**** BEGIN LOGGING AT Mon Mar 26 02:59:57 2007 Mar 26 03:07:08 empty_mind: I never install bitbake. I always get the makefile wrapper to use the latest stable branch. Mar 26 03:28:21 rwhitby, ok, Mar 26 03:30:36 I got bit once by having an installed old bitbake, and not easily being able to override it (or use two different bitbake versions for two different distros) Mar 26 05:04:35 hello, has anyone else had problems with glibc compile basically locking up, or failing when generating the locale binaries? I'm using OE oz354x branch. Mar 26 06:05:24 evening gents Mar 26 06:36:25 morning all Mar 26 06:36:43 morning Mar 26 07:08:50 morning Mar 26 07:13:09 this has me confused and I'm sure I *knew* the answer some time ago.. what's the difference between the image directory and the install directory under a package? I use install for files that will be packaged in my current recipes.. is that not correct? Mar 26 07:24:14 hmm and another slight issue.. if anyone is using ldd with busybox can you let me know? If I run ldd I get no output but stracing ldd shows "Illegal option -o pipefail" returned from a call to set Mar 26 07:30:52 I'm sure I saw that on the wiki before but i'm not seeing it now :( Mar 26 07:42:39 good morning all Mar 26 07:53:36 mornign koen !!! Mar 26 07:58:46 hey gremlin[it] Mar 26 08:03:11 morning Mar 26 08:05:34 hey XorA Mar 26 08:05:45 morning XorA Mar 26 08:09:50 morning Mar 26 08:09:57 After updating my OE checkout and rebuilding an image from scratch I get the error message "terminals database is inaccessible" when running `top` on the target. Mar 26 08:10:11 I have to `export TERM=xterm` to fix Mar 26 08:10:22 what has changed to make this happen? Mar 26 08:10:45 koen: what do you run on your p0 neo? do you build openmoko-devel-image for fic-gta01 and angstrom distro? or some other combination? Mar 26 08:11:15 openmoko-devel-image MACHINE=fic-gta01 DISTRO=angstrom-2007.1 Mar 26 08:11:21 the rest is done via ipkg Mar 26 08:11:41 root@fic-gta01:~$ uptime Mar 26 08:11:41 08:11:35 up 5 days, 17:59, 5 users, load average: 1.00, 1.00, 1.00 Mar 26 08:11:42 :) Mar 26 08:12:01 I have to restart libgsmd everyday, but I don't think that's distro related Mar 26 08:12:58 on a whim I tried that combination, but didn't get the usual touchscreen calibration thingo on boot. I did have some jffs2 node crc errors, so maybe that's the reason. Mar 26 08:13:19 you erased the flash before flashing the image? Mar 26 08:13:41 ah, no. just did the dfu-util write Mar 26 08:13:55 right, I got caught by that one as well Mar 26 08:14:23 * rwhitby makes a note to ask werner about dfu erase capability. Mar 26 08:14:36 iirc there a bugzilla entry for it Mar 26 08:15:00 rwhitby: gpstk-bin takes ~12MB jffs2 :( Mar 26 08:15:16 is that specific to something? I've reflashed many openmoko svn images and they seemed to work. Mar 26 08:15:33 (the erase before write requirement, that is) Mar 26 08:16:18 luck? Mar 26 08:16:23 (I have no idea) Mar 26 08:16:32 probably just similar sized images. Mar 26 08:19:48 koen: no rootfs for the Neo? Mar 26 08:20:08 jselect: the ldd thingy, see http://bugs.openembedded.org/show_bug.cgi?id=1484 Mar 26 08:20:55 tkp: ncurses-terminfo not installed? I noticed it missing for other things, and it looks like it used to be installed by default... Mar 26 08:22:52 v8jlene: yep. that fixes it... thanks Mar 26 08:23:01 any ideas where that fell off the dependency tree? Mar 26 08:23:34 XorA: I looked for exactly the same thing :-) Mar 26 08:23:42 tkp: Nope, only noticed a few days ago and haven't had time to investigate. Mar 26 08:30:50 XorA: I forgot to but it in the buildqueue, will put on up later today Mar 26 08:31:31 v8jlene: OE (used to) include(s) basic terminfo in the main ncurses package Mar 26 08:51:26 03hrw 07org.oe.dev * ra62308ff... 10/ (1 packages/ltrace/ltrace_0.4.bb): ltrace: make it buildable on misc architectures (copied some stuff from linux-libc-headers recipe) - bug reported by Dodji Seketeli@OpenedHand Mar 26 09:00:26 damnation, svn: REPORT request failed on '/!svn/vcc/default' Mar 26 09:05:06 * rwhitby splits slugos sysconf functionality into a separate package ... Mar 26 09:05:12 morning Mar 26 09:05:40 neo1973 images are padded to roper size Mar 26 09:12:43 hrw: Hello! And completely another topic: do you have an idea what's "qte-2.3.12"? Mar 26 09:14:27 hello everyone Mar 26 09:14:55 hello everyone Mar 26 09:16:10 psokolovsky: it is 2.3.12 version of qt/e Mar 26 09:16:27 psokolovsky: not tested as much as 2.3.10 so we do not use it by default Mar 26 09:17:39 hrw: I checked official TT site, 2.3.10 appears to be latest released. ? Mar 26 09:17:59 hrw: and actually, that's why I ask - I'd like to maek it DEF_PREF=-1 Mar 26 09:18:10 psokolovsky: no. 2.3.13 even exists iirc Mar 26 09:18:27 psokolovsky: >2.3.10 are part of qtopia 2.x Mar 26 09:19:09 hrw: ok, but ok for DEF_PREF=-1 too? Mar 26 09:19:32 yes Mar 26 09:19:33 also it lives in qt-embedded-2.3.10-snapshot-20060120.tar.gz somehow Mar 26 09:19:36 ok Mar 26 09:22:22 ~lart pcmcia subsystem Mar 26 09:22:23 * ibot overclocks pcmcia subsystem until pcmcia subsystem burns out Mar 26 09:47:34 ~lart git for being a source uncontrollable system Mar 26 09:47:34 * ibot throws git's poor little doggy off a cliff for being a source uncontrollable system Mar 26 09:56:53 are there any provisions in OE to create a swapfile when creating the rootfs image? Mar 26 09:57:22 likewise: tried bitbake -b glib-2.0_.bb? Mar 26 09:57:31 I'd be interested in what error bitbake will produce Mar 26 09:58:20 koen|lab: will try that when I get home, tnx Mar 26 09:58:50 tkp: not that I know, what exactly would you need? Mar 26 09:59:20 likewise: well, to run the dd command to create a swapfile the required size Mar 26 09:59:25 something like : dd if=/dev/zero of=/swapfile1 bs=1024 count=51200 Mar 26 09:59:38 and an appropriate entry in fstab Mar 26 10:02:17 tkp: for what target is this? Mar 26 10:03:00 this is for an x86 target (epia mini-itx) Mar 26 10:07:02 tkp: and the swapfile resides inside the rootfs, or outside? Mar 26 10:08:07 03hrw 07org.oe.dev * re0342ac3... 10/ (1 packages/xorg-lib/libxaw_X11R7.1-1.0.2.bb): libxaw: split into libxaw6/7/8 as our apps depend on one of them only so why install all Mar 26 10:09:17 likewise: this file would need to reside in the rootfs Mar 26 10:09:22 in my case anyway Mar 26 10:09:42 it could reside on another device or patition though Mar 26 10:09:50 tkp: in that case, yes, it would be easy to do as a bitbake recipe Mar 26 10:10:40 likewise: yes, I could create my own recipie for this tsk... I was wondering if there was already something like this in OE Mar 26 10:12:15 so as far as git is concerned, "merge" means delete all your changes! Mar 26 10:12:24 what a peice of crap of a source control system Mar 26 10:13:05 XorA: but it will delete them OMG FAST! Mar 26 10:13:10 which is what counts Mar 26 10:13:27 koen|away: yeah, it is fast, but not as fast as rm -rf Mar 26 10:13:45 and at least when I call rm -rf there is a hint there that changes will be dumped Mar 26 10:14:57 oh well a weeks worth of work down the drain, next time I shall use Visual Source Safe Mar 26 10:17:43 eek! Mar 26 10:17:54 is life that bad Mar 26 10:20:38 XorA: you can't back out the merge? Mar 26 10:21:21 XorA: you do not use quilt? Mar 26 10:21:33 hrw: in future I will Mar 26 10:28:14 2434 hrw 18 0 5712m 3.5g 2.0g D 22.3 174.5 11:35.52 kopete Mar 26 10:34:14 lunch time! Mar 26 10:56:24 I'm getting "Your Compiler does not do propper IEEE math" when trying to build rrdtool Mar 26 10:56:30 anyone seen this error before? Mar 26 11:00:14 good morning Mar 26 11:00:50 hey florian Mar 26 11:01:03 morning florian Mar 26 11:32:16 03koen 07org.oe.dev * rf246f66d... 10/ (5 files in 2 dirs): gpe-conf: depends on tzdata instead of timezones, makes the USians a bit happier Mar 26 11:32:29 ~lart DST Mar 26 11:32:30 * ibot offers DST some herring Mar 26 11:41:09 ~curse broken toolchains Mar 26 11:41:14 May you be reincarnated as a Windows XP administrator, broken toolchains ! Mar 26 11:41:36 We are not happy about the new DST crap Mar 26 11:41:50 pitch black at 6:30 when I am trying to get up Mar 26 11:43:47 aye Mar 26 11:44:00 train drivers not getting up is also an issue **** BEGIN LOGGING AT Mon Mar 26 11:51:03 2007 Mar 26 12:16:09 ping : koen Mar 26 12:16:43 koen : i found a bug in the sort.sh script for the feed Mar 26 12:18:02 what is the use of CONFFILES_${PN}? Mar 26 12:18:24 koen|lab : if the machine name happens to be part of the filename of the package, then that file goes to the /arch/machine/machinename dir Mar 26 12:19:23 koen|lab : in my case