**** BEGIN LOGGING AT Fri Sep 18 02:59:56 2009 Sep 18 06:03:28 NOTE: Task failed: libsm-1_1.1.0-r0: missing dependencies for the following binary packages: libuuid Sep 18 06:03:33 any idea what provides libuuid? Sep 18 06:03:52 i thought it was util-linux Sep 18 06:05:08 e2fsprogs..? Sep 18 06:11:54 NOTE: Task failed: libsm-1_1.1.0-r0: missing dependencies for the following binary packages: libuuid Sep 18 06:11:57 :-( Sep 18 06:20:55 libuuid-dev? Sep 18 06:25:02 broken? Sep 18 06:44:52 NOTE: Applying patch 'rm_rf_tmp.patch' (util-linux-ng-2.16/rm_rf_tmp.patch) Sep 18 06:52:39 good morning Sep 18 06:53:44 hi mckoan Sep 18 07:34:41 hi. all. how to build armv6 for beagleboard(default is armv7) in OE? Sep 18 07:48:50 03Koen Kooi  07org.openembedded.dev * r41f2f6a124 10openembedded.git/recipes/xorg-xserver/ (2 files in 2 dirs): Sep 18 07:48:50 xserver-xorg-conf: add omap3517-evm support Sep 18 07:48:50 * This fixes the "blank screen" problem Frank Walzer reported Sep 18 08:05:34 03Jan Lübbe  07org.openembedded.dev * r90746f3032 10openembedded.git/ (conf/checksums.ini recipes/perl/libpoe-perl_1.007.bb): libpoe-perl: add recipe Sep 18 08:05:35 03Jan Lübbe  07org.openembedded.dev * rca309af34a 10openembedded.git/recipes/net-snmp/ (5 files): net-snmp: switch to INC_PR Sep 18 08:05:36 03Jan Lübbe  07org.openembedded.dev * r07f750c638 10openembedded.git/recipes/net-snmp/net-snmp.inc: net-snmp: use LDFLAGS (fixes QA error) Sep 18 08:05:37 03Jan Lübbe  07org.openembedded.dev * r6adf15ce21 10openembedded.git/recipes/net-snmp/ (libsnmp-perl_5.4.2.1.bb libsnmp-perl_svn.bb): Sep 18 08:05:40 net-snmp: add recipe for perl bindings Sep 18 08:05:42 It seems hard to convince net-snmp's build system to crossbuild the perl Sep 18 08:05:44 bindings, so i use a new recipe. Sep 18 08:05:46 03Jan Lübbe  07org.openembedded.dev * r277333bc80 10openembedded.git/ (3 files in 2 dirs): net-snmp: move svn srcrev to sane-srcrevs Sep 18 08:05:49 03Jan Lübbe  07org.openembedded.dev * r8f10e06e41 10openembedded.git/recipes/gnupg/ (gnupg.inc gnupg_1.4.2.2.bb): gnupg: split gpgv off into its own package Sep 18 08:05:52 03Jan Lübbe  07org.openembedded.dev * r6063eff6c6 10openembedded.git/ (conf/checksums.ini recipes/perl/libexpect-perl_1.20.bb): libexpect-perl: add recipe Sep 18 08:05:57 03Jan Lübbe  07org.openembedded.dev * r6d66985c26 10openembedded.git/ (2 files in 2 dirs): aufs2-util: add recipe for userspace tools Sep 18 08:06:00 03Jan Lübbe  07org.openembedded.dev * r7a930238ab 10openembedded.git/ (2 files in 2 dirs): libsoap-lite-perl: add recipe Sep 18 08:06:03 03Jan Lübbe  07org.openembedded.dev * r7c54aba805 10openembedded.git/recipes/linux/ (linux-2.6.30/i686/defconfig linux_2.6.30.bb): linux_2.6.30: use debian's defconfig for i686 Sep 18 08:06:08 03Jan Lübbe  07org.openembedded.dev * rb60e3a3e54 10openembedded.git/recipes/linux/ (3 files in 3 dirs): linux_2.6.30: add and enable aufs2 Sep 18 08:06:11 03Jan Lübbe  07org.openembedded.dev * r62cbf42efe 10openembedded.git/ (3 files in 2 dirs): i686-generic: add more devices tables Sep 18 08:14:12 XorA: so, i tracked the error down to being that with my host glibc (2.10), rpcgen fails to generate .h files because it can't find C preprocesor (cpp), which i have alrigth, in /usr/bin, but rpcgen will not work unless you give it "-Y /usr/bin" ... not sure if the bug should be fixed in glibc-2.10 or the recepi though Sep 18 08:22:22 kimrhh: ahah, it should be fixed in OE, we should be able to do that without using host rpcgen Sep 18 08:23:08 cpp I mean Sep 18 08:23:39 XorA: ok, I can't say for sure that it's using my host rpcgen, come to think of it, can only say that it helps adding a "-Y /usr/bin" to the rpcgen line in the recepi Sep 18 08:23:53 kimrhh: Id talk to khem about it Sep 18 08:24:08 XorA: will do, thx Sep 18 08:27:54 Is it possible to define do_install_append() just for a specific machine? Sep 18 08:28:36 Heinervdm: do_install_append_machine Sep 18 08:28:53 hello Sep 18 08:29:06 while building maemo on Oe Sep 18 08:29:09 OE* Sep 18 08:29:18 XorA: thx, found only examples for variables, not for functions Sep 18 08:29:28 there is an OE recipe 'external-toolchain' is taking Sep 18 08:29:32 a lot of tym to build Sep 18 08:29:44 do it generally take a lot of time Sep 18 08:29:53 ? Sep 18 08:30:43 yes Sep 18 08:32:45 ok Sep 18 08:33:10 but it is taking a hell lot of time..... Sep 18 08:33:13 :( Sep 18 08:36:41 florian: good morning Sep 18 08:38:37 good morning Sep 18 08:38:50 mayhem: never tried buiding a toolchain from scratch before? :) Sep 18 08:39:03 no Sep 18 08:44:05 mayhem: what machine are you building with? if you have multicore you can speed it up (if you haven't set the parallel stuff already) Sep 18 08:44:16 omap-3430sdp Sep 18 08:44:48 mayhem: you're building it _on_ the device? Sep 18 08:45:02 what Sep 18 08:45:06 couldnt get you Sep 18 08:45:45 mayhem: what specs are your host machine? i'm assuming you're not building these things using the n8xx Sep 18 08:45:56 no Sep 18 08:46:49 eh ? Sep 18 08:47:26 03Martin Jansa  07shr/import * r3f74153a85 10openembedded.git/ (9 files in 2 dirs): Sep 18 08:47:26 Support for etk/libframeworkd-phonegui-efl/elementary themes and add new nEo theme. Sep 18 08:47:26 Signed-off-by: Klaus Kurzmann Sep 18 08:47:27 03Martin Jansa  07shr/import * rfece468a43 10openembedded.git/recipes/shr/ (7 files in 2 dirs): Sep 18 08:47:27 Add bbfiles for gtk-theme-neo and icon-theme-neo Sep 18 08:47:28 Signed-off-by: Klaus Kurzmann Sep 18 08:47:30 03Martin Jansa  07shr/import * r358609c6b7 10openembedded.git/recipes/gpe-icons/ (gpe-icons.inc gpe-icons_0.25.bb gpe-theme-neo_git.bb): Sep 18 08:47:33 Add bbfiles for gpe-theme-neo and modify gpe-icons to install to subdirectory. Sep 18 08:47:35 Signed-off-by: Klaus Kurzmann Sep 18 08:47:39 03Martin Jansa  07shr/import * r9eb63a532f 10openembedded.git/recipes/ (4 files in 2 dirs): Sep 18 08:47:44 shr-theme-neo: fix typo in DEPENDS, fix regexp for gpe-icons. Sep 18 08:47:46 Signed-off-by: Klaus Kurzmann Sep 18 08:47:48 03Martin Jansa  07shr/import * rcddc415253 10openembedded.git/ (5 files in 4 dirs): Sep 18 08:47:51 Latest xorg stuff 2009-09-18 Sep 18 08:47:53 Signed-off-by: Klaus Kurzmann Sep 18 08:47:55 03Martin Jansa  07shr/import * ra1c78a1def 10openembedded.git/recipes/shr/shr-theme-neo.bb: Sep 18 08:47:58 nEo theme meta-package Sep 18 08:48:00 Signed-off-by: Klaus Kurzmann Sep 18 08:48:04 03Martin Jansa  07shr/import * rb824915f71 10openembedded.git/recipes/images/shr-image.inc: Sep 18 