers. May 31 14:44:16 pb_: of course ;-) May 31 14:44:42 pb_: I'm doing more test, thank you May 31 14:46:13 There is no conspiracy. OE is entirely dependent on people testing the different machines May 31 14:46:25 anarsoul, have a look at the local.conf example it should give you a hint on setting up codesourcery external toolchain May 31 14:47:21 now, it probably is true that the proportion of people actively using armv5 is smaller now than it was a couple of years ago, but that's just a function of hardware progress. it is natural that most of the active development is going to be on the newer platforms. May 31 14:47:23 gchiii: thanks May 31 14:48:23 and, equally, it is natural (and expected) that folks like intel and ti will want to spend their effort on new platforms rather than legacy stuff, because that's where they make their money. but none of that prevents anybody who is interested in keeping armv5 working from doing so. May 31 14:49:29 sure, pxa kernel seems actively maintained May 31 14:50:06 right, and there's the whole meta-handhelds thing going on as well May 31 14:50:18 and slugos, which (iirc) targets armv5 primarily May 31 14:53:17 if ti/intel/whoever were trying actively to force some kind of change into oe which would prevent older platforms from working or make former usecases impossible then I would agree with you, that would definitely be a bad thing and should be fiercely resisted. but, as far as I can tell, that is not happening. May 31 14:53:36 and if it does happen, we will push back May 31 14:53:37 mckoan: fwiw to avoid unneeded variability I still build on 32 bit linux May 31 14:54:03 I'm pretty sure oe-core will be more immune to contamination May 31 14:54:06 and of course, if you think it is happening, you should stand for election to the tsc and then you will be able to stop it. May 31 14:56:15 is someone using udev 171? May 31 14:56:27 it seems to have broken our setup May 31 14:59:14 otavio: I was thinking to try it this evening to log and see why udev refuses here to mount root rw... May 31 14:59:41 ant_work: here it doesn't probe the modules et all May 31 14:59:53 ant_work: even if I do udevadm trigger --action=add May 31 15:00:27 not yet there... could not write /tmp/uname May 31 15:00:39 unless I add tempfs hacking base-files May 31 15:00:50 heavily May 31 15:01:12 * otavio sets udev to 164 and try May 31 15:02:31 otavio: fwiw there is a bigger problem.... many images do not deploy kernel-modules at all May 31 15:02:32 ant_work: would be a good idea :-) May 31 15:02:53 ie. console-, x-11 do not May 31 15:02:56 ant_work: yes. but mine does :P May 31 15:02:58 x11-gpe does May 31 15:03:01 k May 31 15:03:06 ant_work: and without it I even lack network hhehehe May 31 15:03:14 he May 31 15:04:21 I'm under the impression this breakage is recent...shame on me for I did not test runtime for too long ;) May 31 15:06:24 * otavio confirm it works and send an email about it May 31 15:06:34 good news May 31 15:07:06 hm so koen has more todo May 31 15:07:06 * ant_work will probably test an udev-free image instead ;) May 31 15:07:24 woglinde: hopefully May 31 15:07:33 woglinde: finishing test May 31 15:07:46 thsome rules have been separated iirc May 31 15:08:03 ant_work: but modprobe shouldn't May 31 15:08:05 heh, the binconfig mangling in oe-core loses quite badly when prefix is empty. I guess some or other patch needs backporting from oe master for that. May 31 15:09:41 wow, sensible bug May 31 15:10:23 still fighting with usr-less distro? May 31 15:10:55 yeah. well, not much fighting at the moment, though I just discovered that freetype-config in oe-core is coming out all wrong. May 31 15:11:11 I guess there aren't many other binconfig-using packages left. May 31 15:11:36 it's sort of tempting to just burn freetype-config altogether and replace it with pkg-config. May 31 15:12:23 pb_: or do a