**** BEGIN LOGGING AT Thu Oct 08 02:59:58 2009 Oct 08 05:27:18 I've been working in stable/2009 since I setup OE like the "building Angstrom" howto said. Yesterday I was told that most gstreamer-ti is being done in an "unstable" branch. Which branch should I use to follow the latest changes? Oct 08 05:55:53 tasslehoff: org.openembedded.dev Oct 08 05:58:47 akheron: Thanks. I did "git clone git://git.openembedded.org/openembedded.git && git checkout origin/stable/2009 -b stable/2009". Can I just do "git checkout master" to get the development branch? Oct 08 05:58:59 * tasslehoff is new to OE _and_ GIT :) Oct 08 06:05:44 tasslehoff: git checkout -b org.openembedded.dev origin/org.openembedded.dev Oct 08 06:26:19 morning!!! Oct 08 06:33:27 how can create cross compile toolchain for mips? Oct 08 06:34:55 kr_217: I'm not sure how to do the packaging, but compiling a minimal-image will give you a toolchain in tmp. Oct 08 06:44:41 akheron: thanks Oct 08 07:18:34 After the linux->linux-gnu changes in siteinfo.bb and insane.bbclass openssl-native is unable to build. I'm using powerpc and it might also fail on arm. Can someone verify? http://pastebin.com/d5386dfaa Oct 08 07:22:39 sgh: which kind of tool i should use to create a cross compile toolchain for linux Oct 08 07:22:53 sgh: sorry for mips Oct 08 07:23:41 kr_217: bitbake .... is this you first try with openembedded ? Oct 08 07:25:46 sgh: sorry but whats bitbake? Oct 08 07:26:36 kr_217: ok ... seems like you should start here : http://wiki.openembedded.net/index.php/Getting_started Oct 08 07:36:20 http://pastebin.com/m464653ac can one look at this simple bb and tell me why nothing gets installed or a ipkg created ? Oct 08 07:38:25 kr_217, i can give you a little guide ( at least thats what i can do ) in a private chat if you u like Oct 08 07:39:01 rob_w: ok Oct 08 07:39:25 but you will need to read that link from sgh and set up the environment so far Oct 08 07:39:45 so get the git pulled and and i convo you Oct 08 07:39:57 rob_w: ok Oct 08 07:40:05 np Oct 08 07:43:46 hmm i am missing something small that it creates that littel ipkg package .. grrr Oct 08 07:46:20 I'm trying to setup a sensible folder structure that allows me have multiple folders under build/. Is it good practice to set TMPDIR to a subfolder in build (e.g. /build/stable/angstrom-dev)? Oct 08 08:05:27 http://pastebin.com/m5c12a05f i added some FILES but still no package created .. buhuu Oct 08 08:16:31 sgh: openssl-native built for me Oct 08 08:22:56 so just for my clarification .. i dont use autotools and only have some files .. i make a do_install () { } but why does it not make a package out of it ? Oct 08 08:23:41 XorA: have you seen the error ? Oct 08 08:24:12 sgh: waiting for this ancient machine to load it Oct 08 08:25:09 sgh: thats the same error as the insane.bbclass one Oct 08 08:26:17 XorA: argh .... I see why it is called insane.bbclas ;D Oct 08 08:27:18 XorA: But the error occurs somewhere else -> not en openssl-native.bb Oct 08 08:27:31 sgh: thanks for the patch Oct 08 08:27:32 XorA: but in one of its includes/inherits Oct 08 08:27:48 sgh: well if its the same looking code, I guess the same place it fails Oct 08 08:27:58 sgh: though for why I dont know Oct 08 08:28:00 zecke: thanks for helping our with the procedure. Oct 08 08:29:04 XorA: package_qa_walk has something to do with the quality-stuff..... I can remeber how, but I think it can be disabled. Oct 08 08:29:25 XorA: s/can/cant/ Oct 08 08:29:45 morning Oct 08 08:30:34 sgh: what arch is your build machine? Oct 08 08:30:44 XorA: powerpc Oct 08 08:31:02 XorA: sorry .... do you mean HOST Oct 08 08:34:16 ~curse java Oct 08 08:34:17 May you be reincarnated as a Windows XP administrator, java ! Oct 08 08:36:24 sgh: its ok, fails here after a git pull Oct 08 08:37:26 sgh: arse, we need linux in insane.bbclass still for native stuff Oct 08 08:37:32 XorA: ok .... the code that is included in the error is in insane.bbclass. Is it possible to add debugging output to the relevant parts ? Oct 08 08:37:58 sgh: all the info needed is in the error now I read it right Oct 08 08:37:59 XorA: argh .... then linux-gnu could have been avoided for angstrom ? Oct 08 08:39:08 sgh: native gets is TARGET_OS ultimately from uname Oct 08 08:39:42 XorA: then i will recompile uname ;) Oct 08 08:40:15 I really have no idea if its safe to just tag a -gnu on the end of it Oct 08 08:40:58 XorA: where is it done - I can try it. Oct 08 08:41:47 sgh: classes/native.bbclass Oct 08 08:42:28 XorA: TARGET_OS = "${BUILD_OS}-gnu" ?? Oct 08 08:42:58 sgh: yeah, but that looks really wrong to me Oct 08 08:43:47 jo Oct 08 08:44:04 XorA: yes, but I will try anyway. I will build minimal-image from scratch Oct 08 08:44:04 ~curse this whole change Oct 08 08:44:09 May you be reincarnated as a Windows XP administrator, this whole change ! Oct 08 08:44:42 armv4-eabi works Oct 08 08:44:44 cool stuff Oct 08 08:44:57 woglinde: oh, fixing the BX in the linker? Oct 08 08:45:00 XorA: As earlier stated, I do not know why it should change, but I hope there is a good reason. Oct 08 08:45:06 zecke micro has all Oct 08 08:45:10 kernel works Oct 08 08:45:15 userland I will test today Oct 08 08:45:32 micro has all the stuff needed Oct 08 08:45:57 and I ported simpad stuff to 2.6.32 which wasnt hard Oct 08 08:46:25 woglinde: what is still outside the kernel? mq200? Oct 08 08:46:30 yes Oct 08 08:46:35 battery stuff Oct 08 08:46:36 cs3 Oct 08 08:48:23 cs3 and battery stuff needs porting to gpiolib Oct 08 08:48:43 good work of mr_nice and mr_data Oct 08 08:48:49 and thesing Oct 08 08:48:51 for collie Oct 08 08:50:49 wow Oct 08 08:51:57 hrw robert takes the same flight with us to oedem now Oct 08 08:52:07 cool Oct 08 08:52:12 hi florian Oct 08 08:52:17 hi florian Oct 08 08:52:28 good morning Oct 08 08:55:03 good morning florian, all Oct 08 08:55:23 hi dth Oct 08 08:58:06 hrw: I'm looking at your "my-openembedded-environment" posts. do you keep the tmpfiles in the subdir under /build? Oct 08 08:58:34 florian, woglinde: good morning Oct 08 08:58:35 zecke: good evening Oct 08 08:59:57 he pb Oct 08 09:00:06 hm I will bring the simpad to oedem Oct 08 09:01:10 hm can I do git stash; pull ; edit commit; push; unstash; flawless? Oct 08 09:01:24 XorA: http://pastebin.com/d4cf6e80f That is not the way to go. Oct 08 09:02:06 sgh: looks like insane will just have to have replicated linux/linux-gnu sections Oct 08 09:02:28 sgh: unless uber genious zecke has better idea Oct 08 09:02:46 XorA: heh heh ..... should hey be axcaptly the same ? Oct 08 09:03:10 sgh: I think Oct 08 09:04:51 XorA: no idea, we can use references so you can do hash['linux'] = hash['linux-gnu']... but right now I don't really know if it is feasible Oct 08 09:05:50 zecke: XorA: so replication of the linux-gnu section is the best fix for the moment. Oct 08 09:06:22 zecke: well thats an issue, my python knowledge is only slightly better than my Korean Oct 08 09:07:01 03Henning Heinold  07org.openembedded.dev * r4523f9d658 10openembedded.git/recipes/opie-networksettings/opie-networksettings.inc: opie-networksettings: fix install for changedns Oct 08 09:10:01 tasslehoff: yes Oct 08 09:10:12 tasslehoff: devel/build/*/tmp Oct 08 09:11:51 woglinde: about dietibc, It generates bin-i386 and bin-arm but tries to strip them with the same binary... Oct 08 09:12:13 ant yes Oct 08 09:12:18 on debian no problem Oct 08 09:12:19 so, just forcing CROSS=HOST_PREFIX at the beginning doesn't help Oct 08 09:12:29 ant *sigh* Oct 08 09:12:35 that fix is fairly easy Oct 08 09:12:53 woglinde: I ear some x86 libs slipped in... Oct 08 09:12:59 I *fear Oct 08 09:13:02 hm ah okay Oct 08 09:13:08 that would explain the exhagerated size Oct 08 09:13:10 let me think about it Oct 08 09:13:30 http://tinderbox.openembedded.org/public/logs/task/3104257.txt Oct 08 09:13:34 use debian which has multiple-arch strip Oct 08 09:17:23 hrw: ok. what if you want to build some of the stuff in build with stable, and others with dev. do you edit setup-oe.sh each time? I wondered if I could maybe set BBPATH in the script so bitbake could locate my local.conf, and then append the path to my openembedded-tree in local.conf. don't know if that's possible though :) Oct 08 09:18:27 woglinde: all seems fine until strip -R .comment -R .note bin-i386/diet Oct 08 09:18:47 NOTE: make -j 3 -e MAKEFLAGS= all ARCH=arm CROSS= prefix=/oe/build/tmp/staging/armv5te-angstrom-linux-gnueabi/lib/dietlibc Oct 08 09:19:03 ^^ from here the arm-strip is needed imho Oct 08 09:20:54 tasslehoff: setup-oe.sh takes argument - builddir Oct 08 09:21:25 tasslehoff: I do: "cd devel; . setup-oe.sh build/angstrom/" if want to build angstrom. If I want other then I give other dir Oct 08 09:24:10 hm has anyone build an initramfs with uibfs stuff, so you can run mkfs.ubifs from it? Oct 08 09:25:14 why is udev taking sooo long to start Oct 08 09:25:33 robw_w modules loading and device checking Oct 08 09:25:34 zecke pasted "linux and linux-gnu alias" at http://paste.lisp.org/display/88369 Oct 08 09:25:49 XorA: sgh: could you try the above paste? Oct 08 09:25:54 rob_w use static /dev first and start udev later