**** BEGIN LOGGING AT Wed Jul 05 02:59:56 2006 Jul 05 03:22:20 03lenehan 07org.oe.dev * r9fb62df8... 10/packages/conserver/ (5 files in 2 dirs): Jul 05 03:22:20 conserver: Add a proper init script for conserver and use a default file to Jul 05 03:22:20 set the port number so that it does not require an entry in /etc/services to Jul 05 03:22:20 run. Jul 05 03:27:53 03lenehan 07org.oe.dev * r69b99cad... 10/packages/quagga/ (4 files): Jul 05 03:27:53 quagga: Make the default files, which are used for local administrator Jul 05 03:27:53 settings, as configuration files. Jul 05 04:17:46 morning Jul 05 05:20:34 Morning CoreDump|home Jul 05 05:20:59 hi there hvontres Jul 05 05:27:20 03coredump 07org.oe.oz354x * r9ea144b9... 10/packages/altboot/ (altboot_1.0.6.bb altboot_1.0.7.bb): altboot: Update to 1.0.7 final Jul 05 05:27:24 03coredump 07org.oe.dev * r29e89051... 10/packages/altboot/ (6 files in 3 dirs): altboot: Update to 1.0.7 final Jul 05 05:28:50 hey, I see you put in a new altboot.. does that work with 2.6 poodle? Jul 05 05:33:59 yep Jul 05 05:34:03 ~seen RP Jul 05 05:34:07 rp is currently on #gpe #oe #openzaurus #handhelds.org, last said: '(when in headset mode)'. Jul 05 05:35:29 RP: I tried to use the current corgi configutation on poodle. When I try to load the modules, I get this error: Jul 05 05:35:30 insmod: cannot insert `/lib/modules/2.6.17/kernel/sound/soc/snd-soc-core.ko': Unknown symbol in module (-1): No such file or directory Jul 05 05:36:42 RP: Am I correct to asume that the machine specific stuff is in corgi.c? Jul 05 05:37:19 CoreDump|home: neato. I guess its time to do another pull..:) Jul 05 05:37:44 the -rc series from .dev had poodle support for quite some time now ;) Jul 05 05:38:14 CoreDump|home: Arrrgh- not the .dev branch...:) Jul 05 05:41:04 CoreDump|home:It's a shame about the sonund not working... other than that the 2.6 stuff looks great. Jul 05 05:41:55 indeed Jul 05 05:45:15 and it looks like we are really close too... half of the alsa module load ok Jul 05 05:48:23 CoreDump|home:too bad dual booting was kind of a bust... no suspend is a real showstopper.. Jul 05 05:48:44 you mean dualbooting w/ kexec? Jul 05 05:49:31 spitz+kexec=dual boot with 2.4 kernel :-) Jul 05 05:49:52 (would even be Sharp ROM-able if you were so inclined) Jul 05 05:50:33 morning Jul 05 05:50:44 CoreDump|home: yes. Tried to kexec and then altboot to sd. but it would freeze up on suspend...:( Jul 05 05:50:45 ~lart sharp for usb host in tosa Jul 05 05:50:46 * ibot hooks into a hydrant and hoses sharp down for usb host in tosa Jul 05 05:51:03 hvontres|home: :\ Jul 05 05:51:13 morning hrw Jul 05 05:51:24 hrw: Morning...not happy with your sharp/tosa :) Jul 05 05:51:55 CoreDump|home: use 'update-alternative' class! Jul 05 05:52:07 ;D Jul 05 05:52:47 * hrw has to rebuild altboot, then images Jul 05 05:53:31 hrw: I can't for fluxbox Jul 05 05:53:47 CoreDump|home: altboot? Jul 05 05:54:12 dunno actually as postinst works fine ;) Jul 05 05:54:45 CoreDump|home: u-a do the same and I think that it should be used ;) Jul 05 05:54:57 hrw@bitbake:~/devel/build/3541$ for machine in c7x0 spitz akita;do echo "MACHINE = \"$machine\"" > conf/auto.conf;echo "MACHINE = \"$machine\"";bitbake -cclean gpe-image opie-image bootstrap-image;bitbake gpe-image opie-image bootstrap-image ;done Jul 05 05:56:44 hmm... will switch it to bbrebuild next time - faster Jul 05 06:15:00 quote from sig: Jul 05 06:15:04 catholic god himself invented autotools just for amusement Jul 05 06:15:06 first there was the great flood, then the plague, now autotools Jul 05 06:16:00 well - x86 builds seem borked Jul 05 06:16:07 its nothing to do with my fedora 5 rig Jul 05 06:16:19 jkp_: ;( Jul 05 06:16:20 i installed debian in a brand new VM and went from there Jul 05 06:16:26 and i got the same errors Jul 05 06:16:56 hrw: its extremely frustrating...if i can just get a working combination for the toolchain! Jul 05 06:17:14 how can we resolve this issue? Jul 05 06:17:23 is there a bug tracker for OE? Jul 05 06:17:44 bugs.openembedded.org Jul 05 06:18:21 are there more devs than those that hang out in here? Jul 05 06:20:57 someone already filed it Jul 05 06:20:58 http://bugs.openembedded.org/show_bug.cgi?id=1000 Jul 05 06:25:44 ]right, i updated the bug, but im wanting to try and help out solving it Jul 05 06:25:55 is there anyone who might be able to spend some time on it with me later today? Jul 05 06:26:56 not me - too much work at work Jul 05 06:27:16 hrw: thanks Jul 05 06:27:27 ill be back on in a bit, ill shout again and see if anyone else is able Jul 05 06:27:41 hope im not pestering but this will be for the benefit of all in the end i ghuess Jul 05 06:27:48 gtgg Jul 05 06:28:02 ljp: ping Jul 05 06:28:08 ljp: when is Qtopia GPL due? Jul 05 06:32:45 03coredump 07org.oe.oz354x * rd815029f... 10/packages/fluxbox/ (fluxbox/apps.gpe.akita fluxbox_0.99+1.0rc.bb): fluxbox: Add more geometry defaults Jul 05 06:32:50 03coredump 07org.oe.dev * r1c16cc76... 