**** BEGIN LOGGING AT Wed Nov 05 02:59:57 2008 Nov 05 04:12:07 * * OE Bug 4802 has been created by denis(AT)denix.org Nov 05 04:12:09 * * gnome-panel-2.24.0 cannot find scrollkeeper Nov 05 04:12:15 * * http://bugs.openembedded.net/show_bug.cgi?id=4802 Nov 05 04:13:55 cdbot2: good :) Nov 05 07:39:05 hello everybody! Nov 05 07:39:19 I'm new to openembedded, and I already have a question Nov 05 07:39:32 I did the steps in GettingStarted Nov 05 07:39:53 and I ran into this problem when trying to bitbake opie-image Nov 05 07:39:59 Error, Your PACKAGE_ARCHS field contains duplicates. Perhaps you set EXTRA_PACKAGE_ARCHS twice accidently through some tune file? Nov 05 07:41:18 I managed to get rid of the problem (swept it under the rug) by commenting the offending lines (e.g. the error message) from sanity.bbclass Nov 05 07:41:37 it compiles a lot of the packages, but it dies later when trying to compile perl Nov 05 07:42:08 do you think that by ignoring this problem it would create problems while compiling perl? Nov 05 08:08:00 morning Nov 05 09:19:18 hi Nov 05 09:21:02 morning Nov 05 09:21:41 mad_ady: you rather need to fix machine config then hack sanity.bbclass Nov 05 09:27:34 ok, but how do I troubleshoot this problem? I mean, how can I find if PACKAGE_ARCHS is defined elsewhere? Nov 05 09:27:49 I did a recursive grep, but I didn't find out much Nov 05 09:28:09 is there a way to list it's contents, to see what gets doubled up? Nov 05 09:30:08 mad_ady: PACKAGE_ARCHS was recently moved to include/tune-*, but not all machine configs were updated to reflect that change Nov 05 09:30:25 ok Nov 05 09:30:40 I looked into include/tune-*, but I couldn't find my machine Nov 05 09:30:43 let me check again Nov 05 09:31:02 mad_ady: find your machine conf, and then read it to find what tune file it includes Nov 05 09:31:29 I'm using looxc550.conf Nov 05 09:31:41 aha Nov 05 09:31:46 it uses tune-xscale.inc Nov 05 09:32:24 now compare PACKAGE_ARCHS in looxc550.conf and tune-xscale.inc, and remove duplicate entries from looxc550.conf Nov 05 09:32:36 yup, will do, thanks! Nov 05 09:32:42 good luck Nov 05 09:34:32 ok, seems that armv4 and armv4t are duplicates Nov 05 09:34:37 thank you for your help! Nov 05 09:44:45 mad_ady: basically machine configs do not have to play at all with PACKAGE_ARCH variables Nov 05 09:51:34 I removed all the entries from PACKAGE_ARCH from looxc550.conf file Nov 05 09:52:45 and now it seems to have solved the problem Nov 05 09:52:52 cool Nov 05 09:53:02 let's see if it complies :) Nov 05 09:56:12 ~summon zecke and/or RP Nov 05 09:56:12 apt takes out 20 clean, identical-looking phones, some extra hands, and pretends to be a telemarketer for a large corporation, so he gets delivered a phonelist containing zecke and/or RP's coordinates Nov 05 09:57:36 morning Nov 05 09:57:40 hrw whtas wrong? Nov 05 09:58:42 I have svn fetcher patches for review Nov 05 09:58:48 http://pastebin.ca/1245752 Nov 05 09:59:53 hm whats <<< doing? Nov 05 10:01:43 hi woglinde - any hint how i can get further with my glib build problem? Nov 05 10:02:04 woglinde: http://www.arkanis-development.de/weblog/2007/9/7/subversion:-work-around-issuer-is-not-trusted Nov 05 10:02:54 emdete quick fix is to disable pb's patch Nov 05 10:03:21 ah, 'disable'... i understood yesterday it's missing. sry :/ Nov 05 10:04:35 hi stelios Nov 05 10:05:17 hrw ah I understand now Nov 05 10:05:49 emdete hm no, somehow at your config the gatomic.la is not build Nov 05 10:06:15 but thump support is optional and the quicl solution is disbale for the thumb patch on your side Nov 05 10:06:17 woglinde: the bad thing is that this is bashism.. Nov 05 10:06:27 hrw right Nov 05 10:06:43 but a nice feature other common shell should include Nov 05 10:07:05 and I have one recipe which fetch via https with ssl selfsigned.. Nov 05 10:25:22 good morning Nov 05 10:25:27 hi florian Nov 05 10:33:29 florian: good morning Nov 05 10:33:43 hi pb Nov 05 10:33:46 gi robert Nov 05 10:33:56 hi woglinde Nov 05 11:49:12 "swp r0,r1,[r4]" - why gcc say that arm1136 does not supports it? Nov 05 11:50:02 ok.. thumb<>arm Nov 05 11:52:08 "-march=armv6j -mtune=arm1136jf-s -mfloat-abi=softfp -mfpu=vfp -mthumb-interwork -mthumb" is why I had that Nov 05 12:15:43 03Otavio Salvador  07org.openembedded.dev * r301c02d1fe 10openembedded.git/packages/vnc/ (tightvnc/Vncviewer tightvnc_1.3.9.bb): tightvnc: add a default X resources file Nov 05 12:15:50 03Otavio Salvador  07org.openembedded.dev * re90881203c 10openembedded.git/packages/vnc/tightvnc_1.3.9.bb: tightvnc: provide vncviewer alternative to provide more flexibility Nov 05 12:15:52 03Otavio Salvador  07org.openembedded.dev * re196c018e0 