Oct 08 09:26:38 hmm i need to run a watchdog and udev takes too long so the watchdog scripts doesnt get called in time Oct 08 09:26:50 as I said Oct 08 09:26:56 use static dev Oct 08 09:27:00 or mdev from busybox Oct 08 09:27:49 update to 2.6.32-rc and use devtmpfs? Oct 08 09:29:40 hrw: yes, but OEDIR is set in the script and put into BBPATH there. so you Oct 08 09:29:43 argh Oct 08 09:29:46 arg not today .. the proper kernel is still developing Oct 08 09:30:10 don't specify which oe-branch to use in local.conf? Oct 08 09:30:19 mdev or staticdev first then udev Oct 08 09:32:12 zecke: I get some kind of indentation error Oct 08 09:32:40 hrw: what I mean is that BBPATH always points to your org.openembedded.dev. what do you do when you want to build with a frozen repository for release? Oct 08 09:32:46 tasslehoff: no, it uses branch which is in devel/oe/openembedded Oct 08 09:33:04 tasslehoff: my stable branch dir has own copy of OE Oct 08 09:33:15 zecke: ah .. got it .... cmpiling minimal-image now Oct 08 09:35:47 hrw: I think I understand. Have you changed the setup a bit since 2005/2006 when you wrote the blog? Oct 08 09:36:50 tasslehoff: no, still use same Oct 08 09:37:04 tasslehoff: adapted only to bitbake git not svn Oct 08 09:46:35 zecke: ERROR: Error in compiling: Oct 08 09:46:35 ERROR: The lines resulting into this error were: Oct 08 09:46:36 ERROR: 57:IndentationError:' machine_dict['linux'] = machine_dict['linux-gnu']' Oct 08 09:47:15 XorA: ah, space vs. tabs or such :) Oct 08 09:47:50 zecke: thats the thing, I cant see a different, seems to be 8 spaces all the way through Oct 08 09:50:08 xora use syntax higlithing Oct 08 09:50:15 zecke: fixed it Oct 08 09:50:23 zecke: 4 spaces needed Oct 08 09:55:49 zecke: that seems to work here Oct 08 10:00:23 XorA: minimal-image i 3/4 done here... seems to be fine Oct 08 10:06:05 * * OE Bug 5327 has been created by ospite(AT)studenti.unina.it Oct 08 10:06:07 * * gcc-cross-intermediate-4.4.1-r6.1: oe_runmake failed. Oct 08 10:06:09 * * http://bugs.openembedded.org/show_bug.cgi?id=5327 Oct 08 10:09:52 netbase carrys the /etc/network/interfaces and i see machine specific files in its bb directory . but how do those machine specific ones get pulled ? Oct 08 10:10:53 rob_w: via overrides Oct 08 10:12:53 thx Oct 08 10:14:25 anyone familiar with autotools, is Makefile.am required for autoreconf to work or is there another way? Oct 08 10:14:25 hmm how would i activate it ? Oct 08 10:15:42 rob_w: no need to Oct 08 10:16:20 so only the existance of a machine directory and the interfaces file inside netbase/netbase/mymachine is enough ? Oct 08 10:16:25 yes Oct 08 10:16:30 in short Oct 08 10:16:33 k Oct 08 10:16:44 so maybe i missed to rebuild netbase after adding my dir Oct 08 10:30:21 mike_cw: typically, Makefile.am is used but is probably not required. See http://bec-systems.com/site/121/autotools-quick-reference Oct 08 10:34:21 I hope that next change comparable in impact to linux->linux-gnu change will get more testing Oct 08 10:34:39 ... or not happen at all, which might be even better. Oct 08 10:35:20 yeah Oct 08 10:35:25 * Jay7 have strange problem on tosa.. Oct 08 10:35:28 poor change for no gain :( Oct 08 10:35:33 tosa does not boot from mtd Oct 08 10:35:43 right, it seems like an awful lot of disruption for what is at best a slightly prettier TARGET_OS string Oct 08 10:35:43 see no driver for mtdblocks Oct 08 10:36:01 cbrake, that's useful, but it sounds like a missing Makefile.am is definitely going to break autoreconf Oct 08 10:36:32 i'm writing a recipe for xmlrpc-c and it doesn't come with configure.ac or Makefile.am Oct 08 10:36:44 does it come with any autotools bits at all? Oct 08 10:36:44 skipping the autoreconf step seems like it might be a solution, though Oct 08 10:37:12 it comes with configure.in and Makefile.config.in Oct 08 10:37:19 ah right Oct 08 10:37:21 and the pre-generated libtool stuff Oct 08 10:37:28 configure.in == configure.ac, they are basically the same thing Oct 08 10:37:34 so you are using autoconf but not automake Oct 08 10:37:44 Jay7: CFI driver enabled? Oct 08 10:38:07 I thought that autoreconf was smart enough to not run automake for autoconf-only projects, but maybe that is not true Oct 08 10:39:07 MACHINE=native works still ;) Oct 08 10:40:46 zecke: are you going to push the insane.bbclass patch? Oct 08 10:41:27 https://bugzilla.redhat.com/attachment.cgi?id=153816 - someone with X11 experience can look? Oct 08 10:42:56 hrw: MTD_CFI_I1/I2 enabled Oct 08 10:43:01 I4/I8 - not Oct 08 10:43:06 hrw: that seems like a bogus patch Oct 08 10:44:12 pb__: without it I cannot get fbdev working Oct 08 10:44:43 (EE) FBDEV(0): FBIOPUT_VSCREENINFO succeeded but modified mode Oct 08 10:44:43 (EE) FBDEV(1): FBIOPUT_VSCREENINFO succeeded but modified mode Oct 08 10:44:58 without patch it fails on this Oct 08 10:45:27 right, I think you need to debug this properly. Oct 08 10:45:34 the patch is just a lame hack to make it ignore an error condition. Oct 08 10:46:01 pb__: can it be FB driver problem? Oct 08 10:46:09 possibly. Oct 08 10:46:25 more likely it is just a slight mismatch between the video mode that the xserver wants to use and the ones that the framebuffer driver supports. Oct 08 10:46:44 FB is 320x240 Oct 08 10:47:06 geometry 320 240 320 240 16 Oct 08 10:47:06 timings -1 -1 -1 -1 -1 -1 -1 Oct 08 10:47:06 rgba 5/11,6/5,5/0,0/0 Oct 08 10:57:37 hmm.. Oct 08 10:58:06 I've forget what of my public key is on git.oe.org :) Oct 08 11:11:48 XorA: zecke_: It works ..... minimal-image build without errors. Oct 08 11:12:50 who is here having shell access on git.oe.org? :) Oct 08 11:13:30 please do 'id jay7' there :) Oct 08 11:13:30 Jay7: cbrake is git master Oct 08 11:14:17 what is cbrake's timezone? :) Oct 08 11:14:36 pb__, calling oe_runconf directly in do_configure seems to solve my autoconf problem on arm at least. Oct 08 11:14:39 "/ctcp cbrake time" will tell Oct 08 11:15:05 right.. Oct 08 11:15:10 * Jay7 is too slow today Oct 08 11:16:17 hrw: my bootlog: http://pastebin.ca/1603877 Oct 08 11:16:35 hrw: can you have a look/ Oct 08 11:21:49 no idea looks like it finds mtdblocks but not see Oct 08 11:23:17 hmhm.. Oct 08 11:23:28 well.. I'll revert to .29 Oct 08 11:23:47 Jay7: do git bisect? Oct 08 11:24:21 hrw: against linux kernel? Oct 08 11:24:52 yes Oct 08 11:25:15 I'm not ready for this yet :) Oct 08 11:25:41 I'll test .29 and linux-rp .26 before :) Oct 08 11:25:59 re mickeyl Oct 08 11:26:08 morning hrw Oct 08 11:26:48 03Marcin Juszkiewicz  07org.openembedded.dev * rf5c2d8febb 10openembedded.git/recipes/linux/linux_2.6.31.bb: linux 2.6.31: update to 2.6.31.3 Oct 08 11:26:59 03Marcin Juszkiewicz  07org.openembedded.dev * r3b5a46db22 10openembedded.git/recipes/linux/linux_2.6.30.bb: linux 2.6.30: update to 2.6.30.9 Oct 08 11:29:07 does framebuffer driver have to export timings? Oct 08 11:29:28 on BUG I have -1 timing, on at91sam9263ek I have values Oct 08 11:29:56 hm has ipkg or opkg build in stuff which says what is oabi? Oct 08 11:31:24 rather not Oct 08 11:34:07 hms Oct 08 11:34:25 why bitbake isnt searching in armv4 for my packages when building an image Oct 08 11:35:48 EXTRA_PACKAGE_ARCHS? Oct 08 11:36:07 iirc tune-strongarm uses 'arm' not 'armv4' Oct 08 11:36:21 I have already modified this Oct 08 11:36:23 ok Oct 08 11:36:34 check arch.conf used by opkg Oct 08 11:36:41 PACKAGE_EXTRA_ARCHS += "armv4" Oct 08 11:36:46 hm Oct 08 11:36:52 oh where this files resides? Oct 08 11:38:02 somewhere in tmp/work or staging Oct 08 11:38:09 it is generated during rootfs Oct 08 11:38:40 h, Oct 08 11:38:41 hm Oct 08 11:38:46 where is this coming from? Oct 08 11:38:48 pkgarchs='all any noarch arm arm-oabi simpad Oct 08 11:38:57 rootfs_ipkg.bbclass Oct 08 11:39:02 ah Oct 08 11:39:04 gotcha Oct 08 11:39:20 still use simpad? Oct 08 11:39:25 sure Oct 08 11:39:29 tobi likes drawpad Oct 08 11:40:08 ...and asked for eabi Oct 08 11:41:28 woglinde: how old tobi is? Oct 08 11:41:40 3 years and 4 months Oct 08 11:42:04 wher the hack its thinking about oabi Oct 08 11:42:19 PACKAGE_ARCHS is set correct according to bitbake -i Oct 08 11:42:47 woglinde: micro distro? Oct 08 11:42:51 hrw yes Oct 08 11:43:00 but I make it correct Oct 08 11:43:37 woglinde: conf/distro/sane-feed.inc too? Oct 08 11:43:46 # Tweak packaging for strongarm machines since they can't use EABI Oct 08 11:44:04 * hrw -> food Oct 08 11:45:33 sute Oct 08 11:45:35 sure Oct 08 11:45:40 otherwise it woulnd work Oct 08 11:46:09 hm ah no Oct 08 11:46:44 why the hack this sets stuff again Oct 08 11:46:57 why I have to set it in machine/ too Oct 08 12:09:58 sgh: awesome Oct 08 12:10:10 XorA: yes ... Oct 08 12:11:09 sgh: thanks for all your work on this issue Oct 08 12:12:13 XorA: no problem. I was wondering if minimal-image would be a good daily-test to check if the oe-core is working. Oct 08 12:12:27 hi, I know how to use overrides...but how do I add an override for my machine? is it done automatically(some overrides have to be done manually) in the manual they says: "This section is a stub, help us by expanding it" Oct 08 12:12:55 03Graeme Gregory  07org.openembedded.dev * rab16c0637a 