**** BEGIN LOGGING AT Mon Oct 12 02:59:56 2009 Oct 12 07:06:31 good morning Oct 12 07:25:32 I'm have moved from stable to dev, and so far I'm having trouble building anything :). Today it's coreutils that fails. Oct 12 07:28:10 tasslehoff: that is the nature of dev, it's a very fluid environment, you can help fix or wait a little while and see if someone commits a fix but dev will never promise to build out of the box unlike stable. The price you pay for bleeding edge packages. Oct 12 07:28:51 DJW|Home, yup, I'm fine with that :) Oct 12 07:29:12 tasslehoff: that said, what's the error ;-) Oct 12 07:31:37 DJW|Home, too many levels of symbolic links, so actually I think I'm to blame. I stand in ~/src/oe and not /data/src/oe :) Oct 12 07:35:55 tasslehoff: you should know better than to have anything symbolic in your OE path ;-) Oct 12 07:41:09 DJWillis, I actually do know better, but I did some reorg of my code tree on friday, and accidents did happen :) Oct 12 07:44:32 tasslehoff: ;-) Oct 12 07:45:33 * DJWillis cheats with a bound mount *cough* evil *cough* ;-) Oct 12 07:48:32 :) Oct 12 08:02:12 anyone know a command line tool for setting up ip configuration? Oct 12 08:05:51 jovox: ifconfig? Oct 12 08:06:18 and maybe route depending on what you want to do Oct 12 08:14:11 jovox: there's also "ip" Oct 12 08:17:01 Elleo, but ifconfig doesnt write to /etc/network/interfaces (which I want) Oct 12 08:17:28 jovox: afaik only vim/nano/mcedit/... write to /etc/network/interfaces:) Oct 12 08:18:20 if you want something auto-magical/clik-friendly, you probably want network manager or connman Oct 12 08:18:36 (although I'd think it doesn't touch /etc/network/interfaces) Oct 12 08:18:38 morging Oct 12 08:18:48 morgen Oct 12 08:19:10 what I'm doing is that I'm adding a small web ui for configuring the network settings Oct 12 08:21:22 jovox: hm, I'm not aware of any tool that'd produce /etc/network/interfaces, but that doesn't seem to difficult to do on your own Oct 12 08:29:05 zub, oki, I'll write a simple script for parsing/writing it if there isn't already a tool available. Oct 12 08:29:52 also, consider if you want to go that (/etc/network/*) way. This is used by ifup/ifdown, which are just scripts that use ifconfig/route/... (or maybe ip?) to do the actual work Oct 12 08:30:16 maybe for your usecase it might be better to use ifconfig/route/ip... directly Oct 12 08:31:49 yeah.. might be easier Oct 12 08:31:49 (provided your tool has exclusive control of the network settings - i.e. nothing else trying to use ifup/ifdown :) Oct 12 08:40:18 i am currently trying to add things to the EXTRA_OECONF variable; since the variable is set in an included file, i tried to add things with +=; doesnt work; do i have to move the variable from the .inc into the actual recipe, or is there another way to append additional parts ? Oct 12 08:42:14 SRY, solved, was missing a space ... Oct 12 08:51:31 god morning Oct 12 08:54:03 morning Oct 12 08:54:14 hi mckoan Oct 12 08:59:16 hi Oct 12 08:59:26 is there a "right" approach to generate self signed ssl certs via angstrom; openssl seems to lack openssl.cnf and therefore cant process the commands (?) Oct 12 08:59:50 anyone seeing errors building gconf due to polkit-policy-file-validate not being available? Oct 12 09:00:02 anybody knows Cliff's nick? Oct 12 09:00:15 mckoan: cliff brake? that would be cbrake. Oct 12 09:00:36 if some other cliff, you will have to be more specific :-} Oct 12 09:00:51 woglinde: good morning Oct 12 09:00:58 pb_: thx, I needed to add a ssh key for push access Oct 12 09:05:13 morning mckoan Oct 12 09:08:38 gremlin[it]: hi Oct 12 09:09:04 good morning Oct 12 09:09:25 * mckoan has a new patch for wxwidgets, because they don't build anymore Oct 12 09:11:31 Hi ! Oct 12 09:11:41 Which recipe contains "ldd" ? Oct 12 09:11:46 03Jeremy Lainé  07org.openembedded.dev * rbb4cfd56da 10openembedded.git/recipes/gnome/gconf_2.26.2.bb: Oct 12 09:11:46 gconf-2.26.2: don't fail if polkit-policy-file-validate is missing Oct 12 09:11:46 * Do not fail configure step if polkit-policy-file-validate is not Oct 12 09:11:46 present on the build machine. This tool is only used when doing Oct 12 09:11:46 "make distcheck", so we just replace it by "true". Oct 12 09:42:54 Which recipe provides ldd ? Oct 12 09:43:46 libc Oct 12 09:44:59 woglinde: glibc ? Oct 12 09:47:12 glibc, uclibc, eglibc Oct 12 09:47:18 whichever you use Oct 12 09:49:53 where does ipkg/opkg store the list of packages that need to be configured? Oct 12 09:50:30 my postinst only works if I manually remove/install a package. when preinstalled in an image, the postinst appears not to be done. any ideas on how to troubleshoot this? Oct 12 10:14:05 cbrake: hi Oct 12 10:31:26 bye Oct 12 10:53:31 mckoan: hello