10openembedded.git/packages/xfce/ (2 files in 2 dirs): xfdesktop: fix segfault when removable icons are disabled Nov 05 12:18:10 03Otavio Salvador  07org.openembedded.dev * rbe356ca208 10openembedded.git/packages/ica/ica-bin_10.6.bb: ica-bin: add 10.6 recipe Nov 05 12:45:08 bye Nov 05 12:45:54 03Marcin Juszkiewicz  07org.openembedded.dev * rc51ce4a258 10openembedded.git/conf/checksums.ini: checksums.ini: added some entries Nov 05 12:55:57 zecke: hail zecke Nov 05 12:59:58 pb__: hi! are you the pb__ that did the patch that makes glib fail to compile? :) Nov 05 13:03:14 emdete: heh, apparently. what is the failure that you are seeing? Nov 05 13:03:47 it fails to find some atomic lib - give me a seconf Nov 05 13:03:56 what's your MACHINE? Nov 05 13:04:07 gta01/2 Nov 05 13:04:55 arm-angstrom-linux-gnueabi-libtool: link: cannot find the library `libgatomic.la' Nov 05 13:05:10 that is rather odd Nov 05 13:05:21 do you see any trace of libgatomic.* in your build directory? Nov 05 13:05:31 woglinde suggested to remove your patch but i'm not shure which Nov 05 13:05:47 he means the atomic-thumb.patch Nov 05 13:06:03 removing that from SRC_URI probably would make your problem go away, but I would prefer to understand why it fails. Nov 05 13:06:09 without that patch you won't be able to compile for thumb Nov 05 13:06:22 (which probably isn't all that big a deal for you but it would be a shame to break it again) Nov 05 13:06:36 pb__: (still searching) - what ist this thumb & atomic stuff? Nov 05 13:06:58 thumb is a higher-density version of the arm instruction set. Nov 05 13:07:13 compared to arm code it gives you smaller (and, in some cases, faster) binaries. Nov 05 13:07:27 however, there are some operations that you can't do in thumb mode and atomic swaps are one of them. Nov 05 13:07:55 so, when glib is built in thumb mode, it is necessary to build the atomic operation support separately and compile just that one file in arm mode. Nov 05 13:08:00 there is no lib with that name but build/tmp/work/armv4t-angstrom-linux-gnueabi/glib-2.0-2.16.4-r0/glib-2.16.4/glib/.deps/libgatomic_la-gatomic.Plo Nov 05 13:08:10 strange Nov 05 13:08:31 if you go to the build directory (build/tmp/work/armv4t-angstrom-linux-gnueabi/glib-2.0-2.16.4-r0/glib-2.16.4/glib/) and type "make libgatomic.la", what happens? Nov 05 13:08:56 zecke: are you planning to go to c-base this evening? Nov 05 13:09:00 arm-angstrom-linux-gnueabi-gcc: No such file or directory Nov 05 13:09:06 doh Nov 05 13:09:08 pb__: shall i use devshell? Nov 05 13:09:11 yes, please Nov 05 13:10:09 so thumb-mode compiles stuff is faster? Nov 05 13:10:26 not necessarily always, but it seems to be mostly so in my testing Nov 05 13:10:52 I posted some stats to the mailing list. let me find a link. Nov 05 13:11:04 pb__: so everything goes thumbmode - even the kernel? Nov 05 13:11:15 no, not the kernel Nov 05 13:11:31 it would be possible to compile the kernel as mostly thumb, but that would be quite a lot of work Nov 05 13:11:56 likewise glibc, it would be somewhat difficult (though not impossible by any means) to build that as thumb code. Nov 05 13:12:03 but everything else should be buildable in thumb Nov 05 13:12:12 see http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-October/006718.html for some stuff about tis Nov 05 13:12:14 this, even Nov 05 13:14:30 interesting... Nov 05 13:16:23 back to devshell: the package seems to be not configured... Nov 05 13:16:36 what do you mean? Nov 05 13:16:55 there are no makefile only .am... Nov 05 13:16:57 mom Nov 05 13:17:16 er, that is odd Nov 05 13:17:23 are you sure you are looking at the same build directory as before? Nov 05 13:17:35 if you had the .deps/ stuff then it must have been configured at that point Nov 05 13:18:11 oups...no - ah, sry, myfault Nov 05 13:18:40 morning Nov 05 13:19:08 pb__: okay, fixed the dir prob & now it just compiles. Nov 05 13:19:28 pb__: (given your cd glib; make li....) Nov 05 13:19:33 okay Nov 05 13:19:42 it sounds like you must be encountering an automake bug of some kind. Nov 05 13:19:51 pb__: i know left the devshell & build the whole thing Nov 05 13:19:53 I'm not quite sure why it works for me and not for you. what version of automake-native are you using? Nov 05 13:20:36 you could try adding an explicit dependency rule "libglib-2.0.la: libgatomic.la" to the Makefile.am, but I don't think that should be necessary. Nov 05 13:20:37 pb__: yes, build runs through now Nov 05 13:20:50 ok, that's good Nov 05 13:21:07 so, it seems that automake just doesn't realise that it needs to compile libgatomic.la for some reason Nov 05 13:22:20 automake-native is 1.9.6-r0 Nov 05 13:22:42 right, that's the