10openembedded.git/: Merge branch 'org.openembedded.dev' of git+ssh://git@git.openembedded.org/openembedded into org.openembedded.dev Oct 08 12:13:06 03Omar Barron  07org.openembedded.dev * r8f4dd944ec 10openembedded.git/recipes/linux/linux-omap-zoomsync_2.6.31.bb: Oct 08 12:13:06 linux-omap-zoomsync_2.6.31: update commit id Oct 08 12:13:06 updated commit id to most current release. Older Oct 08 12:13:06 commit id does not work on all zoom2 boards. Oct 08 12:13:07 sgh: yes, that is pretty much the definition of oe-core :-) Oct 08 12:14:18 03Graeme Gregory  07org.openembedded.dev * r51c46dafec 10openembedded.git/recipes/linux/linux-omap-zoomsync_2.6.31.bb: linux-omap-zoomsync_2.6.31.bb : increment OEV not PE Oct 08 12:14:40 XorA: in that case I will try to set up a daily build of minimal-image for a couple of machines. If no such daily build exists that is. Oct 08 12:15:11 sgh: not officially Oct 08 12:17:24 XorA: ok .... I'm not able to do an official build, but I can place a log somewhere accessable. Oct 08 12:17:54 sgh: I think tinderbox is back up and running, you can submit your logs to that Oct 08 12:18:01 * XorA calls for the tinderbox experts Oct 08 12:18:05 Ive never used it Oct 08 12:19:27 anyonw for the overrides? Oct 08 12:19:33 maybe what I told was hidden by the commit log Oct 08 12:19:41 s/told/asked Oct 08 12:20:10 XorA: ok ... I'll look into that. Oct 08 12:20:56 Gnutoo: ${MACHINE} is already an override Oct 08 12:21:07 ok Oct 08 12:21:07 Gnutoo: so you can for example to BLAH_beagleboard Oct 08 12:21:36 yes I wondered if BLAH_beagleboard was automatic Oct 08 12:21:51 that is it recognize the beagleboard keyword as an override Oct 08 12:22:05 Gnutoo: grep for OVERIDES variable on conf/ Oct 08 12:22:11 ok Oct 08 12:22:19 thanks so it was in conf Oct 08 12:22:27 Gnutoo: that will show you for which distro which overrides are auto Oct 08 12:23:04 distro/shr.conf:OVERRIDES = "local:${MACHINE}:shr:angstrom:${TARGET_OS}:${TARGET_ARCH}:build-${BUILD_OS}:fail-fast:pn-${PN}" Oct 08 12:23:09 strange...strage... Oct 08 12:23:24 I'll try to debug why my machine config files are not taken into account Oct 08 12:23:32 for xorg and frameworkd.conf Oct 08 12:24:14 btw thanks a lot Oct 08 12:24:20 no probs Oct 08 12:27:11 well.. 2.6.29 have the same problem on tosa Oct 08 12:28:54 Gnutoo: bitbake something -e|grep ^OVERRIDE Oct 08 12:29:36 ok Oct 08 12:29:42 thanks Oct 08 12:40:29 oh meta-toolchain is borked too Oct 08 12:40:37 maybe it is best to revert all linux-gnu changes? Oct 08 12:41:30 zecke_: crap more breakage :-( Oct 08 12:41:56 that is a != eabi build for mips Oct 08 12:42:10 * XorA is getting very sick of the word gnu recently Oct 08 12:44:47 zecke_: I guess a revert and an order to do on a branch for proper merging is in order Oct 08 12:47:09 XorA: do you want to write the mail? Oct 08 12:48:00 hrw: patch 2.6.31.3 have no entry in checksum.ini :) Oct 08 12:48:17 gst-plugins-base_0.10.25.bb does not compile with the latest changes. I do not know if it is linux-gnu related. Oct 08 12:48:24 ops Oct 08 12:48:38 zecke_: I will do, but I wont be able to do the actual revert as in berlin without net access tomorrow :-) Oct 08 12:49:21 XorA: you in berlin tomorrow? Oct 08 12:49:47 03Marcin Juszkiewicz  07org.openembedded.dev * r43556d2323 10openembedded.git/conf/checksums.ini: checksums.ini: added missing entries Oct 08 12:50:14 hrw: yes, ryanair only flies some days Oct 08 12:50:35 oh i need to hurry with my proxy form Oct 08 12:50:47 zecke yes Oct 08 12:53:29 zecke_: pity you are AWOL :-( Oct 08 12:55:08 ah.. e.v. meeting this weekend.. Oct 08 12:55:08 forgot about it Oct 08 12:55:08 so next week I can re-request membership Oct 08 12:55:09 AWOL? Away... something? Oct 08 12:55:22 Absent Without Leave Oct 08 12:55:38 ah Oct 08 12:57:48 argh... me forgot to go out and wait for the trash car to come... Oct 08 12:58:08 zecke whaahaha Oct 08 12:59:40 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=3f7a26b4b9768fe31597d1af35106aa512dc3742 - I love that commit message ;D Oct 08 13:02:00 woglinde: jetzt kroege ich haue Oct 08 13:02:21 zecke stells an die strasse Oct 08 13:02:29 macht man in neukoelln auch so Oct 08 13:02:35 ja, naturlisch Oct 08 13:02:55 hrw :) Oct 08 13:02:59 spaciba Oct 08 13:03:00 haha Oct 08 13:04:27 hm how can fix such errors? Oct 08 13:04:28 * ERROR: Cannot satisfy the following dependencies for task-opie-16mb-settings: Oct 08 13:04:28 | * libqpe-opie (>= 1.2.4) * libopiecore2 (>= 1.2.4) * qte-mt (>= 2.3.10) * Oct 08 13:04:47 chainsaw Oct 08 13:04:49 woglinde: missing the PACKAGE_ARCH for eabi? Oct 08 13:05:09 should be set Oct 08 13:05:27 do I need to extract the ipk? Oct 08 13:05:32 woglinde: /etc/opkg/arch.conf Oct 08 13:05:46 woglinde: to see the arch? no you can see it from the filename Oct 08 13:06:09 there is only opkg.conf Oct 08 13:06:12 no arch.conf Oct 08 13:06:19 in rootfs Oct 08 13:06:36 ah minimal Oct 08 13:06:41 micro Oct 08 13:07:21 hm /devel/arm/oetmp-uclibc/deploy/ipk/Packages is empty Oct 08 13:08:29 normal Oct 08 13:08:36 hm okay Oct 08 13:11:58 will simply try to rebuild Oct 08 13:13:04 woglinde: nah, think... Oct 08 13:13:07 fuck. Intel x11 guys makes me mad Oct 08 13:13:17 woglinde: do you have a libqpe-opie package? Oct 08 13:13:24 zecke yes Oct 08 13:13:32 woglinde: where is it located? Oct 08 13:13:38 armv4 Oct 08 13:13:42 in deploy Oct 08 13:14:19 woglinde: in tmp/work/../rootfs/... how does the opkg.conf that is used by opkg-cl look like? does it list your feed? Oct 08 13:17:06 hm Oct 08 13:17:10 there is no feed Oct 08 13:17:37 but qte-mt was installed Oct 08 13:21:02 mmm SRC_URI = "file://xorg.conf" is not copied into the workdir Oct 08 13:21:20 hm Oct 08 13:21:22 Cannot create directory `/var/lib/opkg': Permission denied Oct 08 13:21:28 hrw: whats up this time? Oct 08 13:21:41 XorA: same as usual - 855GM Oct 08 13:21:58 hrw: at least the speeded up eee 901 slightly Oct 08 13:21:58 hrw is this normal too? Oct 08 13:22:08 hrw: still mega slow compared to win7 drivers though Oct 08 13:22:15 woglinde: yep Oct 08 13:22:26 hehe will my 901 utms to oedem Oct 08 13:22:26 bring Oct 08 13:22:31 hm but where is the error now Oct 08 13:22:58 hm.. .31 fails to build Oct 08 13:23:32 seems just copying defconfig is not enough Oct 08 13:23:35 XorA: what you expect from quite old chipset? Oct 08 13:24:19 hrw: the piss poor intel drivers are what keep nvidia laptop chipsets in business :-) Oct 08 13:24:43 XorA: exactly Oct 08 13:27:30 faillog from .31: http://pastebin.ca/1604117 Oct 08 13:27:51 something is wrong with prism2fw.c Oct 08 13:27:59 staging... Oct 08 13:28:22 Jay7: you know that prism2usb is scheduled for removal from staging? Oct 08 13:28:23 parallel make is disabled here Oct 08 13:28:32 hrw: no :) Oct 08 13:28:42 hrw: are we losing all hope for prism drivers then? Oct 08 13:28:52 XorA: looks like Oct 08 13:29:05 ehmm... Oct 08 13:29:19 looks like no one uses them now Oct 08 13:29:32 ralink etc chips are more popular in usb sticks Oct 08 13:29:34 mostly becuase the drivers never works right :-) Oct 08 13:29:51 I have nothing against usb prism but not cf :) Oct 08 13:30:03 *against removal of Oct 08 13:30:50 Jay7: hostap_cs and orinoco_cs works fine for cf and are in mainline since I do not remember when Oct 08 13:31:20 good to hear that.. Oct 08 13:31:43 just poor tosa is left without wifi Oct 08 13:32:03 is tosa using usb wifi inside? Oct 08 13:32:10 Jay7: yes Oct 08 13:32:15 oh sh@t.. Oct 08 13:32:21 Jay7: never checked how your wifi is connected? Oct 08 13:32:34 hrw: I've no chance before :) Oct 08 13:33:02 Jay7: iirc tosa is really fscked inside. wifi on usb, BT on 115200 port instead of 921600 one Oct 08 13:33:06 I've no time before :) Oct 08 13:33:25 I have no bt at all.. only wifi Oct 08 13:33:32 but the screen... awesome Oct 08 13:33:35 I have no BT either Oct 08 13:33:40 well.. Oct 08 13:33:49 I'll retry bitbaking an image Oct 08 13:33:59 seems we should stick with .30 then Oct 08 13:34:31 but I'll try to build .31 w/o wifi Oct 08 13:35:19 Jay7: we should probably try and fix it, we can always keep prism2usb as an OE patch Oct 08 13:35:30 Jay7: it will never work with WPA though I guess Oct 08 13:38:03 XorA: I'm not an kernel/driver hacker (at least right now) :) Oct 08 13:38:18 lumag have broken screen on his tosa Oct 08 13:38:57 Jay7: he needs to contact with Samuel Ortiz again Oct 08 13:39:01 lumag? Oct 08 13:39:23 aka Dmitry Baryshkov Oct 08 13:39:29 woglinde: Tosa hacker Oct 08 13:39:34 hm Oct 08 13:39:36 aka Dmitry Eremin-Solenikov :) Oct 08 13:40:02 there is a spare OE tosa here Oct 08 13:40:22 if we can get it to him Oct 08 13:40:44 hm.. that can be good idea Oct 08 13:40:55 XorA: Samuel has spare OE tosa too Oct 08 13:41:03 mine is totally wasted in my flat Oct 08 13:41:07 I'll ask him Oct 08 13:47:30 XorA: heh, clearly you are a bad influence. Oct 08 13:47:43 hanging around in your flat all day and getting wasted. Oct 08 13:48:16 I've emailed lumag about our talk Oct 08 13:49:00 quick question: if i would like to use a package that would over some ssl features. the version in the tree however comes without this