10/packages/fluxbox/ (4 files in 3 dirs): fluxbox: Add more geometry defaults Jul 05 06:37:11 hi marcos Jul 05 06:39:09 duh brb Jul 05 06:40:32 ~lart usb wifi on my notebook Jul 05 06:40:32 * ibot throws usb wifi on my notebook's poor little doggy off a cliff Jul 05 06:41:03 hi hrw! sorry, my notebug is weird sometimes Jul 05 06:42:27 happens Jul 05 06:55:45 morning all Jul 05 06:55:52 hi dirk Jul 05 06:56:01 hey Marcin Jul 05 06:56:44 tosa wifi suxx Jul 05 06:57:39 suspend so ifdown wlan0 so usbctl off 1 on down. resume so ifup wlan0 so usbctl on 1 so OOPS.. Jul 05 06:57:49 hehe Jul 05 06:58:30 and wlan-ng lack WPA.. Jul 05 06:58:59 try to insert a ifconfig down before ucbctl Jul 05 06:59:11 yeah wlan-ng is crap Jul 05 07:00:15 btw gpe look sweet on tosa Jul 05 07:00:32 prism2 usb should be handled via the kernel hostap driver, but it needs time and motivation Jul 05 07:01:24 if sharp will add wifi to new models and will do it like it is done in tosa then I will ignore existance of that machine Jul 05 07:02:12 gpe can't configure wifi, they ignore wlan-ng Jul 05 07:02:26 sure you are? Jul 05 07:02:37 try to set a wep key Jul 05 07:03:09 ah.. wlan-ng do it in other way then normal drivers? Jul 05 07:03:23 yes Jul 05 07:03:34 jezz... then it suxx even more Jul 05 07:04:48 I added the systax to the opie networksettings applet, but this need to be done for gpe too Jul 05 07:05:31 also iwconfig won't work (not all functions) Jul 05 07:05:40 fscking shit Jul 05 07:05:55 ~lart wlan-ng Jul 05 07:05:56 * ibot decapitates wlan-ng conan the destroyer style Jul 05 07:06:05 ~kill wlan-ng Jul 05 07:06:08 * ibot shoots a super-inverse tachyon gun at wlan-ng Jul 05 07:06:16 ~kill wlan-ng once more Jul 05 07:06:18 :D Jul 05 07:06:18 * ibot shoots a super-inverse fluxproton gun at wlan-ng once more Jul 05 07:06:37 time to go to work Jul 05 07:28:11 re Jul 05 07:29:32 hrw|work: wb Jul 05 07:31:44 Hello everyone Jul 05 07:44:25 morning Jul 05 07:45:25 hi XorA Jul 05 07:45:54 bloody debian testing has no xserver today :-( Jul 05 07:47:13 morning all Jul 05 07:47:48 XorA: The c7x0 audio input issues should be fixed in 2.6.17. Just that speaker bug remaining Jul 05 07:48:26 koen: If 2.6.16 is still around, that will scare people (for FOSDEM) ;-) Jul 05 07:48:40 RP: I shall upgradew later and test Jul 05 07:52:23 hey Richard Jul 05 07:53:10 hi RP XorA koen Jul 05 07:53:52 RP: udev 0.92 has the input/touchscreen symlink rules? Jul 05 07:54:00 do13_: yes Jul 05 07:54:08 http://kernel.org/git/?p=linux/kernel/git/pavel/zaurus.git;a=commit;h=696a3bb0ee3c88409d614649e7327981fadfea4b Jul 05 07:54:18 http://kernel.org/git/?p=linux/kernel/git/pavel/zaurus.git;a=commit;h=5c557fb7eb1673567eff69f54af1e2af9417fdf4 Jul 05 07:54:29 do13_: Although 2.6.17 has a bug which breaks it (I've pushed a patch) Jul 05 07:54:32 http://kernel.org/git/?p=linux/kernel/git/pavel/zaurus.git;a=commit;h=aac87068ae02031501634f1b105bcb10793fe6fe Jul 05 07:54:41 pavel changes to charging on collie Jul 05 07:55:08 hi Cwiiis Jul 05 07:55:23 RP: thx, I had no symlink :) Jul 05 07:55:23 Hey hrw|work Jul 05 07:55:31 morning all Jul 05 07:55:42 hi ade Jul 05 07:56:01 RP: is the patch in 2.6.17.bb? Jul 05 07:56:05 do13_: yes Jul 05 07:56:19 RP: ok, will rebuild kernel Jul 05 07:56:35 do13_: input_modalias_fix Jul 05 07:56:59 do13_: It should also make the next 2.6.17 stable release Jul 05 07:58:23 good morning all Jul 05 07:58:28 Actually, it was in 2.6.17.2 :) Jul 05 07:58:31 morning koen Jul 05 07:58:57 morning koen Jul 05 07:59:31 hey RP and do13_ Jul 05 07:59:40 morning koen Jul 05 07:59:54 hey ade|desk Jul 05 08:00:04 hey Jul 05 08:00:07 sup people? Jul 05 08:00:25 koen: how is eabi work going ? Jul 05 08:01:23 heh.. need to finally findout why rm_work stops from time to time Jul 05 08:01:54 hrw|work: it doent always run as the last task Jul 05 08:02:43 hrw|work: It needs to be after do_package and do_deploy if I remeber right Jul 05 08:03:29 XorA: 21 addtask rm_work before do_build after do_package Jul 05 08:03:55 hrw|work: do_package isnt always the last task Jul 05 08:04:00 ade|desk: 2.6.17/eabi doesn't mount jffs2 on my c700 adn 2.6.16/eabi doesn't like binaries on my hx4700 Jul 05 08:04:31 XorA: thats too - but failing is other story iirc Jul 05 08:05:19 Working on building glibc-initial-2.4 and I get this: Importing patch /data/oe/build/epia-glibc/packages/glibc/glibc-cvs/arm-longlong.patch (stored as arm-longlong.patch) Jul 05 08:05:31 which is the wrong patch... should be from glibc-2.4... Jul 05 08:05:36 hrw|work: I did work out the other reason as well, but it escapes me currently, I think it might be todo with files with -w set Jul 05 08:05:44 v8jlene: wrong FILES_PATH? Jul 05 08:06:41 I dunno what a FILES_PATH is, let me grep around... Jul 05 08:07:30 BTW, my x86 builds worked yesterday because I had an old glibc cvs checkout... new checkouts of the same date don't work (as I think someone pointed out). Jul 05 08:07:55 koen: bugger perhaps the 2006.9 timeframe is a little hopeful :( Jul 05 08:08:57 koen: You could always try the 3.4.4 csl toolchain Jul 05 08:09:53 v8jlene: glibc-initial-2.4 has wrong FILESPATH Jul 05 08:10:44 do13_: Ahh, I see it now.. I'll change it and test again. Jul 05 08:12:09 mickey|fifa2006: ping Jul 05 08:12:24 pb_: hey, I think I see why GNU needs to be completed Jul 05 08:14:30 Ok, fixed FILESPATH and it looks in the right directory now.. but the patches still fail. I presume someone has actually used glibc-2.4? Jul 05 08:14:46 v8jlene: angstrom does Jul 05 08:17:44 Thanks. Was my fault - you have to unpack the source before you can patch it... Jul 05 08:18:14 ;P Jul 05 08:20:02 Ok, doesn't build on x86... more work than that needed... Jul 05 08:20:17 ade|desk: yeah, maybe bump it up to 2007.1 or so Jul 05 08:21:03 Does someone know what changed to make the glibc cvs checkouts be different now than they were a few months ago? I thought someone mentioned this yesterday? Jul 05 08:22:20 03dirk 07org.oe.dev * r27e5267a... 10/packages/glibc/glibc-initial_2.4.bb: glibc-initial_2.4: FILESPATH changed from glibc-cvs into glibc-2.4 Jul 05 08:23:05 2006.12 Jul 05 08:35:06 v8jlene: they rearrangered their cvs tree Jul 05 08:35:18 v8jlene: and cvs doesn't handle that gracefully Jul 05 08:40:17 koen: Ok. I'll diff the checkouts and see if I can see what needs to be done to make it work. Jul 05 08:41:00 v8jlene: afaik you need a checkout from some months ago, otherwise it won't work Jul 05 08:41:31 koen: Yeah, I have one that works. I'm looking at getting x86 and sh4 toolchain to build from a clean checkout now... Jul 05 08:42:38 glibc-initial fails, the cvs versions you know about, 2.4 has some TLS issue which I'm still looking into... I guess there's older non-cvs glibc-initial versions I could try. Jul 05 08:43:28 getting glibc 2.4 to work would be nice Jul 05 08:50:53 zecke: pöng Jul 05 08:50:55 morning guys Jul 05 08:51:04 morning mickeyl Jul 05 08:51:04 hey mickeyl Jul 05 08:51:11 morning mickeyl Jul 05 08:51:22 hi mickeyl Jul 05 08:51:53 hi mickeyl Jul 05 08:51:53 mickeyl: hmm Jul 05 08:52:20 mickeyl: do you know what I wanted to ask? Jul 05 08:52:20 03coredump 07org.oe.oz354x * r53c8b346... 10/packages/keylaunch/keylaunch_2.0.10.bb: keylaunch: Install configuration using update-alternatives, enables different key-bindings for different window managers Jul 05 08:52:25 03coredump 07org.oe.oz354x * rf2c41c12... 10/packages/fluxbox/ (6 files in 2 dirs): fluxbox: Add custom keylaunchrc, gpe-logout and key-bindings Jul 05 08:52:29 03coredump 07org.oe.oz354x * r4971c475... 10/packages/gpe-session-scripts/gpe-session-scripts_0.66.bb: Jul 05 08:52:29 keylaunch: Jul 05 08:52:29 gpe-session-scripts: Use update-alternatives for gpe-logout to allow for different windowmanagers Jul 05 08:52:33 03coredump 07org.oe.dev * ra952de14... 10/packages/keylaunch/keylaunch_2.0.10.bb: keylaunch: Install configuration using update-alternatives, enables different key-bindings for different window managers Jul 05 08:52:37 03coredump 07org.oe.dev * r36743688... 10/packages/gpe-session-scripts/gpe-session-scripts_0.67.bb: gpe-session-scripts: Use update-alternatives for gpe-logout to allow for different windowmanagers Jul 05 08:52:37 zecke: I'm afraid not. looks like you forgot ? Jul 05 08:52:41 03coredump 07org.oe.dev * ra4530ada... 10/packages/fluxbox/ (6 files in 2 dirs): fluxbox: Add custom keylaunchrc, gpe-logout and key-bindings Jul 05 08:53:05 mickeyl: that is impossible, I never forget anything Jul 05 08:53:09 CoreDump|home: plan to do many changes? I would like to build final images today Jul 05 08:53:34 mickeyl: I will be away until saturday Jul 05 08:53:41 hrw|work: build away. FB is far from perfect but updates can be put into "upgrades" Jul 05 08:53:47 zecke: vacation? Jul 05 08:54:09 vacations... I plan to start them this weekend Jul 05 08:54:56 hrw|work: is mplayer pre8 building for you? It's bitching about a patch here Jul 05 08:55:29 CoreDump|home: added patch to repo - have to push it Jul 05 08:55:39 hrw|work: thanks ;) Jul 05 08:57:01 mickeyl: funeral Jul 05 08:58:08 zecke: oh right :/ Jul 05 09:01:57 03hrw 07org.oe.oz354x * r15159d5b... 10/packages/mplayer/files/pld-onlyarm5.patch: mplayer: added forgotten patch Jul 05 09:07:19 03rpurdie 07org.oe.dev * r8f81c821... 10/packages/linux/ (6 files in 2 dirs): linux-oz: 2.6.16+git -> 2.6.17+git. Refresh several patches against recent changes upstream Jul 05 09:14:00 hello all Jul 05 09:14:04 just reading the topic.... Jul 05 09:14:12 so use HEAD only for development Jul 05 09:14:30 what is the recommended revision / tree to use for a final system? Jul 05 09:14:40 03hrw 07org.oe.oz354x * rb05b107c... 10/packages/gpe-timesheet/ (8 files in 2 dirs): gpe-timesheet: added 0.30 Jul 05 09:14:44 03hrw 07org.oe.oz354x * r65dd0655... 10/packages/gpe-bluetooth/ (17 files in 4 dirs): gpe-bluetooth: added 0.53, dropped old Jul 05 09:14:48 03hrw 07org.oe.oz354x * r99d9f733... 10/packages/libcontactsdb/libcontactsdb_0.4.bb: libcontactsdb: added 0.4 Jul 05 09:15:06 jkp: depends on your handheld. Jul 05 09:15:22 thejapa: its a via c3 based box Jul 05 09:15:36 mini-itx Jul 05 09:15:55 coz im having major issues with HEAD Jul 05 09:16:02 possibly using something more stable might fix them Jul 05 09:16:09 i cant even get a toolchain working Jul 05 09:16:10 jkp: with bitbake head or OE head? Jul 05 09:16:16 OE head Jul 05 09:16:30 jkp: the topic refers to bitbake head Jul 05 09:16:35 ah Jul 05 09:16:36 ok Jul 05 09:16:38 OE head is the place to be Jul 05 09:16:41 well then Jul 05 09:16:50 can someone tell me if they have a clue about this? -> http://bugs.openembedded.org/show_bug.cgi?id=1000 Jul 05 09:17:12 coz its not only me who has had this issue, and it was filed two months ago Jul 05 09:17:21 did you try with DISTRO=generic yet? Jul 05 09:17:25 yup Jul 05 09:17:34 ok, let me try Jul 05 09:17:38 * mickeyl launches a build Jul 05 09:17:40 MACHINE=? Jul 05 09:17:51 MACHINE=x86 Jul 05 09:17:57 or the other, i forget the name Jul 05 09:18:04 epia Jul 05 09:18:04 i have to run to meeting for 30 mins Jul 05 09:18:06 epia? Jul 05 09:18:07 yes Jul 05 09:18:13 ill shout when im back in a bit Jul 05 09:18:16 i appreciate this Jul 05 09:18:28 you can see from my comments ive tried on two different hist rigs Jul 05 09:18:31 *host Jul 05 09:18:34 gotta run Jul 05 09:18:35 bbiab Jul 05 09:18:38 ta Jul 05 09:18:49 i've been suggested to test epia and generic before... but the whole problem was with my local.conf in the end. Jul 05 09:18:56 apparantly not :) Jul 