Oct 12 11:16:54 re Oct 12 11:29:24 morning all Oct 12 11:29:50 morning RP Oct 12 11:31:08 hi rp Oct 12 11:36:06 koobe: see the status file, /var/lib/opkg/status or some such Oct 12 11:38:36 pb__: thanks, but there's no such file. I just found that my postinst fails due to missing adduser, I wonder which package would provide a (native?) adduser. Oct 12 11:42:27 which package calls adduser? Oct 12 11:43:03 woglinde: this is my own package Oct 12 11:43:03 hi XorA Oct 12 11:43:40 hey gremlin[it] Oct 12 11:45:51 what is an offline root mode, and how does it determine which postinst configures to run and which to skip? Oct 12 11:46:18 koobe: it runs all of them Oct 12 11:46:31 koobe: but OE puts an error evit at top of ones not to run Oct 12 11:47:44 XorA: how does it determine which can be run ? Oct 12 11:48:05 koobe: hardcoded into recipes Oct 12 11:48:58 gm Oct 12 11:49:03 yo Crofton|work Oct 12 11:49:09 hi Crofton|work Oct 12 11:49:10 XorA: do you know of a good example to look at? Oct 12 11:49:28 florian: good morning Oct 12 11:49:40 hey pb__ Oct 12 11:49:54 koobe: update-rc.bbclass I think Oct 12 11:54:02 XorA: do you mean that postinst simply exits with error if D is not set? that will delay the postinst until the first boot? Oct 12 11:54:10 koobe: yes Oct 12 11:54:39 XorA: excellent, thank you. Oct 12 11:54:50 What does this line realy mean ? ./Cylinder: /lib/libgcc_s.so.1: version `GCC_4.1.0' not found (required by ./Cylinder) Oct 12 11:54:52 koobe: just copy that snippet to your recipe Oct 12 11:55:26 sgh: pretty much what it says: your version of libgcc doesn't match what the binary was compiled to expect. Oct 12 11:55:36 most likely, you are trying to run with a libgcc that is too old Oct 12 11:55:57 pb__: strange thing is that OE just compiled it. Oct 12 11:56:22 compiled what, libgcc? Oct 12 11:56:37 pb__: no. Vtk Oct 12 11:56:49 pb__: it is a recipe I did myself. Oct 12 11:56:55 and is it compiling against the same libgcc as you are trying to run with? Oct 12 11:57:03 pb__: yes Oct 12 11:57:36 pb__: are there some flags that oe uses that could produce this error ?? Oct 12 11:57:43 no Oct 12 11:57:57 if your versions of libgcc really are identical, this is a bug in either ld or ld.so Oct 12 11:58:01 morning Oct 12 11:58:03 pb__: how about the hashing-alogorithm .... Oct 12 11:58:30 sgh: what about it? Oct 12 11:58:44 Where can I find documentation about busybox ip command? I found the man page but that doesnt give much information about it Oct 12 11:58:58 jovox: it's basically compatible with the "real" ip command from iproute2 Oct 12 11:59:20 pb__: maybe a missmatch in hash could cause it Oct 12 11:59:36 pb__ ok, thanks Oct 12 11:59:54 pb__: do you know in which recipe ldd is contained Oct 12 12:00:02 sgh: if it did, that would fall into the category of "a bug in either ld or ld.so" Oct 12 12:00:14 that's certainly possible, but it doesn't seem all that likely. Oct 12 12:00:18 ldd is in glibc Oct 12 12:00:34 pb__: ok Oct 12 12:00:39 pb__: thanks. Oct 12 12:00:59 you could start by inspecting your libgcc.so to see whether it really does contain that version of that symbol. Oct 12 12:01:16 if yes, this is a ld.so bug and the way to proceed would be to run with LD_DEBUG=all Oct 12 12:01:32 if no, this is an ld bug and you would just have to debug it on the host as usual Oct 12 12:01:39 jovox busybox ip has not all functions of real ip Oct 12 12:01:52 but setting route and ip should work Oct 12 12:01:58 though, before doing that, it would be worth double-checking (e.g. with md5sum) to make sure that your two copies of libgcc.so really are identical. Oct 12 12:02:27 I have never heard of anybody else encountering such a bug and I suspect the most likely cause is that you are not running with the version of libgcc that you think you are. Oct 12 12:03:29 pb__: yes maybe ... I will check the actual linker-command use for linking. Oct 12 12:09:40 jo mickeyl Oct 12 12:10:06 hi woglinde Oct 12 12:10:27 hey mickey|office Oct 12 12:10:41 safely back in Frankfurt then? Oct 12 12:11:30 XorA: hi. yeah, right into the office Oct 12 12:11:43 mickey|office: same here Oct 12 12:16:27 soo tirde Oct 12 12:16:29 err Oct 12 12:16:31 hm right Oct 12 12:16:31 tired as well Oct 12 12:16:36 how was the meeting? Oct 12 12:17:44 productive Oct 12 12:18:35 now I can ask for membership? Oct 12 12:18:44 mickey|office, we will have a summery about the meeting ? Oct 12 12:18:56 germlin :) Oct 12 12:21:25 cbrake: ping Oct 12 12:21:50 mckoan: yes Oct 12 12:22:12 cbrake: sorry for bothering you, did you receive my email? Oct 12 12:23:21 mckoan: ahh, yes -- I don't check my gmail account that often as its just list mail -- adding now Oct 12 12:24:01 cbrake: thank you very much. please add your 'normal' email