support. should i create a new package *-ssl oder just edit the package recipe ? Oct 08 13:49:05 may be give him clamshell too to speedup sharpsl_pm replacing? :) Oct 08 13:49:17 over = offer , ... Oct 08 13:49:33 I'll ask to shr guys...it's not normal... Oct 08 13:49:36 pb__: :-D Oct 08 13:49:43 this time I have: Oct 08 13:49:58 *blank xorg.conf Oct 08 13:50:01 *generic frameworkd.conf Oct 08 13:50:01 pb__: I've been wasted too around XorA's flat ;) Oct 08 13:50:16 and it took a lot of time to re-bitbake the image as it is autorev Oct 08 13:50:20 * lrg casts his mind back to the rum bar Oct 08 13:50:23 and there is the override Oct 08 13:50:30 well.. .31 is compiled w/o prism2usb Oct 08 13:51:56 lrg: your back there in 2 weeks :-) Oct 08 13:53:08 XorA: yeah, but I wont be geting wasted this time ;) Oct 08 13:53:43 lrg: morgen Oct 08 13:54:11 lrg: bring your girl, she can carry you home Oct 08 13:54:32 hey Marcin Oct 08 13:54:34 XorA have expirience ;) Oct 08 13:54:36 XorA: I will Oct 08 13:55:04 Luckilly the bar is close enough I can actually crawl home Oct 08 13:55:28 It's uphill from your flat so rolling is probably another option :) Oct 08 13:56:41 broonie: its pretty much the same level Oct 08 13:56:49 commercial street is flat Oct 08 13:56:58 XorA: you could swim some of the way ;) Oct 08 13:57:19 yeah thats a good way to sober up :-D Oct 08 13:57:32 cant get too wrecked, its Torture Garden the next day Oct 08 13:58:19 03Koen Kooi  07org.openembedded.dev * r8b1f4dc04f 10openembedded.git/recipes/gstreamer/gst-plugins-base_0.10.25.bb: gst-plugins-base: add pango as dep Oct 08 13:58:57 XorA: maybe you can help me out (question, see above) Oct 08 13:59:21 mlip: the answer is yes -ssl version Oct 08 13:59:50 Jay7: do we have clamshells somewhere? Oct 08 14:00:36 morning all Oct 08 14:00:43 hi Richard Oct 08 14:00:48 hey Richard Oct 08 14:01:24 RP: I have pictures of you and Joe balancing on that log...... Oct 08 14:01:29 hrw: I have akita.. Oct 08 14:02:01 lrg: I should have one of you as well. I'll have to check and upload them somewhere :) Oct 08 14:02:19 heh dodgy photo trade :_) Oct 08 14:07:06 morning Oct 08 14:08:38 hi kergoth Oct 08 14:11:02 .31 hanging Oct 08 14:11:56 well.. Oct 08 14:11:58 hi kergoth Oct 08 14:33:55 hmm for some reason my /etc/rcS.d directory gets squashed to a file ?!? Oct 08 14:34:04 morning.... Oct 08 14:35:03 mornnig Oct 08 14:36:36 hi orbarron, hi chouimat Oct 08 14:41:28 being in .ca.us means I miss loads of chatter here :( Oct 08 14:47:36 rob_w: that's probably a careless install command Oct 08 14:47:55 something like 'install -m 0755 myscript.sh /etc/init.d/rcS' without checking that rcS exists first Oct 08 14:52:21 ah you got me Oct 08 14:59:24 well Oct 08 14:59:36 also when i run -DDD at image creation i get this for file in '$(cat ${TESTLAB_DIR}/installed-packages.txt)' Oct 08 14:59:37 ++ find /data/oe/rubicon/tmp/deploy/glibc/ipk -name xdpyinfo_1:1.0.3-r0.3_ppc440e.ipk Oct 08 14:59:37 + head -n1 Oct 08 14:59:37 + du -k Oct 08 14:59:37 du: write error Oct 08 14:59:58 is that critical or what is all the du foo Oct 08 15:00:16 XorA, hrw: lumag is interested :) Oct 08 15:01:06 so.. Oct 08 15:01:17 what is our next steps? Oct 08 15:03:17 Jay7: let he mail Samuel Ortiz about tosa which I gave to him during 'OH is now Intel' event - those words + Cc: to me Oct 08 15:04:18 please 'msg' me Samuel's and your email :) Oct 08 15:04:30 ha Oct 08 15:05:11 Samuel Ortiz + Marcin Juszkiewicz Oct 08 15:05:21 thanks Oct 08 15:05:23 both emails are public and available by simple google search Oct 08 15:05:41 the main issue as always is posting Oct 08 15:06:19 I'll ask about some free collie in ML today Oct 08 15:06:33 we will try to combine delivery Oct 08 15:07:14 hrw: is Samuel from USA? Oct 08 15:08:16 there is a free collie at my flat as well Oct 08 15:08:36 hehe.. Oct 08 15:09:13 Jay7: France Oct 08 15:09:20 I have an OE one + my own one Oct 08 15:09:31 Jay7: he will attend ELC-E next week Oct 08 15:09:40 XorA is antiquarian collector :) Oct 08 15:09:47 I have only c760 Oct 08 15:09:53 my own one Oct 08 15:09:53 they were dumped on me at FOSDEM Oct 08 15:10:07 yep.. Oct 08 15:10:45 it was like 'I took those Zaurii from US - who in Europe will take them?' - no one wanted so XorA got them Oct 08 15:10:47 well.. I have relative in Spain and NL Oct 08 15:10:53 XorA: I always seem to loose kit at shows not gain it, must be doing something wrong ;-) Oct 08 15:11:14 DJWillis: you take too much stuff with you Oct 08 15:11:42 Jay7: getting them to NL should be easy, I could mail koen at his office in UK, he can carry them :-) Oct 08 15:11:51 hrw: could be it Oct 08 15:12:36 I was worried about getting done for PDA smuggling coming back to UK Oct 08 15:15:16 XorA: when no one is against I'll grab your collie :) Oct 08 15:15:32 \o/ Oct 08 15:15:46 XorA: a cavity search for Gumsix devices would not be a nice thought ;-) Oct 08 15:16:01 DJWillis: dunno, could be fun :-D Oct 08 15:16:06 oh no my whole rootfs is not booting anymore .. wtf ! Oct 08 15:16:29 * Jay7 wants to see XorA's flat Oct 08 15:16:38 there are too much devices around ;) Oct 08 15:17:06 Jay7: send an email to oe-devel about device transfers, check no-one objects then we can work out transport Oct 08 15:17:17 ok Oct 08 15:17:34 mailing to .nl should be cheap/easy Oct 08 15:26:09 emailed Oct 08 15:41:57 hm.. what is state of usbnet on Z's? Oct 08 15:44:03 Jay7: all works Oct 08 15:44:22 hrw: what's the status of xmodmap for Z? Oct 08 15:44:28 no idea Oct 08 15:45:19 we gave up for a while and concentrated on bootloader... Oct 08 15:45:20 ant__: would be better if you made a proper keymap file, xmodmap is so 1980's :-) Oct 08 15:45:21 I have problems with .30 on tosa and .30 on my workstation Oct 08 15:45:53 XorA: then we should move to xorg :) Oct 08 15:46:02 XorA: what are the keymaps in kernel for? Oct 08 15:46:03 and use hald Oct 08 15:46:18 ant__: wasting memory, as I understand it :-} Oct 08 15:46:34 same as the rest of the in-kernel vt code Oct 08 15:46:42 ^_^ Oct 08 15:49:14 btw I see installed /all/xserver-common_1.25-r3.3_all.ipk Oct 08 15:49:39 and xserver-kdrive-imageon_1:1.2.0-r8.3_armv5te.ipk Oct 08 15:49:51 isn't that ati imageon used on HTC too? Oct 08 15:50:08 perhaps they did some more development Oct 08 15:50:34 ant__: only in HP hx4700 Oct 08 15:51:44 re Oct 08 15:51:46 he pb Oct 08 15:54:25 woglinde: have you thought how to switch cross-strip in the middle of the build? exporting the new CROSS ? Oct 08 15:54:39 ant_ new make target Oct 08 15:55:05 but didnt have time to test it Oct 08 15:55:07 maybee I am wrong anymway Oct 08 15:55:09 is not urgent but I'm interested by the size you got... Oct 08 15:55:27 which one? Oct 08 15:55:32 ah Oct 08 15:55:33 okaa Oct 08 15:55:35 okay Oct 08 15:55:36 mom Oct 08 15:55:40 will try with armv4 Oct 08 15:56:39 bye Oct 08 15:56:43 bye hrw Oct 08 15:57:38 woglinde: btw is it correct to strip the binaries at this stage (do_compile ?) Oct 08 15:57:42 ant think about it Oct 08 15:57:44 ok, *libc are special case Oct 08 15:58:00 static compile Oct 08 15:58:21 there is no -dbg dietlibc package Oct 08 15:59:53 you shouldn't need to strip the static libs themselves, just the resulting binaries. Oct 08 16:02:06 woglinde: I think I did something insane forcing dietlibc to compile.. Oct 08 16:02:23 possibly I pulled in host libs Oct 08 16:02:54 just by setting CROSS=arm-angstrom-linux-gnueabi- Oct 08 16:03:13 and bitbaking again Oct 08 16:03:54 fwiw dietlibc was built, but the size of kexecboot-dietlibc was >300k Oct 08 16:04:08 (to resume the issue for new readers :) Oct 08 16:04:43 eh..kexec-dietlibc Oct 08 16:04:53 ant where is the seperated patch now? Oct 08 16:05:25 in kexec/files Oct 08 16:05:46 hm oh yo you didnt real split up Oct 08 16:06:24 well, I applied to both shared and static Oct 08 16:06:54 hm dodnt see it here Oct 08 16:07:03 ah okay Oct 08 16:07:07 right inc file Oct 08 16:07:12 in kexec/files/no-getline-nofscanf.patch Oct 08 16:07:38 so let me see if it compiles with armv4-eabi Oct 08 16:07:44 pls do Oct 08 16:08:30 btw right name would be recipes/kexec-tools Oct 08 16:09:19 hehe were is fastes with gdb-7.0 recipe? Oct 08 16:09:47 ant I will apply some debian patches too Oct 08 16:09:47 for dietlibc Oct 08 16:09:55 great Oct 08 16:11:03 so compiled Oct 08 16:11:04 let me see Oct 08 16:11:08 the size Oct 08 16:11:46 157304 8. Okt 18:10 /devel/arm/oetmp-uclibc/work/armv4-oe-linux-uclibceabi/kexec-tools-dietlibc-2.0.1-r0/image/sbin/kexec Oct 08 16:12:01 yeah armv4 needs more operations Oct 08 16:12:11 and no thumb Oct 08 16:12:56 -rwxr-xr-x 1 andrea users 141664 Oct 7 22:52 kexec Oct 08 16:12:56 andrea@mizar /oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/kexec-tools-klibc Oct 08 16:12:56 -static-2.0.1-r1/install/kexec-klibc-static/usr/sbin $ Oct 08 16:13:41 hm mom will try with klibc Oct 08 16:14:54 hm building klibc needs more DEPS Oct 08 16:15:07 perl Oct 08 16:15:11 klcc Oct 08 16:15:13 ah right Oct 08 16:15:15 yes Oct 08 16:15:22 perl is needed for crappy klcc Oct 08 16:19:53 hm intressting Oct 08 16:19:55 142988 8. Okt 18:19 kexec Oct 08 16:30:03 jo crofton Oct 08 16:30:12 gm Oct 08 16:30:18 any luck with patchwork? Oct 08 16:31:23 didnt test Oct 08 16:33:04 k Oct 08 16:33:14 I'll be home next week Oct 08 17:05:20 hms libgcc vs libgcc1 Oct 08 17:08:56 as a package name? it should always be libgcc1. Oct 08 17:09:00 where are you seeing "libgcc"? Oct 08 17:09:33 opie-task Oct 08 17:10:16 ah, the magic word Oct 08 17:10:21 he Oct 08 17:10:32 I like it Oct 08 17:11:09 lol webos has no filebrowser Oct 08 17:11:16 * ERROR: Cannot satisfy the following dependencies for task-opie-16mb-base: Oct 08 17:11:16 | * libgcc (>= 4.4.1) * Oct 08 17:11:34 lol Oct 08 17:11:41 Depends: tslib-conf, libc0 (>= 0.9.30.1), libgcc (>= 4.4.1) Oct 08 17:11:46 now I got it Oct 08 17:12:15 I hope kergoth has a new release soon Oct 08 17:12:57 release? Oct 08 17:13:04 hm Oct 08 17:13:12 wasnt tslib you project? Oct 08 17:13:36 so why the heck tslib thinks it needs libgcc instead of libgcc1 Oct 08 17:13:39 heh, that made it sound as though you wanted a new version of kergoth himself Oct 08 17:14:01 "this kergoth sucks, I will wait for 2.0" Oct 08 17:14:07 pfff Oct 08 17:14:28 that is a bit weird, maybe some kind of shlibs data mixup in staging. Oct 08 17:14:35 the tslib X driver needs a refresh too Oct 08 17:14:39 did you see messages along the lines of "unable to find shared library provider for..."