08:48:07 Fix move gtkrc to gtkrc.image in shr-image generation and add it with priority 1 to update-alternatives. Sep 18 08:48:09 Signed-off-by: Klaus Kurzmann Sep 18 08:50:54 03Martin Jansa  07shr/import * r9e889404cc 10openembedded.git/recipes/update-alternatives/ (3 files in 2 dirs): Sep 18 08:50:54 Use most recently added alternative if there is more alternatives with the same highest priority Sep 18 08:50:54 Signed-off-by: Klaus Kurzmann Sep 18 08:50:57 03Martin Jansa  07shr/import * rf8f8db8031 10openembedded.git/recipes/update-alternatives/update-alternatives-cworth/update-alternatives.use.last.patch: Sep 18 08:50:57 Also remove existing link before linking new, because when its link to directory new link is created inside instead of replaced. Sep 18 08:51:00 Signed-off-by: Klaus Kurzmann Sep 18 09:00:14 03Omar Barron  07org.openembedded.dev * rb2cf1ca47e 10openembedded.git/conf/machine/omapzoom2.conf: omapzoom2.conf: added machine conf for linux-omap-zoomsync kernel Sep 18 09:20:00 hi Sep 18 09:20:15 hi woglinde Sep 18 09:25:54 03Martin Jansa  07shr/import * r8cd407fa49 10openembedded.git/conf/distro/include/shr-autorev-unstable.inc: Sep 18 09:25:54 Lock revision of few fso packages, before release of vala-0.7.6. Sep 18 09:25:54 Signed-off-by: Klaus Kurzmann Sep 18 09:25:57 03Martin Jansa  07shr/import * rabb17340a0 10openembedded.git/conf/distro/ (4 files in 2 dirs): Sep 18 09:26:00 Include only shr-om-gta02-xorg.conf, leave shr-om-gta02-kms.conf for Sep 18 09:26:02 shr-experimental. xf86-video-glamo back to AUTOREV as libdrm patches already Sep 18 09:26:04 pushed. Add AUTOREV for xf86-input-mouse and xf86-input-keyboard shouldn't be Sep 18 09:26:06 needed but some Sep 18 09:26:08 Signed-off-by: Klaus Kurzmann Sep 18 09:47:11 florian: can ACK my mail to oe-members? Sep 18 09:47:20 s/can/can you/ Sep 18 09:49:00 thebohemian: done Sep 18 09:49:09 hi robert and florian Sep 18 09:49:13 florian: thanks! Sep 18 09:49:15 hi woglinde Sep 18 09:49:20 woglinde: moin Sep 18 09:49:31 thebohemian: yw... btw. you should have the password as well :) Sep 18 09:50:37 florian: well... should. can you send it to me by mail? (gpg: 0x6C3012DA) Sep 18 09:51:08 thebohemian: will do... Sep 18 09:52:28 ~seen hrw Sep 18 09:52:31 hrw was last seen on IRC in channel #oe, 8d 19h 7m 10s ago, saying: 'have a nice day(s) guys'. Sep 18 09:52:42 hm Sep 18 09:52:45 oops Sep 18 09:53:08 florian: thanks! Sep 18 09:53:13 again :) Sep 18 09:53:27 thebohemian: no problem... Sep 18 09:54:08 This reminds me that i was interrupted testing the oe paypal account i created with a little donation. Sep 18 09:54:29 hehe Sep 18 09:54:58 loose the pw and your money is gone Sep 18 09:57:43 yeah... that's exactly what I ran into. I choose a random password but do not know it offhand... so no way to accept the payment from here. Sep 18 09:58:05 I really wonder what they earn from people loosing passwords... Sep 18 09:59:06 bbiab Sep 18 09:59:13 can you not use the lost password mechanism? I have to reset my paypal password all the time, I seem to lose it several times a year. Sep 18 09:59:39 pb ask zecke the next time about his paypal odysee Sep 18 10:20:25 hi gnutoo Sep 18 10:20:31 hi Sep 18 10:43:34 hmm. my uclibc build fails for some reason. init dies mysteriously. with user_debug=255 I get this: http://pastebin.com/m15669688 (this is from a micro-base-image build, only change to base is LIBC=uclibc, otherwise it's a micro distro) Sep 18 10:44:02 czr what now uclibc dont builds or init dies? Sep 18 10:44:16 uclibc builds. Sep 18 10:44:24 init dies. although I died with a version from yesterday too Sep 18 10:44:37 eglibc build works ok on this target. Sep 18 10:44:49 same kernel? Sep 18 10:44:51 yes Sep 18 10:45:36 hm maybee its the thumb studd Sep 18 10:45:39 args stuff Sep 18 10:45:43 I'm not even sure what happens with the init. I'd expect an OOPS or something. but there's nothing except the regdump with user_debug=255. Sep 18 10:46:04 how would I go around fixing it? Sep 18 10:46:28 mom Sep 18 10:47:07 * czr_ wonders what his mom has to do with this problem. then hides & runs. Sep 18 10:47:24 np, I'll go and make a sandwich meanwhile. Sep 18 10:47:34 try comment PREFERRED_ARM_INSTRUCTION_SET = "thumb" Sep 18 10:47:45 in conf/distro/micro.conf Sep 18 10:48:40 *g* your cpu has jazell support Sep 18 10:49:57 yes it does, but I don't think it is the root problem ;-) Sep 18 10:50:06 nope it isnt Sep 18 10:50:07 but that will turn everything into arm right? Sep 18 10:50:22 where is the problem? Sep 18 10:50:35 binarys are faster and bigger Sep 18 10:50:42 at 10% or so Sep 18 10:50:42 I have 4 MiB NOR only Sep 18 10:50:56 but we want get the root of the problem Sep 18 10:51:03 so maybee its thumb or not Sep 18 10:51:05 yes. I'll do the arm, np :-) Sep 18 10:51:15 * czr_ is using this: http://www.digi.com/products/embeddedsolutions/digiconnectme9210.jsp#overview Sep 18 10:51:18 my other tip would be floating point stuff Sep 18 10:51:29 but as eglibc works with the same kernel Sep 18 10:51:35 yes Sep 18 10:51:52 and I have a small hello world which uses thumb and it runs fine. Sep 18 10:52:09 although obviously it's not an exhaustive test :-) (it just prints hello world) Sep 18 10:52:28 yes Sep 18 10:52:38 another step would be threading stuff Sep 18 10:53:11 can I put the PREFERRED_ARM_INSTRUCTION_SET directly into local.conf? Sep 18 10:53:33 nope Sep 18 10:53:36 its hardcoded Sep 18 10:53:51 maybee pb should change it to ?= Sep 18 10:53:56 ah, so I see. Sep 18 10:54:58 ~ping pb__ Sep 18 10:54:59 pong pb__ Sep 18 10:55:03 *g* Sep 18 10:55:18 that's handy ;-) Sep 18 10:55:54 ok, started churning on micro-base-image Sep 18 11:01:42 03Graeme Gregory  07org.openembedded.dev * r0d0ff49487 10openembedded.git/recipes/linux/linux-omap-zoomsync/omapzoom2/defconfig: linux-omap-zoomsync/omapzoom2/defconfig : enable twl4030 keypad driver Sep 18 11:02:05 so time for lunch now Sep 18 11:08:42 woglinde, I'll get back to you once I can test it again Sep 18 11:16:08 re Sep 18 11:39:49 woglinde, do you uses uclibc+mips+the bad gcc version? Sep 18 11:40:11 czr_, update your oe copy Sep 18 11:40:30 czr_, mario-goulart,and me + help from others fixed that Sep 18 11:40:51 s/fixed that/we should at least Sep 18 11:41:06 btw I've not my router serial cable with me Sep 18 11:42:50 czr, woglinde sorry I didn't saw it was on arm Sep 18 11:42:59 Hey, I've seen RP post on the mailing list, and I was wondering what was a canadian sdk. Are we really building