05 09:19:08 i can help out if you can suggest some things Jul 05 09:19:59 well, i was just playing. ended up using MACHINE=native Jul 05 09:20:07 for x86 Jul 05 09:20:18 but the host isnt necessarily x86 Jul 05 09:20:24 it was AMD_64 in the first case Jul 05 09:20:35 then i used a VM (x86) and had the saem issue Jul 05 09:21:22 thejapa: native is for other things Jul 05 09:21:28 oops Jul 05 09:21:51 thejapa: native is for using OE to build missing stuff - not whole systems Jul 05 09:21:55 right - ive seen some stuff about some things having moved in glibc after that snapshot Jul 05 09:22:04 hrw|work: thanks for pointing out. :) Jul 05 09:22:17 the one from 2004, actually it was koen who posted the reply to that issue Jul 05 09:22:33 but the 2004 snapshot is still being used for the initial build of it Jul 05 09:24:45 thejapa: for example I use native to develop OPIE on x86, used it to build some tools when I was debugging problems of amd64 build and lacked mc, screen etc Jul 05 09:25:28 jkp: I've verified that you cannot build for x86 from a current checkout... still working on trying to fix it. You'd need the old glibc checkouts to build at the moment... Jul 05 09:26:14 v8jlene: ok - thats good news i guess. i was just about to try making the initil build use the later snapshot to see what it did Jul 05 09:26:22 can you tell me how to get those old checkouts? Jul 05 09:26:51 jkp: I've tried... no luck with any of them for various reasons. I'm working on glibc-initial-4.0 at the moment. Jul 05 09:27:10 yeah, i got that one to build... Jul 05 09:27:14 but then it failed elsewhere Jul 05 09:27:34 did you find a working combination with glibc-initial-4.0? Jul 05 09:27:40 or are you stuck later on as well? Jul 05 09:28:59 glibc-initial_2.4 is failing... original due to applying the wrong patches, now due to TLS support? Jul 05 09:29:50 yeah - i remember now Jul 05 09:29:56 it failed applying the patches :/ Jul 05 09:30:00 did you fix that? Jul 05 09:30:40 jkp: Yeah, that's fixed. But it has other issues now. Jul 05 09:30:48 sucky / Jul 05 09:31:01 is there anything i can do to help? Jul 05 09:31:23 it was wierd, i couldnt work out why it was applying the wrong patches - id love to know... Jul 05 09:31:28 it was pulling the CVS patches Jul 05 09:31:36 rather than the ones in the 2.4 folder Jul 05 09:31:54 but i couldnt see how you controlled that - the only way i could get past that was to move the cvs folder Jul 05 09:31:59 id like to know the proper way Jul 05 09:32:32 btw - im kinda assuming that once i get a working toolchain things should be a it smoother than this right?! Jul 05 09:33:59 I got glibc-initial_2.3.2 to build Jul 05 09:34:05 tkp: really? Jul 05 09:34:06 jkp: The .bb file had: FILESPATH = "${@base_set_filespath([ '${FILE_DIRNAME}/glibc-cvs', '${FILE_DIRNAME}/glibc', '${FILE_DIRNAME}/files', '${FILE_DIRNAME}' ], d)}" - note the -cvs, should have been -2.4... FILEPATH tells it where to search for patches.. Jul 05 09:34:07 how? Jul 05 09:34:32 jkp: bitbake -b ~/stv/oe/org.openembedded.dev/packages/glibc/glibc-initial_2.3.2.bb bitbake -b glibc-initial_2.3.2.bb Jul 05 09:34:52 not sure if that's relevant to using the bootstrap thing tho Jul 05 09:34:55 FILESPATH that is.. Jul 05 09:35:05 tkp: When did you check it out of CVS though? Jul 05 09:35:19 tkp: Ignore that.. .it's not in CVS ;) Jul 05 09:35:27 jkp: or rather: bitbake -b ~/stv/oe/org.openembedded.dev/packages/glibc/glibc-initial_2.3.2.bb Jul 05 09:35:47 tkp: how is this different to what normally happens? Jul 05 09:36:04 ~lart DOM and JavaScript Jul 05 09:36:04 * ibot beats DOM and JavaScript senseless with a 50lb Unix manual Jul 05 09:36:11 jkp: it's a different version for a start Jul 05 09:36:26 ah - thats the norml version, not the cvs version Jul 05 09:36:37 it ends up doing: glibc-initial-2.3.2-r6 Jul 05 09:36:38 tkp: did you alter the distoro file? Jul 05 09:36:52 you know you can specify preffered versions Jul 05 09:36:54 jkp: no, I just ran that as a standalone command Jul 05 09:36:58 ok Jul 05 09:37:06 jkp: I tried but couldn't get it working Jul 05 09:37:12 im going to alter the distro file, biuild and see what effect that has Jul 05 09:38:14 jkp: If 2.3.2 worked then you'd be able set your glibc versions to that in the distro. But you'd also be limited to 2.3.2 for the actual glibc it uses. Could be enough to get you past the toolchain for now. Jul 05 09:38:54 you mean the final glibc would have to be that version? Jul 05 09:39:08 jkp: a couple of useful pages I found too: http://www.uv-ac.de/openembedded/openembedded-4.html and http://www.uv-ac.de/openembedded/openembedded-3.html... Jul 05 09:39:13 specifically the "Speed up package build (in case of errors)" Jul 05 09:39:23 jkp: Yeah, since none of the other ones build yet... Jul 05 09:39:23 and "Cleaning" Jul 05 09:39:41 hang on - i know that would be in my toolchain Jul 05 09:39:41 good morning Jul 05 09:39:52 but how does that affect my final distro? Jul 05 09:39:55 jkp: Anyway, I got to go. I'll continue work on getting glibc 2.4 working tommorow... Jul 05 09:40:05 v8jlene: ok - ta :) Jul 05 09:40:13 so just to clairfy Jul 05 09:40:33 ill be limited to glibc2.3.2 all over right? toolchain + final distro? Jul 05 09:41:19 jkp: Your limited to what you can build... we know the cvs versions don't work and the 2.4 ones don't work at the moment. If 2.3.2 worked the you'd be limited to that for now... Jul 05 09:41:53 I'm having a strange problem with RC4 for my akita, and a usb memory