? Oct 08 17:14:40 * CosmicPenguin runs Oct 08 17:14:59 hm not after rebuilding Oct 08 17:15:28 maybee its a problem when gcc is at intermediate? Oct 08 17:16:50 possibly. I don't think you should be able to build anything against intermediate gcc due to deps, but who knows. Oct 08 17:16:59 * pb__ go home now Oct 08 17:17:05 bye pb Oct 08 17:17:09 or, more accurately, go to grocery store to deal with large shopping list from my wife Oct 08 17:17:12 later all Oct 08 17:17:19 yeah now it worked Oct 08 17:17:22 strange Oct 08 17:58:56 hmm, openssl-native fails to build Oct 08 17:59:03 is this related to the gnu-linux changes? Oct 08 17:59:11 cbrake: whats the error Oct 08 17:59:51 khem: http://pastebin.ca/1604782 Oct 08 18:01:17 what is your build machine running Oct 08 18:01:28 yes seems to be Oct 08 18:02:12 this is sakoman's gumstix repo Oct 08 18:02:51 oh Oct 08 18:02:59 he khem Oct 08 18:03:02 have there been any recent fixes? Oct 08 18:03:05 cbrake: check openssl.inc do_configure Oct 08 18:03:13 hi cbrake Oct 08 18:03:18 woglinde: howdy Oct 08 18:03:21 woglinde: servus Oct 08 18:03:32 khem armv4 eabi kernel works Oct 08 18:03:39 :) good news Oct 08 18:03:48 now I am trying userland Oct 08 18:03:51 a nail in coffin for oabi :) Oct 08 18:03:58 khem yes yes Oct 08 18:04:37 once it works for you we should make sure to obsolete oabi in OE Oct 08 18:04:55 khem hm Oct 08 18:05:26 hm I will raise the topic at oedem Oct 08 18:05:31 cbrake: compare to os Oct 08 18:05:33 woglinde: plz Oct 08 18:06:11 khem: no differences Oct 08 18:06:13 afternoon question, i have a verdex board that was purchased 2 years ago compiled with buildroot, can i now use OE Oct 08 18:06:21 and redo the board Oct 08 18:06:26 hm someonefamiliar with the ubifs options here? Oct 08 18:13:47 I don't think we want to remove support for oabi in OE. That would be silly. Oct 08 18:14:28 hm because I am the only one using simpad I will switch to eabi for it Oct 08 18:15:04 right, that makes sense Oct 08 18:18:04 How does makedevs work? Oct 08 18:18:47 in what sense? Oct 08 18:18:57 it just reads a config file and creates the nodes that it mentions Oct 08 18:18:59 erm i think it uses oe/files/device_table*.txt Oct 08 18:19:03 I don't think there's much more to it than that. Oct 08 18:19:42 hey, either of you run into the 8 character crypt() limit for passwords? Oct 08 18:20:06 ok.. sounds more or less trivial Oct 08 18:21:13 m4t: not quite sure what you mean, but classical crypt() does indeed have an 8-char limit. Oct 08 18:21:25 the des algorithm that it uses only works with a 56-bit key Oct 08 18:21:34 hms I need newer ubifs stuff Oct 08 18:21:43 heh i think the md5_crypt.c in libcrypt does 8 char as well Oct 08 18:21:49 florian: yeah, it is fairly trivial. the source is in the makedevs directory if you want to look at it. Oct 08 18:21:55 eg. if i set my pass to 1234567890 Oct 08 18:22:11 12345678 works as the pass, as does 12345678..... Oct 08 18:22:46 pb_: ah.... i see. and the nodes that it is not allowed to create are just ignored i guess Oct 08 18:22:56 * florian takes a look at the source Oct 08 18:22:56 this is with tinylogin and uclibc, but i think that uclibc libcrypt is the issue Oct 08 18:22:58 m4t: that is a bit odd. is this glibc libcrypt, or some other? Oct 08 18:23:01 ah, uclibc Oct 08 18:23:04 yup Oct 08 18:23:28 this is with md5 /etc/shadow as well Oct 08 18:23:53 busybox internal shadow functions shouldn't be of relevance since this is openssh linked to uclibc Oct 08 18:24:23 hm, md5.c in uclibc doesn't obviously seem to have an 8-char limit Oct 08 18:25:25 I guess it would be easy enough to write a little test program to exercise it and check the results. Oct 08 18:26:40 oh, actually, crypt() manpage does say that md5 is limited to 8 significant chars for password Oct 08 18:26:55 apparently you need to use SHA to make the whole password significant Oct 08 18:27:16 yea i found the line in md5_crypt.c i think Oct 08 18:27:42 grep sp md5_crypt.c|grep 8 Oct 08 18:27:57 that is odd, my uclibc doesn't have an md5_crypt.c Oct 08 18:28:01 it's just md5.c in this version Oct 08 18:28:09 hm Oct 08 18:28:10 ubifs Oct 08 18:28:19 why it have so complicated Oct 08 18:28:23 erm Oct 08 18:28:24 http://git.uclibc.org/uClibc/tree/libcrypt/md5.c?h=0_9_30 Oct 08 18:28:25 yea Oct 08 18:28:27 soryry Oct 08 18:28:35 for(ep=sp;*ep && *ep != '$' && ep < (sp+8);ep++) Oct 08 18:28:36 that line? Oct 08 18:28:43 I think that's to do with the salt, not the key itself Oct 08 18:28:54 yup Oct 08 18:28:55 oh Oct 08 18:29:03 that actually is probably right Oct 08 18:29:13 yeh sp = salt Oct 08 18:29:14 hrm Oct 08 18:30:34 pb_: any hashing algorithm the longest meaningful password is the length of the hash Oct 08 18:31:59 well, that's not strictly true. Oct 08 18:32:25 pb_: it is, once password is longer than hash there are more than one solution to the problem Oct 08 18:32:31 obviously if you allow longer passwords then there will be a 1:n mapping from passwords to hash values, but unless your password is a string of completely random bits you are probably still better off with a longer password. Oct 08 18:33:12 i might just have to use ssh keys Oct 08 18:33:13 all the attacks on hashes are to quicken the finding of one of the possible solutions Oct 08 18:33:33 you dont normally care which one you get Oct 08 18:34:02 right, but you do care that your password is not susceptible to dictionary attack. Oct 08 18:34:24 do_testlab what is this actually doing ? Oct 08 18:35:04 if you accept that many users will choose dictionary-like words for their passwords, you are better off allowing them to use long ones in order to make them less guessable, even if this means that there is a shorter "alias" password which would work as well. Oct 08 18:35:13 the shorter password is liable to be much more random and less guessable. Oct 08 18:35:33 interesting Oct 08 18:36:01 yes that makes sense, that overlap occurs if the hash space is smaller than the key space Oct 08 18:36:07 if your users do invent completely random bit-patterns for their passwords in the first place then you're right, there is no benefit in allowing the password length to exceed the amount of available hash space. Oct 08 18:39:17 pb hm do you remeber how I can erease a partiton with old simpadbootloader? Oct 08 18:41:49 the hh.org bootloader, or the original siemens one? Oct 08 18:41:52 is it possible to get opkg to prioritise one repository over another, regardless of package version? Oct 08 18:41:55 hh,org Oct 08 18:42:09 "flash erase"? something like that, let me look at the source Oct 08 18:42:22 alright thanks, Oct 08 18:42:34 hm I dont want delete the partition from partition table Oct 08 18:42:42 only cleaning thats why I am asking Oct 08 18:42:56 ah, seems to be "eflash " according to howto Oct 08 18:43:26 http://www.handhelds.org/handhelds-faq/bootldr.html Oct 08 18:43:32 ah okay Oct 08 18:43:33 thanks Oct 08 18:49:10 03Stanislav Brabec  07org.openembedded.dev * r6cb4ec0ce5 10openembedded.git/recipes/matchbox2/ (2 files in 2 dirs): matchbox-desktop-2: Make fallback folder compatible with current matchbox-common. Oct 08 18:50:52 pb_: so oe uses it to create the initctl fifo only? Oct 08 18:51:16 * florian just ran it on his workstation and admired the boring result Oct 08 18:54:40 ah Oct 08 18:55:08 * florian just spotted the fakeroot call :-) Oct 08 18:55:34 now things are much more lear Oct 08 18:55:37 +c Oct 08 18:55:41 heh Oct 08 18:56:15 i wondered how this thingy could do more use than creating the fifo :-) Oct 08 19:01:12 florian: get more sleep! Oct 08 19:01:19 * florian -> home, bbiab Oct 08 19:03:29 hi, from tinderbox is possible to obtain info about the local.conf file used to build ? Oct 08 19:50:19 pb_: about makedevs, there is a patch around