a crosscompiler to build the sdk crosscompiler? Sep 18 11:43:48 czr, woglinde do you have a working userland or 2 serial ports? Sep 18 11:43:56 tardyp: canadian cross is quite usefull Sep 18 11:44:16 tardyp: e.g. on a x86 host, build a compiler for an amd64 system that creates arm binaries Sep 18 11:44:29 working userland means another embedded something/Linux userland that works Sep 18 11:44:55 Gnutoo, update? I just updated some hours back Sep 18 11:45:01 what changed? Sep 18 11:45:11 czr_, nothing I said that after: Sep 18 11:45:19 czr, woglinde sorry I didn't saw it was on arm Sep 18 11:45:31 thebohemian, yes, but on x86, gcc-native supports both architecture. Sep 18 11:45:33 ah, yeah, serves me right to read all the lines before replying ;l-) Sep 18 11:45:36 :-) even. Sep 18 11:45:46 tardyp: thats not the point Sep 18 11:45:57 it wont work if we want to build an sdk for powerpc on x86? Sep 18 11:46:01 czr_, do static binaries work? can you make something work? Sep 18 11:46:01 I think I should have two serials, but for some reason only one is enabled at the moment in the kernel Sep 18 11:46:09 Gnutoo, yes. eglibc build worked ok. Sep 18 11:46:14 uclibc build didn't. Sep 18 11:46:24 although I'm now trying uclibc with thumb disabled. Sep 18 11:46:37 czr_, ok...so you can have static eglibc things that works Sep 18 11:46:40 (to clarify, micro + uclibc + thumb = init borks) Sep 18 11:46:48 tardyp: the sdk is something you ship and use outside of OE/Poky Sep 18 11:46:52 (micro + eglibc + thumb = ok) Sep 18 11:46:58 tardyp: its the compiler which itself cannot run on the machine that will build the arm binaries Sep 18 11:47:10 czr_, ok and what about micro+uclibc+thumb+static? Sep 18 11:47:32 tardyp: sure you can run 32bit gcc on a amd64 system with compatibility libs. but not everyone wants those. Sep 18 11:47:55 Gnutoo, how would I test that? Sep 18 11:47:56 yes, I know what is the sdk. but I was just wondering if we were really compiling 2 cross compiler to build the canadian sdk. Sep 18 11:48:02 czr_, (just try a static hello world compiled with uclibc) Sep 18 11:48:10 ah, give me a sec. Sep 18 11:48:27 and point init= to the hello world Sep 18 11:48:39 tardyp: Yes, you need a cross tooltchain to build a candian sdk Sep 18 11:49:39 One compiler to cross from BUILD to SDK and then a candian to run on SDK and build for TARGET Sep 18 11:49:41 RP. ok. thats the power of oe. Sep 18 11:50:18 tardyp: If you feel a certain part is unneeded you can always ASSUME_PROVIDED but I think you'll hit suitble problems ;-) Sep 18 11:50:28 RP and probably another cross to build libs for target. Sep 18 11:50:29 czr, helloworld_1.0.0.bb should have the -static inside it Sep 18 11:50:42 tardyp: Thats just the usual cross compiler Sep 18 11:50:46 Gnutoo, micro + uclibc + thumb + static hello works Sep 18 11:50:55 czr_, ok good Sep 18 11:51:06 (although the kernel panics right after that since "init" dies, but that was to expected) Sep 18 11:51:09 czr_, so now I had something similar on mips Sep 18 11:51:22 RP, btw, I'm back at mozembed Sep 18 11:51:24 czr_, ah you didn't add while(1) in your hello world? Sep 18 11:51:40 Gnutoo, no. just a single printf :-) Sep 18 11:51:46 ok Sep 18 11:52:08 I got weird compiler bugs with the optimisation option there where. Sep 18 11:52:56 czr_, me and mario-goulart have had the same problem with mips: basically it was uclibc's fault,it was in the ld.so some things weren't inlined and they should have been Sep 18 11:53:03 Gnutoo, anyhow, woglinde said that I should build a micro + uclibc with arm only to rule arm vs thumb out. Sep 18 11:53:05 czr_, I'll try to find the commit Sep 18 11:53:13 ah, I saw that commit Sep 18 11:53:28 czr_, ok Sep 18 11:53:37 you see, I built microbase already yesterday with uclibc. and today again after pulling. Sep 18 11:53:49 uclibc & hello world were binary same, but the ld-uclibc wasn't Sep 18 11:53:59 the newer version was couple of K's larger. Sep 18 11:54:07 both behave in exactly the same way though. init crashes. Sep 18 11:54:37 ok Sep 18 11:55:29 anyhow, I'll try the uclibc + arm only thingy next. Sep 18 11:57:43 ok Sep 18 11:57:50 hmm Sep 18 11:58:01 micro + uclibc + arm-only = works. Sep 18 11:58:10 woglinde, see above. Sep 18 11:58:17 mmm maybe your hello world wasn't thumb Sep 18 11:58:20 so it's thumb. Sep 18 11:58:21 it was Sep 18 11:58:27 well, let me just check it. Sep 18 11:59:23 huh. weird. Sep 18 11:59:35 main is arm. puts is thumb & init and fini. Sep 18 11:59:54 and the trampolines are all thumb Sep 18 12:00:36 weird. some functions are arm, some are thumb Sep 18 12:01:15 __GI_strchr = thumb. __aeabi_uidiv = arm. __GI_fseek = thumb. Sep 18 12:02:18 03Robert Schuster  07org.openembedded.dev * rf380fe99da 10openembedded.git/recipes/xml-commons/ (4 files): Sep 18 12:02:18 jaxp1.3 1.3.04: New recipe (from Jalimo SVN). Sep 18 12:02:18 jaxp1.3-native 1.3.04: Dito. Sep 18 12:02:18 xml-commons-resolver1.1 1.2: Dito. Sep 18 12:02:18 xml-commons-resolver1.1-native 1.2: Dito. Sep 18 12:02:20 03Robert Schuster  07org.openembedded.dev * rf4a8de9123 10openembedded.git/recipes/jakarta-libs/ (regexp-native_1.5.bb regexp_1.5.bb): Sep 18 12:02:25 regexp 1.5: New recipe (from Jalimo SVN). Sep 18 12:02:27 regexp-native 1.5: Dito. Sep 18 12:02:29 03Robert Schuster  07org.openembedded.dev * r6c6d824e98 10openembedded.git/recipes/xalan-j/ (xalan-j-native_2.7.1.bb xalan-j_2.7.1.bb): Sep 18 12:02:32 xalan-j 2.7.1: New recipe (from Jalimo SVN). Sep 18 12:02:34 xalan-j-native 2.7.1: Dito. Sep 18 12:02:35 hmph. why is there a __gnu_Unwind_Restore_WMMXD in uclibc? Sep 18 12:02:36 03Robert Schuster  07org.openembedded.dev * re815e80bdf 10openembedded.git/recipes/jlex/ (files/jlex jlex-native_1.2.6.bb jlex_1.2.6.bb): Sep 18 12:02:39 jlex 1.2.6: Added (from Jalimo SVN). Sep 18 12:02:41 jlex-native 1.2.6: Dito. Sep 18 12:02:43 03Robert Schuster  07org.openembedded.dev * r224b851262 10openembedded.git/recipes/xerces-j/ (xerces-j-native_2.9.1.bb xerces-j_2.9.1.bb): Sep 18 12:02:46 xerces-j 2.9.1: Added (from Jalimo SVN). Sep 18 12:02:48 xerces-j-native 2.9.1: Dito. Sep 18 12:02:50 03Robert Schuster  07org.openembedded.dev * rd9fef0c7c9 10openembedded.git/recipes/cup/ (cup-native_0.10k.bb cup_0.10k.bb files/cup): Sep 18 12:02:50 good morning Sep 18 12:02:57 cup 0.10k: New recipe (from Jalimo SVN). Sep 18 12:02:59 cup-native 0.10k: Dito. Sep 18 12:03:01 03Robert Schuster  07org.openembedded.dev * re48d77d6dc 10openembedded.git/recipes/bcel/ (bcel-native_5.2.bb