stink reader - the reader seems to be asking for way more power than it needs, so config #1 doesn't get detected and has to be forced. Anyone had similar problems? Jul 05 09:42:04 tkp: i managed to alter the distro file...what was your issue with it? Jul 05 09:42:18 * v8jlene is really gone now Jul 05 09:42:23 v8jlene: :) Jul 05 09:42:27 speak tomorrow dude Jul 05 09:42:32 jkp: I didn't spend too long trying tbh... got distracted with real work Jul 05 09:42:38 jkp: commit your conf Jul 05 09:42:38 tkp: heh Jul 05 09:42:42 and distro file Jul 05 09:42:49 ~lart real work Jul 05 09:42:49 * ibot executes killall -KILL real work Jul 05 09:42:57 jkp: did it work then? Jul 05 09:42:58 just wait till i know it will build Jul 05 09:43:02 k Jul 05 09:43:02 ibot: lol Jul 05 09:43:05 well, lol is stands for Laughs Out Loud. It is grammatically incorrect to use LOL in the first person; use 'heh' or 'haha' instead. If you want to use LOL, do '/me lol' instead. Jul 05 09:43:22 im at "supposed real work" now - Jul 05 09:43:27 ibot: lol Jul 05 09:43:28 from memory, lol is stands for Laughs Out Loud. It is grammatically incorrect to use LOL in the first person; use 'heh' or 'haha' instead. If you want to use LOL, do '/me lol' instead. Jul 05 09:43:35 ibot: stfu Jul 05 09:43:39 i heard stfu is Shut the F*** Up!, or http://www.linuks.mine.nu/stfu-noob.jpg. Sucks To Feel U, or or http://beerandspeech.org/tmp/STFU_cup_Soldier.jpg Jul 05 09:43:43 :p Jul 05 09:43:46 RP: ping Jul 05 09:44:50 ibot: LOL is Laughing On Line Jul 05 09:44:52 mikearthur: cannot alter locked factoids Jul 05 09:45:13 tkp: it built, but i want to see if it craps out now Jul 05 09:45:27 im sure i did this yesterday, but actually, what i did before was to move to 2.4 Jul 05 09:45:36 which worked in the first instance Jul 05 09:45:42 so lets just cross our fingers Jul 05 09:46:03 is .dev merged currently? Jul 05 09:46:23 jkp: you think 2.3.2 is going to work with the other packages we are wanting to install? Jul 05 09:46:29 CoreDump|home: pull and find out Jul 05 09:46:31 * jkp hopes so Jul 05 09:46:37 it's not for me... Jul 05 09:46:50 but i may have fubar'ed my db Jul 05 09:48:13 it only has one head on the servers Jul 05 09:48:19 ok thanks Jul 05 09:48:23 CoreDump|home: 8f81c821d4522c9367e23584313e4bbf36557144 rpurdie@openembedded.org 2006-07-05T09:22:26 Jul 05 09:48:39 * Bernardo thinks the usb power limit is set way too low in our latest akita kernel - even his usb memory key doesn't work anymore without a hack, and it always worked well with the akita Jul 05 09:48:47 whats the cheapest linuxable PDA that OE supports? Jul 05 09:48:51 http://www.openembedded.org/viewmtn/headofbranch.psp?branch=org.openembedded.dev Jul 05 09:49:39 Bernardo: its set to 150mA iirc Jul 05 09:49:51 * CoreDump|home wget's a new .db Jul 05 09:50:39 mikearthur: Sharp SL5000, Sharp SL5500 Jul 05 09:50:55 only the latter can be recommended though Jul 05 09:51:23 (if at all.... the clamshell models like SL-C750 or SL-C1000 are much better) Jul 05 09:51:39 mickeyl: thanks, googling Jul 05 09:51:42 avoid 5000/5500/5600 anyway Jul 05 09:51:53 mikearthur: h22xx is nice Jul 05 09:52:00 small screen though Jul 05 09:52:05 sharp h22xx? Jul 05 09:52:15 mikearthur: hp ipaq h2210/2215 Jul 05 09:52:18 * thejapa envies vga res Jul 05 09:52:42 vga on tosa is awesome Jul 05 09:53:07 aw Jul 05 09:53:08 hrw|work: I have my usb key loading oly after forcing it, and dmesg complains it is 50ma Jul 05 09:53:18 hrw|work: I have my usb key loading oly after forcing it, and dmesg complains it exceeds the power limit by 50ma Jul 05 09:53:37 and my memory stink reader exceeds it by 100ma Jul 05 09:54:01 I'll have to install lsusb and check what power those claim to drain, as both work Jul 05 09:54:05 contact RP about it then Jul 05 09:54:09 lsusb is not installed? Jul 05 09:54:27 hrw|work: ooo, I like the look of the keyboard Jul 05 09:54:45 Bernardo: The limits in the device are correct as far as I know. Jul 05 09:54:51 hrw|work: not by default Jul 05 09:55:35 Bernardo: it is in usbutils? Jul 05 09:55:52 I will add it into images Jul 05 09:56:04 it seems it didn't build usbutils too... Jul 05 09:56:21 anyone has a usbutils for the akita rc4 that can email me? :) Jul 05 09:56:34 Bernardo: its in .3.5.4.1 feed ;) Jul 05 09:57:08 RP: the strange part is that I am pretty sure that the usb key worked even when running the akita from batteries (it is charging now as it was completely drained out) Jul 05 09:57:33 RP: hey Jul 05 09:57:45 hrw|work: thanks - I didn't even know that 3.5.4.1 was out... I do need to get this project ruinning so I have some free time Jul 05 09:58:34 Bernardo: it will be out this week Jul 05 09:58:54 failed where it did for me before tkp Jul 05 09:59:01 Bernardo: thx to you it now contain usbutils Jul 05 09:59:01 building intermediate glibc Jul 05 09:59:07 ill try some other versions in there now Jul 05 09:59:07 hmm. Jul 05 09:59:29 jkp: http://www.handhelds.org/hypermail/oe/87/8777.html Jul 05 09:59:33 i seem to remember i had trouble getting it to pick up my request to use another intermediate version Jul 05 10:00:07 yeah - thats interesting Jul 05 10:00:19 note the changes to local.conf Jul 05 10:00:21 ill resort to this if what im doing doesnt work Jul 05 10:00:38 yeah - that can be in local.conf of distro.conf Jul 05 10:00:41 im using distro.conf Jul 05 10:01:16 koen: http://lists.gnu.org/archive/html/monotone-devel/2006-07/msg00018.html Jul 05 10:01:39 koen: I will check if my colinux tree will be 'syncable' :) Jul 05 10:01:56 03hrw 07org.oe.oz354x * rd617ef09... 