fixing the count from 0 Oct 08 19:50:31 pb_: mickey|sofa hacked it, nobody else Oct 08 19:50:54 pb_: being this is core-thing, another core-dev should ack Oct 08 19:51:39 ant__: core devs are dead Oct 08 19:51:43 imho lot of interesting patches are starving in patchwork Oct 08 19:52:31 XorA|gone: what to do? Just go ahead and commit? If anyone is disturbed they'll revert? Sort-of? Oct 08 19:52:47 ant__: can it all wait until OEDEM? Oct 08 19:53:02 ant__: I think we are just in the period of transition to eV Oct 08 19:53:04 of course Oct 08 19:53:20 cbrake: ping :) Oct 08 19:53:24 ant__: OE came more alive after last OEDEM as people felt empowered Oct 08 19:53:43 Jay7: yes Oct 08 19:53:46 ant__: personally Im just crazy busy Oct 08 19:54:04 XorA|gone: as you know I'm just hobbist/enthusiast, but I feel lot of devs flow away... Oct 08 19:54:07 cbrake: are you the guy for ssh keys for git these days? Oct 08 19:54:12 cbrake: seems I've forget my login to oe git :) Oct 08 19:54:20 XorA|gone: sure Oct 08 19:54:24 re Oct 08 19:54:25 cbrake: check please presence of jay7 user Oct 08 19:54:33 Jay7: ok Oct 08 19:54:40 ant__: not flew, we all got jobs based on our OE experience which perversly meant less time to work on OE Oct 08 19:54:54 cbrake: cool, not for me, mine is still valid Oct 08 19:55:00 I've found right ssh key but I don't know what login I have :) Oct 08 19:55:15 XorA|gone: perfectly licit Oct 08 19:55:20 Jay7: login does not matter -- ssh key is the only thing it checks Oct 08 19:55:25 hm.. Oct 08 19:55:33 Jay7: it would be nice to be consistent though for records Oct 08 19:55:41 ant__: feel free to apply for TSC if you think you can bring some drive Oct 08 19:55:50 Jay7: are you listed here: http://wiki.openembedded.net/index.php/OpenEmbedded_Developers Oct 08 19:55:57 cbrake: well.. lets replace key then :) Oct 08 19:56:14 XorA|gone: sorry...no time for this...and there are better candidates :) Oct 08 19:56:43 Jay7: sure, send me the key and tell me who you are :-) Oct 08 19:56:51 florian: about makedevs, look in patchwork, there is a nice fix Oct 08 19:56:58 ack and commit :) Oct 08 19:57:04 sounds sane Oct 08 19:58:44 Submitter: Antonio Ospite Oct 08 19:58:47 RQ33LOh313nUTcoxr9ROz3wU7AAAAFQDw3qyYqFPmr2ghr9L7oCzFV05mswAAAIAqWpmUAHvxgHCCmR28jkAT5Sul4uufIrd4hYokUW1rC3hB1DA+46RG5FGQtsd78yv4+N6RA3LgomMPmUOfXBDVbfRfobrMafzQT/BukkFte2DZA2 Oct 08 19:58:50 69ut5VMC9dKhGEDAyx7eqB9pCf+F1GkzT9pAldth8Qoq74kOwtZKFPGt+abwAAAIBG77HPrV5NyCJUwtpka4d3aSetT11bB9QmBlJ/Rc22BIKlGKSB9Yb9ZNCtjKVly2oJSEig5AooqY1Ot+jbTSLTJmEWWa+BWz3zMqxQVEQCs42aL Oct 08 19:58:59 (doh!) Oct 08 20:03:03 hehe :) Oct 08 20:03:26 * kergoth chuckles Oct 08 20:03:32 * Jay7 now have magic power to push Oct 08 20:16:43 jay7 hehe Oct 08 20:16:45 great Oct 08 20:17:24 Jay7: telekinesis? :) Oct 08 20:17:41 denix: kinda :) Oct 08 20:17:42 London stock exchange is moving to ubuntu :) Oct 08 20:18:01 dumping windoze Oct 08 20:18:07 mickey|sofa: ping Oct 08 20:25:50 khem und solaris Oct 08 20:27:22 kergoth: Can I get the link to your personal OE wishlist/ good to have stuff that you had pointed me to sometime back? /me is unable to hunt that page back :( Oct 08 20:27:50 the one on the web isn't completely up to date at the moment, but sure, give me a sec Oct 08 20:28:49 http://dl.getdropbox.com/u/112715/Documents/OpenEmbedded%20Tasks.html/index.html Oct 08 20:29:45 there's also http://dl.getdropbox.com/u/112715/Documents/OpenEmbedded%20Usability%20Concerns.html/index.html Oct 08 20:30:03 and like 5 other random omnioutliner docs on oe related stuff.. i should really organize and consolidate Oct 08 20:30:03 . Oct 08 20:30:05 heh Oct 08 20:30:08 kergoth: have you had any time to unbreak packaged-staging? Oct 08 20:30:28 kergoth: thanks :) Oct 08 20:30:29 nope, quite busy with work, haven't had time to look at upstream OE builds or pushing anything upstream or anything Oct 08 20:30:41 khem hm eabi userland seems not to start Oct 08 20:31:08 kergoth: np Oct 08 20:31:48 khem -> http://pastebin.ca/1605076 Oct 08 20:32:45 someday i'll find some time and motivation... someday... Oct 08 20:32:47 * kergoth sighs Oct 08 20:34:03 kergoth: if you would hav ebeen in EU I'd have had the way to compensate your efforts: an 1996 Alfa GTV V6TB for cheap... Oct 08 20:34:24 Thank you for your help - would you like to buy a used car from me? Oct 08 20:34:44 kergoth: my co-worker sold it for 700 Euro Oct 08 20:36:16 woglinde: is it uclibc ? Oct 08 20:36:23 khem yes Oct 08 20:36:39 do you have BX enabled in .config Oct 08 20:36:44 no Oct 08 20:36:52 should I Oct 08 20:36:58 no Oct 08 20:38:50 hm I have oabi compat enabled in the kernel Oct 08 20:38:55 might this be a problem? Oct 08 20:39:03 I dont think so Oct 08 20:39:20 but the error itself doesnt give much clue what could be wrong Oct 08 20:41:19 hm init isnt working Oct 08 20:41:24 hm ah let mem remeber Oct 08 20:41:37 there was a debug option showing how init faile Oct 08 20:41:39 d Oct 08 20:42:48 ah there Oct 08 20:42:50 CONFIG_DEBUG_USER=y Oct 08 20:43:02 user_debug=31 as commandline Oct 08 20:45:18 user_debug=-1 Oct 08 20:45:24 yeah Oct 08 20:45:52 woglinde: could you try eglibc by any chance Oct 08 20:46:10 khem hm yes tomorrow maybee Oct 08 20:47:07 woglinde: uclibc/eabi has not been tested on non interworking stuff Oct 08 20:47:12 so it could have issues Oct 08 20:47:20 woglinde: what binutils are you using Oct 08 20:47:27 micro one Oct 08 20:47:34 kernel is working fine Oct 08 20:47:42 ok 2.19.51 Oct 08 20:47:53 with oabi compat I could start oabi userland Oct 08 20:48:02 I think most probably it is uclibc crapping out Oct 08 20:48:20 btw. do you have thumb turned on ? Oct 08 20:48:26 no Oct 08 20:48:40 ok so all userspace is compiled arm Oct 08 20:49:04 in local.conf PREFERRED_ARM_INSTRUCTION_SET = "arm" Oct 08 20:49:45 I am not sure of uclibc eabi asm code was written keeping armv4 in mind Oct 08 20:49:58 hm Oct 08 20:50:06 I only looked at the BX stuff Oct 08 20:50:07 if you can get traces where it fails in userspace that could help Oct 08 20:50:30 yeah mainly thats the one but you dont seem to get illegal instruction so seems something else Oct 08 20:51:58 ah Oct 08 20:51:59 yes Oct 08 20:52:06 init (1): undefined instruction: pc=4000940c Oct 08 20:52:07 Code: e0812002 e5832000 e28bd000 e8bd0800 (e12fff1e) Oct 08 20:52:48 hm Oct 08 20:52:55 cool Oct 08 20:52:59 ???? Oct 08 20:53:02 can you disassemble uclibc.so Oct 08 20:53:15 will do Oct 08 20:53:26 and also ld.so Oct 08 20:53:45 opcodes look like arm instructions Oct 08 20:54:09 I hope its not one of the linker inserted veneers Oct 08 20:54:25 where should I look for? Oct 08 20:54:42 pc=? Oct 08 20:54:50 in uclibc.so search for address 940c Oct 08 20:54:53 yes Oct 08 20:55:09 args Oct 08 20:55:11 hm Oct 08 20:55:12 ldrbtvs r7, [r5], #-1134 ; 0x46e Oct 08 20:56:10 section .dynstr Oct 08 20:57:56 hmmm section does not seem correc Oct 08 20:57:57 khem whats ldrbtvs ? Oct 08 20:58:13 its a byte load with transpose Oct 08 20:58:23 its available instruction on all arms Oct 08 20:58:35 hm I can search the instrcution right? Oct 08 20:59:04 I think it could have branched wrongly before if this address is in .dynstr Oct 08 20:59:10 args Oct 08 20:59:14 the sucker included bx Oct 08 20:59:29 iow the fault could be elsewhere Oct 08 20:59:29 b324: e12fff1e bx lr Oct 08 20:59:31 damn it Oct 08 20:59:37 oh hmm Oct 08 20:59:41 why this Oct 08 20:59:55 look at your build uclibc config file Oct 08 20:59:59 back to the config Oct 08 21:00:16 this looks like a asm macro Oct 08 21:00:25 what is the function the bx is in Oct 08 21:00:51 TARGET_ARCH="arm" Oct 08 21:00:51 FORCE_OPTIONS_FOR_ARCH=y Oct 08 21:00:51 # CONFIG_ARM_OABI is not set Oct 08 21:00:51 CONFIG_ARM_EABI=y Oct 08 21:01:29 look for USE_BX Oct 08 21:01:36 I have Oct 08 21:01:41 args Oct 08 21:01:43 sorry Oct 08 21:01:55 FORCE_OPTIONS_FOR_ARCH=y Oct 08 21:01:55 # CONFIG_ARM_OABI is not set Oct 08 21:01:55 CONFIG_ARM_EABI=y Oct 08 21:01:55 # USE_BX is not set Oct 08 21:01:57 so Oct 08 21:02:05 forgot to paste latest Oct 08 21:02:07 line Oct 08 21:02:25 ok what is the function name where bx appears Oct 08 21:02:45 there are manys Oct 08 21:03:00 hmm Oct 08 21:03:01 starting with _dl_pagesize Oct 08 21:03:17 each function has it Oct 08 21:03:18 hm Oct 08 21:03:26 binutils error? Oct 08 21:04:14 dl_pagesize it a variable Oct 08 21:04:17 hm but why the kernel isnt effected? Oct 08 21:04:41 depends what options are passed on commandline Oct 08 21:05:40 okay Oct 08 21:05:45 other libs has it to Oct 08 21:05:49 so its not uclibc Oct 08 21:05:59 but give me a function around a bx instruction Oct 08 21:06:09 so I can trace back Oct 08 21:06:20 how function names start in objdump? Oct 08 21:07:08 ah breaking is in ld.so Oct 08 21:07:27 yeah thats what I was hinting on next Oct 08 21:07:27 Disassembly of section .text: Oct 08 21:07:28 00000a00 <_dl_linux_resolver>: Oct 08 21:08:25 I see the problem Oct 08 21:08:35 binutils? Oct 08 21:09:01 there is a uncondioned bx ip Oct 08 21:09:13 in _dl_linux_resolve Oct 08 21:09:19 hm oh Oct 08 21:09:28 but there are more functions Oct 08 21:09:33 and other libs has bx