i had a package named perl-module-text-wrap_5.8.7-r21_ppc405.ipk Mar 26 12:19:45 and also defined machine 'wrap' under i486 Mar 26 12:19:52 ah, right Mar 26 12:20:06 feel free to fix it :) Mar 26 12:20:21 koen|lab : thanks. you are so kind :) Mar 26 12:20:35 * koen|lab wonders why googling for 'multi-mode filter' only returns "OMG SYNTH!" pages Mar 26 12:21:35 should I do ${PN}-conf to package up configuration files seperately... or do I use CONFFILES_${PN} somehow? Mar 26 12:24:59 What's wrong with this line?: INHERIT_append_nslu2 = " nslu2-image" Mar 26 12:25:33 and why would that line set INHERIT correctly (according to bitbake -e), but fail to include nslu2-image.bbclass (according to bitbake -DDD)? Mar 26 12:25:50 whereas this line works fine: INHERIT += "nslu2-image" Mar 26 12:26:35 sorry.. will someone please explain the usage of CONFFILES... I can find noting about it in the mailing list or in the oe manual Mar 26 12:26:52 tkp_: use CONFFILES, see http://www.openembedded.org/user-manual&dpage=recipes_conffiles Mar 26 12:27:04 v8jlene: doh! ta Mar 26 12:27:09 (MACHINE is nslu2le, but nslu2le.conf sets OVERRIDES to include 'nslu2') Mar 26 12:27:53 Is it the case that OVERRIDES does not affect variable expansion for INHERIT? Mar 26 12:29:54 koen|lab : fiexed it . You need to change in sort.sh in the machine loop from -> for i in `find . -name "*.ipk"| grep $machine` to ->for i in `find . -name "*_$machine.ipk"| grep $machine` Mar 26 12:31:57 (this is with bitbake 1.6 - I guess I'll try with 1.8 and trunk before calling it a bug) Mar 26 12:32:39 koen|lab : let me know if you want me to change it to the angstrom rep also Mar 26 12:33:16 Ifaistos: please change that as well :) Mar 26 12:59:17 koen|lab : done Mar 26 13:00:55 hmmm anyone knows why perl-modules depends on perl-dev and perl-dbg ? Mar 26 13:02:24 Ifaistos: buggy packaging Mar 26 13:06:49 ~lart perl Mar 26 13:06:49 * ibot decapitates perl conan the destroyer style Mar 26 13:07:18 it seems its more than buggy packaging with perl Mar 26 13:07:50 http://rafb.net/p/IouuQ263.html Mar 26 13:13:53 could somebody look at this and see if it is a bug? http://builds.hackndev.com/builds/angstrom/daily-1174867801/build-palmz72.-nodebug.log Mar 26 13:15:23 03rwhitby 07org.oe.dev * r4b62fa79... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: Added upslug2, devio and slugimage Mar 26 13:19:51 ping ~rwhitby Mar 26 13:20:02 ~ping rwhitby Mar 26 13:20:18 pong rwhitby Mar 26 13:26:41 hi Cliff Mar 26 13:42:42 slapin_nb: you'd have to hassle psokolovsky for that Mar 26 13:43:07 koen, thanks! Mar 26 13:46:06 hi all Mar 26 13:46:53 ~seen mrdata Mar 26 13:47:25 mrdata was last seen on IRC in channel #oe, 18h 44m 27s ago, saying: 'have a ARM kernel developer some time for me, help to find the reason for a kernel-2.6.20 Oops - bad syscall problem '. Mar 26 13:48:24 slapin_nb: please submit that as bug together with local.conf and other relevant configs Mar 26 14:03:48 is anyone using OE build perl on their target ? Mar 26 14:04:12 Ifaistos: I used Mar 26 14:04:41 Ifaistos: I used OE under Angstrom few months ago Mar 26 14:04:54 hrw: hello Marcin Mar 26 14:05:06 hrw: can you try to run a perl program to see if it works on your machine ? Mar 26 14:05:58 hrw: download this-> http://www.asternic.org/files/op_panel-0.27.tar.gz Mar 26 14:06:15 hrw: unpack and try to run op_server.pl Mar 26 14:06:34 hrw: I am getting this http://rafb.net/p/IouuQ263.html Mar 26 14:10:06 Can't load '/usr/lib/perl5/5.8.7/powerpc-linux/auto/IO/IO.so Mar 26 14:10:24 this tell a lot.. you have all needed perl modules? Mar 26 14:10:52 Ifaistos: I do not have that chroot under which I had perl Mar 26 14:13:41 hrw: i did an ipkg install perl-modules Mar 26 14:13:59 hrw: it installs the kitchen sink Mar 26 14:15:23 !oebug 2046 Mar 26 14:15:25 * * Bug 2046, Status: NEW, Created: 2007-03-26 07:13 Mar 26 14:15:26 * * slapinid(AT)gmail.com: glibc-2.5 build problem Mar 26 14:15:27 * * http://bugs.openembedded.org/show_bug.cgi?id=2046 Mar 26 14:15:32 psokolovsky, Mar 26 14:16:45 !oebug 2046 Mar 26 14:16:46 * * Bug 2046, Status: RESOLVED, Created: 2007-03-26 07:13 Mar 26 14:16:47 * * slapinid(AT)gmail.com: glibc-2.5 build problem Mar 26 14:16:48 * * http://bugs.openembedded.org/show_bug.cgi?id=2046 Mar 26 14:16:54 slapin_nb: ^ Mar 26 14:17:01 too bad