10/conf/distro/openzaurus-3.5.4.1.conf: oz 3.5.4.1: install usbutils on devices with USB Host Jul 05 10:01:58 interesting it wasnt rolled in though tkp Jul 05 10:02:01 wonder why that was Jul 05 10:02:32 hehe Jul 05 10:02:38 RMK: "Sounds like Montavista broke something - this works with the standard Jul 05 10:02:38 kernel. Direct your bug report to Montavista, thanks." Jul 05 10:03:30 hehe Jul 05 10:06:12 hrw|work: stupid question, where are the 3.5.4.1 feeds? I can only find the rc4, and 3.5.4 feeds, should I use the one in the 3.5.4 feed? Jul 05 10:06:30 tkp: failed - trying that solution Jul 05 10:06:34 looks like it could work that Jul 05 10:06:43 almost the same config as ours Jul 05 10:06:55 good find Jul 05 10:07:44 PREFERRED_VERSION_glibc ?= "2.3.5+cvs200506272.4" ... Jul 05 10:07:48 not sure if that'sa typo? Jul 05 10:07:53 200506272? Jul 05 10:07:59 extra 2 Jul 05 10:08:17 then again, he has done it twice so it can't be Jul 05 10:08:24 it is Jul 05 10:08:27 it should be 27 Jul 05 10:12:18 RP: it seems the limit is 150ma, as the usb memory key has a MaxPower of 200ma, and the reader 250ma. Is that what it should be? Jul 05 10:13:07 I can force them both to work with the "echo -n 1 >/sys/bus/usb/devices/2-2.1/bConfigurationValue" hack Jul 05 10:15:30 03coredump 07org.oe.oz354x * ra4464d25... 10/packages/altboot/altboot_1.0.7.bb: altboot: Upload correct .bb for altboot 1.0.7 Jul 05 10:15:34 03mickeyl 07org.oe.dev * raa68ff4c... 10/packages/binutils/ (24 files in 5 dirs): Jul 05 10:15:34 binutils 2.15.94 add gcc4 patch Jul 05 10:15:34 remove binutils 2.15.* except 2.15.94 Jul 05 10:15:38 03coredump 07org.oe.dev * rbfd44a04... 10/packages/altboot/altboot_1.0.7.bb: altboot: Upload correct .bb for altboot 1.0.7 Jul 05 10:18:48 Bernardo: you can take from 3.5.4 Jul 05 10:19:04 Bernardo: unstable/.3.5.4.1/feed are 3.5.4.1 currently Jul 05 10:19:06 hrw|work: already did, thanks. Jul 05 10:23:50 Bernardo: The device is only capable of 150mA, yes Jul 05 10:24:17 RP: can you disable the squashfs module in the oz kernels? Jul 05 10:24:27 RP: gcc 4.x can't build it Jul 05 10:24:34 koen: I wasn't the one who added it Jul 05 10:24:43 but you are the maintainer Jul 05 10:24:49 RP: that is what is speced, right? As mine seems to support 250mA when charging Jul 05 10:24:52 koen: There is a patch in the bugzilla which works around it - we should probably apply that Jul 05 10:25:32 Bernardo: Its unlikely your device uses all 250mA so it probably used to work. The figure of 150mA is right though Jul 05 10:27:09 RP: probably. It is frustrating to have to hack it to work, but I should blame the manufacturers who have specced their power demands too high. Jul 05 10:27:35 Bernardo: right. Its something we just have to live with I'm afraid :-( Jul 05 10:27:59 time to add another entry to the faq... Jul 05 10:28:14 Bernardo: If I change the default, devices which really need 250mA won't work and I'll get complaints so there is nothing I can do Jul 05 10:28:41 I know Jul 05 10:29:11 like the wifi key I tryed to make work Jul 05 10:29:21 the usb part would work, the wifi part wouldn't Jul 05 10:31:53 jkp: any joy? Jul 05 10:32:00 looking positive tkp Jul 05 10:32:08 passt the intermediate? Jul 05 10:32:13 just waiting for the glibc-intermediate compile to finish Jul 05 10:35:12 RP: I will start creating a oh-test soonish Jul 05 10:36:32 03mickeyl 07org.oe.dev * rcc177b76... 10/packages/inotify-tools/inotify-tools_2.1.bb: add inotify-tools, set of command line utilities for the Linux inotify filesystem change notification system. Jul 05 10:38:04 zecke: oh-test? Jul 05 10:39:35 automatic booting of the qemu image Jul 05 10:39:41 + whatever crack I'm smoking Jul 05 10:40:16 RP: I will disable squashfs in .17 kernels Jul 05 10:40:37 so koen will stop despair about 'oz kernel does not build in .dev' Jul 05 10:40:45 hrw|work: Or apply the patch in the bugzilla Jul 05 10:40:52 or Jul 05 10:41:03 zecke: sounds good :) Jul 05 10:41:49 zecke: For autobuilding, you could use oh-image-sdk to fully stretch the image Jul 05 10:42:40 RP: okay. I'm restarting the building anyway. Somehow our Arcor firewall fascist threw too big log files away (which blocked in glibc then) Jul 05 10:42:43 RP: packages/linux/linux-openzaurus/ will be created Jul 05 10:43:24 hrw|work: Why? Jul 05 10:43:49 RP: same patch(es) for 3 kernels - why to duplicate them? Jul 05 10:44:18 hrw|work: They are not the same patches in some cases and not the same defconfigs Jul 05 10:45:03 RP: ok - will then patch all copies of squashfs Jul 05 10:45:56 now time to build 3 kernels... Jul 05 10:45:56 tkp: looking good Jul 05 10:46:01 :) Jul 05 10:46:12 commit and let me try this end Jul 05 10:46:22 if you want ok Jul 05 10:46:33 hrw|work: Just add a fix patch on top of the squashfs one. If it applies for one, its likely to apply for all Jul 05 10:46:38 jkp: actually, I'll try it on the server since that may not have been the issue anyway Jul 05 10:46:43 it was Jul 05 10:46:46 hrw|work: We just need to be careful - I just know I had to rediff 50% of the patches for the latest git kernel this morning :-/ Jul 05 10:46:50 it should be ok with these configs Jul 05 10:46:56 i saw exactly the same issue Jul 05 10:47:01 ok Jul 05 10:47:03 fwiw, a .bb for busybox 1.2.0 is in the bugtracker Jul 05 10:47:13 koen: add it into repo as not-default? Jul 05 10:47:38 jkp: so it should work on the server now too? Jul 05 10:47:43 think so Jul 05 10:47:51 coolio, i'll try Jul 05 10:48:09 mikearthur: did you install all the RequiredSoftware? Jul 05 10:48:24 for kernel in ~/devel/oe/org.openembedded/packages/linux/linux-openzaurus_2.6.1*.bb;do bbrebuild $kernel;done Jul 05 10:48:32 koen: yep, I'll just triple check Jul 05 10:50:03 tkp: done Jul 05 10:50:10 cool Jul 05 10:50:19 jkp: so we have a toolchain?! Jul 05 10:50:24 not yet Jul 05 10:50:32 its still buildin Jul 05 10:50:39 looking good though - fingers crossed Jul 05 10:51:36 koen: I seem to have Jul 05 10:52:59 03rpurdie 07org.oe.dev * r1f2a8090... 