wrapper against it that uses pkg-config May 31 15:12:43 yeah, that's what I meant May 31 15:14:05 ant_work: yes; udev 164 works May 31 15:16:58 heh, yeah, oe-core is missing 952e5e2b7a5c1deefc939594d40b81a71fb16a54 May 31 15:17:02 from october 2008 :-} May 31 15:17:05 lol May 31 15:17:08 !!!!!!!!!!!!!!! May 31 15:17:09 woot May 31 15:17:32 well, the nice thing about fixing everything twice is that it's always much easier the second time around May 31 15:19:09 * woglinde waits for the pull request from pb May 31 15:32:26 hm, and glib-2.0_git (in meta-oe) seems to lack a dependency on python-argparse May 31 15:32:37 * pb_ builds the one from oe-core instead May 31 15:42:42 pb_: does it? http://git.openembedded.org/cgit.cgi/meta-openembedded/commit/?id=10d8c7ae116e231ed175e3a021e8e22d70a2141c May 31 15:44:12 pb_: and strange if it was picked for you even with D_P -1 May 31 15:50:06 ah, that's strange. May 31 15:50:20 this was regular glib-2.0, not the -native version. I didn't think my meta-oe tree was that old but maybe it is. May 31 15:50:36 a fresh rebuild is still in progress, thank you pb_ woglinde ant_work May 31 15:50:43 have anice rest of the day May 31 15:50:50 okay, good luck May 31 15:50:50 mckoan no prob May 31 15:50:57 yeah from me too May 31 15:51:09 hm so now I should have all ready May 31 15:51:18 for uclibc and CPU funcs May 31 15:51:26 let us try to compile it May 31 15:51:28 hm, yeah, ddc40247cb772c59b98b95111dc3ad09d7c38375 is the latest rev I have in meta-oe. I guess I should pull the new bits. May 31 15:52:58 but it is still funny that it picked that one with D_P=-1. May 31 15:54:34 morning kergoth May 31 15:54:59 hi kergoth May 31 15:59:13 03Tom Rini  07master * rd61ba0ce8f 10openembedded.git/recipes/iproute2/ (iproute2-2.6.38/configure-cross.patch iproute2_2.6.38.bb): May 31 15:59:13 iproute2: Add 2.6.38 version May 31 15:59:13 Signed-off-by: Tom Rini May 31 17:02:23 hm whats the diffrent btw execve and execvpe? May 31 17:04:09 hm ah May 31 17:04:10 okay May 31 17:04:17 ones search the actual dir May 31 17:04:26 the other searches in path May 31 17:06:16 khem`: there? May 31 17:06:55 khem`: you can probably help me ... which var I can use to gather the CC for building for native, not target? May 31 17:09:04 khem`: got it May 31 17:20:21 read bitbake.conf May 31 17:28:16 jo khem May 31 17:34:55 Hi woglinde May 31 17:35:23 kergoth: I fail to get the info I need May 31 17:35:49 khem: basically I need to know the var to use that will be with right values for cross and target builds May 31 17:35:58 khem: that I can use in place of TARGET_SYS May 31 17:36:14 khem: and also the CC_ARCH var to use. May 31 17:38:06 "the var to use that will be with right values"? May 31 17:38:28 kergoth: I am building a cross and target binaries May 31 17:38:30 usually when you inherit cross it should do the right things for you May 31 17:38:47 kergoth: yes but I need to discover the vars to use May 31 17:38:56 kergoth: what the CC_ARCH to use, for example May 31 17:39:04 * woglinde fights with ifdef and if defined May 31 17:39:05 otavio: what problems do u need to solve now May 31 17:39:21 khem: I got it built and installed into sysroot May 31 17:39:23 woglinde: look for if.*def :) May 31 17:39:30 khem: but it got built to target, not host May 31 17:39:49 otavio: ok is that the cross recipe ? May 31 17:39:56 khem: yep May 31 17:39:56 khem I am trying to hack execvpe into uclibc May 31 17:43:44 otavio: ok you can export CC="${BUILD_CC}" May 31 17:44:06 * otavio checks May 31 17:44:41 * otavio test May 31 17:44:45 and it could be a better if chicken supported some sort of sysroot option but not must May 31 17:45:07 khem: it doesn't but the install seems right. May 31 17:45:17 khem: now I am trying to build it for right arches May 31 17:45:21 k May 31 17:47:58 so now lets see May 31 17:48:04 if the hack works May 31 17:49:59 I have