same one I have Nov 05 13:22:44 very weird Nov 05 13:23:02 ...and i did a clean recently Nov 05 13:23:10 clean == rm -rf tmp... Nov 05 13:23:24 right Nov 05 13:23:28 I wonder if it is failing for anybody else Nov 05 13:25:06 so how mu Nov 05 13:25:18 oups - so how much is compiled in that mode now? Nov 05 13:26:03 I've built the whole of fso-image as thumb, though I haven't actually tested all the resulting binaries. Nov 05 13:26:14 so far the only one I've actually booted is an fso-console-image. Nov 05 13:26:49 I was side-tracked by a crash in opkg which turned out to be a gcc-4.1 compiler bug. it seems that you need gcc-4.2 for correct thumb code generation. Nov 05 13:27:11 is 4.1 used by default? Nov 05 13:27:28 4.1 does seem to be the default for openmoko on 4.1 Nov 05 13:27:31 er, on gta01 Nov 05 13:27:52 I guess mickeyl would know to what extent that is a deeply entrenched policy. Nov 05 13:27:55 gta02 too? Nov 05 13:28:08 not sure. I don't have a gta02 so I haven't bothered trying to build for that platform. Nov 05 13:28:16 I suspect it's probably the same though. Nov 05 13:29:06 so if i build now i get a corrupt/crashing version? Nov 05 13:29:22 if you turn on thumb, yes, some things might get miscompiled. Nov 05 13:29:35 the only place I have actually observed a problem was in libcurl, which caused "opkg update" to segfault. Nov 05 13:29:45 ah, but thumb is off by dfault.. right? Nov 05 13:29:48 right Nov 05 13:29:52 gcc 4.1 seems to be fine for arm code Nov 05 13:38:48 hi! Nov 05 13:42:31 http://www.theonion.com/content/news_briefs/black_man_given_nations Nov 05 13:42:36 only in America .... Nov 05 13:45:43 I preferred the "they like having a black person clean up for them" variant on that one. Nov 05 13:46:38 heh Nov 05 13:47:14 Crofton: probably you mean usa, he wont want to cleanup whole america, right? Nov 05 13:49:49 how many bloody xservers do sun need Nov 05 13:51:55 emdete, good point :) Nov 05 13:52:26 dont worry, we shall power up the LHC and point it at the americans, they will then be hoovered up Nov 05 13:52:27 but it is a fake news site Nov 05 13:52:41 I had wondered what that was really for Nov 05 13:53:19 black holes make good waste disposal :-D Nov 05 13:58:43 moo Nov 05 13:59:33 who could update checksum.ini with an entry for http://downloads.sourceforge.net/pyopenssl/pyOpenSSL-0.6.tar.gz? Nov 05 14:00:07 ...and http://trevp.net/tlslite/tlslite-0.3.8.tar.gz Nov 05 14:01:03 emdete: you could :-) Nov 05 14:01:22 no, i have no commit/push rights to oe Nov 05 14:01:38 yeah, but you can do git-format-patch and post it :-) Nov 05 14:01:39 XorA: shure i can give you my patch Nov 05 14:02:00 I cant apply patches right this moment, working on a different branch Nov 05 14:02:18 XorA: harharhar... `git checkout ...`! Nov 05 14:02:44 Id lose all my work :-( Nov 05 14:02:54 git stash :) Nov 05 14:03:15 only 43 recipes still to go Nov 05 14:07:29 http://pyneo.org/downloads/0001-add-various-checksums.patch Nov 05 14:08:17 WTF is a vermilion? Nov 05 14:09:43 hi kergoth Nov 05 14:09:45 a specific red color I think Nov 05 14:15:06 * * OE Bug 4803 has been created by  Nov 05 14:15:08 * * libxkbfile-1.0.4-autobuild Nov 05 14:15:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4803 Nov 05 14:28:40 re Nov 05 14:28:55 hrw: wb Nov 05 14:29:40 re robert und hrw Nov 05 14:30:02 gcc 4.2 != 4.3 fight is argh Nov 05 14:30:11 hm? Nov 05 14:30:40 I use debian 'sid' and have gcc 4.3. buildbot machine has ubuntu 8.04 and gcc 4.2 Nov 05 14:30:52 some stuff not builds on my system but builds on bbot one Nov 05 14:32:22 update to intrepid Nov 05 14:32:38 easiest solution Nov 05 14:33:21 no way Nov 05 14:33:32 I tried Poky trunk on 8.10 and it fails Nov 05 14:33:51 hrw why it fails? Nov 05 14:34:10 because of the missing patches for O_CREATE issue? Nov 05 14:34:45 thats too Nov 05 14:36:45 so poky is broken *duck and hide* Nov 05 14:37:07 and I do not work on Poky direct so.. Nov 05 14:52:53 is OE known working with Ubuntu 8.10? Nov 05 14:53:02 cbrake hehe Nov 05 14:53:04 yes Nov 05 14:53:19 kergoth: morning Nov 05 14:53:21 gerwinin: long time no see. you see? Nov 05 14:53:41 gerwinin: you at ELCE tomorrow? Nov 05 14:53:55 hi kergoth Nov 05 14:54:12 hi all Nov 05 14:54:25 hi likewise Nov 05 14:55:16 cbrake there can be native packages which can brake with the gcc-4.3 compiler from ubuntu Nov 05 14:55:24 but I or others will fix this asap Nov 05 14:55:33 morning Nov 05 14:55:47 hmmm Nov 05 14:55:49 woglinde: ok, thanks Nov 05 15:02:25 re Nov 05 15:05:05 * * OE Bug 4804 has been created by sledz(AT)dresearch.de Nov 05 15:05:07 * * Some more checksums missing Nov 05 15:05:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4804 