bcel_5.2.bb): Sep 18 12:03:04 bcel 5.2: New recipe (from Jalimo SVN). Sep 18 12:03:06 bcel-native 5.2: Dito. Sep 18 12:03:47 czr_, ok Sep 18 12:03:49 re Sep 18 12:03:57 woglinde: jalimo time! :) Sep 18 12:04:15 robert yeah Sep 18 12:04:23 czr thumb has not float support Sep 18 12:04:29 if I remeber correctly Sep 18 12:05:28 woglinde, but neither does my ARM core. or am I missing something? Sep 18 12:05:49 czr if it works now, fine Sep 18 12:06:00 I have to discuss it with pb and khem Sep 18 12:07:05 woglinde, uclibc + thumb = broken. uclibc + arm = works. Sep 18 12:07:24 if there's something I can do to help, drop me a line :-) Sep 18 12:07:29 I'm compiling for davinci-dvevm, in stable tree, with BitBake Build Tool Core version 1.8.13. I have a problem in rootfs script: http://pastebin.com/m6c1be0cb . Any idea? Sep 18 12:07:31 it's a fairly generic ARM926TE core. Sep 18 12:08:50 hm stable Sep 18 12:09:02 recalcati only few people here builds for stable Sep 18 12:09:14 ah Sep 18 12:09:18 let me think Sep 18 12:09:30 recalcati you need the bitbake which is in stable trunk Sep 18 12:09:38 1.8.12 ? Sep 18 12:09:39 otherwise you will get many errors Sep 18 12:09:56 thats why there is a bitbake dir in stable Sep 18 12:11:02 woglinde: I dind't see it Sep 18 12:11:04 sorry Sep 18 12:11:32 ok, is 1.8.12 Sep 18 12:11:45 thx . I try Sep 18 12:12:49 recalcati, check the buiding Angstrom instructions on the angstrom site, that should get you pointed at the correct bitbake Sep 18 12:13:03 hi crofton Sep 18 12:13:14 Crofton|work: I have to check my old setup from scratch. Sep 18 12:13:20 hi woglinde Sep 18 12:13:32 I built stable for the beagle the other day Sep 18 12:13:42 crofton oh? Sep 18 12:13:44 why Sep 18 12:13:52 stable dont have the dsp stuff Sep 18 12:17:28 some guy had a build issue and I wanted to make sure it built Sep 18 12:18:08 hm ah okay Sep 18 12:18:30 he had a symlink in BBPATH :) Sep 18 12:18:31 I hope I get the answer form the igep-sales stuff Sep 18 12:18:42 do you have an url for the board? Sep 18 12:18:57 I am curious about likewise's remark about it being industrial? Sep 18 12:19:24 sure Sep 18 12:19:33 http://www.myigep.com/ Sep 18 12:19:40 its like the beagle board Sep 18 12:19:52 but whith ethnernet, wifi and bt Sep 18 12:20:07 Crofton|work: well, I meant it seems more function complete than beagleboard. forget industrial Sep 18 12:20:26 ah Sep 18 12:20:28 s/industrial/commercial/ Sep 18 12:20:33 *g* Sep 18 12:20:39 http://www.igep-platform.com/index.php?option=com_content&view=article&id=46&Itemid=55 Sep 18 12:21:54 they said shipping should start in september Sep 18 12:22:04 how much? Sep 18 12:22:16 149 Sep 18 12:22:18 euros Sep 18 12:22:35 cool Sep 18 12:22:51 their margin must be grim Sep 18 12:22:59 dont know Sep 18 12:23:05 maybee the price isnt final Sep 18 12:25:01 I found via the beagle board wiki Sep 18 12:25:42 looks like they may make real money from OEM products, so this is a good way to get proto boards into customer hands Sep 18 12:29:37 what provides -lbluetooth these days in OE? Sep 18 12:29:52 robert dont ask Sep 18 12:29:56 bluez4 Sep 18 12:29:57 bluez4 I guess Sep 18 12:30:01 I have a package that depends on bluez-libs but angstrom prints all kind of funny messages ... Sep 18 12:30:04 and its still a mess Sep 18 12:30:17 we should get rid of bluetooth Sep 18 12:30:20 :) Sep 18 12:30:21 :) Sep 18 12:30:37 thebohemian: I ran into this too... no idea how to fix this yet. Sep 18 12:30:45 cool ... Sep 18 12:30:47 thebohemian: bluez-lis in angstrom is provided by bluez4 Sep 18 12:30:59 ok Sep 18 12:31:01 thebohemian: it should just all work Sep 18 12:31:11 But it pretty much looks like I'll have some time to spend on OE in near future Sep 18 12:31:11 the message are just notes of interest Sep 18 12:31:26 XorA: so I should change DEPENDS="bluez-libs" to DEPENDS="bluez4" ? Sep 18 12:31:27 florian, I hope this is good news :) Sep 18 12:32:08 thebohemian: leaving it should work Sep 18 12:32:16 XorA: ok Sep 18 12:32:26 pb has started again to split all up Sep 18 12:32:52 koen will dont work on it, because he dont care if he has to build 500 or 2000 packages Sep 18 12:32:56 thebohemian: bluez4 has PROVIDES = buez-libs bluez-utils Sep 18 12:33:45 woglinde: I dont think koen invented the bluez4 recipe Sep 18 12:34:09 Crofton|work: Yes in this context it is :-) I did not want to suggest that I'll have a lot of free time ;-) Sep 18 12:34:12 but koen decided to use bluez4 only in angstroem Sep 18 12:34:45 woglinde, distros need to make decisions about what they support Sep 18 12:34:57 woglinde: that choice was kind of made for us, mix and matching didnt work Sep 18 12:35:02 in one extreme you have poky that is a subset of recipes Sep 18 12:35:29 since we use oe.dev, we cannot delete recieps that create support requests from users :) Sep 18 12:36:46 you could create a distro that uses the bluez versions you like Sep 18 12:37:08 *sigh* Sep 18 12:37:14 thats was not the point Sep 18 12:37:35 bluez causes problems, that we all understand Sep 18 12:37:46 Crofton|work: I admit it was nice just to be able to remove non bluez4 :/ Sep 18 12:39:38 yeah, the advantage of forking Sep 18 12:39:46 make that "forking" Sep 18 12:40:48 Crofton|work: right, I don't consider Poky a fork - I still talk to you OE people ;-) Sep 18 12:41:05 RP, do you have enough time to keep up with OE and the TSC? Sep 18 12:41:15 rthat is why I added the ""'s Sep 18 12:41:20 er rp Sep 18 12:41:22 Crofton|work: I'd hope so Sep 18 12:41:40 Crofton|work: My problem is not being able to run actual OE builds, not keep up with the development Sep 18 12:41:56 have you employer let you test their hw .... Sep 18 12:42:09 Crofton|work: I do that on Poky Sep 18 12:42:13 surely they have some multi-core machines that need stress testing :) Sep 18 12:42:27 Crofton|work: What do you think is sitting next to me? :) Sep 18 12:42:47 Crofton|work: I just can't even stress test poky enough with the time available Sep 18 12:42:48 heh Sep 18 12:43:36 Crofton|work: I did think about the time commitment thing re: the TSC quite hard Sep 18 12:43:41 good Sep 18 12:46:00 * Crofton curses logicpd and their teaser linux support Sep 18 12:46:08 I think they exist to drive biz to timesys Sep 18 12:56:16 Crofton: hehe, move to zoom2 :-) Sep 18 12:56:34 I need to see the gpmc pins Sep 18 13:17:10 woglinde: you're right that thumb doesn't have fp instructions, but unless your cpu has hardware fp (99% don't) this doesn't make any