**** BEGIN LOGGING AT Wed Jul 29 02:59:57 2009 Jul 29 04:07:00 khem: hi, what's the problem with eglibc and thumb instruction set ? Jul 29 04:07:12 (I just saw some irc log) Jul 29 04:39:59 hello Jul 29 04:40:12 is there ANY way i can get rid of madwifi-ng from being built? Jul 29 04:40:22 it just spews non-stop errors and i don't need it at all Jul 29 04:40:27 I am angry and at the end of my rope. Jul 29 04:44:33 TSCHAKeee: why dont you hang yourself on the rope then ? :-) Jul 29 04:44:42 TSCHAKeee: anyway ... what distro are you building ? Jul 29 04:44:51 angstrom-2008, trying to anyway Jul 29 04:44:53 x86 target Jul 29 04:45:14 trying to build to a very specific geode based embedded machine that i'm trying to sculpt eventually into a machine configuration file. Jul 29 04:45:22 using the x86 machine conf for now Jul 29 04:45:37 is there any way i can tell madwifi-ng not to build? Jul 29 04:45:41 i don't need it for this device. Jul 29 04:45:52 and i am trying to build base-image Jul 29 04:46:52 Im not an OE guru, for starters, what about just plain deleting it from recipes ? ;-) Jul 29 04:47:48 that's a scary thought, wouldn't that cause the dependency chain to go apeshit? Jul 29 04:48:45 dunno, you'll see Jul 29 04:49:48 sigh Jul 29 04:49:51 it'll at least show you what needs madwifi Jul 29 04:50:10 argh, I told you Im no OE guru ... you should wait for someone more into it ;-) Jul 29 04:50:11 * TSCHAKeee goes back to reading what little documentation exists Jul 29 05:27:45 hello, any experts alive/ Jul 29 06:28:52 good morning Jul 29 06:55:40 good morning Jul 29 07:19:30 Has enyone here experience with mpc51xx from freescale ?? Jul 29 08:25:31 morning Jul 29 08:26:48 jo hrw Jul 29 08:28:03 hi henning Jul 29 08:33:39 jo florian Jul 29 08:34:34 good morning Jul 29 08:40:11 heh... 2 years old metadata is like digging in trashcan Jul 29 08:41:14 hehe Jul 29 08:41:51 hrw: i see - you want to build a really small filesystem ;) Jul 29 08:44:16 florian: no, I want to use some acceleration available in HW. but it is for xserver-kdrive 1.3.0.0 which do not builds in angstrom anymore Jul 29 08:44:23 too old it is Jul 29 08:46:53 hrw: I'm tempted to start a more flexible distribution that doesn't enfoce all latest stuff but allows more long time maintainable filesystems. Jul 29 08:47:32 I'm not happy with latest Angstrom... its way too much focused on fast platforms like the Beagle Jul 29 08:53:00 XorA: people is wondering about the uImage duplicate code... Jul 29 08:53:47 florian latest angstroem is too much focused on koen *g* Jul 29 08:54:10 ant_work: and once again I answered, but once again my answer will be ignored Jul 29 08:54:19 he he Jul 29 08:54:47 woglinde: hum yes, that might be the point Jul 29 08:55:18 bah, Ill point out I use Angstrom as well as koen Jul 29 08:55:37 use and dictator is something different Jul 29 08:56:07 ok, develop then Jul 29 08:56:29 * ant_work foresees a flame-war for the USE-Flags thing Jul 29 08:56:37 not again Jul 29 08:56:43 ant_work: USE_FLAGS is coming Jul 29 08:58:22 I have no strong opinion (but I abuse them in Gentoo...) Jul 29 08:58:54 ant_work: from the Angstrom point of view, we need some way to make sure builds are invalidated when use flags change Jul 29 08:59:11 ant_work: and this is the one feature mising in gentoo as well Jul 29 09:01:37 well, the actual debian-like package splitting/naming is sometimes confusing for me. Still is perfectly valid for some purposes (routers,...). Jul 29 09:02:01 btw in Gentoo with the --newuse option one can track the changes Jul 29 09:02:18 ant_work: how does that markup the binary packages though? Jul 29 09:02:46 ok, you would need an hash of the useflags string ... Jul 29 09:03:04 ant_work: the problem we are very repidly approaching is we are going to end up with EXTRA_OECONF_${DISTRO} = "blah" in pretty much every recipe Jul 29 09:03:14 ant_work: or blah-angstrom repeated a million times Jul 29 09:03:23 I see Jul 29 09:03:36 as I bet micro and minimal want different options than Angstrom Jul 29 09:04:04 but Angstrom being a distro with paying customers want to maintain some sense of order on binary feeds Jul 29 09:05:14 xora didnt you set libtool-2.2.6a to def_pref -1? Jul 29 09:05:49 woglinde: erm I had meant to, Im guessing the question you ask means I fucked up Jul 29 09:06:32 xora just wonderd why micro builds it Jul 29 09:07:03 woglinde: it should harm it, works in my tests here, but curses if I forgot to set that line Jul 29 09:07:21 *g* I know that it works Jul 29 09:07:31 but your lovely koen might be confused Jul 29 09:07:36 ups our Jul 29 09:08:12 woglinde: Angstrom is already scheduled to upgrade Jul 29 09:08:16 XorA: ant_work: did you ever look at macports? and their variants? Jul 29 09:08:30 zecke: macports? Jul 29 09:09:11 XorA: google, it is a port system for OSX Jul 29 09:09:23 zecke: ah cool, they solved the issue I guess Jul 29 09:09:45 XorA: and it has variants. E.g. I have git, and sometimes I want to have git-svn and git-server-daemon... Jul 29 09:09:52 XorA: no, mi