it doesn't show resolution Mar 26 14:17:38 morning Mar 26 14:18:13 psokolovsky: pester CoreDump|home to add it Mar 26 14:18:36 ;-) Mar 26 16:02:21 ERROR: Build of /home/koen/OE/monotone/org.openembedded.dev/packages/abiword/abiword-plugins_2.4.6.bb do_qa_configure failed Mar 26 16:03:35 koen: abiword 2.5.x is not in OE? Mar 26 16:03:47 it is, but noone updated the plugins Mar 26 16:03:58 no one Mar 26 16:04:07 (and 2.5 exposes a bug in glib-2.0) Mar 26 16:10:31 * rob_w started new channel #geode to cover AMD geode cpu discussions Mar 26 16:35:15 hello Mar 26 16:53:54 mr_nice: hi Mar 26 16:54:00 hi all Mar 26 17:08:21 HopsNBarley: hi there Mar 26 17:08:25 Hello all Mar 26 17:08:44 hey likewise Mar 26 17:08:51 Jin^eLD: Hey! Mar 26 17:13:40 03koen 07org.oe.dev * rd4ed6162... 10/ (1 packages/librsvg/librsvg_2.16.1.bb): librsvg: add 2.16.1 Mar 26 17:17:26 please help Mar 26 17:18:12 I've just pulled OE db (the complete one) and I can't see the folder org.openembedded.dev Mar 26 17:18:24 where can I find it? Mar 26 17:19:43 Good evening Mar 26 17:20:49 djay-il: http://monotone.ca/monotone.html Mar 26 17:20:53 hey sirfred Mar 26 17:22:44 hi likewise! Mar 26 17:23:12 likewise, thanks again for the visit Mar 26 17:23:57 mr_nice: hi Mar 26 17:27:39 koen: thanks, reading now Mar 26 17:32:37 Whom of the bitbake designers are in? Mar 26 17:33:29 koen: hm, do you have anything to do with OE wiki? Mar 26 17:33:49 everyone has something to do with the wiki! Mar 26 17:34:11 djay-il, doesn't getting started cover that? Mar 26 17:34:32 maybe I'm too stupid, but I had to checkout the branch to see the folder. It misses this step in "getting started" Mar 26 17:34:40 hmmm Mar 26 17:34:45 :-) Mar 26 17:35:06 "Checkout your local copy of the OpenEmbedded tree. Mar 26 17:35:07 $ mtn --db=/stuff/OE.mtn checkout --branch=org.openembedded.dev Mar 26 17:35:07 " Mar 26 17:35:16 checkout is there :) Mar 26 17:35:19 halfway down at http://www.openembedded.org/wiki/GettingStarted Mar 26 17:35:34 http://www.openembedded.org/wiki/GettingStarted ? Mar 26 17:35:36 mmmm Mar 26 17:35:38 ooops Mar 26 17:35:53 sorry, you right, slipped directly to updating Mar 26 17:35:55 my bad Mar 26 17:35:56 heh Mar 26 17:36:35 thanks dudes Mar 26 17:36:39 :-) Mar 26 17:37:09 Curious, g_spawn_sync is complaining about SIGCHLD action being set to SIG_IGN. But I'm inspecting the source code of gpe-conf and I don't see such call to signal. Any idea of where could this be made? Mar 26 17:37:29 sirfred: maybe gpe-dm Mar 26 17:37:32 If bitbake 'skips' packages, how can I find out *why* it does that? Mar 26 17:38:10 koen: In a different process? But signal handling is not inherited, is it? Mar 26 17:38:15 sirfred: we had to do http://www.openembedded.org/repo/org.openembedded.dev/packages/dbus/dbus-1.0.2/fix-waitpid.patch to dbus to fix waitpid Mar 26 17:38:31 sirfred: which implies it is inherited Mar 26 17:38:43 (gpe-dm is just 20 lines of C) Mar 26 17:38:59 likewise: some versions of bitbake told you that IIRC Mar 26 17:39:15 koen: But, since signals are managed by functions in the process scope, how can they be inherited. THe case SIG_IGN is trivial. But what happens when we set a signal handler function? Mar 26 17:39:32 sirfred: no idea, sorry Mar 26 17:39:46 I just got caught by the dbus thing and robtaylor solved it for me :) Mar 26 17:40:00 koen: OK. Thanks. I'm going to investigate a little. Mar 26 17:47:37 koen: It seems that signals set to the default handler remains the same, signals set to be caught are set to the default handler, after exec. Mar 26 17:48:14 koen: So, if gpe-dm is setting SIGCHLD to SIG_IGN, it makes sense that spawned processes are broken in g_spawn_sync() Mar 26 17:49:06 could be Mar 26 17:49:13 * koen <- signal illiterate Mar 26 17:49:43 So, perhaps it must be set to SIG_DFL just after forking and before exec'ing. I'm going to see gpe-dm source code. Mar 26 17:50:27 mrdata: hi Mar 26 17:51:15 sirfred: http://projects.linuxtogo.org/plugins/scmsvn/viewcvs.php/trunk/base/gpe-dm/gpe-dm.c?rev=8312&root=gpe&view=auto Mar 26 17:51:41 koen: It doesn't seem to play with SIGCHLD, at least in the code I'm using Mar 26 17:52:08 sirfred: which is why robtaylor was cursing as well Mar 26 17:52:18 koen: :) Mar 26 17:52:25 koen: It should be in another place... Mar 26 17:52:28 sirfred: gdm has a lot