Nov 05 15:21:58 morning all. Nov 05 15:22:44 which bitbake should I be using with the current head? Nov 05 15:23:14 i'm using 1.8.10, but it is not seeing one of my PREFERRED_PROVIDER s Nov 05 15:23:30 hm, the newly generated packages complain while installing: " * ERROR: libdbus-1-3.postinst returned 1" - is that a known problem? Nov 05 15:27:06 HopsNBarley: last released one (1.8.10 iirc) Nov 05 15:27:15 HopsNBarley: you can also use 1.8-svn Nov 05 15:28:24 re hm now radeonhd works Nov 05 15:28:28 crappy flgrx Nov 05 15:29:14 windows xp + printer shared by cups == total disaster Nov 05 15:29:44 hrw hm? Nov 05 15:29:51 I often got it worked Nov 05 15:30:01 hrw: cups == total disaster... Nov 05 15:30:14 even 3/4 years ago where macos had a crappy cups version Nov 05 15:30:45 woglinde: my epson shoots me with 5 empty pages per each printed one Nov 05 15:30:47 emdete please show me your bettere printing solution for unix Nov 05 15:30:57 hrw ink? Nov 05 15:30:59 hrw. thanks, and hmmm. i'm just trying this: PREFERRED_PROVIDER_java2-runtime = "jamvm". bb "sees" both options, but picks cacao anyway, telling me I should define a provider... Nov 05 15:31:00 anyone an idea about my ".postinst returned 1" ? Nov 05 15:31:00 woglinde: ink Nov 05 15:31:04 woglinde: lprng Nov 05 15:31:04 hrw muahahahaaa Nov 05 15:31:17 never buy a ink Nov 05 15:31:31 even the cheapest ink is nt it worth Nov 05 15:31:32 woglinde: sure - show me color laser under 200 eur Nov 05 15:31:47 or rather color laser+scanner+copier under 200 eur Nov 05 15:32:06 for new one Nov 05 15:32:11 let me se what the hp colojert 1605 now is Nov 05 15:32:44 249 Nov 05 15:32:54 and we have this at home Nov 05 15:32:59 and we are happy with it Nov 05 15:33:48 if I would have to buy next printer then it would have to have printserver inside Nov 05 15:34:20 hehe Nov 05 15:34:23 anyway with older cups this one was working nicely Nov 05 15:34:29 hm Nov 05 15:34:31 okay Nov 05 15:34:44 linux + printer + sharing === disaster Nov 05 15:34:53 hm, `for n in /usr/lib/opkg/info/*.postinst; do echo exit 0 >> $n; done` did it - is the checking of retuncodes a new feature? Nov 05 15:34:54 nono Nov 05 15:35:17 woglinde: yesyes Nov 05 15:35:18 at home the brother hp1440 and the hp 1605 working fine Nov 05 15:35:21 to win2000 Nov 05 15:35:34 and to win 2003 server in vmware Nov 05 15:35:37 other network Nov 05 15:35:40 with cups Nov 05 15:37:05 Do you think the AP7000 is capable of running asterisk ? Nov 05 15:38:12 gerwinin hm depends on the ram you habve Nov 05 15:38:24 and how many phones you will use Nov 05 15:38:35 but for the price you are better with a integreated box Nov 05 15:38:56 steliosk gave me a cool link Nov 05 15:39:02 unforantly this is at home Nov 05 15:40:05 which package contains/builds xserver-security-policy? Nov 05 15:40:09 woglinde: I have some good oppurtunity to make some cheap oe router/boards :) Nov 05 15:41:37 gerwinin hm okay you could ask steliosk too if he is around, he has some expierence with runnig asteriks on embedded stuff Nov 05 15:49:06 on the angstrom filesystem , "starting GPE display manager" fails. can anyone give a starting point to debug? Nov 05 15:56:27 does kexecboot live only in the OE tree, or does it have its own little project space? Nov 05 15:57:04 Sundar: Try to run Xserver from a console and check its output Nov 05 15:57:18 cosmic hm you wait for ant or thesing Nov 05 15:57:36 florian I told this him yesterday Nov 05 15:57:44 seems he didnt mange it yet Nov 05 15:58:23 woglinde: oh... Nov 05 16:03:52 woglinde:i probably missed your note.I will run and get back Nov 05 16:05:03 Sundar login onto ssh Nov 05 16:05:13 yes ,i hav Nov 05 16:05:17 and run /etc/init.d/ Nov 05 16:05:24 okay Nov 05 16:05:27 where the scripts for starting resides Nov 05 16:05:35 do not know which are used for x and gpe Nov 05 16:06:34 resetting my target.give me a few. Nov 05 16:12:50 florian:woglinde i dont have permission to /etc/init.d/ even as root from ssh Nov 05 16:14:11 hello everyone Nov 05 16:15:12 woglinde : did you mean /etc/init.d/gpe-dm start ? Nov 05 16:16:48 sundar yes Nov 05 16:17:05 How can I "install" some file (gdal-config) of some builded package (gdal) when it is needed to build another package (grass), please? (Talking about bitbake.) Nov 05 16:17:39 i get a status 'Starting GPE display manager: gpe-dm " . but the same old blue screen. dmesg doesnt give me any onfo Nov 05 16:18:27 And how to I refer to it in bb recipe... Nov 05 16:18:28 Blackhex RDEPENDS = moo Nov 05 16:18:37 in .bb file Nov 05 16:18:45 Sundar: you are using NFS? Nov 05 16:18:47 sharedlibrary are found automagicly Nov 05 16:18:56 sd card