wife could kill me if I touch the HFS+ partition :) Jul 29 09:10:03 haha Jul 29 09:10:19 it's already a miracle it boots 10.5.7 Jul 29 09:10:19 XorA: so you do port install git +svn +server, and you can query the variants of each provider Jul 29 09:10:31 ant_work: use zfs? Jul 29 09:10:51 well, the bootloader is special Jul 29 09:10:58 zecke: reading the docs now Jul 29 09:11:03 and magically git +svn +server will replace git, git+svn Jul 29 09:11:14 (our problem is how to map this onto ipkg...) Jul 29 09:13:14 khem: the build failed in the end last night Jul 29 09:13:33 khem: apparently uclib-initial is later in the build Jul 29 09:13:35 zecke: the defining of variants and their handling looks like it would fit into bb style well Jul 29 09:14:30 XorA: yes, I like that part too, the selection of which variant you want when building "bluez" is a bit different (for packaged things) Jul 29 09:18:23 XorA: https://trac.macports.org/ticket/10618 Jul 29 09:22:47 anyway I need to return to paid for code Jul 29 09:23:04 XorA: look at this after work: http://gobo.kundor.org/wiki/Compile#Useflags_and_Compile Jul 29 09:23:29 "the poor-man's portage" Jul 29 09:24:03 it looks like there are a lot of ways to achieve the result Jul 29 09:24:34 ant_work: really the compiling is not an issue, packaging is the issue Jul 29 09:55:42 ~kill gtk+ developers which maintain file requester Jul 29 09:55:43 * ibot shoots a excited photon gun at gtk+ developers which maintain file requester Jul 29 10:07:05 * * OE Bug 5288 has been created by superchaits(AT)gmail.com Jul 29 10:07:07 * * md5/sha256 checksum mismatch for alsa-utils-1.0.20 Jul 29 10:07:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5288 Jul 29 10:41:38 morning all Jul 29 10:43:38 Has anyone actually come up with a proposal for "USEFLAGS" in OE? Jul 29 10:46:18 hi RP, sort of Jul 29 10:48:44 hi RP Jul 29 10:49:32 * RP waits for the detailed proposal with all the issues thought through Jul 29 10:51:07 hrw: it is dangerous to fire in a black hole Jul 29 10:51:14 RP: by the way the parser speed is way better now (commit revert + cache-patch ) Jul 29 10:52:03 RP: no, just been kicking a few ideas around recently Jul 29 10:52:28 RP: the packaging one is the biggest I feel Jul 29 10:54:05 XorA: packaging one? Jul 29 10:54:16 ant_work: cool :) Jul 29 10:54:30 RP: how to tell the difference between two package generated with different flags/variants/blah Jul 29 10:54:34 The question is now whether those patches can be backported to the 1.8 series Jul 29 10:55:48 XorA: What worries me more is how we manipulate DEPENDS and EXTRA_OECONF based on these magic flags Jul 29 10:56:01 more anonymous python? Jul 29 10:56:19 It stands a good chance of making the recipes unreadable Jul 29 10:56:58 RP: zecke pointed me at macports which has a nice syntax Jul 29 10:58:07 XorA: Do you have an example? Jul 29 10:58:56 RP: http://guide.macports.org/#using.variants Jul 29 10:59:38 RP: sorry wrong link http://guide.macports.org/#development.variants Jul 29 11:02:09 XorA: and how does this differ from a recipe which does "inherit base-recipe.bb\n EXTRA_OECONF=foo\n DEPENDS=bar\n" ? Jul 29 11:03:27 That way PN changes too meaning different package names. The main tricky part will be selecting between different virtual providers Jul 29 11:04:06 RP: that is another way to look at it Jul 29 11:04:09 <_dash__> Do people talk about oe and bitbake related issues here ? Jul 29 11:04:31 Provides: blah, blah-ssl, blah-fishing Jul 29 11:04:41 Depends: blah-fishing Jul 29 11:04:46 hmmm Jul 29 11:05:09 _dash__: yes, in general just ask and see if you get a question Jul 29 11:06:05 although that starts to get messy when a package has loads of options Jul 29 11:06:22 XorA: It gets messy regardless of how you do it... Jul 29 11:06:43 RP: well, PV bumps are way more easy with macports variants Jul 29 11:06:58 RP: no way of forgetting half of the old recipes when moving them Jul 29 11:07:05 RP: something new to consider though Jul 29 11:07:22 zecke: What is it about their design that makes it easy though? Jul 29 11:07:27 RP: and I see it like this One Template Recipe < Recipe Family < PAckages Jul 29 11:07:44 RP: everything in one file, just move the file.. all variants are "migrated" to the new base already Jul 29 11:07:45 XorA: certainly, I'm just playing devils advocate Jul 29 11:08:03 zecke: ah, right. So you want everything in one file Jul 29 11:08:29 RP: I think it is nice, they somehow belong together. Or variants/flags are attached to a given PN-PV-PR Jul 29 11:08:50 zecke: In a way its kind of like my BBCLASSEXTEND stuff Jul 29 11:08:51 (but in contrast to gentoo and macports... distributing packages is something we have to do and they don't) Jul 29 11:08:57 that merges sdk/native/normal versions Jul 29 11:09:14 I'm not aware of that, something for me to read up on :} Jul 29 11:09:41 zecke: There were posts to the list about it a while back and its now merged into bitbake Jul 29 11:09:47 <_dash__> ok ... I am facing issue while building oe kernel .... "angstrom-dev/staging/i686-linux/usr/bin/mkimage: Can't map arch/arm/boot/uImage: No such device" Jul 29 11:10:28 <_dash__> My Image and zImage get created in kernel mother dir Jul 29 11:10:33 I think one file with all variants is easier to track changes in Jul 29 11:11:00 <_dash__> zecke: this is what my problem is Jul 29 11:11:45 _dash__: sorry, that is not my area of knowledge Jul 29 11:11:56 RP: no worry, I just need to read up on it Jul 29 11:12:40 * hrw -> food Jul 29 11:12:48 XorA: I think there could be some neat syntax to create a number of varients from one file like the BBCLASSEXTEND stuff Jul 29 11:12:57 <_dash__> zecke: So who will be the right person? Jul 29 11:13:30 _dash__: no idea, the joy of irc :) Jul 29 11:13:38 RP: yes, not saying we should use exactly macports syntax, but its a nice example of one way, only at fledgling thoughts at moment Jul 29 11:13:51 * XorA is busier writing drivers than OE code at moment Jul 29 11:13:55 BBVARIENTS = "ping bing pong" DEPENDS_var-pong = "" DEPENDS_var-ping = "" Jul 29 11:14:28 That would be a very OE way to do it :) Jul 29 11:14:44 using overrides and mostly existing infrastructure Jul 29 11:16:20 <_dash__> Anybody here to dicuss bitbake issue while building OE kernel ?? Jul 29 11:17:46 _dash__: well, I can help with common sense and hints. that is which task of which recipe? Jul 29 11:17:52 RP: the USE flags are powerful because you can have them deselected (USE="ctype gd imap jpeg mysqli php png ppds samba slang spell vhosts" \ Jul 29 11:17:54 USE="${USE} -ldap -snmp -mailwrapper -cups") Jul 29 11:18:04 _dash__: what do you have in arch/arm/boot/? do you have a zImage but no uImage? Jul 29 11:18:27 ant_work: how do you see which ebuild supports which flags? Jul 29 11:18:49 I don't know.. : ) but I imagine -bluez is one of the most-awaited Jul 29 11:19:06 zecke: Do have a look at the BBCLASSEXTEND stuff, I was quite pleased with its perfornance :) Jul 29 11:19:11 zecke: ypu mean in Gentoo? Jul 29 11:19:11 zecke: Its used a lot in Poky now Jul 29 11:19:37 zecke: equery Jul 29 11:21:04 It is somewhat worrying that most USE flags systems don't generate package feeds Jul 29 11:21:52 ant_work: and in the ebuild. how do you mention which flags you provide/honor? Jul 29 11:22:02 RP: i looked at the bitbake diff Jul 29 11:22:19 RP: stupid question. is your poky mostly done in the spare time? Jul 29 11:22:25 inser a "work" there Jul 29 11:22:40 zecke: at times it has been, at the moment, no Jul 29 11:22:50 RP: ah cool Jul 29 11:23:30 zecke: all subject to change ;-) Jul 29 11:25:19 zecke: Do you like the idea of BBCLASSEXTEND or not? Jul 29 11:25:59 RP: looks sensible, is conceptually really close to the variants Jul 29 11:26:08 zecke: there are 'defaults' set in the environment (global flags) + individual flags in /etc/portage/package.use Jul 29 11:26:29 zecke: It could be made even closer too... Jul 29 11:26:42 <_dash__> zecke: I have Image zImage as well as uImage in arch/arm/boot directory Jul 29 11:27:12 ant_work: that is not what I asked for. How do I find out what mplayer ebuild supports, I'm really stupid, e.g. paste the equery call you would make Jul 29 11:27:41 equery hasuse slang Jul 29 11:27:58 equery uses mc Jul 29 11:28:16 _dash__: is this a custom kernel recipe? so from which directory is this mkimage called (I assume you know you can go to tmp/work/arm*/linux-*) to look at the scripts driving mkimage Jul 29 11:28:33 RP: do we have an example use in OE? Jul 29 11:28:52 zecke: first will show packages using slang flag, second shows the useflags of mc Jul 29 11:29:09 zecke: I don't think its caught on in OE yet so no Jul 29 11:29:54 <_dash__> zecke: Tt is getting called from $OETREE/angstrom-dev/staging/i686-linux/usr/bin/mkimage Jul 29 11:30:02 ant_work: do you know how it works? I'm mostly after the syntax in an ebuild to declare what it supports. Jul 29 11:30:05 Hi guys - I've just tried compiling klibc for a powerpc machine - it failed because KLIBCARCH needs to be set to ppc or ppc64. I've got a patch which sets it to ppc for ARCH=powerpc. Are there any ppc64 targets in OE and is there anyway to fix it for them at the same time? Jul 29 11:30:14 _dash__: no, that is what is getting called, what is the $PWD? Jul 29 11:30:29 _dash__: e.g. if the file is there, the mkimage command is found but it complains the uImage is not there Jul 29 11:30:35 zecke: http://git.pokylinux.org/cgit.cgi/poky/tree/meta/packages/xorg-proto/fontsproto_2.0.2.bb is an exmaple in Poky Jul 29 11:30:47 _dash__: you either have fun with apparmor/SELINUX, or you are in the wrong directory Jul 29 11:30:51 zecke: there were some enhancements needed to sdk.bbclass and native.bbclass Jul 29 11:31:26 zecke: http://git.pokylinux.org/cgit.cgi/poky/tree/meta/classes/native.bbclass the anonymous function at the end :/ Jul 29 11:31:36 zecke: have a look at http://www.gentoo-portage.com/app-misc/mc (llook at USE flags / Deps) Jul 29 11:33:05 zecke: http://www.gentoo.org/doc/en/gentoolkit.xml Jul 29 11:33:12 that's all python Jul 29 11:33:18 <_dash__> zecke: ohh .. So what is needed to be done if it's SELINUX problem (I think most probably i is)?? Jul 29 11:33:18 zecke: http://git.pokylinux.org/cgit.cgi/poky/tree/meta/packages/fakeroot/fakeroot_1.9.4.bb has a more interesting RDEPENDS Jul 29 11:35:00 _dash__: well, I'm just guessing. I you have something like a directory tmp/work/arm*/*linux*/. And then you have a temp/ directory in there Jul 29 11:35:05 * * OE Bug 5289 has been created by jon(AT)jfpossibilities.com Jul 29 11:35:07 * * Toupad stuck key on hx4700 Jul 29 11:35:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5289 Jul 29 11:35:18 _dash__: for the statge that failed you have a log + a shell script Jul 29 11:35:38 _dash__: call the shell script again, check if it is in the right directory (e.g. add a ls $PWD) above the mkimage Jul 29 11:35:52 _dash__: if it should find it, but it doesn't, strace will be your friend Jul 29 11:43:29 <_dash__> zecke: U mean I should have tmp dir in OETREE ; probably not since TMPDIR has been overriden somewhere while doing config Jul 29 11:44:27 _dash__: right, I have no idea about your layout... in the $WORKDIR of your kernel, you will have a temp/ directory Jul 29 11:44:51 _dash__: actually the failure message will point you to the log file, next to the log file is the script that lead to the error Jul 29 11:45:28 <_dash__> yup that script is something like run.blahblah Jul 29 11:45:33 <_dash__> right ? Jul 29 11:45:48 <_dash__> I have temp and log in kernel dir Jul 29 11:46:20 _dash__: well, I already told you what I would do now ;) Jul 29 11:47:34 <_dash__> ok i WILL TRY IT Jul 29 11:47:35 _dash_: agin, google finds out: http://www.mail-archive.com/u-boot-users@lists.sourceforge.net/msg05299.html Jul 29 11:48:18 aging = 2 options: 'aging' (you are asking since hours), again (I already posted the link) :) Jul 29 11:50:50 zecke: iirc Mike Frysinger / vapier was seen in #oe ML. He could give valid hints about USE Jul 29 11:51:51 cool. i've gotten further building minimal-uclibc on the dev branch than I did on stable Jul 29 11:53:37 thaytan: I'm building some images with micro-uclibc... the build error of util-linux-ng is known Jul 29 11:53:52 (two different patches exist) Jul 29 11:54:29 zecke: my error on stable was in uclibc-initial, about --hash-style Jul 29 11:54:41 RP: BBCLASSEXTEND... looks sensible with the "virtual" we can have different do_stage/_install/whatever methods, so it is quite nice Jul 29 11:54:48 trying to build a mipsel minimal-uclibc Jul 29 11:58:02 <_dash__> zecke: Can u tell me where linux.bin file is situated ? Jul 29 11:58:31 RP: while you're there: what's the right way to stage the mtd/mtd-user.h , mtd/mtd-abi.h headers? I added a recipe with its own headers... Jul 29 11:58:47 _dash__: I have no idea, and the message you pasted didn't involve linux.bin Jul 29 12:00:24 _dash__: did you mean vmlinux.bin? Jul 29 12:02:02 <_dash__> zecke: nope mkimage uses linux.bin file in it's -d option. if I caome to know wher is it then I can give options to mkimage and can build uImage manually Jul 29 12:07:03 _dash__: but why is the script failing? wrong directory? you said the uImage got created by the kernel? Jul 29 12:07:37 <_dash__> yup uImage I can find in arch/arm boot Jul 29 12:07:49 _dash__: so why did the script fail? Jul 29 12:07:49 <_dash__> only mkimage is failing Jul 29 12:08:10 <_dash__> mkimage is unable to find uImage Jul 29 12:08:25 _dash__: but it is there? why can't it find it? Jul 29 12:09:28 zecke: Can't read !? Can't map Jul 29 12:10:54 ant_work: haha, I'm not the one having the problem. the term "map" is not really unique, I assume it talks about mmap (and I hate the author of mkimage for not strerror) and this is why besides wrong $PWD I put SELinux/apparmor in the class of possible problems Jul 29 12:11:45 I've read horror-stories about beagleboard vs. u-boot Jul 29 12:12:16 ant_work: I want to encourage _dash_ to find the root cause of his problem, that is why I repeated my question Jul 29 12:12:38 zecke: you're fair :) Jul 29 12:13:15 <_dash__> in my mother kernel directory I have another gir called git Jul 29 12:13:31 <_dash__> inwhich entire code resides Jul 29 12:14:02 <_dash__> in that I am having arch/arm/boot/uImage Jul 29 12:14:23 I'm having a problem building uclibc-initial. Gcc complains about cross compile badness because /usr/include is in the path. I've tracked it down to the KERNEL_HEADERS var that is set to /usr/include. Can someone tell me if this var should have that value or if it should be replaced as it is in the uclibc build? Jul 29 12:14:26 _dash__: http://www.nabble.com/OE-errors-to17154279.html#a17157059 Jul 29 12:14:33 _dash__: what file system do you use? Jul 29 12:14:36 _dash__: so is the script referring to the wrong directory? Jul 29 12:16:31 XorA: that's the 'polite, better than nothing' kind of answers :) Jul 29 12:17:05 ant_work: following on from zeckes guess, it wont work on filesystems that dont support mmap Jul 29 12:17:25 I hope beagleboard is using modern u-boot Jul 29 12:17:31 <_dash__> XorA: I use fedora core 9 and my code is on NTFS mounted drive Jul 29 12:17:55 _dash__: then you lose, use a unix filesystem or write a non mmap version of mkimage Jul 29 12:18:07 XorA: we have mkimage even working for Zaurus, so you see... :D Jul 29 12:19:13 ant_work: WTF has u-boot got to do with the colour of the sky today? Jul 29 12:19:34 I mean the mkimage creation and the paths are verified by a moltitude Jul 29 12:20:24 I could find immediately the mmap() potential issue...probably I did not explain myself very well, sorry Jul 29 12:20:50 I just sent _dash__ the link Jul 29 12:23:10 XorA: btw if something wrt uImage get deleted/moved we'll have to retest anew :d Jul 29 12:23:48 ant_work: youve lost me Jul 29 12:24:10 bah..too warm here...better to move to the see Jul 29 12:24:36 my english skill are dangerously low atm... :) Jul 29 12:24:40 <_dash__> ALL who r helping me : script is not referring to wrong dir Jul 29 12:24:43 bbl Jul 29 12:24:57 <_dash__> I think the problem may be with NTFS Jul 29 12:27:26 <_dash__> ant_work: I have ssen that link So how do i update my mkimage ?? Jul 29 12:27:47 im surprised the build got that far on an NTFS drive Jul 29 12:28:47 <_dash__> XorA: me too Jul 29 12:29:09 _dash__: you like the pain :-D Jul 29 12:29:16 <_dash__> XoRA: But definitely got some lessons Jul 29 12:29:46 <_dash__> XorA: If pain teaches me then I like it Jul 29 12:30:57 <_dash__> But thank U all XorA, ant_work, and specifically zecke for encouraging me Jul 29 12:49:32 zecke: ahr - I assume this undefined reference to `__tls_get_addr' error is the one you meant Jul 29 12:51:31 thaytan_: yes, you can rebuild your gcc with a patch sent to the mailinglist, or hack m4/tls.m4 in util-linux-ng Jul 29 12:54:32 zecke: ta, I'll check the archives Jul 29 12:54:38 I only subbed to oe-devel this morn Jul 29 12:54:46 woglinde: ping? Jul 29 13:00:12 zecke: Another nice bonus is the parsing speedup ;-) Jul 29 13:00:45 sorry Jul 29 13:00:50 RP: could one be as crazy as attempting to build mutliple PV Jul 29 13:00:57 woglinde: don't worry! I know you need to work Jul 29 13:01:56 some developers should not even try to build rootfs Jul 29 13:03:03 -sh: can't access tty; job control turned off Jul 29 13:03:49 sounds familiar Jul 29 13:04:07 good that uboot has mmc support Jul 29 13:04:29 zecke: It could work Jul 29 13:05:00 zecke: technologically its possible. Whether its sane in practise I will not comment on Jul 29 13:05:49 RP: to answer your question from earlier, yes, I think BBCLASSEXTEND is a sane idea Jul 29 13:06:44 zecke: cool :) Jul 29 13:06:57 zecke: Somehow it felt like a good thing to do Jul 29 13:13:30 I'm having some trouble building uclibc-initial. It looks like it has to do with uclibc's .config file not being set up correctly, causing KERNEL_HEADERS to be /usr/include and when I fix that it starts compiling the crt1.S for an alpha. Jul 29 13:14:10 Any idea what could be causing this? Jul 29 13:24:33 nokia is poor company. I reported bug on ovi.com (so wrote in English). got reply (in English) so replied (in English). I got very fast reply with "please describe problem in Polish". Did they fired all English knowing people? Jul 29 13:28:31 Griffon26: which machine Jul 29 13:31:09 x86-32-nocona Jul 29 13:31:41 Griffon26: yes, see the mailinglist this month Jul 29 13:31:48 Griffon26: I sent a RFC no one has replied yet Jul 29 13:32:19 Griffon26: make a git log -p recipes/uclibc and for now you will be forced to create a x86-32-nocona director Jul 29 13:33:28 I see Jul 29 13:39:27 gotta figure out why my oe is using 0.9.30-r7 Jul 29 14:07:08 morning Jul 29 14:25:54 hi chouimat Jul 29 14:46:52 hi Jul 29 14:47:19 I have some mad fun Jul 29 14:47:57 I do bitbake -c mrproper xserver-xorg and then I do bitbake xserver-xorg and it jumps straight to the compile stage without unpack/configure, and thus of course won't build.. Jul 29 14:48:08 what's wrong? http://pastebin.ca/1511106 Jul 29 14:48:14 replace mrproper with clean Jul 29 14:48:22 I did that first and also didn't work Jul 29 14:48:44 then I tried mrproper thinking maybe more tough remedies were needed Jul 29 14:48:58 I should probably mention, that my machine crashed during the previous build Jul 29 14:49:15 but I thought clean/mrproper would take care of any residues.. Jul 29 15:03:49 th1_: bitbake libx11-native -c clean Jul 29 15:04:01 since its libx11-native thats failing Jul 29 15:05:19 morning kergoth Jul 29 15:05:32 hey Jul 29 15:15:11 zecke: with your gcc patch, and after rebuilding all the gcc instances, I'm still getting: http://pastebin.ca/1511127 in the util-linux-ng build Jul 29 15:15:34 I have no idea what that's pointing to Jul 29 15:18:09 thaytan_: you need to clean util-linux-ng? Jul 29 15:18:27 happened after a clean too Jul 29 15:18:37 of util-linux-ng, not the whole tree Jul 29 15:22:20 thaytan_: the test case is the following. compile a simple application with int main(void) { static int __thread foo = -2; if (foo == -2) foo = 1; } with your-compiler-gcc -fPIC -o foo foo.c Jul 29 15:22:33 