too ;-) Oct 12 12:24:41 gremlin[it]: for sure, we have a full fledged protocol that's going to be sent this week Oct 12 12:26:32 woglinde: wait until the details of internet voting are published, we cant carry out the votes until this is done Oct 12 12:29:25 mickey|office, good Oct 12 12:34:34 morning Oct 12 12:36:43 hi jay7 Oct 12 12:43:40 if I do : git push git@git.openembedded.org:openembedded I get the following message : Enter passphrase for key '/home/koan/.ssh/id_dsa', what could be the problem? Oct 12 12:44:14 mckoan: are you sure that scheme is git but not ssh? Oct 12 12:44:32 check your git config Oct 12 12:45:08 git config remote.origin.url Oct 12 12:45:34 Jay7: git://git.openembedded.org/openembedded Oct 12 12:45:39 hm.. Oct 12 12:45:56 can you push via git@? Oct 12 12:46:04 mckoan: you set a password on your ssh key Oct 12 12:46:27 XorA: shit, i don't remember Oct 12 12:47:01 hm.. but how ssh is used here? Oct 12 12:47:08 git:// Oct 12 12:47:56 mckoan: dont worry, it'll only take a few years to brute force :-) Oct 12 12:48:47 XorA: I could regenerate a new key Oct 12 13:10:03 any news on OE e.V. application yet? Oct 12 13:10:20 mickey|office, meeting was successful? Oct 12 13:17:17 dcordes-kais: absolutely Oct 12 13:24:44 03Marco Cavallini  07org.openembedded.dev * rd5e3e74b1b 10openembedded.git/recipes/wxwidgets/wxwidgets-2.8.10/gsockgtk-cpp-wx2810.diff: Oct 12 13:24:44 Added recipes/wxwidgets/wxwidgets-2.8.10/gsockgtk-cpp-wx2810.diff Oct 12 13:24:44 without this patch wxwidgets do not build using latest dependecies Oct 12 13:24:44 for details see http://trac.wxwidgets.org/ticket/10883 Oct 12 13:24:53 03Marco Cavallini  07org.openembedded.dev * r61efe54629 10openembedded.git/: Merge branch 'org.openembedded.dev' of git://git.openembedded.org/openembedded into org.openembedded.dev Oct 12 13:25:14 re Oct 12 13:29:32 03Marco Cavallini  07org.openembedded.dev * r48d1a89604 10openembedded.git/recipes/wxwidgets/wxwidgets_2.8.10.bb: Oct 12 13:29:32 Modified recipes/wxwidgets/wxwidgets_2.8.10.bb due to build failure Oct 12 13:29:32 This patch includes a new patch coming from recipes/wxwidgets/wxwidgets-2.8.10/gsockgtk-cpp-wx2810.diff Oct 12 13:29:32 Recipe version bumped Oct 12 13:30:38 hms Oct 12 13:30:49 * woglinde waits for koen rejectiong Oct 12 13:31:00 Marco are you here? Oct 12 13:31:19 woglinde: yes, I was wondering if everything is ok Oct 12 13:31:52 commit message are not oe styled Oct 12 13:32:48 if koen is in bad mood he will revert the commits Oct 12 13:32:49 http://wiki.openembedded.net/index.php/Commit_log_example Oct 12 13:33:25 I thought I respected it Oct 12 13:33:49 BTW I saw patched worst commented ;-( Oct 12 13:34:24 maybe the problem is the empty line? Oct 12 13:34:46 nope Oct 12 13:34:56 recipename should be the first Oct 12 13:35:09 wxwidgets: blalsalsdllsddfs Oct 12 13:35:19 * XorA wonders what is so damn hard about that example, no-one gets it Oct 12 13:35:20 then *modified bla because foo Oct 12 13:35:28 * bumped PR Oct 12 13:35:44 xora hm style Oct 12 13:35:55 woglinde: IMHO Commit_log_example in unclear Oct 12 13:36:04 s/in/is Oct 12 13:36:06 maybee first line should be bold Oct 12 13:36:15 mckoan whats unclear? Oct 12 13:36:43 args Oct 12 13:38:03 Xora maybee no ones read the text after the example Oct 12 13:38:32 or the text above Oct 12 13:38:40 woglinde: I politely think that is not well explained Oct 12 13:38:43 what should be first line Oct 12 13:38:54 mckoan what exactly is unclear? Oct 12 13:41:01 Quick question, is anyone actively using upstart as a replacement for sysvinit in OE (I know the recipe is there, been playing with it) but it seems incomplete. Oct 12 13:41:27 mckoan its more clear now? please reload Oct 12 13:41:41 djwillis webos Oct 12 13:42:04 but unfornatly the dont give something back Oct 12 13:42:54 hm I see I mostly forget the empty line btw. summary and long descriotion too Oct 12 13:42:59 woglinde: yep, looked into that, hit a dead end and a 3MB patch of garbage ;-). Oh well, i'll hack on it when I have some time. Oct 12 13:44:17 nice Oct 12 13:44:19 First word should be recipe name, normaly without the version. << How about >> The first word should be recipe name, normally without the version then a colon and space followed by a short (no more than a few works) description of the change. Oct 12 13:44:26 I would like to see this in oe Oct 12 13:45:01 djwillis you have the power to edit it Oct 12 13:45:04 woglinde: it looks like there is not a lot wrong with it, just needs update alts and an upgrade to the latest 0.3 code (0.6 code needs a dbus upgrade in OE). Oct 12 13:45:14 my english is a little bit limited at style Oct 12 13:45:29 woglinde: do I? I thought the OE pages had been protected :-o, sorry, I did not