Nov 05 16:19:00 http://trac.project-builder.org/ is cute Nov 05 16:19:06 but i can ,if it would be useful Nov 05 16:19:14 woglinde: you mean RDEPENDS = gdal in grass.bb? I'll try, thx Nov 05 16:21:13 woglinde: and the reference? ${D}usr/bin/gdal-config? Nov 05 16:22:26 Sundar: Then your mount options are bad Nov 05 16:23:42 florian : you mean theway i mount my FS in uboot? Nov 05 16:24:38 Sundar: no, the way you mount the fs in linux Nov 05 16:25:21 Blackhex???? the REDEPENDS let opkg/ipkg install gdal when you run opkg/ipkg install grass Nov 05 16:25:40 and I guess gdal-config is in gdal Nov 05 16:26:11 Blackhex or are you in another context buildung .ipk? Nov 05 16:26:21 florian : i did not understand.bootargs=console=ttyS0,115200n8 root=/dev/mmcblk0p1 rw ip=dhcp rootdelay=1 mem=64M . r u talking abt this? Nov 05 16:27:29 Sundar: no... check what 'mount' tells you Nov 05 16:30:06 among the many, i also have at the end : "/dev/mmcblk0p1 on /media/mmcblk0p1 type ext3 (rw,data=ordered)" Nov 05 16:31:27 Sundar: so you have /dev/mmcblk0p1 mounted to / and to /media/mmcblk0p1 - not good. Nov 05 16:31:45 woglinde: I want to build grass using "bitbake -c build grass" but to properly detect location of gdal library I need to add --with-gdal=/bin/gdal-config to configure script execution. gdal-config was build using "bitbake -c build gdal" but it's only in resulted packages and nowhere in building environment... Nov 05 16:32:47 woglinde: So I want to know how to "install" some builded package like it was a toolchain Nov 05 16:34:23 Blackhex then use DEPENDS = gal Nov 05 16:34:29 in the .bb Nov 05 16:34:39 DEPENDS = is for building depends Nov 05 16:34:39 woglinde: I had it already Nov 05 16:34:47 florian: i realize something is wrong.but nt enuf , to fix it :D some more tips? should i umount /media/mmcblk0p1 Nov 05 16:34:47 Blackhex please read teh oe manual Nov 05 16:34:55 Blackhex to write sane and clean .bb Nov 05 16:35:18 blackhex http://bec-systems.com/oe/html/index.html Nov 05 16:35:26 woglinde: but the problem is that there is no gdal-config installed during installation step Nov 05 16:35:58 Sundar: Might be a good idea... and then check the binaries in your fs if alll are executable. Same for the Xserver script... Nov 05 16:36:03 Blackhex add inherit binconfig to you gal.bb Nov 05 16:36:18 Blackhex read the manual section about do_stage Nov 05 16:36:58 woglinde: binconfig is there; I read something about staging, but I thoguth it's only for build speedup Nov 05 16:37:16 Blackhex????? Nov 05 16:37:30 staging is where are stuff lays for native and crosscompiling Nov 05 16:37:45 install is only for installing so it fits to the .ipk Nov 05 16:38:08 woglinde: OK now it makes little sense Nov 05 16:38:45 woglinde: this one confused me a little: http://blog.haerwu.biz/2008/07/01/packaged-staging-and-what-it-gives/ Nov 05 16:39:20 Blackhex: why it confused you? Nov 05 16:39:31 Blackhex: thats my blog Nov 05 16:40:13 hrw I think he tought the package staging are done automagiclly Nov 05 16:40:17 from install Nov 05 16:40:19 hrw: :-) I was looking what is staging about and there is said that's it's about build speedup, not setting up build environment Nov 05 16:40:42 i unmount media/mmcblk0p1. and as a root i have rwx on all scripts in /etc/init.d/ . is tht what u meant?still no success Nov 05 16:41:24 Blackhex: with packaged-staging my builds can be faster Nov 05 16:41:39 woglinde, hrw: But thank you both. Nov 05 16:42:48 hrw: I don't dispute, but this blog was only kind of relevant information about staging google gave me and I think it don't explain real purpose for those who do know nothing about it. Nov 05 16:43:21 florian:i did umouunt media/mmcblk0p1. and as a root i have rwx on all scripts in /etc/init.d/ . is tht what u meant?still no success Nov 05 16:45:19 hrw: ah now I see the difference: I din't realize that staging and packaged staging are two different things Nov 05 16:45:54 Sundar: you will have to check all scripts and binaries they can call - e.g. from /bin, /usr/bin, /etc/X11 etc.. Nov 05 16:48:24 03Otavio Salvador  07shared/xorg-7.4-update * r301c02d1fe 10openembedded.git/packages/vnc/ (tightvnc/Vncviewer tightvnc_1.3.9.bb): tightvnc: add a default X resources file Nov 05 16:48:24 03Otavio Salvador  07shared/xorg-7.4-update * rbe356ca208 10openembedded.git/packages/ica/ica-bin_10.6.bb: ica-bin: add 10.6 recipe Nov 05 16:48:24 03Otavio Salvador  07shared/xorg-7.4-update * re90881203c 10openembedded.git/packages/vnc/tightvnc_1.3.9.bb: tightvnc: provide vncviewer alternative to provide more flexibility Nov 05 16:49:03 florian : okay and i shud make sure, i have execute permission on all? on a