of code to fixup signal handling Mar 26 17:53:37 Curious, I've finished looking at this, because my zaurus backlight is setting off unexpectly. Mar 26 17:53:50 It seems some gpe-conf issue, but I'm not sure. Mar 26 18:08:07 It seems a matchbox problem: matchbox-session sets SIGCHLD to SIG_IGN, and is the one forking the process launched using the GUI Mar 26 18:09:46 I mean, matchbox-desktop Mar 26 18:11:35 g'evening all Mar 26 18:12:22 gremlin[it]: Good evening Mar 26 18:22:35 mr_nice: hi too Mar 26 18:24:46 mr_nice: i hade modify the mq200-patch a little bit, insert a default backlight value, was zero before Mar 26 18:27:20 mr_nice: the work for suspend problem is in progress, i have found that the kernel tell a Internal error: Oops - bad syscall: Mar 26 18:27:35 mr_nice: after apm --suspend Mar 26 18:30:31 mr_nice: i need the help of a arm kernel developer, who could give me a hint, to find the reason for this Mar 26 18:33:45 mr_nice: i hope zecke or somthing else have some time for me Mar 26 18:42:45 morning all Mar 26 18:43:13 RP: morning Mar 26 18:43:41 RP: can i ask you to a arm kernel problem i have Mar 26 18:44:02 mrdata: sure although I can't promise an answer :) Mar 26 18:44:28 RP: i use kernel-2.6.20 for simpad Mar 26 18:44:44 RP: witheabi-angstrom-linux Mar 26 18:45:10 RP: morning Mar 26 18:45:41 RP: when i do apm --suspend i get Internal error: Oops - bad syscall: 5f5353 [#1] Mar 26 18:46:00 mrdata: Is APM support compiled in? Mar 26 18:46:09 RP: yes Mar 26 18:46:52 Any oops in dmesg? Mar 26 18:47:51 RP: checked - CONFIG_APM=y and no, no oops in dmesg Mar 26 18:48:30 RP; i could use apm -v and get a true response Mar 26 18:49:08 RP: you could look at http://pastebin.ca/409123 Mar 26 18:49:56 RP: i have no idea whats going wrong Mar 26 18:50:25 mrdata: Its not a bad syscall, the syscall is suceeding but crashing somewhere in the kernel Mar 26 18:50:43 Probably one of the simpads devices isn't suspending properly and oopsing when its suspend method is called Mar 26 18:52:04 RP: so i should look at any simpad related suspend function Mar 26 18:52:46 mrdata: Yes, you could instrument the suspend calls to print the module names they're calling to see which one fails Mar 26 18:53:09 I've created a bug report about libmatchbox regarding SIGCHLD handling and g_spawn_sync() errors. It is bug 2048. Mar 26 18:53:22 Or 2^11 Mar 26 18:53:23 RP: this is a posibility, give me the 5f5353 any extra hint Mar 26 18:53:44 Also attaching a proposed patch. Mar 26 18:54:00 sirfred: That should probably go into the OH bugzilla Mar 26 18:54:10 mrdata: The number means nothing to me Mar 26 18:54:16 RP: Oh. Mar 26 18:54:42 RP: Too many bugzillas around. :) Mar 26 18:55:17 sirfred: :) Mar 26 18:55:23 RP: Hmm, matchbox is an OH project,isn't it? Mar 26 18:56:08 RP: thx for your help, i will change the suspend calls Mar 26 18:56:10 RP: I expected that the maintainer (assuming that figure exists) could propagate it upstream, if it is valid. Mar 26 18:57:35 sirfred: matchbox is an OH project, correct. Its not something I look after though so you need to get that bug to others. The OH bugzilla is the place to do it Mar 26 18:57:58 RP: I will do. Mar 26 18:58:31 RP: Thanks Mar 26 19:22:49 Hi Mar 26 19:23:47 I guess I am not supposed to run bitbake as root, why does the GettingStarted give /stuff as example? Mar 26 19:24:47 It seems I'm suffering an annoying problem with ipaq-sleep. Mar 26 19:25:36 I've found that it's calling /usr/bin/bl off when X is starting, I wonder why Mar 26 19:25:56 keesj: if you have a /stuff directory, who says you have to be root to write inside of it? never heard of chown and chmod? Mar 26 19:26:06 keesj: You guess correctly, root is a bad idea. /stuff can be replaced with /whereever/you/would/like Mar 26 19:26:59 sirfred: same problem here. I wonder what caused that change. Mar 26 19:27:16 pH5: Annoying. First time, I thought the machine had hunged. Mar 26 19:28:37 pH5: Looking at the source code of ipaq-sleep, perhaps the cause could be that the brightness is negative... Mar 26 19:28:39 kergoth is just think is is a bad example , the faq tels me no to use the system as root. Mar 26 19:29:00 pH5: For brightness val < 0, it calls: return system (BL " off"); Mar 26 19:29:40 /stuff is obviously an example path. you took it literally. and it was your