mean to insult. Oct 12 13:45:33 woglinde: the problem is how to change commit comment now :-( Oct 12 13:45:45 mckoan as I said Oct 12 13:46:02 pray koen that koen dont revert it Oct 12 13:46:09 and make it better next time Oct 12 13:46:39 woglinde: ok thanks Oct 12 13:47:39 * XorA buys koen a ticket to Italy :-) Oct 12 13:50:19 probably time for (someone other than koen) to post the commit message format to the dev list .... Oct 12 13:50:30 mckoan, will this fix wswidgets on the beagle? Oct 12 13:50:37 I am building now Oct 12 13:51:08 Crofton|work: I don't know which error are you facing on Beagle Oct 12 13:51:14 ok Oct 12 13:51:18 I am checking now Oct 12 13:51:32 2.9 did build for me, but I had to mess with the recipes some Oct 12 13:51:48 need to sort it out and commit that as DEF PREF -1 Oct 12 13:52:40 hi Oct 12 13:52:45 hi,all Oct 12 13:53:10 hi crofton Oct 12 13:53:12 I'm facing a matchbox-window-manager, could one give me a hand? Oct 12 13:53:14 hi derekhe Oct 12 13:53:21 ^_^ Oct 12 13:53:30 derekhe there was patches for mb2 Oct 12 13:53:40 I pulled the latest version Oct 12 13:54:29 And use bitbake x11-gpe-image Oct 12 13:55:01 The whole process ran well expect a small source code compile issue in ImageMagick package Oct 12 13:55:07 I fixed it and it's OK. Oct 12 13:55:21 Then I downloaded the image into my MINI2440 board and started Oct 12 13:55:28 I saw Angstrom logo Oct 12 13:55:36 and some first-run configuration Oct 12 13:55:53 Finally, it give a console not GUI stuff. Oct 12 13:56:05 I logged using serial port as root Oct 12 13:56:17 and I just run : Xorg & Oct 12 13:56:42 X runs with some configuration error, but it still ok. I saw a blank screen. Oct 12 13:56:51 Then I run: export DISPLAY=:0.0 Oct 12 13:57:12 And then, dbus-launch matchbox-windows-manager Oct 12 13:57:48 And X server failed Oct 12 13:57:58 Backtrace: Fatal server error: Caught signal 11. Server aborting Oct 12 13:58:16 And matchbox giveme: matchbox-wm: can't open display! check your DISPLAY variable. Oct 12 13:58:26 I tried again and again, no success. Oct 12 13:58:53 So if there any ideas about this? Oct 12 13:59:24 woglinde: I had a quick edit of that commit log page, do you think it is better or worse now? Oct 12 14:00:01 * DJWillis resists the temptation to add some guidelines on what good commit messages contain (as well as format) ;-) Oct 12 14:00:14 oh Oct 12 14:00:21 I edited it too Oct 12 14:01:11 hm Oct 12 14:01:33 I hope dont overwritten yours Oct 12 14:01:45 woglinde: ahhh, no worries, yep, just noticed I lost ;-). No probs. Oct 12 14:02:02 sorry Oct 12 14:02:58 I'll try again ;-), with your updates in there. Oct 12 14:03:07 thanks Oct 12 14:03:37 Any one has any ideas about my questions? Thanks!! Oct 12 14:04:34 derekhe: it sounds like your xserver is broken but it is hard to say any more than that from what you have mentioned. Oct 12 14:05:01 OK. I just tried run x-terminal-emulator Oct 12 14:05:13 And I got a console and I could input text into it Oct 12 14:05:14 hm if you started X server Oct 12 14:05:21 try DISPLAY=:0.0 foobar Oct 12 14:05:22 So X server is OK Oct 12 14:05:26 form the console Oct 12 14:05:34 well, if the xserver is crashing with signal 11, it is clearly not ok Oct 12 14:05:47 dbus-launcher crap should die Oct 12 14:06:03 try DISPLAY=:0.0 foobar ? Oct 12 14:06:15 export DISPLAY=:0.0 foobar??? Oct 12 14:06:35 nope either DISPLAY=:0.0 foobar Oct 12 14:06:41 or export DISPLAY=:0.0 Oct 12 14:06:43 foobar Oct 12 14:06:49 ok, Let me check Oct 12 14:07:49 This time, I run Xorg& and export DISPLAY=:0.0 and then dbus-launch x-window-manager, X crashes Oct 12 14:08:00 Backtrace: 0: Xorg(xorg_backtrace+0x28) [0xeb2c0] Fatal server error: Caught signal 11. Server aborting Oct 12 14:08:16 I have seen the 0xeb2c0 before. Oct 12 14:08:56 whats happen when you do xterm only Oct 12 14:09:03 whithout dbus crap Oct 12 14:09:07 Let me check, wait a moment Oct 12 14:09:20 Thanks for your patient! Oct 12 14:10:17 Here I only have x-terminal-emulator and it works fine. Oct 12 14:10:40 A terminal opened and a mouse cursor is shown. Oct 12 14:10:40 ah so either dbus suckz Oct 12 14:10:49 or x-window-manager Oct 12 14:10:55 Maybe is dbus Oct 12 14:11:04 try starting x-window-manager without dbus Oct 12 14:11:08 ok Oct 12 14:11:13 or start xterm with dbus Oct 12 14:12:19 A lot of d-bus issue Oct 12 14:12:29 But I could see matchbox gui Oct 12 14:12:31 this time Oct 12 14:12:48 I could even choose the icons Oct 12 14:12:54 Seem it is running OK. Oct 12 14:12:59 *g* Oct 12 14:13:10 so forget about dbus stuff Oct 12 14:13:29 OK. Then how about the GConf-WARNING **: Client failed to connect to the D-BUS daemon: dbus-launch failed to autolaunch D-Bus session: Failed to waitpid() for babysitter