sidenote : i can run the same FS on beagleboard , no problem , all works fine.shud i try nfs?will tht b of any use? Nov 05 16:49:32 Sundar: get a new keyboard first ;) Nov 05 16:50:42 florian : point noted :D .whts the latest typo,i did? don see one in the last line :-? Nov 05 16:50:58 03Koen Kooi  07shared/xorg-7.4-update * r44a09d14b3 10openembedded.git/packages/sysvinit/sysvinit_2.86.bb: sysvinit: avoid u-a race condition by raising 'init' to 60 Nov 05 16:50:58 03Graeme Gregory  07shared/xorg-7.4-update * r553b4bd158 10openembedded.git/packages/xorg-driver/xf86-video-intel_2.4.2.bb: xf86-video-intel_2.4.2.bb : new version for X11R7.4 Nov 05 16:50:58 03Graeme Gregory  07shared/xorg-7.4-update * r199b245113 10openembedded.git/packages/xorg-driver/xf86-video-sunffb_1.2.0.bb: xf86-video-sunffb_1.2.0.bb : new version for X11R7.4 Nov 05 16:51:00 03Holger Hans Peter Freyther  07shared/xorg-7.4-update * rc1401f7b18 10openembedded.git/ (3 files in 3 dirs): (log message trimmed) Nov 05 16:51:03 python-pygame: Blacklist smpeg and build pygame without smpeg Nov 05 16:51:05 We can not build smpeg when enterprise is on. Put it into the Nov 05 16:51:07 BBMASK to make sure it is not buildable. Nov 05 16:51:09 Use base_conditional to not add the dependency to the DEPENDS Nov 05 16:51:14 when building for the enterprise branch. Remove the movie support Nov 05 16:51:18 (the one actually requiring smpeg) and only echo it into the Setup Nov 05 16:51:22 03Graeme Gregory  07shared/xorg-7.4-update * rd3d3ad1cca 10openembedded.git/conf/checksums.ini: checksums.ini : more new checksums for X11R7.4 Nov 05 16:58:03 bye Nov 05 17:11:25 florian :got a new kb :P /bin /usr/bin has execute permissons. in /etc/X11/ the scripts to start X are executable binaries. do gpe-login* files and xmodmap need to be executable binaries? Nov 05 17:12:37 Sundar: yes Nov 05 17:13:29 florian : will modify. curiosity ,how did the bootup on beagleboard dint have these issues? Nov 05 17:14:09 Sundar: And Xserver still does not bring up anything? What's the error message? Nov 05 17:16:27 florian :/etc/init.d/gpe-dm start -this says "starting display manager "with no error message. Nov 05 17:16:45 florian:55 to all scripts in /etq/x11/ Nov 05 17:17:13 Sundar: Run /etc/X11/Xserver Nov 05 17:17:16 florian:gave 755 to all scripts in /etq/x11/ Nov 05 17:19:02 florian :(EE) Failed to load module "glx" (module does not exist, 0) Nov 05 17:19:23 florian : Fatal server error: no screens found Nov 05 17:20:02 oh Nov 05 17:20:51 Sundar: That's an Xorg error message. Either configure Xorg for your device or replace it with kdrive. Nov 05 17:21:46 florian : by configure ,you mean xorg.conf? can you throw some more info on kdrive? Nov 05 17:22:45 florian : am i missing a kernel loadable module? is there a kernel option, that would statically link this into my kernel for now? Nov 05 17:24:59 bye Nov 05 17:28:09 florian : in a mail archive.koen kool mentioned :"All recent versions of angstrom should have kdrive patched to support 800x480, which version are you running? " i use "angstrom-2008.1" and cloned "org.openembedded.stable" . should i run a different version? Nov 05 17:28:39 re Nov 05 17:28:47 Sundar: 2008.1 is in .dev not in .stable Nov 05 17:34:12 hrw: i did my first build yesterday. i have mis-matched a devleopment release and org.openembedded.stable. this is a mistake. do i have to redo? Nov 05 17:59:05 * * OE Bug 4805 has been created by enrico.scholz(AT)sigma-chemnitz.de Nov 05 17:59:07 * * ffmpeg: missing 'install -s' case Nov 05 17:59:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4805 Nov 05 18:20:01 anybody? i'm trying this in my distro .conf -> PREFERRED_PROVIDER_java2-runtime = "jamvm" <- bb "sees" both jamvm and cacao as providers, but picks cacao anyway, telling me I should define a provider... Nov 05 18:20:20 how do I debug this? I tried all the -D's in the world - time to learn python? Nov 05 18:21:39 other preferred i have defined ARE working. is there something special about "-runtime" ?? Nov 05 18:29:36 hm HopsNBarley could grep overall config maybee it is unset somewhere Nov 05 18:37:35 does anyone know if busybox is supposed to be suid root? Nov 05 18:37:53 oilford hm Nov 05 18:37:54 i.e can someone with a working rootfs where they haven't manually set it check for me? Nov 05 18:38:13 is not set by default Nov 05 18:38:27 I am heading home Nov 05 18:38:29 how are things like su supposed to work then? Nov 05 18:38:29 bye Nov 05 18:38:43 thanks though Nov 05 18:38:44 bye Nov 05 18:39:05 anyone else know how it's supposed to work? Nov 05 19:01:13 hmm, looks like its an ongoing, hack time then Nov 05 19:16:44 gdk-pixbuf does not build... any