difference. Sep 18 13:17:21 and working hardware fp at that, the crazy cirrus maverick thing doesn't count Sep 18 13:18:04 czr_: micro-uclibc as thumb does work for me. where is it crashing for you? Sep 18 13:18:48 hi pb Sep 18 13:18:49 mom Sep 18 13:19:25 pb -> http://pastebin.com/m15669688 Sep 18 13:25:27 hm, that is rather unfortunate Sep 18 13:25:39 let me run a qemu build now and check that it really does still work for me. it was fine last time I tried though. Sep 18 13:26:50 did thumb interworking got fixed in 2.19.XX binutils, it was broken in 2.19 itself Sep 18 13:27:37 pb when did you test? Sep 18 13:27:45 mickeyl raised the binutils version Sep 18 13:27:55 a while ago, but I was already using 2.19.51 Sep 18 13:28:01 ah okay Sep 18 13:28:07 assuming that's what mickey raised it to, it should be ok Sep 18 13:33:07 right, build running now. I need to go out for 45 minutes or so but I guess I will be back before the binary is ready :-} Sep 18 13:33:25 hehe Sep 18 13:33:27 I should get a futer. Sep 18 13:33:29 qemuqrm? Sep 18 13:33:33 faster computer, that is Sep 18 13:33:34 qemuarm Sep 18 13:33:36 right Sep 18 13:33:40 how can I test it Sep 18 13:33:52 here it only builds tgz and jffs2 Sep 18 13:33:53 just build it and then run it under qemu :-) Sep 18 13:34:09 oh right yeah, that's a pain. you need to fiddle IMAGE_FSTYPES to get ext2, or repack the tgz as ext2 Sep 18 13:34:19 and both I cannt run directly under qemuarm Sep 18 13:34:36 then something like "qemu-system-arm -M versatilepb -hda image.ext2 -kernel zImage.bin -append root=/dev/sda" Sep 18 13:34:38 okay I will try again Sep 18 13:34:44 yeag Sep 18 13:34:53 the qemu line I know Sep 18 13:34:54 thanks Sep 18 13:35:14 good luck Sep 18 13:35:15 * pb__ -> Sep 18 13:44:29 hi Sep 18 13:44:35 hi sushisan Sep 18 13:45:00 I have another annoying question :-) Sep 18 13:45:23 I'm trying to compile tightvnc (the last src) Sep 18 13:46:50 but have an Imakefile that must be processed by xmkmf Sep 18 13:47:33 I have compiled imake and put a do_compile_prepend in the recipe Sep 18 13:47:45 but the system can't find xmkmf Sep 18 13:47:47 ! Sep 18 13:48:12 dont know in which recipe it is Sep 18 13:48:29 maybe there isn't one because x now uses autotools Sep 18 13:48:37 my problem is tha if I do : bitbake -b ../../openembedded/recipes/xorg-util/imake_1.0.2.bb Sep 18 13:48:52 xmkmf isn't installed Sep 18 13:49:07 in the system tree for futures compilations Sep 18 13:49:20 why bitbake imake dont work? Sep 18 13:49:53 is, that work Sep 18 13:49:56 why not bitbake tightvnc??? Sep 18 13:50:18 because tightvnc doesn't have a Make file Sep 18 13:50:28 only an Imakefile Sep 18 13:50:42 sushisan: did you look in recipes/vnc Sep 18 13:50:44 ?? Sep 18 13:50:50 yes Sep 18 13:51:03 in this recipe Sep 18 13:51:18 copy a Makefile pre confired Sep 18 13:51:21 configured Sep 18 13:51:39 don't create the original Makefile from Imakefile! Sep 18 13:51:57 sushisan: so why not the same trick? Sep 18 13:52:00 to do that you must to run xmkmf that is in imake Sep 18 13:52:26 but...if I have xmkmf I can do: do_compile_prepend() { cd ${S} xmkmf } Sep 18 13:52:35 hm qemuarm works with thumb Sep 18 13:53:03 but I don0t know how to install the imake package in the main develope path Sep 18 13:53:18 sushisan: did you actually read the makefile? Sep 18 13:53:54 the Makefile in the original source? Sep 18 13:54:13 sushisan: re you trying to build vnc server? Sep 18 13:55:23 yes, client and server. After xmkmf you can do : make World Sep 18 13:55:52 sushisan: I wrote the Makefile by hand, no need of xmkmf Sep 18 13:55:59 but, my problem is that imake isn't in the develope tree of OE Sep 18 13:56:00 sushisan: for client anyway Sep 18 13:56:21 XorA: actually I need both of them Sep 18 13:56:34 I always used x11vnc as the server :-) Sep 18 13:57:16 Yes, it's work Sep 18 13:57:46 but I want my tightvnc!! :-) Sep 18 13:59:01 the question is: How I can install imake that the xmkmf can be executed inside a BB (like in do_compile_prepend) ?? Sep 18 13:59:26 sushisan: you will need to make an imake-native Sep 18 14:00:46 mmmm Sep 18 14:01:42 with inherit native ? Sep 18 14:01:58 no Sep 18 14:02:13 hm maybee it works Sep 18 14:02:19 haha Sep 18 14:02:27 trying... Sep 18 14:02:47 but do_stage may need some tweaks Sep 18 14:04:27 ? Sep 18 14:05:03 some bb for example please? Sep 18 14:05:19 hm Sep 18 14:05:34 try bitbake -c build imake-native Sep 18 14:05:57 so you see in the oetmp dir what will ends up in staging Sep 18 14:06:54 hi Sep 18 14:07:03 hi ulitu Sep 18 14:07:05 im doing a little reading on bitbake source code Sep 18 14:07:17 which one? Sep 18 14:07:26 or better from which source Sep 18 14:07:28 and which version Sep 18 14:07:31 git Sep 18 14:07:34 okay Sep 18 14:07:39 master Sep 18 14:07:53 woglinde: there are not any imake-native bb Sep 18 14:08:18 *sigh* Sep 18 14:08:24 than make one Sep 18 14:08:38 i wonder wether execute_runqueue_internal busywaits if no task is ready to be processed Sep 18 14:08:44 imake goes down in my books of build systems from hell :-D Sep 18 14:08:56 xora *g* Sep 18 14:08:56 in make-native_3.81.bb: Sep 18 14:08:58 inherit native do_stage() { install -d ${STAGING_BINDIR} install -m 755 make ${STAGING_BINDIR} } Sep 18 14:09:15 I merged the sdk stuff into poky master Sep 18 14:09:23 rp cool Sep 18 14:09:59 it actually looks like there is nothing that makes bb idle if no new task is ready to be process yet. Is that known? is it true? or plain wrong? Sep 18 14:11:29 ulitu: plain wrong - see waitpid in runqueue.py Sep 18 14:12:11 that does not wait, it's WNOHANG Sep 18 14:14:21 ulitu: ah, in bitbake master it will wait in a socket select Sep 18 14:14:50 could it be, that select has a timeout of 0? Sep 18 14:15:09 where can I find that select? Sep 18 14:16:22 ulitu: See xmlrpcserver.py serve_forever, self.timeout Sep 18 14:16:53 ulitu: Thats new code and its possible there is actually a bug in it Sep 18 14:17:55 ulitu: A timeout of 0.5 was what we used to have... Sep 18 14:18:40 till later Sep 18 14:18:49 is this a temporary soultion? Or is there any notification scheme planned? Sep 18 14:19:20 RP: right now self.timeout = 0 Sep 18 14:19:31 ulitu: feel free to find a way of making pythons xmlrc server play nicely with waitpid ;-) Sep 18 14:19:37 ulitu: or None Sep 18 14:19:52 yep Sep 18 14:20:23 ulitu: 0 does sound wrong Sep 18 14:20:24 RP: thanks. im still in the phase of understanding the problem :-) Sep 18 14:22:08 RP: lurking in your last