10/packages/linux/linux-openzaurus-2.6.17+git/ (00-hostap.patch 10-pcnet.patch): linux-oz-2.6.17+git: Drop merged patches Jul 05 10:53:05 mikearthur: do you have makeinfo in your host? Jul 05 10:53:40 s/in/on/ Jul 05 10:54:28 RP: after 2.6.18-rc1 we need to pester pavel to send us extracted collie charge patch Jul 05 10:54:48 koen: bash: makeinfo: command not found Jul 05 10:54:59 hrw|work: He posted his kernel patch yesterday. We'll just have to wade through that and find the interesting bits :-/ Jul 05 10:55:11 o.. nice to know Jul 05 10:55:12 mikearthur: install makeinfo and add it to RequiredSoftware :) Jul 05 10:55:25 hrw|work: The code quality is horrible :-( Jul 05 10:55:34 RP: ;( Jul 05 10:56:01 koen: what supplies makeinfo? Jul 05 10:56:18 fetching git kernel.. Jul 05 10:56:26 texi2html? Jul 05 10:56:46 hrw@bitbake:~$ dpkg -S `which makeinfo` Jul 05 10:56:46 texinfo: /usr/bin/makeinfo Jul 05 10:57:20 hrw|work: I've forwarded you the message. See what you think... Jul 05 10:57:28 ok Jul 05 10:57:35 RP: lkml? Jul 05 10:57:43 hrw|work: It was cc'd there I think Jul 05 10:58:16 so I will read it on gmail later too Jul 05 11:00:46 koen: installed texinfo, added to Ubuntu and Debian sections of OEandYourDistro Jul 05 11:00:59 hrw|work, RP: I have extracted some pieces from pavels tree. It boots and the touchscreen is working Jul 05 11:01:16 do13_: great Jul 05 11:01:24 mikearthur: thanks Jul 05 11:06:17 03tmbinc 07org.oe.dreambox * r78d916cd... 10/packages/gstreamer/gst-plugin-dvbmediasink.bb: gst-plugin-dvbmediasink: workaround automake install problem until this is fixed upstream Jul 05 11:06:21 03tmbinc 07org.oe.dreambox * r30c29ff4... 10/packages/enigma2/enigma2.bb: enigma2: update to 20060705, remove python-pyxml dependency as python-xml works fine Jul 05 11:06:28 03tmbinc 07org.oe.dreambox * r8546a545... 10/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules/dm7025: update to 20060627 Jul 05 11:09:19 do13_: Can you add that patch to the kernel .bb file? (as conditional for collie) Jul 05 11:10:35 eh? thats weird my pref'd uclibc svn got switched to 0.9.28 during a build by i cant find what caused that to change, something during a x11 build but i dont know what Jul 05 11:11:07 RP: I splitt it into smaller pieces, but it isn't pavels latest code :-( Jul 05 11:11:11 RP wait a sec Jul 05 11:13:35 RP: common/sharpsl_pm.c need some love to compile for collie Jul 05 11:16:14 do13_: I know. It has quite a few pxa assumptions which I was hoping to abstract once Pavel got things roughly working Jul 05 11:17:08 pushed Jul 05 11:18:40 03dirk 07org.oe.dev * r051c0951... 10/packages/linux/ (2 files in 2 dirs): Jul 05 11:18:40 linux-openzaurus_2.6.17.bb: update collie patches Jul 05 11:18:40 - update collie defconfig Jul 05 11:18:40 - update pavels collie patches (clean) Jul 05 11:20:51 RP: we need to discuss the collie.conf. I added the collie part to zaurus-clammshell-2.6.conf, but there are too many modules for the small collie flash Jul 05 11:22:53 RP: his pm code needs to be implemented in a clean way Jul 05 11:26:09 do13_: I'd guess collie shouldn't use the clamshell-2.6.conf file but use its own instead Jul 05 11:26:28 do13_: I'm just reading through those patches. There is a fair bit of cleanup needed :-/ Jul 05 11:26:38 do13_: Does audio work? Jul 05 11:27:13 do13_: That battswitch driver needs to die for a start - the pm code has support for that... Jul 05 11:28:51 RP: audio doesn't work. The battswitch driver isn't enabled. Jul 05 11:29:40 do13_: for 2.6.16 which ver of wlan-ng-modules are ok? 0.2.3 will be ok? (for spitz/akita) Jul 05 11:30:57 hrw|work: 0.2.3 is ok (with all patches applied) Jul 05 11:31:15 k Jul 05 11:33:25 do13_: Even the backlight fix looks wrong :-/ Jul 05 11:34:05 do13_: It would break every other device at the very least... Jul 05 11:34:22 RP: locomolcd-backlight-r0? Jul 05 11:34:28 do13_: yes Jul 05 11:35:28 RP: I think he has a new version, we need to rediif the backlight driver Jul 05 11:35:53 do13_: wlan-ng-modules are broken Jul 05 11:36:07 do13_: they need tmp/work copy of kernel ;( Jul 05 11:36:49 RP: I haven't looked at the backlight stuff, I cleaned collie.c and the mcp code. It was enough work Jul 05 11:37:51 03koen 07org.oe.dev * r7ac1989c... 10/packages/gpe-taskmanager/gpe-taskmanager_0.20.bb: gpe-taskmanager: add missing dep on libgpelaunch Jul 05 11:38:15 do13_: I can well imagine Jul 05 11:38:36 do13_: I suspect the backlight patch can just be dropped Jul 05 11:39:07 hrw|work: read mickeys comment inside the bb file Jul 05 11:39:48 do13_: Looking at the code, it needs a locomo-backlight device adding to collie.c and then any resulting problems with the driver fixing Jul 05 11:40:03 do13_: ok Jul 05 11:40:09 03coredump 07org.oe.oz354x * r471ae2af... 10/packages/fluxbox/ (fluxbox/fluxbox-gpe-session fluxbox_0.99+1.0rc.bb): fluxbox: Fix session script to install key file with correct name Jul 05 11:40:14 