thaytan_: if you have the new gcc installed it should reject the code Jul 29 15:23:59 zecke: I'm not sure how to do that properly Jul 29 15:24:07 can it drop me into a shell or some such? Jul 29 15:24:19 <- very new to OE Jul 29 15:25:51 thaytan_: you can just invoke the compiler, tmp/cross/arch-os/bin/arch-os*-gcc Jul 29 15:27:10 seems correct Jul 29 15:27:13 (test.c:1: error: thread-local storage not supported for this target) Jul 29 15:28:38 thaytan_: okay, so util-linux-ng should compile, are you sure you cleaned util-linux-ng Jul 29 15:29:02 yep, and confirmed the build dir was gone from tmp/work afterward Jul 29 15:29:03 thaytan_: alternatively bitbake -f -c configure -b Path/to/util-linux-ng_version.bb; and compile again Jul 29 15:29:23 thaytan_: you can also look at the config.log and search for __thread Jul 29 15:30:35 isn't there Jul 29 15:30:42 I suspect this is another bug entirely Jul 29 15:51:22 thaytan_: it is ___tls_get_addr undefined? it is the same. Jul 29 15:51:25 anyway, sleeping now Jul 29 15:51:39 zecke: no, it's a different undef'd symbol Jul 29 15:51:56 __fini_array_end Jul 29 15:52:09 goodnight :) Jul 29 15:53:04 oh Jul 29 15:57:21 hmm Jul 29 16:02:36 have a nice rest of day Jul 29 16:16:57 mickey_away: ping, whenever you're around Jul 29 16:32:28 * kergoth wonders how much work itd be to glue together the test recipes from the old OpenEmbedded bitbake repo to the oe-test repo and all Jul 29 17:01:19 thaytan_: once you add --disable-tls to gcc configure you have to rebuild uclibc Jul 29 17:01:52 and uclibc-initial? Jul 29 17:01:59 thaytan_: or even try a complete rebuild Jul 29 17:02:02 not sure of the relationship between those too Jul 29 17:02:17 khem: ja - it just takes so long Jul 29 17:02:24 thaytan_: I think just uclibc should be enough Jul 29 17:02:28 I was planning to a complete rebuild run overnight :) Jul 29 17:02:51 ok try to clean up uclibc and then rebuilt just uclibc Jul 29 17:04:32 ok, trying that Jul 29 17:13:33 Dotster is following me on twitter now Jul 29 17:13:49 They heard me complain about the dns problem yesterday. Jul 29 17:18:50 Crofton: Today it seems to be better I think Jul 29 17:19:39 yeah Jul 29 17:19:56 basically, they said it would get better, just take time Jul 29 17:20:17 * khem loves foxyproxy Jul 29 17:29:26 hmmmm Jul 29 17:31:28 khem: bingo - fresh uclibc and util-linux-ng succeeded Jul 29 17:40:14 thaytan: enjoy Jul 29 17:41:56 khem: not sure if you saw my first messages this morning Jul 29 17:42:04 I ended up switching to the dev branch Jul 29 17:42:10 because the build last night did fail after all Jul 29 17:42:17 it just got to uclibc-initial later in the cycle Jul 29 17:42:35 (the --hash-style thing) Jul 29 17:44:07 no i did not Jul 29 17:45:06 setting PREFERRED_LIBC didn't actually fix it Jul 29 17:46:50 I thought so because there were hard assignments inside the conf files Jul 29 17:49:12 in minimal.conf, you mean? Jul 29 18:00:26 no in the distro config file Jul 29 18:17:11 re Jul 29 18:17:24 khem: foxyproxy? what this extension do? Jul 29 18:17:56 03Stanislav Brabec  07org.openembedded.dev * r8d8036ab7d 10openembedded.git/recipes/gqview/ (4 files in 2 dirs): gqview: Use motion hints for smooth panning. Sanitize rc file parser. Fixed gimp call. Jul 29 18:18:28 hrw: You can separate out traffic to your work and your personal stuff when working from home Jul 29 18:18:52 khem: proxy switcher? Jul 29 18:19:04 hrw: actually it has patterns Jul 29 18:19:12 you can add proxy configs Jul 29 18:19:36 and proxy configs have regexp or pattern match for urls that should go thru it Jul 29 18:20:01 so you can direct all work traffic thru work proxy Jul 29 18:25:25 ah.. Jul 29 18:27:38 03Steffen Sledz  07org.openembedded.dev * r5fe09b5028 10openembedded.git/recipes/linux/linux-2.6.24/hipox/defconfig: linux-2.6.24: some pin assignements fixed in defconfig for hipox machine Jul 29 18:27:47 03Graeme Gregory  07org.openembedded.dev * r8bd33c32ad 10openembedded.git/recipes/linux/ (2 files in 2 dirs): Jul 29 18:27:47 linux-omap-pm_git.bb : update the commit to the latest SRCREV Jul 29 18:27:47 defconfig is customized for this release as two drivers are broken. Jul 29 18:27:47 mtd - broken in compile patches in progress Jul 29 18:27:49 musb - oopses on boot Jul 29 18:27:58 03Felix Domke  07org.openembedded.dreambox * rb450e9f2f2 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.opendreambox.org/git/acidburn/opendreambox-1.5 into org.openembedded.dreambox Jul 29 18:28:09 03Klaus Kurzmann  07shr/import * r185a2c9484 10openembedded.git/recipes/python/python-phoneutils_git.bb: Jul 29 18:28:09 python-phoneutils: add missing DEPENDS for libphone-utils Jul 29 18:28:09 Signed-off-by: Klaus Kurzmann Jul 29 18:28:18 03acid-burn  07org.openembedded.dreambox * rce144ea70e 10openembedded.git/packages/images/dreambox-image.bb: add CrashlogAutoSubmitter to dreambox-image Jul 29 18:28:30 03ghost  07org.openembedded.dreambox * rabb5f23de0 10openembedded.git/packages/tuxbox/ (2 files in 2 dirs): tuxbox-tuxtxt-32bpp: add fix for hardware