assumption that you'd have to build as root to write to such a path if it did exist verbatim. Mar 26 19:29:50 It seems it can be launched with -d for debug output. Mar 26 19:32:26 kergoth I did not run it a root , I first tried without, But I had to check the faq before being sure. is the example world show ~/oe everybody whould have oe installed in ~/oe and the Getting started would just be yeasier Mar 26 19:32:39 is = if Mar 26 19:32:43 this is the first i've heard it complained about. Mar 26 19:32:59 again, it's obviously your assumptions that are the problem. Mar 26 19:33:00 and therefore it's wrong? Mar 26 19:34:50 for me it was just not obvious alright? Mar 26 19:35:42 ~oe invites symlinks Mar 26 19:35:57 (I symlinked it and got burned by binutils) Mar 26 19:37:45 koen , yes and I also know that becasue there is a good explanation in the local.conf about that Mar 26 19:38:15 koen: If you're interested, the problem with SIGCHLD before seems to have its origin in matchbox-desktop. It is the process setting SIGCHLD to SIG_IGN, and that is inherited by all processes started from the GUI Mar 26 19:38:36 sirfred: aha! Mar 26 19:38:46 koen: I don't know if that's the case for your dbus problem, I suppose it's not. Mar 26 19:39:07 sirfred: dbus should be launched earlier Mar 26 19:39:16 koen: I suppose so. Mar 26 19:39:28 hi Mar 26 19:39:44 sirfred: it does not work with autolaunched dbus sessions I guess Mar 26 19:40:37 koen: I have opened a bug in oe , and later, in oh Mar 26 19:42:32 What are people's usual graphical terminal programs out of interest? Mar 26 19:43:24 on OE-built devices: rxvt-unicode and the stock osx term on osx (which I dislike) Mar 26 19:43:39 Nothing beats xterm, but gnome-terminal is tolerable in a pinch. Mar 26 19:43:39 koen: I mean on the machines doing the building Mar 26 19:44:02 bash + screen Mar 26 19:44:06 (other than unattended builds, obviously) Mar 26 19:44:11 heh Mar 26 19:44:15 PuTTY client if via ssh. Mar 26 19:44:16 are there any other kind? Mar 26 19:44:37 koen: vnc of course :) Mar 26 19:45:37 Is there a way to tell bash to run in a new terminal in an existing screen session? Mar 26 19:46:35 RP: what do you mean? Mar 26 19:48:17 I'm wondering if I can do something like 'gnome-terminal --disable-factory -x bash' for screen Mar 26 19:48:27 RP: what if you run screen inside the screen? Mar 26 19:50:21 I guess this is a call for opinions on http://www.rpsys.net/openzaurus/temp/oe-terminal.patch Mar 26 19:53:47 RP: screen -t "OE term" Mar 26 19:54:11 that will open a new 'window' named "OE term" Mar 26 19:55:11 koen: ok, so that means that patch should just need SCREEN_TERMINAL_COMMAND = "screen -t "$TERMINAL_WINDOW_TITLE"' Mar 26 19:55:11 SCREEN_TERMINAL_RUNFILE = '${SCREEN_TERMINAL_COMMAND} -x bash --rcfile $TERMINAL_RUNFILE' Mar 26 19:55:49 and then you can set TERMINAL_CMD = ${SCREEN_TERMINAL_CMD} in local.conf Mar 26 19:56:15 This could work out quite well... Mar 26 19:57:10 or just a few sample TERMINAL_CMDs in local.conf.sample and/or bitbake.conf Mar 26 19:57:38 koen: The idea was to define the common ones and then just have the user select which one they wanted in local.conf Mar 26 19:58:12 koen: If we cover screen, gnome and I guess kconsole, that should cover most options and we can add more if requested... Mar 26 19:58:34 The plan here is to fix interactive tasks in bitbake 1.8 and make them suitable for the UI too Mar 26 19:59:12 devshell is broken in trunk Mar 26 19:59:24 koen: This will fix it Mar 26 19:59:30 http://rafb.net/p/zfO1Om45.html Mar 26 20:00:06 koen: This is the fix - a rethink on how to do it Mar 26 20:00:15 * RP -> back in an hour Mar 26 20:11:13 Can someone explain the dependencies with regard to the non-chronological dates: http://www.openembedded.org/viewmtn/revision.psp?id=a6a1d87397b13dbd8f0440381acfd5344a012b60 Mar 26 20:12:45 likewise: bug in viewmtn and/or wrong time one peoples workstations Mar 26 20:12:46 I do not understand how the changes go from 2007-02-09 to ...-08 then to ...