hints? it says "reduce-dependencies.patch does not apply "... Nov 05 19:29:34 <_diego_> emdete, try to remove the patch ;) Nov 05 19:36:59 shure, why not just remove everything ;) Nov 05 20:06:21 oliford: su doesn't work. Nov 05 20:07:03 oliford: if you think the security risk of running busybox setuid root is acceptable, then you can set it so -- but the built image will not make that assumption for you. Nov 05 20:41:38 re Nov 05 20:42:34 hi, is there an open source OSEK (Open Systems and their Interfaces for the Electronics in Motor Vehicles) operating system emulator ? Nov 05 20:44:15 kantor: you're off topic, and sending the same message to multiple channels is usually frowned upon, particularly at the exact same time Nov 05 20:45:15 * mwester didn't believe the disk diag program, when it reported that SMART indicated the drive was about to fail... :( Sigh. Nov 05 20:45:40 :( Nov 05 20:46:50 It's only my build/tmp directory, so it could be a lot worse. Nov 05 20:47:27 :) But there's no doubt that OE builds will find marginal disks! Nov 05 20:48:17 hehe Nov 05 21:00:28 woglinde, hi...the mplayer package doesn't compile anymore because it uses ffmpeg not from svn...how can i force it to use that...http://pastebin.com/m7bd188a3 Nov 05 21:00:42 s/uses/does not use/ Nov 05 21:01:12 s/s\/uses\/does not use\/ Nov 05 21:03:58 Gnutoo guntoo rais the DEFAULT_PREFRENCES in ffmpeg_svn Nov 05 21:04:17 woglinde, thanks a lot Nov 05 21:05:50 mwester: ok, thanks. But then why did busybox get configured to provide su if it doesnt work? Surely it should have been left as tinylogin? Nov 05 21:06:17 As it is, building x11-gpe-image gives me an unusable system Nov 05 21:27:55 oliford: there is some discussion about the issues. But tinylogin is unmaintained. As far as an unusable system, if you are building a "one-time-use" build, just set the setuid bit manually. If you are building a distro, it's really quite easy for a distro to set that bit as the image is being assembled. Nov 05 21:28:41 From the OE point of view, the default should be safe. And not everyone is prepared to agree that it is known always to be safe to run busybox setuid root -- it has been less-than-completely stable in the past ;) Nov 05 21:29:25 ok, so should I pester angstrom to set it then? (thats what I'm building) Nov 05 21:30:11 It seems silly to have to do it specifically for my device Nov 05 21:32:11 Certainly discuss it with the angstrom folks; I know they have participated in the discussions previously, so they may be able to provide better insight into their rationale for that distro. Nov 05 21:32:21 ok Nov 05 21:32:44 IIRC, they don't trust busybox ;) Nov 05 21:33:53 with good reason Nov 05 21:34:47 There was a recent discussion regarding this. There's no way to make sure busybox can be safely set suid root. Nov 05 21:36:48 in -theory- itd be fine, since busybox drops privs for applets that dont need it, but i don't know that the code's been audited or anything Nov 05 21:36:51 heh Nov 05 21:37:33 and any libraries used by busybox would also need to be audited, right? Nov 05 21:38:00 you mean, the c library? I'm pretty sure people have looked into that, given the number of suid binaries on your average Linux desktop Nov 05 21:38:30 It doesn't use anything other than uclibc? Nov 05 21:38:34 glibc? Nov 05 21:38:53 I didn't really want to get into that, I just think its not right to leave it config'ed in a way that doesn't work Nov 05 21:39:36 the tinylogin binary is present in my /bin so why not config it, if it works and is safe Nov 05 21:39:45 NAiL: the busybox folks tend to be pretty minimalist. they work to trim out a couple K from a module's size. i couldn't see them adding unnecessary deps Nov 05 21:39:46 so yea.. angstrom issue Nov 05 21:51:32 I need to check the busybox mount utils to see if we can use them in Angstrom Nov 05 21:51:43 the older versions had some missing options that Angstrom used Nov 05 21:59:04 I was able to get everything working that SlugOS needs, and SlugOS uses probably everything. There were a couple of mount options that busybox didn't support, but it turned out that the variant supported by busybox is actaully the prefered one. Nov 05 21:59:40 * mwester sighs at his spelling. Nov 05 22:22:27 It does seem a bit silly to ship an su binary that isn't going to work. It seems like it would be better to just turn su off in the busybox configuration and let folks who want su install tinylogin instead. Nov 05 22:23:43 indeed. it should be considered a bug as is, definately Nov 05 22:37:05 * * OE Bug 4806 has been created by  Nov 05 22:37:07 * * ffmpeg-0.4.9+r35+gitrafb98868e19e63cbca6c9f0ed9e6cfa48d40277d-autobuild