intermediate process: No child processes Oct 12 14:13:37 GConf Error: No D-BUS daemon running Oct 12 14:13:58 should be clear? Oct 12 14:14:05 dbus was not started? Oct 12 14:14:15 D-Bus, that just loves to cause my pain in my X images ;-) Oct 12 14:14:23 I don't know ~ Oct 12 14:14:28 type ps -ef? Oct 12 14:14:32 OK Oct 12 14:14:35 if there is no dbus Oct 12 14:14:42 then dbus was not started Oct 12 14:14:49 or crashed at start Oct 12 14:14:51 OK Oct 12 14:14:57 Let me check Oct 12 14:15:28 I could see couples of dbus-deamon process Oct 12 14:15:59 If I'm running without dbus-launcher , any problems? Oct 12 14:18:38 I think there will be no problems Oct 12 14:19:03 OK. That's really confuse me. Oct 12 14:19:15 OK. Now it's working well by start gpe-confd Oct 12 14:19:25 lol Oct 12 14:19:40 I could see the good GUI stuff. Thanks for your help! Oct 12 14:25:59 no prob Oct 12 14:57:15 Hello, I implemented sha256 sum check in opkg is anyone interested in the patch ? Should I post it on googlecode ? Is googlecode repo still alive ? Oct 12 15:08:58 pix_: OE still builds from the GoogleCode SVN but as for OPKG being alive, hmmm. I guess you could try it's mailing list and maybe CC oe-devel? Oct 12 15:11:26 I have several projects involving opkg such as adding x509 support to the package list verification, I will send a mail, I need to do some work before. Oct 12 15:12:05 pix_: it sounds like you have been doing a bit of work on it, do send that mail as it would be a crime to loose patches and work like that. Oct 12 15:14:14 I just started to work on it last few days, I plan to start x509 support in a day or so ;-) Not so much work for the moment besides sha256 support. Oct 12 15:43:45 mckoan, fix worked for beagleboard, thanks Oct 12 16:00:01 Crofton|work: glad to be helpful :-D Oct 12 16:01:52 have a nice rest of the day Oct 12 16:04:31 bye mckoan Oct 12 16:33:04 * * OE Bug 5328 has been created by van.bourgognie(AT)gmail.com Oct 12 16:33:06 * * gconf do_configure fails due to missing polkit-policy-file-validate Oct 12 16:33:08 * * http://bugs.openembedded.org/show_bug.cgi?id=5328 Oct 12 17:02:07 The FOSDEM call for stands is up Oct 12 17:02:15 http://www.fosdem.org/2010/ Oct 12 17:05:45 morning all Oct 12 18:08:49 03Otavio Salvador  07org.openembedded.dev * r925dea6086 10openembedded.git/recipes/xorg-app/xvidtune_1.0.1.bb: Oct 12 18:08:49 xvidtune: include app-defaults otherwise it fails to run Oct 12 18:08:49 Signed-off-by: Otavio Salvador Oct 12 18:12:39 Hello Oct 12 18:12:53 I'd like to know when the next stable branch is planned Oct 12 18:13:37 I believe that currently .dev has a lot of nice work that would be nice to have as basis for a new stable release and allow for more widely usage of ig Oct 12 18:13:40 it Oct 12 18:15:47 otavio: I suspect (at a guess) the new year Oct 12 18:16:46 otavio, any features in particular? Oct 12 18:17:07 otavio: that said, a lot of stuff is backported to stable once the feature is nice and stable in dev. Oct 12 18:17:19 Crofton|work: well; It has received a lot of cleanup Oct 12 18:17:41 anything that can be proposed for stable? Oct 12 18:17:43 Crofton|work: and specially the INC_PR usage makes our life much easier Oct 12 18:18:14 Crofton|work: We had to backport a lot of stuff to get a mips board working recently, for example Oct 12 18:18:28 Crofton|work: and all that was done in .dev already Oct 12 18:18:35 hey otavio Oct 12 18:18:40 I think that can be backported on a case by case basis, since I do not think it needs any support from bitbake etc Oct 12 18:18:42 Tartarus: hello :-) Oct 12 18:18:43 speaking of MIPS, did you guys build eglibc 2.10? Oct 12 18:18:46 or just glibc? Oct 12 18:18:58 eglibc failed, but i didn't have time to dig Oct 12 18:19:20 Tartarus: nowadays glibc; I didn't look now difficult would be to move to eglibc yet Oct 12 18:19:26 k Oct 12 18:19:30 eglibc 2.9 was fine Oct 12 18:19:34 but 2.10 broke Oct 12 18:19:37 (in .dev) Oct 12 18:19:37 Crofton|work: well; yes, I guess so Oct 12 18:19:47 Tartarus: what I need to do to go to eglibc? Oct 12 18:19:58 And I've got some ORBit2 mips-common site stuff, once i get a chance to make sure it actually works, heh Oct 12 18:20:08 otavio, LIBC=eglibc instead of LIBC=glibc Oct 12 18:20:19 Tartarus: heh too .dev specific Oct 12 18:20:29 Tartarus: I'll need to check how to port it to stable Oct 12 18:20:47 er? Oct 12 18:20:48 Tartarus: our mips board is using uclibc atm Oct 12 18:20:54 LIBC= should be in stable, heh Oct 12 18:20:59 and, ah Oct 12 18:21:08 Tartarus: really? Will look at it Oct 12 18:51:55 hi all Oct 12 18:52:10 someone have try angstrom/uclibc on atngw100 ? Oct 12 19:01:39 someone ? Oct 12 19:16:09 someone have try angstrom/uclibc on atngw100 ? Oct 12 19:16:23 gremlin[it]: there's no point in just asking the same question again and again. Oct 12 19:16:28 gremlin[it], apparently not Oct 12 19:16:36 if nobody answered, presumably it means the answer is "no" Oct 12 19:17:00 pb_ i reask cause a couple of persons come in ... Oct 12 19:17:30 btw : on stable blahch the result image is too big Oct 12 19:17:45 on .dev branch is incomplete (missing libintl.so) ... Oct 12 19:17:55 you might have better luck asking on the mailing list, or in in #angstrom Oct 12 19:18:08 you were missing libintl.so also? Oct 12 19:18:19 martinmeba, yes Oct 12 19:18:29 I had a similar problem that I am still fighting with Oct 12 19:18:40 I am using uclibc/avr32 Oct 12 19:18:49 i built a console-image Oct 12 19:18:59 martinmeba, same architecture Oct 12 19:19:11 i had to add gettext to the image to pull in the libintl.so dependency Oct 12 19:19:15 martinmeba, same image and angstrom distro Oct 12 19:19:19 then at first boot, i had to run ldconfig Oct 12 19:19:25 but that has caused a new problem Oct 12 19:19:35 the thing gets through init but I never get a login prompt now Oct 12 19:19:39 I am totally stumped Oct 12 19:19:46 happen me too Oct 12 19:19:54 without libintl.so, i get login and a mostly working system minus /proc and mount capability Oct 12 19:20:02 now it hangs right when it should allow me to log in Oct 12 19:20:05 I am totally lost here Oct 12 19:20:18 long long udev init and hang few step after ... Oct 12 19:20:34 [17179571.832000] eth0: link up (100/Full) Oct 12 19:20:34 INIT: version 2.86 booting Oct 12 19:20:34 Please wait: booting... Oct 12 19:20:34 Starting udev Oct 12 19:20:36 Oct 12 19:20:38 udevadm settle - timeout of 0 seconds reached, the event queue contains: Oct 12 19:20:40 Remounting root file system... Oct 12 19:20:42 root: mount: mount point /proc/bus/usb does not exist Oct 12 19:20:44 Oct 12 19:20:46 ALSA: Restoring mixer settings... Oct 12 19:20:48 Configuring network interfaces... /usr/sbin/alsactl: load_state:1608: No soundc. Oct 12 19:20:50 done. Oct 12 19:20:52 Starting portmap daemon: portmap. Oct 12 19:20:54 martinmeba, are using .dev branch right ? Oct 12 19:20:54 net.ipv4.conf.default.rp_filter = 1 Oct 12 19:20:56 net.ipv4.conf.all.rp_filter = 1 Oct 12 19:20:58 hwclock: can't open '/dev/misc/rtc': No such file or directory Oct 12 19:21:00 date: invalid date '%2m%2d%2H%2M2009' Oct 12 19:21:02 hwclock: can't open '/dev/misc/rtc': No such file or directory Oct 12 19:21:06 that is where it dies Oct 12 19:21:08 udev takes forever and then times out Oct 12 19:21:09 martinmeba, dont paste all !!! Oct 12 19:21:10 no Oct 12 19:21:12 stable Oct 12 19:21:14 oh, wait Oct 12 19:21:16 yes Oct 12 19:21:18 I guess that I am Oct 12 19:22:02 mhhhh strnage ... stable for me generate a .jffs2 image of 8.3Mbyte ... too big for my flash (8MByte) Oct 12 19:22:27 strnage -> strange Oct 12 19:22:27 so you are saying that with the stable branch everything is fine? Oct 12 19:22:43 otherwise Oct 12 19:22:48 gremlin[it], look in the testlab directory in deploy Oct 12 19:23:03 that should have some information what is inside the image and how large it is Oct 12 19:23:21 should be helpful for creating an image you can use Oct 12 19:24:23 thanks Crofton|work Oct 12 19:26:37 if you are not using angstrom, you may need to INHERIT += testlab Oct 12 19:27:14 Crofton|work, the directory is present ... so testlab seem anabled Oct 12 19:27:25 ok Oct 12 19:28:45 florian: good evening Oct 12 19:31:38 gremlin[it], is the information helpful? Oct 12 19:31:50 re Oct 12 19:32:23 Crofton|work, no ... only con config libintl.so isn't installed on .dev bramch ... but i continue investigate Oct 12 19:32:27 gm Oct 12 19:33:34 gremlin[it]: proxy-libintl provided libintl.so too if you dont need so much of gettext Oct 12 19:33:36 gremlin[it]: so you are saying that the stable branch builds and boots correctly except that it is too big for your requirement of fitting in the 8MB onboard flash? Oct 12 19:34:25 khem, yes i know ... i try to make a dumb image as console-image + proxy-libintl Oct 12 19:34:45 martinmeba, about yes ... except i never boot with .stable cause is too big for me ... Oct 12 19:35:01 martinmeba, so i think it should boot ... Oct 12 19:35:19 gremlin[it]: this combo booted ok for me on arm Oct 12 19:35:26 gremlin[it]: ok - I will try that - I have no space requirements - I just want it to boot correctly Oct 12 19:35:42 khem, ok let's try ... :) Oct 12 19:36:19 ping XorA|gone Oct 12 20:02:44 martinmeba, ok .dev branch plus proxy-libintl don't work ... used spend long long time to start and never got login ..... Oct 12 20:03:16 gremlin[it]: I just don't understand what is going on there Oct 12 20:03:32 when you read through the rcS script, there is nothing left in it after