too Oct 08 21:10:28 so I guess more its the linker Oct 08 21:10:36 or assembler Oct 08 21:10:38 yeah Oct 08 21:10:54 shouldnt it take care? Oct 08 21:11:03 hm which one makes the exchange? Oct 08 21:11:05 linker? Oct 08 21:11:27 as I understood some tool in the chains relpaces the bx with mov Oct 08 21:11:46 --fix-v4bx Oct 08 21:12:01 gcc option? Oct 08 21:12:04 is that option passed by gcc to asm Oct 08 21:12:08 its a gas opt Oct 08 21:12:19 ah okay Oct 08 21:12:39 let me see if gas has it Oct 08 21:13:28 okay yes Oct 08 21:13:35 --fix-v4bx Allow BX in ARMv4 code Oct 08 21:13:44 so let me see why gcc is failing Oct 08 21:14:20 this option should be passed by gcc to both gas as well as ld Oct 08 21:14:32 |march=armv4:--fix-v4bx Oct 08 21:14:36 okay spec has it Oct 08 21:15:12 ccache arm-oe-linux-uclibceabi-gcc -march=armv4 Oct 08 21:15:12 try cross-gcc -v -x c /dev/null Oct 08 21:15:17 maybee a ccache problem? Oct 08 21:15:27 or rather cross-gcc -march=armv4 -v -x c /dev/null Oct 08 21:15:44 and see if this option is actually passed down to assembler and linker Oct 08 21:17:22 khem hm http://pastebin.com/m732a02aa Oct 08 21:17:29 oh I think linker does not get it Oct 08 21:18:07 khem whats? --fix-v4bx-interworking Oct 08 21:18:21 Rewrite BX rn branch to ARMv4 interworking veneer Oct 08 21:18:35 if you want to run armv4 code on interworking arches Oct 08 21:19:06 hm where in the output I see the linker options? Oct 08 21:19:12 woglinde: I think the problem is that we need to pass --fix-v4bx to linker Oct 08 21:19:15 too Oct 08 21:19:31 the options passed to collect2 Oct 08 21:19:59 --fix-v4bx-interworking is not needed if the code stays on armv4 machine Oct 08 21:20:11 yes Oct 08 21:20:13 okay Oct 08 21:20:23 I can do it in the specs right? Oct 08 21:20:34 by hand for now Oct 08 21:20:51 I think you can all -Wl,--fix-v4bx to linker flags Oct 08 21:21:01 to make sure if thats the problem Oct 08 21:21:09 I will Oct 08 21:22:03 so I think assembler did the job but its the linker who should rewrite the bx instructions Oct 08 21:22:14 and it wont do it unless we say --fix-v4bx Oct 08 21:22:43 hm is the gcc spec hardcoded inside gcc? Oct 08 21:22:55 yes Oct 08 21:23:01 args okay Oct 08 21:23:12 you can still supply one on commandline Oct 08 21:23:13 I will but it in tunes-strongarm fornow Oct 08 21:23:17 yes Oct 08 21:23:27 but thats also a pain Oct 08 21:23:34 yes Oct 08 21:23:53 we need to fix gcc Oct 08 21:23:59 but that should be easy Oct 08 21:24:06 if it is the root of the evil Oct 08 21:24:09 try it out if it works then we know what to do Oct 08 21:24:17 exactly Oct 08 21:24:38 although I am hopeful but it might not be all :) Oct 08 21:24:54 we will know in some minutes Oct 08 21:24:59 cool Oct 08 21:25:05 are you just compiling uclibc ? Oct 08 21:25:07 parsing recipes Oct 08 21:25:09 yes Oct 08 21:25:19 hmmm ok Oct 08 21:25:37 but this might need you to recompile whole userspace Oct 08 21:25:43 I know Oct 08 21:25:45 or atleast telink it Oct 08 21:25:47 but my machine is fast Oct 08 21:25:49 but lets see Oct 08 21:27:29 * Jay7 is trying first push :) Oct 08 21:27:59 jay7 dont break something Oct 08 21:28:29 I'm pushing only tosa additions for linux-2.6.30 Oct 08 21:29:04 woglinde: I could easily recompile my Gentoo strip but that's not the solution :) Anything to test yet for dietlibc? Oct 08 21:29:16 ant sorry dont know Oct 08 21:29:22 ask khem Oct 08 21:30:05 ant__: anyway to use cross strip ? Oct 08 21:30:22 bitbake.conf:BUILD_OS = "${@os.uname()[0].lower()}" Oct 08 21:30:28 see, half onf the binaries are i386, half are arm Oct 08 21:30:28 khem I think the bx are now 1310: fffef278 undefined instruction 0xfffef278 Oct 08 21:30:30 hmmm thats the root of problems Oct 08 21:30:38 is that okay? Oct 08 21:30:41 I think not Oct 08 21:31:22 you have to disassemble Oct 08 21:31:31 thats from Oct 08 21:31:36 diassamble Oct 08 21:31:44 but I see the mov pc, lr Oct 08 21:31:47 in the end too Oct 08 21:31:51 right Oct 08 21:32:15 khem but what will happen at undefined instruction 0xffff7698 ? Oct 08 21:32:26 I think it wont get executed Oct 08 21:32:37 that seems like a thumb bx to me Oct 08 21:32:49 hm Oct 08 21:33:00 but I have no thumb in Oct 08 21:33:02 enabled Oct 08 21:33:21 do you see any bx in the dump Oct 08 21:33:31 no bx are gone Oct 08 21:33:36 ok then its ok Oct 08 21:33:43 or post the disassembly somewhere Oct 08 21:33:48 I can see what it means Oct 08 21:34:36 it was something with bx in Oct 08 21:35:54 0xffff7698 looks like garbage. if the linker is producing those then I think it is broken. Oct 08 21:36:07 pb as I thought Oct 08 21:36:19 but I can diff the instruction with old out put Oct 08 21:36:33 you're using --fix-v4bx, right, not the interworking one? Oct 08 21:36:57 yes Oct 08 21:37:36 hi pb_ Oct 08 21:37:39 its not comming from the linker I guess Oct 08 21:37:47 old compile has it too Oct 08 21:38:41 in that case, it should always be replacing "bx rN" with "mov pc, rN". Oct 08 21:38:56 thats is doing fine Oct 08 21:38:58 now Oct 08 21:39:04 I set -Wl, Oct 08 21:39:13 hi khem Oct 08 21:39:13 okay Oct 08 21:39:59 so, where exactly are the garbage instructions showing up? Oct 08 21:40:01 hm I should look at other eabi output maybee I will get a clue Oct 08 21:40:04 khem: about dietlibc: http://tinderbox.openembedded.org/builds/32833/ Oct 08 21:40:08 pb overall Oct 08 21:40:15 it seems to me two strip are needed Oct 08 21:40:34 native for the first part, cross for the second Oct 08 21:40:44 problem is CROSS=" " Oct 08 21:40:45 ant thats what I said Oct 08 21:41:05 how did other *libc solve this? hmm Oct 08 21:41:49 I find bad when a recipe builds only by chance on some hosts... :) Oct 08 21:42:46 do we need a strip-native (multiarch)? Oct 08 21:43:00 or just Makefile slaughtering? Oct 08 21:43:45 the latter I think... Oct 08 21:45:28 ant__: you should pass CROSS=arm-blah-blah- on commandline Oct 08 21:45:35 if you are building cross Oct 08 21:45:44 yea, but not at the beginning... Oct 08 21:46:22 make -j 3 -e MAKEFLAGS= all ARCH=arm CROSS= prefix=/oe/build/tmp/staging/armv5te-angstrom-linux-gnueabi/lib/dietlibc Oct 08 21:46:27 this is no good Oct 08 21:46:32 CROSS is empty Oct 08 21:46:41 it should be armv5te-angstrom-linux- Oct 08 21:47:00 or CROSS=armv5te-angstrom-linux-gnueabi- Oct 08 21:47:03 actually Oct 08 21:47:04 so I will build toolchain for sheeva Oct 08 21:47:04 see, we build bin-i386 and bin-arm on my host Oct 08 21:47:09 and differ the output Oct 08 21:47:46 khem: if I change this in the Makefile after the failure (after the i386 bins) do_compile can finish... Oct 08 21:48:57 ant__: so how does dietlibc build Oct 08 21:49:12 it is that some parts are native x86 and some parts are cross Oct 08 21:49:26 it's HOST_PREFIX iirc Oct 08 21:49:47 the $(CROSS) what I need for my arm-target Oct 08 21:50:33 iirc arm-angstrom-linux-gnueabi- Oct 08 21:50:38 not armv5te Oct 08 21:50:54 yes whatever your tools are prefixed with Oct 08 21:51:13 well, there is a discrepancy it seems Oct 08 21:52:57 you can use CROSS=$HOST_SYS Oct 08 21:53:22 you mean in the recipe? Oct 08 21:53:27 yes Oct 08 21:54:01 in do_compile Oct 08 21:54:36 ok, so it will pick up the right prefix after host switch Oct 08 21:54:43 let's try Oct 08 21:54:59 CROSS=${HOST_SYS}- Oct 08 21:55:07 is the right one I guess you need that dash Oct 08 21:55:44 not for HOST_PREFIX iirc Oct 08 21:56:32 right you should use that infact Oct 08 21:57:03 CROSS=${HOST_PREFIX} Oct 08 21:57:37 it was failing with that.. Oct 08 21:57:53 so it is now Oct 08 21:58:10 make: arm-angstrom-linux-gnueabi-arm-angstrom-linux-gnueabi-gcc: Command not found Oct 08 21:58:44 openssl-native fails in my dev.oe.org beagleboard build as well, hmm Oct 08 21:59:33 cbrake_away: what does uname -a give you on the build machine Oct 08 22:00:20 cbrake_away: I think I know what the issue is Oct 08 22:00:48 ant reset CC Oct 08 22:01:06 btw Oct 08 22:01:19 www.openembedded.org != openembedded.org Oct 08 22:01:31 second showing ltg site Oct 08 22:06:25 khem: seems successful Oct 08 22:07:17 khem why log.do_compile is empty for eglibc-initial? Oct 08 22:07:31 ah stage again Oct 08 22:07:37 yeah Oct 08 22:07:38 * woglinde hms Oct 08 22:07:50 stage should staging Oct 08 22:07:53 not compile Oct 08 22:08:11 only for crappy buildsystems which have no relocation Oct 08 22:08:40 khem: woglinde: Oct 08 22:08:43 do_compile () { Oct 08 22:08:43 <------>oe_runmake all CC="${BUILD_CC}" CFLAGS="${BUILD_CFLAGS}" prefix=${STAGING_DIR_TARGET}/lib/dietlibc Oct 08 22:08:43 <------>oe_runmake all ARCH="${TARGET_ARCH}" CC="${BUILD_CC}" CROSS="${HOST_PREFIX}" prefix=${STAGING_DIR_TARGET}/lib/dietlibc Oct 08 22:08:43 } Oct 08 22:08:54 compiles fine Oct 08 22:09:20 ant okay Oct 08 22:09:32 fine you figure and worked it out Oct 08 22:09:36 commit it Oct 08 22:09:42 many thx Oct 08 22:09:59 but before..let see the size of kexec :P Oct 08 22:11:20 *kexec │ 211620│Oct 9 00:10 Oct 08 22:11:22 urgh Oct 08 22:11:31 hm strange Oct 08 22:11:41 stripped or unstripped? Oct 08 22:11:46 ibc-2.0.1-r0/install/kexec-dietlibc/usr/sbin $ file kexec Oct 08 22:11:46 kexec: ELF 32-bit LSB