without lcd/oled Jul 29 18:28:39 03Klaus Kurzmann  07shr/import * rd54dbdb73d 10openembedded.git/recipes/connman/mokonnect_svn.bb: Jul 29 18:28:39 mokonnect_svn: bump PV to 0.4 Jul 29 18:28:39 Signed-off-by: Klaus Kurzmann Jul 29 18:28:50 03Koen Kooi  07org.openembedded.dev * rb006455c12 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Jul 29 18:29:03 03Koen Kooi  07org.openembedded.dev * r956487eebf 10openembedded.git/ (conf/checksums.ini recipes/ntfs-3g/ntfs-3g_2009.1.1.bb): ntfs-3g: update to 2009.4.4 Jul 29 18:29:14 03Henning Heinold  07org.openembedded.dev * r977539b14a 10openembedded.git/recipes/libxine/ (libxine-1.1.16/ffmpeg_headers.patch libxine_1.1.16.3.bb): Jul 29 18:29:14 libxine: remove old/new header configure stuff from 1.1.16.3 Jul 29 18:29:14 * now buildable again Jul 29 18:29:14 * make 1.1.16.3 the default again Jul 29 18:29:16 * bump PR Jul 29 18:29:24 03Henning Heinold  07org.openembedded.dev * rd9c4308549 10openembedded.git/recipes/libsndfile/libsndfile1_1.0.20.bb: Jul 29 18:29:24 libsndfile: remove the libtool.m4 from libtool-2.x Jul 29 18:29:25 * now compileable with older libtool versions again Jul 29 18:29:27 * bump PR Jul 29 18:29:35 03Henning Heinold  07org.openembedded.dev * rf17f904673 10openembedded.git/conf/distro/chinook-compat.conf: chinook-compat.conf: use slighty newer gstreamer version Jul 29 18:29:45 03Henning Heinold  07org.openembedded.dev * r53131062ae 10openembedded.git/recipes/libsndfile/libsndfile1_1.0.20.bb: libsndfile: now really bump PR Jul 29 18:29:56 03Andrea Adami  07org.openembedded.dev * r8361ab6537 10openembedded.git/recipes/nandlogical/ (3 files in 2 dirs): Jul 29 18:29:56 nandlogical: add nandlogical-klibc-static for Zaurus Jul 29 18:29:56 - code to be used in kexecboot to detect the size of the rootfs in nand Jul 29 18:30:06 03Koen Kooi  07org.openembedded.dev * r6664caa939 10openembedded.git/classes/qt4x11.bbclass: qt4x11 bbclass: only match first 12 chars in automagical dependency adder Jul 29 18:30:17 03Koen Kooi  07org.openembedded.dev * r69a41a50ac 10openembedded.git/recipes/qt4/ (2 files): Jul 29 18:30:17 qt4-x11-free-gles: reinstate it now the weird dependency problems have been resolved Jul 29 18:30:17 * also switch to gles2, requires less patching Jul 29 18:30:29 03Koen Kooi  07org.openembedded.dev * r10f96a16b0 10openembedded.git/recipes/linux/ (72 files in 2 dirs): linux-omap-pm git: start updating it to .31rc Jul 29 18:30:39 03acid-burn  07org.openembedded.dreambox * r177228f80c 10openembedded.git/packages/python/ (python-2.5.1-manifest.inc python_2.5.1.bb): Jul 29 18:30:39 add missing MimeWriter to python-mime Jul 29 18:30:39 bump PR Jul 29 18:30:50 03acid-burn  07org.openembedded.dreambox * rcc3a0a4922 10openembedded.git/packages/enigma2/ (enigma2-plugins.bb enigma2.bb): add CrashlogAutoSubmit and push SRCDATE Jul 29 18:31:00 03ghost  07org.openembedded.dreambox * r9f0f04dfe1 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.opendreambox.org/git/acidburn/opendreambox-1.5 into org.openembedded.dreambox Jul 29 18:37:12 haha cia Jul 29 18:38:56 ? Jul 29 18:39:21 kergoth he got lost Jul 29 18:39:32 and now spilled out the commits from yesterday Jul 29 18:39:36 ahh Jul 29 18:40:08 i wonder how much work itd be to change the cia notifications to only notify for objects that were only pushed to the repo this time, so merges wouldn't re-notify, or something Jul 29 18:40:09 hrm Jul 29 18:41:16 not so much I gess Jul 29 18:43:08 bye Jul 29 18:47:06 bye Jul 29 19:09:53 hmm, woglinde's right, i bet itd be pretty easy actually, I'm going to have to look over our cia notification scripts Jul 29 19:13:36 * kergoth works on converting the old oe/bitbake test metadata to a git repo Jul 29 19:53:56 kergoth: Did you see our conversations earlier about BBCLASSEXTEND? Jul 29 19:54:03 nope Jul 29 19:54:18 kergoth: Have you played with that at all? Jul 29 19:55:01 kergoth: Basically it lets you create sdk and native recipes in one line to an existing recipe :) Jul 29 19:57:30 re Jul 29 19:58:38 haven't played with it yet, i like the idea, haven't had the time to try it out though Jul 29 20:31:53 * kergoth kicks CIA-1 Jul 29 20:31:54 ow Jul 29 20:39:07 03Chris Larson  07kergoth/setuptools * r6410cf3d9c 10bitbake.git/MANIFEST.in: Jul 29 20:39:07 Add ez_setup.py, doc/manual/*, contrib/vim/*/*, and HEADER to MANIFEST.in. Jul 29 20:39:07 Signed-off-by: Chris Larson Jul 29 20:42:06 Crofton: they broke DNS again :-( Jul 29 20:42:54 grrr Jul 29 20:43:13 was working for me earlier as well Jul 29 20:43:14 I'm ok .... Jul 29 20:43:25 03Graeme Gregory  07org.openembedded.dev * r40cf96ddd2 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: Jul 29 20:43:25 angstrom-2008-preferred-versions.inc : update the libtool version used. Jul 29 20:43:25 Tool are starting to use this newer version, in testing it didn't seem Jul 29 20:43:25 to break anything Jul 29 20:43:44 My hosted server is ok Jul 29 20:43:51 hence the commit :-) Jul 29 20:57:11 03Chris Larson  07master * rb335437993 10bitbake.git/ (5 files in 2 dirs): Jul 29 