a problem, bitbake virtual/kernel seems to building everything (all the other stuff like tcl, perl etc) May 31 17:50:23 Is that normal? I want to build just the kernel in OE May 31 17:50:50 hmmm so origen board is on similar lines as panda but has samsung soc May 31 17:51:16 joelagnel: oe has to build all dependencies too May 31 17:51:26 that one would need to build kernel May 31 17:51:30 e.g. toolchain May 31 17:51:39 and then toolchain might have other deps pulle din May 31 17:51:42 hum May 31 17:51:42 oh, I actually already have the toolchain May 31 17:52:09 ok then you need to check deptree and make sure if dependencies are right May 31 17:52:09 khem: any suggestion how I can pass 'BUILD_CC' on C_COMPILER param for make? May 31 17:52:21 sometimes one wrong dependency can screw the depchain May 31 17:52:37 The thing is I can build the kernel manually without OE May 31 17:52:42 otavio: EXTRA_OEMAKE I would think May 31 17:52:45 I just need OE because it nicely applies patches May 31 17:53:03 khem: the problem is BUILD_CC has spaces May 31 17:53:03 joelagnel: heh it does a lot more than patching May 31 17:53:03 so I want to have OE do it for me without building perl/tcl etc May 31 17:53:09 you can use quilt for patching May 31 17:53:14 its same what oe does May 31 17:53:42 otavio: quote it May 31 17:53:49 in \"\" May 31 17:53:57 do you think if I tell OE where the toolchain is in the conf, will it build _only_ the kernel? May 31 17:54:09 that is when I say bitbake virtual/kernel May 31 17:54:34 joelagnel: you can but you have to provide the dependency information somehoe May 31 17:54:45 may be a lonh list of ASSUME_PROVIDED May 31 17:54:57 ah May 31 17:55:14 OE_TRUST_ME=1 :p May 31 17:55:15 http://www.origenboard.org/index.php?mid=about_page May 31 17:56:06 khem: it calls: make 'C_COMPILER="ccache' gcc '"' BINDIR... May 31 17:56:08 khem do you know how uclibc determines __GI_execvp symbols? May 31 18:02:48 woglinde: those are hidden symbols you need to add libc_hidden_def and libc_hidden_proto May 31 18:03:04 for this function May 31 18:06:27 khem can you take a look http://pastebin.ca/2072886 May 31 18:06:44 i get May 31 18:06:45 libc/libc_so.a(execvpe.os): In function `*__GI_execvp': May 31 18:06:46 | /devel/arm/oetmp-ang/work/beagleboard-angstrom-linux-uclibceabi/uclibc-git-r35.11+gitr0+cf0b3a3342f24dbf601ba639bdce5b2a2f001c7a/git/libc/unistd/exec.c:237: multiple definition of `__GI_execvp' May 31 18:06:50 at linktime May 31 18:08:31 hm I will try without the double defines May 31 18:11:43 hm ah good May 31 18:11:46 now its working May 31 18:11:58 gogo systemd May 31 18:13:46 argssggsss!!!! May 31 18:13:56 systemd needs unshare too May 31 18:13:58 hrms May 31 18:21:10 03Simon Busch  07master * r99c3e4062a 10openembedded.git/recipes/freesmartphone/qfsodbusxml2cpp-native_git.bb: May 31 18:21:10 qfsodbusxml2cpp-native: add initial recipe May 31 18:21:10 Signed-off-by: Simon Busch May 31 18:21:20 03Simon Busch  07master * rbad11dc46b 10openembedded.git/recipes/freesmartphone/aurora_git.bb: May 31 18:21:20 aurora: make it suitable for the latest version and bump PR/SRCREV May 31 18:21:20 Signed-off-by: Simon Busch May 31 18:24:52 03Koen Kooi  07org.openembedded.dev * r8dfc85fe6e 10openembedded.git/recipes/linux/ (8 files in 3 dirs): May 31 18:24:52 linux-omap 2.6.39: sync with meta-texasinstruments for bootup fixes May 31 18:24:52 Signed-off-by: Koen Kooi May 31 18:38:25 khem: I got it built but now the no-cross one builds for host too heh May 31 18:38:40 he stefan_schmidt May 31 18:41:00 hi woglinde May 31 18:41:01 hi all May 31 18:44:14 otavio: hmmm May 31 18:44:20 otavio: show me your recipes May 31 18:44:49 fedora 15 will have systemd hmm May 31 18:51:00 khem you mean has May 31 18:51:05 fc15 is already out May 31 18:51:12 with broken imap stuff gedit May 31 18:51:14 khem: I am doing a last test; let me see how it