executable, ARM, version 1 (SYSV), statically linked, stripped Oct 08 22:14:49 woglinde: andrea@mizar /oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/kexec-tools-dietl Oct 08 22:14:49 ibc-2.0.1-r0/install/kexec-dietlibc/usr/sbin $ 2/staging-pkg/cross/armv5te/bin $ file diet Oct 08 22:14:49 diet: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), statically linked, stripped Oct 08 22:14:51 ?? Oct 08 22:16:13 ant you are wrong again Oct 08 22:16:13 haha Oct 08 22:16:23 arg no Oct 08 22:16:25 thats okay Oct 08 22:16:33 diet sets some gcc options Oct 08 22:16:43 and is for cross compiling Oct 08 22:16:48 the diet bin runs on i386 for target arm Oct 08 22:16:52 so it needs to bee i386 Oct 08 22:16:52 ok Oct 08 22:17:01 like the cross compiler sel Oct 08 22:17:01 f Oct 08 22:17:06 ok Oct 08 22:17:49 for a moment I thougth I had again the hybridation... Oct 08 22:18:17 woglinde: how can I check why it's soo big? Oct 08 22:18:28 ant dissasemble? Oct 08 22:18:41 btw. we have paypal now :) Oct 08 22:18:44 klibc stages own headers... Oct 08 22:18:57 florian nice Oct 08 22:18:57 perhaps the include-chain is much longer Oct 08 22:19:52 for dietlibc-static kexec-tools Oct 08 22:19:57 hm Oct 08 22:20:06 what you mean with include? Oct 08 22:20:17 include " " Oct 08 22:20:21 no Oct 08 22:20:26 locally staged by patch Oct 08 22:20:33 thats only for getting the right types and defs Oct 08 22:21:00 include stuff is done by preprocessor Oct 08 22:21:20 result is the file which will be compiled Oct 08 22:21:23 well, all headers have to be copied iirc Oct 08 22:21:27 once Oct 08 22:21:32 ???? Oct 08 22:22:01 in the code the cpp subsitutes the #include with code, isn't? Oct 08 22:22:10 * Jay7 can be commercial partner of OE in Russia :) Oct 08 22:22:30 no its searchin the headers and is replacing macros and types Oct 08 22:23:20 ok, I'm confusing languages at this hour... Oct 08 22:25:40 khem ah intressting Oct 08 22:25:45 armv5 has the same too Oct 08 22:26:55 so the undefined stuff should be okay Oct 08 22:32:12 woglinde: kexec-dietlibc still fails with 'unknown option --command-line' Oct 08 22:32:30 and several Alignment traps Oct 08 22:32:33 ant with the getline patch? Oct 08 22:32:35 hm okay Oct 08 22:32:36 yes Oct 08 22:32:47 then eabi stuff isnt working right Oct 08 22:32:51 I guess Oct 08 22:33:42 I have logs Oct 08 22:33:44 1 mom Oct 08 22:34:04 sorry Oct 08 22:34:13 khem has more knowledge Oct 08 22:34:33 http://fr.pastebin.ca/1605391 Oct 08 22:35:33 Jay7: btw ^^^ Oct 08 22:37:01 strange is it works in glibc and even after my manipulations on klibc... Oct 08 22:37:07 :D Oct 08 22:37:22 ant__: for me more interesting is Unknown keyword: 'NOINITRD DEBUG ROOT' Oct 08 22:37:44 this should be separated by whitespace Oct 08 22:38:05 hm..see cmdline Oct 08 22:38:11 I see Oct 08 22:38:27 I should check this case Oct 08 22:38:35 the spaces before debug are old-known crap Oct 08 22:38:58 always forgetting to fix it Oct 08 22:39:12 it's not critical Oct 08 22:39:30 ant__: what should I expect Oct 08 22:39:43 I do not know about kexec Oct 08 22:40:13 now you can compile but it does not run right ? Oct 08 22:40:29 yep Oct 08 22:40:39 pb_: I have a patch which reverts the -gnu stuff Oct 08 22:40:50 fals if linked against dietlibc Oct 08 22:40:59 only in this case Oct 08 22:42:40 ant__: hmmm what does kexec do exactly Oct 08 22:42:57 so its a static arm binary in this case right Oct 08 22:43:24 kexec uses syscalls for restarting a new kernel Oct 08 22:43:55 mostly used in initramfs Oct 08 22:45:03 kexec_load failed: Bad file number Oct 08 22:45:11 what does that error convey Oct 08 22:45:44 I suppose it ignores all args Oct 08 22:45:49 invalid option `command-line'. Oct 08 22:46:27 hmm Oct 08 22:46:27 or perhaps is my dietlibc broken... Oct 08 22:46:47 is it a regression ? Oct 08 22:47:02 fact is, the vanilla recipe works when linked against glibc Oct 08 22:47:21 the klibc-static one works at least for arm Oct 08 22:47:52 so the code is supposedly sane enough Oct 08 22:47:53 ok so kexec itself seems to be ok or atleast working with glibc Oct 08 22:47:58 :) Oct 08 22:48:16 now you tried it with dietlibc and it borks Oct 08 22:48:26 did you also try it with uclibc ? Oct 08 22:48:34 but more than this, I get double-sized binary Oct 08 22:48:50 thats interesting Oct 08 22:48:59 khem: Linux happy 2.6.28-15-generic #49-Ubuntu SMP Tue Aug 18 19:25:34 UTC 2009 x86_64 GNU/Linux Oct 08 22:49:01 dietlibc must be drinking diet coke Oct 08 22:49:22 woglinde gets reasonable sizes, though Oct 08 22:49:48 cbrake_away: I am reverting the -gnu stuff it should start working once I commit that patch I will send that to list soon Oct 08 22:49:56 cbrake_away: do you need a workaround ? Oct 08 22:50:08 khem: no, these are just my routine builds Oct 08 22:50:54 cbrake_away: if you are stuck and need a workaround then I can find one for you Oct 08 22:51:00 khem: stuff like this is never easy :-\ Oct 08 22:51:07 khem: no, nothing critical Oct 08 22:51:11 khem: thanks Oct 08 22:51:30 khem: need to go help with dishes now :-) Oct 08 22:52:33 oh I see the problem Oct 08 22:52:50 its one liner Oct 08 22:53:27 khem: so you build on a 32-bit machine? Oct 08 22:53:56 no I have 64bit box and I did not see this error Oct 08 22:53:59 thats strange Oct 08 22:55:27 woglinde: should I bump the PR? Oct 08 22:57:23 ant yes Oct 08 23:00:27 woglinde: did it go further with bx stuff Oct 08 23:00:46 03Yuri Bushmelev  07org.openembedded.dev * r50f5f9ae80 10openembedded.git/recipes/linux/ (linux-2.6.30/tosa/defconfig linux_2.6.30.bb): Added SHARP Tosa support to linux_2.6.30 recipe Oct 08 23:01:10 well Oct 08 23:04:32 jay7 looks good Oct 08 23:04:33 *g* Oct 08 23:04:41 khem I am recompiling now Oct 08 23:04:58 woglinde: thanks ;) Oct 08 23:05:09 but as I said with armv5 it has the same unkown intstrcutions Oct 08 23:08:43 woglinde: hmm interesting Oct 08 23:08:50 I wonder why those are in there Oct 08 23:09:04 post the disasm somewhere Oct 08 23:09:07 or part of it Oct 08 23:09:22 03Andrea Adami  07org.openembedded.dev * rb696acdd72 10openembedded.git/recipes/dietlibc/dietlibc.inc: dietlibc.inc: use the right prefix for cross tools. Bump PR. Oct 08 23:09:22 khem armv5 was with eglibc build Oct 08 23:09:23 ok, thx again khem Oct 08 23:09:47 and woglinde Oct 08 23:09:47 good nite Oct 08 23:09:47 nite ant Oct 08 23:10:10 nite Oct 08 23:10:13 ant Oct 08 23:24:03 * Jay7 -> sleep Oct 08 23:25:11 nite jay7 Oct 08 23:28:20 hi mwester Oct 08 23:30:36 Greetings Oct 08 23:45:24 woglinde: how do we use dietlibc in oe Oct 08 23:45:24 inherit dietlibc Oct 08 23:45:24 but its not perfect jet Oct 08 23:45:25 ah Oct 08 23:45:29 eabi userland works now Oct 08 23:45:30 oh so we dont have distro choice Oct 08 23:45:34 right Oct 08 23:45:46 I only support static builds now Oct 08 23:45:49 as I read thru it it seems we should give it a shot Oct 08 23:45:55 and only few software will compile Oct 08 23:46:32 woglinde: right but it will be cooler to have it as yet another system library :) Oct 08 23:46:46 oh so now armv4 +eabi is working well ? Oct 08 23:46:51 booted ? Oct 08 23:47:07 yes Oct 08 23:47:08 thats a good news Oct 08 23:47:09 sir Oct 08 23:47:12 jepp Oct 08 23:47:35 khem sure but as I said only few software will compile Oct 08 23:47:37 so send a note to ml and encourage armv4 devs to start using it Oct 08 23:47:41 with dietlibc Oct 08 23:47:57 so now I like to use ubifs Oct 08 23:48:18 ofcourse gcc 4.4.x + binutils 2.19.51 is way to go Oct 08 23:48:23 but I need to figure out parameters Oct 08 23:48:29 woglinde: oh we need to fix gcc driver Oct 08 23:48:30 yes Oct 08 23:48:37 -Wl Oct 08 23:48:39 I will take care of that Oct 08 23:48:43 aeh Oct 08 23:48:50 collect2 needs it in spec Oct 08 23:48:52 fine Oct 08 23:48:56 yes Oct 08 23:49:08 I could reproduce that part here Oct 08 23:50:37 slowly we should obsolete oabi Oct 08 23:51:44 hm Oct 08 23:52:21 woglinde: calling oe_runmake twice in dietlibc.inc doesnt seem right to me Oct 08 23:52:44 why do we need first one ? Oct 08 23:52:50 oe_runmake all CC="${BUILD_CC}" CFLAGS="${BUILD_CFLAGS}" prefix=${STAGING_DIR_TARGET}/lib/dietlibc Oct 09 02:54:43 i am having a strange issue Oct 09 02:54:55 i am running on an avr32 with uclibc Oct 09 02:55:11 i build the console-base-image and get no proc filesystem Oct 09 02:55:24 I checked the kernel .config file and it is configured to build the kernel with proc support Oct 09 02:56:23 in line with that, the system complains that it cannot find 'libintl.so' Oct 09 02:56:33 mount: can't load library 'libintl.so' grep: /proc/filesystems: No such file or directory Oct 09 02:56:51 i checked the staging direcory and this library definitely gets built Oct 09 02:57:46 when I simply copy it from staging to the /lib directory, i get further - it seems to find the library but hangs on boot Oct 09 02:57:58 any thoughts? Oct 09 02:59:27 martinmeba: why isn't libintl.so in the rootfs? is the rootfs too big and some parts were dropped? Oct 09 02:59:51 that is what I don't understand Oct 09 02:59:55 i don't think so Oct 09 03:00:00 I am nfs mounting it **** ENDING LOGGING AT Fri Oct 09 03:00:00 2009