commits I've seen a couple are pstage related. Is it only for sdk? Sep 18 14:23:09 RP: looks like a combination of pselect/sigchld is the right driection Sep 18 14:23:20 hi zecke Sep 18 14:23:38 RP: I'll think about it Sep 18 14:23:48 thanks for your support Sep 18 14:28:39 ant_work: General improvements really Sep 18 14:29:03 ant_work: related to some of the sdk work though Sep 18 14:29:15 can I cherry-pick just the p.s. class? Sep 18 14:29:31 for testing in OE? Sep 18 14:30:32 ant_work: no point tbh Sep 18 14:31:04 ok, the arch bug is still there Sep 18 14:31:44 ant_work: yes, no fix for that yet Sep 18 14:32:02 btw there is a second bug.. Sep 18 14:32:11 ant_work: I am working towards some pstaging work fwiw Sep 18 14:32:13 probably one mkdirhier Sep 18 14:32:22 as in the last p.s. commit Sep 18 14:32:37 ? Sep 18 14:32:48 http://patchwork.openembedded.org/patch/978/ Sep 18 14:32:54 the dir is missing Sep 18 14:36:59 ant_work: Probably a valid fix for OE, that code doesn't exist in poky Sep 18 14:37:02 like that anyway Sep 18 14:39:24 03Robert Schuster  07org.openembedded.dev * re90cec1338 10openembedded.git/recipes/classpath/ (3 files in 2 dirs): Sep 18 14:39:24 inetlib 1.1.1: New recipe (from Jalimo SVN). Sep 18 14:39:24 inetlib-native 1.1.1: Dito. Sep 18 14:39:25 03Robert Schuster  07org.openembedded.dev * r9360aebebc 10openembedded.git/recipes/classpathx/ (6 files in 3 dirs): Sep 18 14:39:27 gnujaf 1.1.1: New recipe (from Jalimo SVN). Sep 18 14:39:29 gnujaf-native 1.1.1: Dito. Sep 18 14:39:31 gnumail 1.1.2: Dito. Sep 18 14:39:33 gnumail-native 1.1.2: Dito. Sep 18 14:39:35 03Robert Schuster  07org.openembedded.dev * r02d7af7ebb 10openembedded.git/recipes/jcraft/ (4 files): Sep 18 14:39:38 jsch 0.1.40: New recipe (from Jalimo SVN). Sep 18 14:39:40 jsch-native 0.1.40: Dito. Sep 18 14:39:42 jzlib 1.0.7: Dito. Sep 18 14:39:44 jzlib-native 1.0.7: Dito. Sep 18 14:39:46 03Robert Schuster  07org.openembedded.dev * r7b686c7c52 10openembedded.git/recipes/jdepend/ (jdepend-native_2.9.bb jdepend_2.9.bb): Sep 18 14:39:49 jdepend 2.9: New recipe (from Jalimo SVN). Sep 18 14:39:51 jdepend-native 2.9: Dito. Sep 18 14:39:53 03Robert Schuster  07org.openembedded.dev * r5443425676 10openembedded.git/recipes/mozilla/ (4 files in 2 dirs): Sep 18 14:39:56 rhino 1.7r1: New recipe (from Jalimo SVN). Sep 18 14:40:06 rhino-native 1.7r1: Dito. Sep 18 14:40:08 03Robert Schuster  07org.openembedded.dev * r8987e7ad7d 10openembedded.git/recipes/junit/ (4 files): Sep 18 14:40:11 junit 3.8.2: New recipe (from Jalimo SVN). Sep 18 14:40:13 junit-native 3.8.2: Dito. Sep 18 14:40:15 junit4 4.3.1: Dito. Sep 18 14:40:17 junit4-native 4.3.1: Dito. Sep 18 14:40:19 03Robert Schuster  07org.openembedded.dev * r4fa1b8b074 10openembedded.git/recipes/jacl/ (jacl-native_1.4.1.bb jacl_1.4.1.bb): Sep 18 14:40:24 jacl 1.4.1: New recipe (from Jalimo SVN). Sep 18 14:40:26 jacl-native 1.4.1: Dito. Sep 18 14:40:28 03Robert Schuster  07org.openembedded.dev * r16d63d9910 10openembedded.git/recipes/servlet-api/ (4 files): Sep 18 14:40:31 jsp2.0 5.5.26: New recipe (from Jalimo SVN). Sep 18 14:40:33 servlet2.3 4.1.37: Dito. Sep 18 14:40:35 servlet2.3-native 4.1.37: Dito. Sep 18 14:40:39 servlet2.4 5.5.26: Dito. Sep 18 14:40:45 03Robert Schuster  07org.openembedded.dev * rf807b0604d 10openembedded.git/recipes/hsqldb/hsqldb_1.8.0.10.bb: hsqldb 1.8.0.10: New recipe (from Jalimo SVN). Sep 18 14:40:48 03Robert Schuster  07org.openembedded.dev * r9ffda150b3 10openembedded.git/recipes/jakarta-libs/ (oro-native_2.0.8.bb oro_2.0.8.bb): Sep 18 14:40:51 oro 2.0.8: New recipe (from Jalimo SVN). Sep 18 14:40:53 oro-native 2.0.8: Dito. Sep 18 14:40:59 03Robert Schuster  07org.openembedded.dev * r41f689d2e8 10openembedded.git/recipes/icommand/ (files/makefile.patch icommand-bluez_0.7.bb): icommand-bluez 0.7: New recipe (from Jalimo SVN). Sep 18 14:41:02 03Robert Schuster  07org.openembedded.dev * r3b6557acfe 10openembedded.git/recipes/jakarta-libs/ (log4j1.2-native_1.2.15.bb log4j1.2_1.2.15.bb): Sep 18 14:41:11 log4j 1.2.15: New recipe (from Jalimo SVN). Sep 18 14:41:13 (29 lines omitted) Sep 18 14:44:35 03Robert Schuster  07org.openembedded.dev * r74fc79f49e 10openembedded.git/recipes/xml-commons/jaxme_0.5.2.bb: jaxme 0.5.2: New recipe (from Jalimo SVN). Sep 18 14:44:39 03Robert Schuster  07org.openembedded.dev * rf6c552dcea 10openembedded.git/recipes/jakarta-libs/poi_3.0.bb: poi 3.0: New recipe (from Jalimo SVN). Sep 18 14:49:37 RP: bitbake needs an event lopp ;-) Sep 18 14:50:20 RP: and python needs pselect Sep 18 14:50:38 for now i increased the xmlrpc server timeout to 0.5 again Sep 18 14:51:58 which made "Handling BitBake files:" run at 2 recipies per second :-( Sep 18 14:54:42 ulitu: Ah, that was why it changed :/ Sep 18 14:54:57 ulitu: Have the idle handlers report the timeout they'd like? Sep 18 14:56:36 how about sleeping in execute_runqueue_internal() "if task is None" Sep 18 14:57:06 actually, buildFileIdle in cooker.py already returns 0.5. But server_forever does not handle that Sep 18 14:57:33 ulitu: Obviously I go so far with implementing it and got distracted Sep 18 14:57:45 rp: yep. that's what i thought Sep 18 14:58:32 RP: would you take the max of all idle return values? Sep 18 15:01:00 mickey|patio: good afternoon Sep 18 15:01:25 hi pb__ Sep 18 15:09:08 RP: I think, the idle functions should return, wether they did any work Sep 18 15:09:34 RP: And if all report false, the timeout should be set >0. Sep 18 15:09:48 RP: If any work was done, the timeout should be set = 0 Sep 18 15:10:22 RP: the problem is. the functions already return, wether they want to be called again Sep 18 15:10:48 ulitu: hola Sep 18 15:11:24 zecke: hi Sep 18 15:11:37 ulitu: i will arrive in berlin the 22nd Sep 18 15:12:30 re Sep 18 15:12:32 he zecke Sep 18 15:12:47 woglinde: hm, right, init does crash for me as well. clearly something is amiss there. Sep 18 15:12:52 zecke: hail zecke Sep 18 15:12:56 pb hm her it dont crash Sep 18 15:12:57 ah Sep 18 15:13:02 I didnt recompile uclibc Sep 18 15:13:03 okay Sep 18 15:13:07 ah, maybe that's it Sep 18 15:13:12 I built from scratch Sep 18 15:13:15 yeah Sep 18 15:13:25 hm so what it is quick fix Sep 18 15:13:34 pb I would make it ovveridable Sep 18 15:13:42 dunno, let me have a look at the crash Sep 18 15:13:44 so everyone can set in local.conf Sep 18 15:19:05 ah, drat, default kernel has user-debug turned off Sep 18 15:19:06 * pb__ recompile Sep 18 15:19:22 hm Sep 18 15:19:49 pb the crash-pastebin from crm is not enough? Sep 