goes May 31 18:51:15 and some other May 31 18:51:24 khem: so I don't waste your time May 31 18:56:01 khem: OK; it built fine. Now moving to the compiler itself May 31 18:59:51 woglinde: yes it released last week May 31 19:06:20 *sigh* May 31 19:06:27 now I have support for unshare too May 31 19:06:37 let's see if systemd compiles now May 31 19:20:17 khem I got it finally compiled May 31 19:20:36 now its time for watching rest of basketball May 31 19:36:50 woglinde: ok May 31 19:44:33 anyone know why the init scripe installed with asterisk does not actuall start asterisk? May 31 19:46:13 Crofton|work: sorry, no idea. does it try to read /etc/default/asterisk? May 31 19:46:55 not sure, but dropping the -U asterisk -G asterisk from the scrip0t lets it start May 31 19:47:04 some kind of permission issue May 31 19:47:50 Crofton|work: ah.. do you have the group/user? May 31 19:47:56 they exist May 31 19:48:32 oh.. no idea then May 31 19:51:06 [pid 526] open("/var/log/asterisk/queue_log", O_WRONLY|O_CREAT|O_APPEND|O_LARGEFILE, 0666) = -1 EACCES (Permission denied) May 31 19:51:07 :) May 31 19:51:16 fails trying to open the logfile May 31 19:52:45 bluelightning: hi May 31 19:52:56 hi anarsoul May 31 19:53:03 bluelightning: what's status with .39 on your h1940? May 31 19:53:52 anarsoul: I've still yet to create a uimage... I wasn't sure what the correct addresses were May 31 19:56:00 Crofton: I think your /var/log/asterisk/ is not writable by user asterisk May 31 19:56:54 yes May 31 19:57:17 I do not think it gets created right May 31 20:02:18 jupp May 31 20:19:07 03Simon Busch  07master * r9e2ffc85de 10openembedded.git/recipes/freesmartphone/aurora_git.bb: May 31 20:19:08 aurora: supply QT_DIR_NAME to configure script and bump PR/SRCREV May 31 20:19:08 Signed-off-by: Simon Busch May 31 20:30:25 03Tom Rini  07master * r1a1f68d70a 10openembedded.git/recipes/lmbench/ (lmbench-3.0-a9/obey-ranlib.patch lmbench_3.0-a9.bb): May 31 20:30:26 lmbench: Obey ranlib May 31 20:30:26 In some cases the host may have too old of a ranlib to work. Since it's May 31 20:30:26 not exactly a great idea to not be using the cross ranlib anyhow, fix the May 31 20:30:26 Makefile so we can override ranlib and then override it. May 31 20:30:26 Signed-off-by: Tom Rini May 31 20:31:53 Tartarus: dont use RHEL 3.x :) May 31 20:32:30 5 :) May 31 20:32:40 even 5 May 31 20:32:43 hmm May 31 20:32:50 Tell $customer, of course, heh May 31 20:32:57 And then tell fray's customers too May 31 20:33:23 (of course, this is also a correctness thing, don't call host ranlib, call your cross ranlib) May 31 20:33:28 heh do they still want cygwin May 31 20:34:28 Oh someone somewhere does I'm sure May 31 20:34:35 sure the patch is betterment is any case May 31 20:34:37 I know May 31 20:34:56 or solaris May 31 20:34:59 or HPUX May 31 20:36:31 ah hpux May 31 20:37:20 I have a collegue here who is very sentimental about HPUX May 31 20:37:28 since he developed that for 22 years May 31 20:37:45 I just remember there was an mvista customer that sent in an hpux box May 31 20:37:50 I saw it one of my trips there May 31 20:38:33 yeah I did a bit of work on HPUX well it was not much May 31 20:38:50 getting intel's compiler to build on it May 31 20:39:44 why does meta-zaurus belong to meta-smartphone ? I don't see the smartphone part in zauri May 31 20:40:00 it doesn't May 31 20:40:07 dcordes: its just meta meta layer May 31 20:40:39 dcordes: read meta-smartphone/meta-zaurus/README May 31 20:40:40 kind of place holder like meta-openembedded holds many layers beneath May 31 20:44:16 hmm so now you can read a iWorks on your iPods May 31 20:44:43 well for iPad it totally makes sense May 31 20:49:59 JaMa|Wrk: ok May 31 20:51:18 JaMa|Wrk: do u use qt as well ? May 31 20:51:25 we need a meta-qt layer May 31 20:51:27 as well May 31 20:51:38 to complete the mosaic May 31 20:53:37 khem: only qml for