-10 ? Mar 26 20:12:59 koen: arghh. Mar 26 20:13:02 we have a 'special' viewmtn Mar 26 20:13:21 koen: I am trying to binary-search for a problem and am using dissection-by-date. :-) Mar 26 20:14:21 koen: OK, so when looking from top to bottom, I go back in real time. Mar 26 20:15:11 likewise: 'mtn log' with mtn 0.33 should give you a better view Mar 26 20:15:24 that includes a tree on the left Mar 26 20:16:27 RP: I'm a few minutes late, but I use Konsole and ssh into a bash shell on my dev box Mar 26 20:20:06 likewise: http://www.openembedded.org/~koen/log-tree.txt Mar 26 20:21:32 koen: yup, got mtn 0.33 and the logs. will help me the next time. Mar 26 20:21:49 koen: I'm almost there, only 2log8 steps to go :-) Mar 26 20:23:15 * koen is compiling for efika now Mar 26 20:26:31 g'night Mar 26 20:35:18 good evening all Mar 26 20:38:16 apaulsen: ok, that shouldn't be a problem :) Mar 26 20:43:13 koen|away: Can you create a screen window and wait on its output? Mar 26 21:06:09 wow i killed my netbook flash \o/ Mar 26 21:08:15 hum i need to manage better boost bootloader Mar 26 21:18:38 meister: ! Mar 26 21:19:53 nand 0x.... ecc bad: read failed Mar 26 21:30:56 RP: can you explain where the "task dependency information that was added to OE.dev" can be found? Mar 26 21:36:11 likewise: Remind me what context that was in :} Mar 26 21:36:26 RP: bitbake 1.8.0 Mar 26 21:37:01 likewise: It means things like rdeptask in debian.bbclass Mar 26 21:41:35 RP: thanks Mar 26 21:42:59 RP: doesn't seem to be used much. When is it needed? Mar 26 21:44:28 likewise: Also look for deptask and recrdeptask Mar 26 21:46:11 rwhitby: good morning rod Mar 26 21:46:25 likewise: They're used to establish dependecies between tasks ie a given tasks DEPENDS must have executed up to task X for the task Y to be runable (do_Y[deptask] = "do_X") Mar 26 21:49:53 RP: thanks, copy that (literally). Mar 26 21:49:54 pH5: I think I've found the root problem: /usr/bin/gpe-conf task_backlight is returning 'off number' even with the backlight on. Mar 26 21:50:04 goot nite all, need some sleeping on this Mar 26 21:50:19 pH5: That's what confuses ipaq-sleep daemon, that sets the backlight off. Mar 26 21:52:31 sirfred: interesting. so it's either a bug in gpe-conf or some missing initial configuration? Mar 26 21:52:38 pH5: It seems so. Mar 26 21:55:15 brbn Mar 26 21:58:42 pH5: At a first sight, it seems that gpe-conf expects /sys/class/backlight/*/brightness to be 0 to consider backlight on and 4 to consider backlight off. But... Mar 26 21:58:55 Perhaps I need to sleep a little. Mar 26 22:00:17 It seems somebody has confused SYS_POWER and SYS_BRIGHTNESS. The first is the one that is 0 for light on and 4 for light off. Mar 26 22:01:14 likewise: hi Mar 26 22:01:16 RP: ping Mar 26 22:02:04 rwhitby: pong Mar 26 22:02:29 RP: did you happen to see my bitbake INHERIT_append_nslu2 problem last night? Mar 26 22:02:38 That little change seemed to fix it. Mar 26 22:02:44 rwhitby: I'm afraid I didn't Mar 26 22:03:26 RP: http://logs.nslu2-linux.org/livelogs/oe.txt at Mar 26 12:24:59 Mar 26 22:03:53 haven't had a chance to test on anything other than bitbake 1.6 yet. Mar 26 22:04:20 rwhitby: I think the INHERIT variable is a little strange in that you can just set it several times. I doubt it honors overrides Mar 26 22:05:43 Good night. Mar 26 22:06:34 RP: that line is in conf/distro/mokoslug.conf - in general what is the preferred way to set up machine-specific image creation whilst still allowing a distro to be used with any machine? Mar 26 22:07:26 What's in nslu2-image.bbclass? Is this in .dev? Mar 26 22:07:30 yes Mar 26 22:07:43 code which creates an 8MB flashable image for an NSLU2. Mar 26 22:08:04 A rootfs post process command might be a better idea Mar 26 22:08:24 RP: nslu2-image.bbclass does exactly that, doesn't it? Mar 26 22:08:55 oh, sorry, I use an IMAGE_POSTPROCESS command Mar 26 22:09:09 (cause it needs the rootfs.jffs2 that image creates) Mar 26 22:09:53 I want to get this right, cause it's supposed to be the way that I remove nslu2 image creation out of slugos and make it generic for use with any distro. Mar 26 22:10:12 (e.g. make Angstrom flashable on nslu2) Mar 26 22:11:01 RP: how would I use a rootfs post process command, and why would that be different from the image post process command that I use now? Mar 26 22:11:20 Right, just needed to look at the file. You should declare the post process command in general but only append it to the image creation process in the nslu2 case Mar 26 22:12:01 The extra image depends then become nslu2 specific too Mar 26 22:13:20 !oebug 2046 Mar 26 22:13:21 * * Bug 