20:57:11 Initial switch from distutils to setuptools. Jul 29 20:57:11 Signed-off-by: Chris Larson Jul 29 21:01:59 XorA|gone, no updates from Dotster Jul 29 21:03:42 Crofton|work: obviously the bad zone file is still scooting around the internet Jul 29 21:06:09 yeah Jul 29 21:22:54 hi,just to know...is anybody working on making recipes for the palm pre? so I don't duplicate someone's work for the 3rd time Jul 29 21:25:00 maybe mickey Jul 29 21:25:02 is Jul 29 21:25:42 I'll ask him tomorrow Jul 29 21:27:57 03Graeme Gregory  07xora/angstrom-srcpv * r453a13c7a6 10openembedded.git/ (398 files in 133 dirs): Jul 29 21:27:57 recipes/* : resurect the SRCPV always upgradeable git recipes work that Jul 29 21:27:57 Holger Hans Peter Freyther did for the OM git tree Jul 29 21:27:57 that was never merged back into OE. Jul 29 21:34:50 damnation we still have git recipes using SRCDATE Jul 29 22:13:54 XorA|gone: eek. I'm slowly moving to SRCPV now for poky :) Jul 29 22:14:34 kergoth: We were postulating about a way of creating "use flag" like functionality from a new varient of BBCLASSEXTEND Jul 29 22:14:56 kergoth: Worth keeping the idea in mind if use flags rear their head again Jul 29 22:15:04 * kergoth nods, thought about that too Jul 29 22:24:12 * kergoth grumbles, going to have to pull bits from 3 different bkcvs->git conversions to make a coherent history of our old unit test recipes Jul 29 22:42:02 jo re Jul 29 22:42:14 hm is TLS a synomym for __thread? Jul 29 22:42:22 synonym Jul 29 23:59:25 Question if I may, is there a way to have Bitbake/OE use a linux_2.6.28.bb in my "oe/build/recipies/linux" folder instead of one in "oe/openemebbed/recipies/linux"? I tried upping the PR from r9 to r10, and added a 'DEFAULT_PREFERENCE_at91sam9261ek = "28"' in my version that isn't in the orig, but it still always defaults to orig unless I delete the orig, then it uses mine. Jul 30 00:00:44 I've also tried added 'PREFERRED_PROVIDER_virtual/kernel = "linux-mycustom"', to my local.conf, but I guess the Machine.conf overrites it which sets it to "linux", because it still uses the orig. I'm I going at this wrong? Jul 30 00:04:26 NvrBst, have you tried setting DEFAULT_PREFERENCE to a high number in your linux_2.6.28.bb ? Jul 30 00:05:10 you could also copy the machine conf file to your build/conf/machine/ directory and modify it Jul 30 00:06:05 it sets "DEFAULT_PREFERENCE = -1" and "DEFAULT_PREFERENCE_at91sam9261ek = 28". My machine is at91sam9261ek. It didn't seem to work... should it work? Orig doesn't have at91sam9261ek in it. Jul 30 00:06:53 I'll try to copy the machine file and edit it :) Jul 30 00:07:09 im not sure... im new here too Jul 30 00:07:37 i know that copying the machine file will work though Jul 30 00:11:09 i suppose the assignments for things like PREFERRED_PROVIDER_virtual/kernel should use '?=' instead of '=' so that users can override them in their local.conf Jul 30 00:14:53 ahh copy'ing the machine config to build/conf/machine/* worked for me as well. Thank you kindly grg. Jul 30 00:25:30 I've changed the provider for virtual/xserver and rebuilt it. Anyone know what target to specify in order to rebuild the image files that get spit out in deploy/*/images/*/ ? Jul 30 00:25:45 (its a gpe-image) Jul 30 00:32:01 I want to build toolchain for linux ver 2.6.15, I am getting error any help available Jul 30 00:32:22 2.6.15 is old and unsafe Jul 30 00:33:18 try an older gcc Jul 30 00:33:21 yeah but the code I have require that and i am not capable to port all the code to the latest kernel version Jul 30 00:34:02 having a gcc newer than your kernel will *probably* not work Jul 30 00:34:28 how to try an older gcc, I am not expert in all this, but little help will make my job easier Jul 30 00:34:48 dunno, sorry. haven't had to do that (yet) Jul 30 00:35:32 set PREFFERED_VERSION_gcc-cross = "2.2" Jul 30 00:35:38 in local.conf Jul 30 00:35:49 maybe look at PREFERRED_VERSION_gcc and friends in your distro's conf for ideas Jul 30 00:36:05 2.2 :) Jul 30 00:37:46 How does one ensure a _svn recipe gets default preference over any specific version ? Jul 30 00:39:03 03Henning Heinold  07org.openembedded.dev * rd894b5a78e 10openembedded.git/recipes/util-linux-ng/ (util-linux-ng-2.16/tls.patch util-linux-ng_2.16.bb): Jul 30 00:39:03 util-linux-ng: fix tls detection Jul 30 00:39:03 * breaks msvc support for now Jul 30 00:39:03 * bump PR Jul 30 00:39:11 so last commit Jul 30 00:39:13 good nite Jul 30 00:40:32 mankash, a gcc 4.0.x version might be old enough for 2.6.15 kernel. If not, try the most recent 3.x gcc Jul 30 00:43:13 ok thx Jul 30 00:44:11 grg: do you have some knowledge about ssh, I am remote login to the system and it keep logging me out Jul 30 00:44:33 like an idle timeout? Jul 30 00:45:17 it says software caused connection abort Jul 30 00:45:28 * grg shrugs Jul 30 00:45:48 i'm sure there's a #openssh or #ssh around here somewhere Jul 30 00:46:00 ok thx I will try that Jul 30 01:54:31 What are "mindate" and "maxdate" in the SRC_URI for ? **** ENDING LOGGING AT Thu Jul 30 02:59:57 2009