18 15:22:39 http://pastebin.com/m15669688 Sep 18 15:25:00 not quite, I need to be able to match up the addresses with the code Sep 18 15:25:08 I'm not 100% confident that his code is bit-for-bit identical with mine Sep 18 15:25:08 okay Sep 18 15:25:14 re zecke Sep 18 15:36:59 ah, so at least part of the problem is that some functions (e.g. __errno_location) have been built as thumb but somehow marked as arm Sep 18 15:38:20 hm Sep 18 15:38:31 error in binutils or uclibc? Sep 18 15:40:05 binutils, looks like Sep 18 15:40:10 hms Sep 18 15:40:34 odd, I'm sure this version was working for me before Sep 18 15:40:37 oh well, gotta update I guess Sep 18 15:40:43 ??? Sep 18 15:41:15 newer binutils would probably be better Sep 18 15:41:22 I think drow fixed a bug in this area quite recently. Sep 18 15:41:45 hm Sep 18 16:20:45 03Henning Heinold  07org.openembedded.dev * r876a2f7420 10openembedded.git/ (17 files in 2 dirs): Sep 18 16:20:45 ti: switch download url from dir.com to dir.local Sep 18 16:20:45 * dir.com is a registered domain Sep 18 16:20:45 * no bump of versions needed Sep 18 16:58:56 woglinde: binutils_cvs does seem to link libc.so correctly Sep 18 16:59:07 hm Sep 18 16:59:09 haven't checked yet whether that actually fixes czr's crash, but I think it will Sep 18 17:03:23 hi everybody Sep 18 17:03:49 pb__: is it uclibc/ld.so crash in thumb mode ? Sep 18 17:07:16 khem: it's crashing in libc.so, not actually ld.so, but yes Sep 18 17:07:42 __errno_location is compiled as thumb but for some reason has been marked as an arm function in libc.so, so it goes bang when trying to call that function Sep 18 17:08:12 it's correct in __errno_location.os, just goes wrong during final link. I think drow checked in a fix for this recently, which I had thought was in 2.19.51 but it seems not. Sep 18 17:08:13 pb____: which binutils ? Sep 18 17:08:27 it fails with 2.19.51, works with current 2.20-branch cvs Sep 18 17:08:42 that was my hope Sep 18 17:09:05 there were problems with veneers in ld for lonh Sep 18 17:11:08 ah yes, the patch was applied on 2009-09-09, but 2.19.51 was released 2009-09-04 Sep 18 17:12:06 actually I planned to put together a recipe for 2.20 which is 2.19.90 as of now Sep 18 17:12:34 it's probably a bit premature to write a 2.20 recipe, might as well wait for it to actually be released. Sep 18 17:12:43 I'll just add that one patch to 2.19.51, I think that should be good enough for now Sep 18 17:13:46 woglinde: anyway, in the meantime, you might like to check that binutils-cross_cvs really does fix the problem. have to clean and rebuild uclibc of course. Sep 18 17:14:00 going home now, I'll add that patch later this evening Sep 18 17:14:02 * pb__ -> Sep 18 17:14:25 yes thats better Sep 18 17:16:22 anyone seen errors with addgroup/adduser/update-modules like this: http://pastebin.com/d2b19beed Sep 18 17:17:23 might be something simple, i dunno tho Sep 18 17:20:42 i let the build run overnight, but i don't recall seeing anything like that before Sep 18 17:23:59 will the adduser/addgroup be taken care of on first boot? Sep 18 17:30:55 hrm well bitbake update-modules fixes that. im assuming users/groups are created on firstboot when the opkg rc script runs Sep 18 17:35:32 03Koen Kooi  07org.openembedded.dev * r72d9084abf 10openembedded.git/ (conf/checksums.ini recipes/pidgin/pidgin_2.6.2.bb): pidgin update to 2.6.2 Sep 18 17:42:49 re Sep 18 17:51:18 wb :) Sep 18 17:53:42 hey rkirti Sep 18 17:54:40 hello florian, how its going ? Sep 18 17:55:00 * rkirti is happy -- weekend time again. time for some OE dabbling Sep 18 17:55:37 rkirti: busy but fine - more or less like always :-) Sep 18 17:59:20 florian: good evening Sep 18 17:59:54 hey pb____ Sep 18 18:00:29 Anyone had a problem building uclibc stuff on an x86_64 host? Sep 18 18:01:18 Tartarus: what error ? Sep 18 18:01:19 Tartarus: no; I build on x86-64 all the time and it works for me. Sep 18 18:01:23 what goes wrong for you? Sep 18 18:01:41 native packages are being done for x86_64-linux-uclibc Sep 18 18:02:02 heh, that is no good Sep 18 18:02:17 could be that this is happening for me and I just haven't noticed; I guess most native packages don't care Sep 18 18:02:25 gmp does, heh Sep 18 18:02:31 Tartarus: hmmm Sep 18 18:02:37 trying to undo one thing i've got locally, maybe thats it Sep 18 18:03:10 hm, no, that doesn't seem to be happening for me Sep 18 18:03:21 gmp is building ok, and all the directories in question are correctly named x84_64-linux Sep 18 18:03:21 not for me as well Sep 18 18:03:46 on x86_64 as well as i686 its ok Sep 18 18:04:04 Blah, it was my local change Sep 18 18:04:11 damn self-inflicted bugs :) Sep 18 18:04:38 doh :-) Sep 18 18:09:05 rkirti: did your build from scratch work well Sep 18 18:09:30 khem: the last I saw, glibs still had issues. Will check today Sep 18 18:09:50 khem: th ABI version change broke everything last time Sep 18 18:11:04 delete tmp Sep 18 18:11:13 and redo the build Sep 18 18:11:21 what is your host Sep 18 18:11:37 rather build machine running (OS) Sep 18 18:13:36 I had a fresh TMPDIR already. ubuntu 32bit intrepid Sep 18 18:16:44 ok I have hardy on one x86 machine and it build fine on that one your is newer so should work Sep 18 18:21:28 03Phil Blundell  07org.openembedded.dev * r37ccccb96d 10openembedded.git/recipes/binutils/ (binutils-2.19.51/thumb-func.patch binutils_2.19.51.bb): apply patch from drow to fix linking of thumb binaries Sep 18 18:28:04 morning kergoth Sep 18 18:28:26 mroning Sep 18 18:28:29 morning, even Sep 18 18:31:54 morning kergoth Sep 18 18:44:03 pb____, evenink. is that the fix for the thumb/uclibc-problem? Sep 18 18:45:13 yeah, I think so Sep 18 18:45:32 you'll need to clean and rebuild uclibc after pulling that patch Sep 18 18:45:36 * czr_ nods Sep 18 18:45:44 I've been doing clean builds all the time now anyway Sep 18 18:46:14 right, with uclibc it isn't too painful to rebuild from scratch every time Sep 18 18:46:18 gets a bit tiresome with glibc though Sep 18 18:46:22 I'm going afk for some time now, but will try asap, maybe tomorrow morning. Sep 18 18:46:27 heh :-) Sep 18 18:46:42 well, I've now got 5 different toolchain+libc-version+thumb/arm-variants lying on the disk Sep 18 18:46:58 righto, later Sep 18 18:47:12 laters, will report back as soon as can. thanks a bunch :-) Sep 18 19:29:32 re Sep 18 19:33:42 woglinde: wb Sep 18 19:33:52 jo florian Sep 18 20:15:27 re crofton Sep 18 20:15:39 re Sep 18 20:17:43 Do we use generate ipks to populate staging