Nov 05 22:37:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4806 Nov 05 22:37:20 sigh, hardcoded paths in hplip source :-( **** ENDING LOGGING AT Wed Nov 05 22:48:18 2008 **** BEGIN LOGGING AT Wed Nov 05 22:48:39 2008 Nov 05 22:51:29 kergoth: the problem with busybox security is not so much external libs, it's more the amount of code sharing within busybox. Nov 05 22:52:30 kergoth: it's all very well to say that applets which don't need setuid will drop the privs, but because the whole thing is linked together into one monolithic binary it is fairly hard to be certain which functions might possibly be called in a setuid context and hence are security-critical. Nov 05 22:53:22 nod. i'd expect most of that to be in libbb though.. hopefully, at least. having a lot of inter-applet dependency would be bad in general.. I'd expect you'd have to go over libbb with a fine tooth comb Nov 05 22:53:23 plus, the rate of code churn within busybox is high enough that you would end up having to re-audit it on a frequent basis. Nov 05 22:54:11 true, that wouldn't be fun at all Nov 05 22:54:13 as for tinylogin being "dead", I don't really buy that argument. sure, it isn't being updated, but it doesn't seem to require any updates. it does what it says on the box. Nov 05 22:54:40 and, it has the advantage of being small, self-contained, and easy to audit with low code churn. Nov 05 22:55:45 on the angstrom file system ,i need some advice configuring my Xserver for an OMAP target with an lcd display. Xserver fails with loading omapfb & no devices detected. how do i make it identify the screen i have?should i edit xorg.conf to recognize this display? Nov 05 22:56:08 hi ho Nov 05 22:56:43 emdete is getting this http://rafb.net/p/fGsLd321.html when trying to build an image which contains glib Nov 05 22:57:06 Building it by hand after this failure it goes through and the image also can be copntinued Nov 05 22:57:13 Threading bug? Nov 05 22:57:18 hm no Nov 05 22:57:35 i have not THREAD -j something set - what's the default? Nov 05 22:57:37 somehow Makefile.am has to be better Nov 05 22:57:48 woglinde: Known bug? :) Nov 05 22:57:55 stefan_schmidt nope Nov 05 22:58:04 only emdete has the problem Nov 05 22:58:05 stefan_schmidt: woglinde gave mit the make in devshell hint ;) Nov 05 22:58:11 woglinde: ok, so best to fill one then. :) Nov 05 22:58:23 ah, woglinde is already helping out. Good. Nov 05 22:58:41 woglinde: but i see a strange behavior with bb - if i press ^Z i see output going on Nov 05 22:58:48 somehow the .la file is not build Nov 05 22:58:49 I have no idea how to fix this and I'm to lazy to dig into this tonight. Nov 05 22:59:24 emdete: Best is to fill a bug about it with the error pasted and a description that it works in the second run Nov 05 22:59:54 woglinde: So sloopy dep handling in Makefile.am Nov 05 23:00:38 woglinde: whats the cause for output after ^Z? any idea? Nov 05 23:06:19 emdete what ^Z you are talking off? Nov 05 23:08:49 woglinde: if you let a program run in the shell and press Ctrl-Z Nov 05 23:09:12 woglinde: it gets immediatly supended... Nov 05 23:11:55 hm Nov 06 00:00:13 emdete still there? Nov 06 00:00:22 found the error Nov 06 00:01:43 woglinde: I think he left some time ago Nov 06 00:01:49 woglinde: What is it? Nov 06 00:02:53 dependency Nov 06 00:03:11 look one line below in the patch Nov 06 00:04:03 woglinde: patch? Nov 06 00:04:12 * stefan_schmidt misses some infos it seems Nov 06 00:05:19 ackages/glib-2.0/files/atomic-thumb.patch Nov 06 00:06:26 woglinde: Ah, introduced with pb_'s thumb work Nov 06 00:07:04 woglinde: Nicely spotted Nov 06 00:08:27 03woglinde  07org.openembedded.dev * rda9083a094 10openembedded.git/packages/glib-2.0/ (files/atomic-thumb.patch glib-2.0_2.16.4.bb): Nov 06 00:08:27 glib-2.0: fix atomic-thumb.patch Nov 06 00:08:27 * is only in included in glib-2.0_2.16.4 Nov 06 00:08:27 * missed depenedcy on .la file Nov 06 00:08:27 * bump PR of glib-2.0_2.16.4 Nov 06 00:08:59 okay spelling error is okay for this time Nov 06 00:09:01 good nite Nov 06 00:09:16 woglinde: heh, exactly now a bug report about this came to openmoko-devel :) Nov 06 00:48:03 anyone have any experience running OE on CentOS 4.6? Nov 06 02:00:29 4.6? Nov 06 02:02:36 I just checked; 5.1 on a quad-core w/4GB RAM worked pretty darn well. But IIRC I had to do a lot of building of things myself (git, monotone, and such). I expect 4.6 would need even more pre-reqs built. Nov 06 02:03:26 rwhitby's NSLU2 Master Makefile http://www.nlsu2-linux.org/Makefile may have some hints; I know he uses CentOS. Nov 06 02:03:52 mwester: I don't build on CentOS any more. **** ENDING LOGGING AT Thu Nov 06 02:59:57 2008