2046, Status: RESOLVED (INVALID), Created: 2007-03-26 07:13 Mar 26 22:13:22 * * slapinid(AT)gmail.com: glibc-2.5 build problem Mar 26 22:13:23 * * http://bugs.openembedded.org/show_bug.cgi?id=2046 Mar 26 22:13:41 hrw|gone: added the resolution thingy, good idea btw =) Mar 26 22:13:42 ah, so you're saying move the conditional from where I inherit nslu2-image.bbclass to inside nslu2-image.bbclass, so that anyone can include nslu2-image.bbclass and it only get triggered if your machine is an nslu2? Mar 26 22:13:56 rwhitby: right Mar 26 22:14:14 re Mar 26 22:14:22 and then I just add inherit nslu2-image.bbclass unconditionally to any distro which is compatible? Mar 26 22:14:43 rwhitby: yes, or remove the file and just add the function to the end of image.bbclass Mar 26 22:15:05 or have nslu2.conf inherit nslu2.bbclass Mar 26 22:15:55 distro's shouldn't have to inherit things, it should just work, its a machine specific config problem Mar 26 22:16:11 RP; the thing is that I wasn't sure whether it would be compatible with all distros (i.e. the distro rootfs might be too big, which would cause an error). I guess if that is the case then having this error out is the right thing to do anyway. Mar 26 22:16:50 e.g. if building for generic ixp4xx this won't get called, so it won't error out anyway. (where generic ixp4xx may have larger rootfs capacity than nslu2) Mar 26 22:17:30 rwhitby: Erroring out sounds appropriate Mar 26 22:18:14 RP: I will shoot for adding this inherit to nslu2{le,be}.conf, after checking that all current compatible distros work with it. Mar 26 22:18:40 RP: thanks for your (as always) wise counsel. Mar 26 22:19:49 rwhitby: np :) Mar 26 22:39:28 Hi, all! Do anybody knows of any on-device test suite to aid QA? Mar 26 22:40:39 something which could be automated to write testcases and unit tests for all subsystems? Mar 26 22:41:50 not about to auto-click something, but at least to test that apps compiled properly and do not dump core at start, have no unresolved dynamic symbols, broken symlinks, etc.? Mar 26 23:05:41 <__kamera__> i built a kernel image for a arm based board, the output created was a file called linux-kernel-2.6.20.3-arm i'm trying to load it on the board using uboot/tftp any idea if I have to do mkimage on it? Mar 27 00:06:47 03nail 07org.oe.dev * r032f0954... 10/ (5 files in 3 dirs): foonas: Update foonas distro Mar 27 00:06:58 03nail 07org.oe.dev * r7afb35cd... 10/ (1 conf/machine/storcenter.conf): storcenter: Make storcenter work with foonas Mar 27 00:07:02 03nail 07org.oe.dev * r6a73c5c1... 10/ (4 files in 2 dirs): update n2100, lsppchg and lspphcd images Mar 27 00:18:51 NAiL: FYI, I'm splitting out sysconf from turnup, so that sysconf can be used on other distros. Mar 27 00:19:10 v8jlene thanks for the ldd tip Mar 27 00:19:51 rwhitby: ok, I'll look into it :) Mar 27 00:20:24 foonas actually includes sysconf/turnup stuff, but it doesn't use it Mar 27 00:20:35 Haven't gotten that far yet Mar 27 00:21:43 I'm separating sysconf from turnup, cause turnup should be replaced by altboot. Mar 27 00:21:56 (but sysconf is generally useful, even without turnup/altboot) Mar 27 00:30:47 rwhitby: yeah, if there is a sysconf partition available Mar 27 00:33:43 There's room for a sysconf-partition on the ts101, and I think there is on the storcenter too Mar 27 00:34:01 not sure about them linkstations yet, will see when I get one Mar 27 01:10:01 having a play with aufs and unionfs.. am i blind or is there no docs on the wiki? Mar 27 02:21:07 03rwhitby 07org.oe.dev * r2800af4a... 10/ (1 classes/nslu2-image.bbclass): nslu2-image.bbclass: Made operation conditional upon an nslu2 override. Mar 27 02:21:15 03rwhitby 07org.oe.dev * r2e622ef1... 10/ (1 conf/machine/nslu2be.conf conf/machine/nslu2le.conf): nslu2[bl]e.conf: Inherit NSLU2 image creation class Mar 27 02:21:22 03rwhitby 07org.oe.dev * r85b8018e... 10/ (3 files in 3 dirs): unslung-kernel, unslung-image: Changed to use the generic nslu2_pack_image class. Mar 27 02:21:29 03rwhitby 07org.oe.dev * rf2ee250b... 10/ (1 packages/images/unslung-image.bb): unslung-image: Bump the PR Mar 27 02:21:38 03rwhitby 07org.oe.dev * rea9eb1ab... 10/ (1 packages/nslu2-binary-only/unslung-rootfs/unsling): unslung-rootfs: Removed cpio and findutils stuff from unsling **** ENDING LOGGING AT Tue Mar 27 02:59:56 2007