navit (from oe-core qt4-x11-free) May 31 20:53:49 but haven't tested it in runtime yet May 31 20:55:57 JaMa|Wrk: there is a qml frontend for navit? May 31 20:58:53 morphis: yes May 31 21:38:37 03Stanislav Brabec  07master * rd26f4f6958 10openembedded.git/recipes/gammu/gammu_1.26.1.bb: May 31 21:38:37 gammu: Upgrade to version 1.29.0 and fix build. May 31 21:38:37 Signed-off-by: Stanislav Brabec May 31 21:40:18 03Stanislav Brabec  07master * r936cf6aac9 10openembedded.git/recipes/navit/navit_svn.bb: May 31 21:40:18 navit: Update to SVN revision 4503. May 31 21:40:18 Signed-off-by: Stanislav Brabec May 31 21:43:52 03Stanislav Brabec  07master * rdbc1e12e03 10openembedded.git/recipes/mtkbabel/ (files/fast-logging.patch mtkbabel_0.8.bb): May 31 21:43:52 mtkbabel: Upgrade to version 0.8.2. May 31 21:43:52 Signed-off-by: Stanislav Brabec May 31 21:53:04 bitbake with gitorious SRC_URI .. anybody experienced problems before ? May 31 21:58:53 otavio: hm, about the cross recipe, we are trying to put klcc-cross in shape. Same issues have been faced I think. HAve a look at that May 31 22:01:35 (there is still that strange 'fixmepath' in sysroot and 'FIXMESTAGINGDIR' in th eruntime pkgdata May 31 22:01:38 ) May 31 22:03:02 hm parsing seems so mega smooth with the streamlined metadata May 31 22:04:55 03Richard Purdie  07master * r606f1acc6f 10bitbake.git/lib/bb/providers.py: (log message trimmed) May 31 22:04:55 providers.py: Correct PREFERRED_VERSION handling May 31 22:04:55 Overrides on the right are the highest priority and in this case, pn-PN May 31 22:04:55 and PN should take priority over any other override so fix the code to May 31 22:04:55 do this. May 31 22:04:56 Also, since overrides will have been processed by bitbake, we shouldn't May 31 22:04:57 then be specifically looking up PREFERRED_VERSION_${PN} but just using May 31 22:29:45 03Stanislav Brabec  07master * r06d31dc06b 10openembedded.git/recipes/lynx/lynx_2.8.7.bb: May 31 22:29:45 lynx: New recipe, version 2.8.7. May 31 22:29:45 Signed-off-by: Stanislav Brabec May 31 22:46:17 03Stanislav Brabec  07master * r7aa5d79911 10openembedded.git/recipes/lynx/ (files/locale-charset.patch lynx_2.8.7.bb): May 31 22:46:17 lynx: New recipe, version 2.8.7. May 31 22:46:17 Signed-off-by: Stanislav Brabec May 31 22:46:27 03Stanislav Brabec  07master * r07a373d50c 10openembedded.git/: May 31 22:46:28 Merge branch 'master' of git.openembedded.net:openembedded (adding missed patch). May 31 22:46:28 Conflicts: May 31 22:46:28 recipes/lynx/lynx_2.8.7.bb May 31 23:19:33 where to send patches for meta-shr and meta-smartphone ? Jun 01 01:23:18 03Tom Rini  07master * re173373ca5 10openembedded.git/recipes/perl/perl-native_5.10.1.bb: Jun 01 01:23:18 perl-native: Bump PR Jun 01 01:23:18 This will make the utils.bbclass change that fixes using perl${PV} Jun 01 01:23:18 take effect for all users. Jun 01 01:23:18 Signed-off-by: Tom Rini Jun 01 01:23:23 03Andreas Müller  07master * rd8d2e1730a 10openembedded.git/classes/utils.bbclass: (log message trimmed) Jun 01 01:23:23 utils.bbclass: modify create_wrapper to correctly follow symlinks Jun 01 01:23:23 * taken from OE-Core [1] Jun 01 01:23:23 * thanks Khem for pointing out[2] Jun 01 01:23:23 * fixes build failure perl Jun 01 01:23:23 | env: ./perl5.10.1.real: No such file or directory Jun 01 01:23:24 | make[1]: *** [pod/perlapi.pod] Error 127 Jun 01 01:23:28 03Tom Rini  07master * rfffb908d5a 10openembedded.git/recipes/lm_sensors/lmsensors-apps_3.2.0.bb: Jun 01 01:23:28 lmsensors-apps: Various important fixes Jun 01 01:23:28 - Add DEPENDS = "bison-native flex-native" Jun 01 01:23:28 - Pass in CC/AR so that we build for the target Jun 01 01:23:28 - Use EXTRA_OEMAKE to pass in args Jun 01 01:23:28 - Set PACKAGE_ARCH = ${MACHINE_ARCH} since we reference kernel stuff. Jun 01 01:23:28 Signed-off-by: Tom Rini **** ENDING LOGGING AT Wed Jun 01 02:59:56 2011