it hangs Oct 12 20:04:19 at now last line is "Configuring ppp" ... Oct 12 20:05:42 martinmeba: that kind of thing usually means that your inittab is wrong, either not starting getty at all or starting it on the wrong console. Oct 12 20:06:00 interesting Oct 12 20:07:10 ok Oct 12 20:07:17 pb_, but so why console come up if linintls.so is missing ??? Oct 12 20:07:27 pb_: at the bottom of inittab, there is a fall-through sequence Oct 12 20:07:38 it seems that when libintl.so is not there, it falls-through and brings up the console Oct 12 20:08:02 when libintl.so is present, it boots through most items normally and falls on its face somewhere so never brings up getty Oct 12 20:08:04 oh, right, you get a login prompt with the same image minus libintl? Oct 12 20:08:10 yeah Oct 12 20:08:20 in that case, it must be getting jammed somewhere earlier on. you might need to instrument the scripting to find out exactly where. Oct 12 20:08:28 well Oct 12 20:08:36 it goes through all of the rcS scripts Oct 12 20:08:39 that is the funny part Oct 12 20:08:46 do you get as far as "entering runlevel 2"? Oct 12 20:08:46 it never brings up the console at the end Oct 12 20:08:54 (or 5, or any other number) Oct 12 20:09:01 let me look Oct 12 20:09:47 it never reports that Oct 12 20:10:02 in that case, it sounds like /etc/init.d/rc is never returning Oct 12 20:10:11 might be worth instrumenting that script to find out what is happening Oct 12 20:10:20 but as to your point - the fallthrough specifically launches getty on a specific terminal - there is nothing talking about that for the normal case in the inittab script Oct 12 20:10:42 nor does that happen in the final rcS script Oct 12 20:11:43 pb_: noob question - how do I instrument the boot? Oct 12 20:11:48 that does sound a bit fishy, but even if your inittab was missing the right getty call I would have expected you to see the runlevel change messages. Oct 12 20:12:15 martinmeba: the things that init calls are just shell scripts, you can edit them and insert strategic "echo" statements. Oct 12 20:12:25 I think /etc/init.d/rc and/or /etc/init.d/rcS would be the operative ones here Oct 12 20:12:33 plus any other scripts that might be named in /etc/inittab Oct 12 20:12:44 here is the fall-through: Oct 12 20:12:50 # Normally not reached, but fallthrough in case of emergency. Oct 12 20:12:50 z6:6:respawn:/sbin/sulogin Oct 12 20:12:50 S:2345:respawn:/sbin/getty 115200 ttyS0 Oct 12 20:13:22 ah, that is a misleading comment Oct 12 20:13:28 but this is for normal: Oct 12 20:13:28 # What to do in single-user mode. Oct 12 20:13:28 ~~:S:wait:/sbin/sulogin Oct 12 20:13:29 only the "z6" line is the fallthrough Oct 12 20:13:32 this seems incorrect Oct 12 20:13:46 the "S:2345" line that you pasted is the one that is used in normal operation Oct 12 20:14:21 if I move that up in the inittab file, will it execute that? Oct 12 20:14:28 yes, that /sbin/sulogin line probably is wrong, but unless you have requested single user mode it will be harmless. Oct 12 20:14:33 what does your "initdefault" line say? Oct 12 20:14:51 not much: Oct 12 20:14:52 # The default runlevel. Oct 12 20:14:52 id:5:initdefault: Oct 12 20:14:58 okay, the 5 is the important number Oct 12 20:15:20 so, you should be going to runlevel 5 (unless your kernel commandline says otherwise, which I assume it doesn't) Oct 12 20:16:19 in that case, you can ignore all lines where the second field (i.e. between the first and second : signs) isn't empty and doesn't contain a 5. Oct 12 20:16:24 hi .. i am having proxy issue with wget and but http_proxy="bla" doesnt seen to work for do_fetch in bitbake ?? Oct 12 20:16:27 is that true ? Oct 12 20:17:07 is there a proxy parameter for bitbake itself ? Oct 12 20:17:10 pb_: ok - so it is falling over in the rc5.d scripts then and I can start instrumenting there? Oct 12 20:17:27 well, if it doesn't even print "entering runlevel 5" then i think it is falling over before that. Oct 12 20:17:47 because I seem to be getting boot output from the rcS scripts all the way through Oct 12 20:18:05 the last file is the only one that i have not been able to track Oct 12 20:18:10 it does not seem to do much Oct 12 20:18:12 try adding something to the end of rcS itself to see if it ever completes Oct 12 20:18:20 the rcS script, that is, not the rcS.d directory Oct 12 20:18:58 what were the errors that you saw without libintl installed? that might give a clue as to where it is likely to be getting stuck. Oct 12 20:19:16 it can't mount anything Oct 12 20:19:20 it can't create /proc Oct 12 20:20:12 aaah Oct 12 20:20:15 may have found something Oct 12 20:20:31 pb_, basicallu without libintl.so only busybox work ... Oct 12 20:20:31 /etc/init.d/rc: eval: line 1: can't create /dev/tty0: No