always ? Sep 18 20:17:54 generated Sep 18 20:20:40 no Sep 18 20:20:55 you have to setup it in local.conf Sep 18 20:21:02 and its broken since some time Sep 18 20:21:03 it would be nice if everything in staging was using it Sep 18 20:21:14 woglinde: yeah I use to have it Sep 18 20:21:55 I could in theory remove eglibc ipks and then reinstall eglibc-initial ipks to get to stage before eglibc build Sep 18 20:22:19 eglibc-initial should be reinstalled after removing eglibc Sep 18 20:22:32 khem: is broken Sep 18 20:22:44 otherwise gcc wont find the headers that it needs even to rebuild eglibc Sep 18 20:23:00 khem: but for native and cross should work Sep 18 20:23:49 well, you don't need to recompile the toolchain Sep 18 20:24:14 I have a couple of fixes pending Sep 18 20:24:27 waiting that RP unbreaks the rest Sep 18 20:24:49 (arch mismatch) Sep 18 20:38:12 these are harmless..let's go Sep 18 20:38:24 03Andrea Adami  07org.openembedded.dev * rba4bad720f 10openembedded.git/classes/packaged-staging.bbclass: packaged-staging: fix populate_staging_postamble after staging-split by base package arch (introduced by f876463527bd9ae92a09159fdb4899b4a4d19121) Sep 18 20:38:33 03Andrea Adami  07org.openembedded.dev * ra95c355af4 10openembedded.git/classes/package_ipk.bbclass: package_ipk.bbclass: force recreation of DEPLOY_DIR_IPK. The dir is removed wiping tmp. Sep 18 20:39:07 khem: now, after arebuild from scratch, your toolchain will be found Sep 18 20:39:21 on subsequent rebuilds from pstage Sep 18 20:39:59 cool Sep 18 20:40:44 the first one comes from poky. there is more to cherry-pick there Sep 18 20:44:48 yeah Sep 18 20:45:08 khem: I merged all and don't see any issue Sep 18 20:45:08 I wonder if there are ipks around that I can poke for contents Sep 18 20:45:19 look in /deploy/pstage Sep 18 20:45:39 ant__: does your change need a rebuild Sep 18 20:45:51 yes Sep 18 20:45:55 ant__: thats I know but I mean some other packages which I did not build Sep 18 20:46:13 ah. no Sep 18 20:46:37 I suppose they get package-staged in the right location Sep 18 20:47:55 khem: atm you don't have /tmp/cross/armv5te/bin Sep 18 20:48:07 but /tmp/cross/bin Sep 18 20:48:16 (e.g.) Sep 18 20:48:46 I mean, after deleting all but pstage Sep 18 20:48:47 I have /cross/armv5te/bin Sep 18 20:48:57 on rebuild :) Sep 18 20:49:22 oh ok Sep 18 20:55:21 khem: now we should see why the machine and multimach_arch in the staging ipks are wrong... Sep 18 20:55:37 well, I have an idea Sep 18 20:56:05 we set PSTAGE_PKGARCH = "${BUILD_SYS}" Sep 18 20:56:26 and where is PSTAGE_PKGARCH used Sep 18 20:56:32 echo "Architecture: ${PSTAGE_PKGARCH}" Sep 18 20:56:36 should it point to build machine arch Sep 18 20:56:47 or the target machine arch Sep 18 20:57:16 perhaps smthg like if bb.data.inherits_class('native', d) or bb.data.inherits_class('cross', d) or bb.data.inherits_class('sdk', d): Sep 18 20:57:59 for the machine and multimach_arch (and for 'all' ?) Sep 18 20:58:15 yes it should depend upon the type of package it is going to stage right Sep 18 21:06:22 khem: ther is another (small) thing: how to combine "/${OELAYOUT_ABI}/${DISTRO_PR}/" in PSTAGE_EXTRAPATH Sep 18 21:08:34 khem: see the comments about this in my merged-with-poky-packaged-staging here http://fr.pastebin.ca/1571234 Sep 18 21:09:12 "/${OELAYOUT_ABI}/${DISTRO_PR}/" today is /3/.3/ Sep 18 21:09:30 "/${OELAYOUT_ABI}${DISTRO_PR}/" today is /3.3/ seems better, isn't? Sep 18 21:09:42 yes Sep 18 21:09:55 but see..is overwritten Sep 18 21:11:20 khem: having $LIBC here has sense only if we (finally) move pstage out of deploy Sep 18 21:11:33 (out of tmp, of course) Sep 18 21:12:10 we already have /deploy/$LIBC Sep 18 21:12:23 sometimes we have uclibc and eglibc and glibc Sep 18 21:12:28 all build in same tmp Sep 18 21:12:43 yes, separated one level before Sep 18 21:13:32 khem: I ended up with Sep 18 21:13:33 andrea@mizar /oe/build/tmp/deploy/glibc/pstage/angstrom/3/.3 $ ls Sep 18 21:13:33 glibc glibc-oe-build-tmp Sep 18 21:15:55 khem: atm there are 4-5 packages using package_stagefile_shell so deploy must be inside tmp :/ Sep 18 21:16:19 s/packages/recipes/ Sep 18 21:16:24 bootloaders Sep 18 21:16:52 this func make assumptions Sep 18 21:17:38 hmmm cleaning wud be nice Sep 18 21:20:23 I'm behind one of them (zaurus-updater) Sep 18 21:20:34 it seems poky solves differently Sep 18 21:20:57 btw, when using package_stagefile_shell PARALLEL_MAKE needs to be disabled :/ Sep 18 21:21:14 some race/populating issue Sep 18 21:22:21 khem: I've seen in poky there are fixes for recipes not staging anything and for autotooled Sep 18 21:22:45 I don't understand 100% but my merged version did not show new bugs... Sep 18 21:23:04 and was a bit faster if I'm not wrong Sep 18 21:23:52 well, I mean, reached the ipk-arch-error earlier (Can't find package xy) :P Sep 18 21:26:49 khem: if you look at the logs during rebuild from pstage you'll see Sep 18 21:26:52 * Package staging-bzip2-armv5te-angstrom-linux-gnueabi version 1.0.2-r2 has no architecture specified, defaulting to i686-linux. Sep 18 21:26:53 * Package staging-klibc-c7x0-angstrom-linux-gnueabi version 1.5.15-r2 has no architecture specified, defaulting to i686-linux. Sep 18 21:27:25 hmmm thats because PSTAGE_ARCH is not defined ? Sep 18 21:28:08 well, is fix-defined as i686-linux Sep 18 21:29:36 khem: it happens for native too Sep 18 21:30:09 http://fr.pastebin.ca/1571259 Sep 18 21:30:09 but native would work Sep 18 21:30:18 by chance :D Sep 18 21:34:36 arch = bb.data.getVar('PACKAGE_ARCH_%s' % pkg, d, 1) Sep 18 21:34:36 if not arch: Sep 18 21:34:36 arch = bb.data.getVar('PACKAGE_ARCH', d, 1) Sep 18 21:37:10 hmm Sep 18 22:44:13 bonne soir, zecke Sep 18 22:48:37 why do we use LEAD_SONAME Sep 18 22:48:54 khem: he he Sep 18 22:49:01 pb_ can't remember Sep 18 22:49:13 why he introduced it Sep 18 22:49:33 but now in his opinion was an error Sep 18 22:50:18 khem: I was sitting the kid in the bed...have you found anything about that wrong arch? Sep 18 22:54:13 ant__: no I am looking into unrelated problems right now Sep 18 22:59:06 np Sep 19 00:14:14 03Khem Raj  07org.openembedded.dev * r82e740b54d 10openembedded.git/ (7 files in 3 dirs): Sep 19 00:14:14 tzcode-native_2009k.bb, tzdata_2009m.bb: New timezone data Sep 19 00:14:14 * Add new recipes for latest zoneinfo data Sep 19 00:14:14 * Resort to INC_PR Sep 19 00:14:14 * Abstract out common parts into .inc files. Sep 19 00:14:16 Signed-off-by: Khem Raj **** ENDING LOGGING AT Sat Sep 19 02:59:58 2009