such device or address Oct 12 20:23:45 guys .. do you need to tell bitbake about a http proxy ? Oct 12 20:26:28 gremlin[it]: not entirely true - I got an entire python stack on it that 95% worked... it was the 5% not working that caused me to notice this problem... Oct 12 20:26:44 ahhh Oct 12 20:30:13 why is bitbake do_fetch using wget ignoring my proxy even ignoring /etc/wgetrc ?? Oct 12 20:30:18 my suspect is that a deamon is hang ... dbus or avahi Oct 12 20:31:14 yeah - that is what I am thinking looking at the rc5 directory Oct 12 20:31:17 that is the first thing to go Oct 12 20:31:21 i am going to start there Oct 12 20:33:59 martinmeba, i'm not able to find the messages i see on the /etc scripts ... so i suspect messages came from a binary ... Oct 12 20:40:35 re Oct 12 20:40:40 re Oct 12 20:40:47 hi woglinde Oct 12 20:41:03 hi woglinde Oct 12 20:41:18 woglinde: hello Oct 12 20:41:23 jo khem Oct 12 20:41:33 we will have uclibc 0.9.30.2 soon Oct 12 20:41:37 woglinde: I committed the patch other day did you try it out Oct 12 20:41:37 woglinde, khem is the armv4 eabi stuff ready to use? Oct 12 20:41:39 yes Oct 12 20:41:55 khem the patch seems to work Oct 12 20:41:59 I have bx in the libs Oct 12 20:42:00 woglinde: ok Oct 12 20:42:02 args no Oct 12 20:42:29 you might have to fix packages where it does not use gcc to link Oct 12 20:42:41 yes Oct 12 20:42:46 I will deal with it Oct 12 20:43:04 ok Oct 12 20:43:23 thesing: tools seems to work now woglinde knows more about target parts Oct 12 20:43:45 what do I need to do to test it? Oct 12 20:43:53 ah Oct 12 20:44:02 change some files for collie Oct 12 20:44:24 latest .dev and make sure you use gcc >= 4.4.1 Oct 12 20:44:34 use micro as distro Oct 12 20:44:37 which works Oct 12 20:44:42 yeah or minimal Oct 12 20:45:19 thesing I will pastebin my changes for simpad Oct 12 20:45:27 you have to do similar for collie Oct 12 20:49:07 thesing http://paste.debian.net/48892/ Oct 12 20:59:30 dumb question : we are sure udev work with uclibc ??? Oct 12 20:59:41 gremlin yes works Oct 12 20:59:45 atleast 124 Oct 12 21:00:07 udev_141-r11.3_avr32.ipk Oct 12 21:00:08 udev-utils_141-r11.3_avr32.ipk Oct 12 21:00:12 should work ? Oct 12 21:00:21 seemed to compile Oct 12 21:00:26 so it should work Oct 12 21:01:04 i try to boot in single user ... udev spend minutes ... Oct 12 21:01:12 hm Oct 12 21:01:27 you could try 124 Oct 12 21:01:41 gremlin[it]: I am having the same problem Oct 12 21:03:32 I get this error message: udevadm settle - timeout of 0 seconds reached, the event queue contains: Oct 12 21:05:44 i think al hand on S98configure ... Oct 12 21:06:41 i think alal hang on S98configure Oct 12 21:14:16 djwillis are you coming to oedem? Oct 12 21:14:31 martinmeba, if u want to test try to add to bootargs init='/sbin/getty 115200 ttyS0' in u-boot Oct 12 21:15:00 woglinde: I would like to but as my good lady has a broken leg at the moment it might be rather hard to disapear for a few days. Oct 12 21:15:07 oh Oct 12 21:15:09 okay Oct 12 21:15:50 woglinde: you going? I'll confess all being equal I would def. go. Oct 12 21:16:00 yes Oct 12 21:16:02 I will be there Oct 12 21:16:13 flying with hrw and robert from berlin Oct 12 21:17:23 woglinde: so there should be a good OE turnout then? Oct 12 21:17:33 yeah Oct 12 21:17:37 20 people or so Oct 12 21:17:41 pb knows more Oct 12 21:39:36 woglinde, DJWillis: Its about 20 people, yes Oct 12 21:41:39 he rp Oct 12 21:41:52 f.d.o is now restored Oct 13 01:53:28 gremlin[it]: This is my bootargs variable: root=/dev/nfs nfsroot=192.168.1.101:/srv/nfs ip=192.168.1.150::192.168.1.101:::eth0:off console=ttyS0,115200n8 init='/sbin/getty 115200 ttyS0' Oct 13 01:53:39 That yields the boot message: Oct 13 01:53:55 Failed to execute '/sbin/getty. Attempting defaults... Oct 13 02:01:05 i got rid of the quotes Oct 13 02:01:05 the console comes up! Oct 13 02:01:05 Thanks, gremlin[it]! Oct 13 02:16:28 Very strange though - Oct 13 02:16:33 no /proc filesystem Oct 13 02:16:58 yeah Oct 13 02:17:03 you don't want getty as your init :) Oct 13 02:17:08 you want an init as your init Oct 13 02:17:28 what happens when you don't pass an init= line? Oct 13 02:17:47 it hangs at the end of rcS Oct 13 02:17:54 console never comes up Oct 13 02:18:05 that means your inittab is probably not right Oct 13 02:18:09 check /etc/inittab Oct 13 02:18:16 yeah Oct 13 02:18:30 it seems to be on the boundary between rcS and runlevel 5 Oct 13 02:19:40 pb_ helped me track it down to there Oct 13 02:41:16 i think that inittab may be pointing off at nothing Oct 13 02:41:50 So i guess that this maps runlevel 5 to the rc script: l5:5:wait:/etc/init.d/rc 5 Oct 13 02:42:38 there are no rc scripts in that directory though **** ENDING LOGGING AT Tue Oct 13 02:59:56 2009