03coredump 07org.oe.dev * r6859af4f... 10/packages/fluxbox/ (fluxbox/fluxbox-gpe-session fluxbox_0.99+1.0rc.bb): fluxbox: Fix session script to install key file with correct name Jul 05 11:40:46 do13_: Actually, that's wrong as its a locomo device. I was right the first time - it should just work without that patch :) Jul 05 11:42:34 ~lart cvs Jul 05 11:42:35 * ibot stuffs cvs into a shiny new tin can and vacuum seals it Jul 05 11:42:44 RP: another idea is to use the corgibl driver and handle the lowlevel stuff via bl_machinfo Jul 05 11:42:44 hmm Jul 05 11:42:55 I hate this world, I had hoped to bring arm support to klibc :( Jul 05 11:43:30 it even has EABI support Jul 05 11:43:43 zecke: whats up ? Jul 05 11:45:01 ade|desk: I want to do some low core ARM stuff, but everything got arm support already :( Jul 05 11:45:21 lol Jul 05 11:45:52 CoreDump|home: will you integrate kinit into altboot? Jul 05 11:46:08 RP: btw audio: the buzzer is connected via ucb device and the headphone via the ssp bus Jul 05 11:46:50 zecke: since I don't even know what that is I would think not Jul 05 11:46:54 ;) Jul 05 11:47:11 CoreDump|home: http://lwn.net/Articles/7117/ Jul 05 11:49:19 zecke: you want to do low level ARM stuff, are you mad ? Jul 05 11:49:31 ade|desk: sure Jul 05 11:49:44 who is not? Jul 05 11:49:45 zecke: geee, another fancy name for initrd.bin isn't it? Jul 05 11:49:45 zecke: if you want to do that make ARM6/7 for qemu Jul 05 11:50:09 CoreDump|home: not really Jul 05 11:51:28 zecke: but dont show it to anyone with NDA with ARM Jul 05 11:51:52 else they can't dev for qemu any more Jul 05 11:52:14 heh Jul 05 11:52:14 heh? Jul 05 11:52:49 ade|desk: is that the same thing as arm paying for the debian armel port? Jul 05 11:53:24 are they? how strange Jul 05 11:54:00 ade|desk: martin guy and aleph0 (wookey et al) are getting paid afaik Jul 05 11:56:06 RP: what do you think about enabling more stuff in 2.6.16 kernel so users will not complain about lack of 'usb-audio' for example? Jul 05 11:56:16 ARM6 and above has docs that say arm6 devs can't simulate ARM in anyway.So qemu is stuffed cos all the arm folks have the docs but can't contribute to arm6 on qemu Jul 05 11:56:47 03koen 07org.oe.dev * r2e750f1f... 10/packages/angstrom/angstrom-gpe-image.bb: angstrom-gpe-image.bb: initialize vars before using Jul 05 11:59:44 wow.. wlan-ng-utils need kernel source too.. Jul 05 12:00:04 ~seen noodles Jul 05 12:00:20 noodles was last seen on IRC in channel #oe, 102d 21h 9m 4s ago, saying: 'What about any of the mips stuff?'. Jul 05 12:00:41 hrw|work: welcome in the wonderful world of wlan-ng Jul 05 12:00:52 ~die sharp die! Jul 05 12:00:58 * ibot takes two shots to the head, murders sharp die! then crumples to the ground, lifeless... Jul 05 12:02:50 hrw|work: The goal of the linux-wlan(tm) Project is to develop a complete, standards based, wireless LAN system using the GNU/Linux operating system. What differentiates this project from the Linux wireless extensions and other Linux wireless projects is that everything is based on the IEEE 802.11 standard Jul 05 12:03:38 blebleble Jul 05 12:04:17 IEEE 802.11 define WPA, prism firmware is WPA capable. wlan-ng is not WPA capable... Jul 05 12:04:48 my tosa wifi has 1.7.4 firmware - iirc best version for using but I had to open home network for it ;( Jul 05 12:07:06 hrw|work: same with atmel berlios drivers Jul 05 12:08:46 03koen 07org.oe.dev * rc31e4927... 10/packages/angstrom/angstrom-gpe-image.bb: angstrom-gpe-image: work around some weirdness involving the xserver Jul 05 12:12:02 hi cbrake. Jul 05 12:12:23 Gerrath_: morning Gerrath_ Jul 05 12:13:52 hi Cliff Jul 05 12:14:05 hi Marcin Jul 05 12:14:11 hey cbrake Jul 05 12:14:18 koen: hello! Jul 05 12:14:54 Gerrath_: when is the last time you pulled OE for your logicpd builds? Jul 05 12:15:01 hrw|work: I'm tempted to wait for a complaint first ;-) Jul 05 12:15:32 RP: already requested usb-audio ;) Jul 05 12:15:42 NOTE: package wlan-ng-utils-0.2.3-r2: task do_build: completed Jul 05 12:15:55 RP: I want usb-dishwasher module built please ;) Jul 05 12:16:01 oh and the device too Jul 05 12:16:05 hrw|work: ah. I guess we could enable the modules then. More cruft to slow down my kernel builds... Jul 05 12:16:27 do13_: corgi_bl and locomolcd are probably disimmilar enough to make merging them not worthwhile... Jul 05 12:16:57 cbrake, this morning, I'm doing a fresh build with the patches you pushed for me :-) Jul 05 12:16:59 do13_: The audio connections sound interesting... Jul 05 12:17:14 cbrake, why? Jul 05 12:17:14 RP: only in 2.6.16 which is used in 3.5.4.1 only currently... Jul 05 12:17:30 do13_: I was wondering if ASoC could help. We'd have to write a fair chuck of audio drivers though... Jul 05 12:17:51 hrw|work: We should try and keep 2.6.17 the same, otherwise we lose data when we next upgrade Jul 05 12:18:35 RP: I would like to avoid it Jul 05 12:18:42 Gerrath_: just curious why I'm running into bootstrap-image problems and you are not. Jul 05 12:18:51 Gerrath_: are you building the bootstrap-image target? Jul 05 12:20:02 hrw|work: Add it them all. I'll have to start using my own defconfigs if the compile time bothers me enough... Jul 05 12:20:23 ok Jul 05 12:21:34 cbrake, well I will let you know how things work out this morning on my new build with the latest OE. Jul 05 12:21:39 RP: I was first thinking about enabling all stuff in 2.6.16 only, building, releasing, disabling Jul 05 12:21:58 RP: I think ASoC is the preferred way. I'ver never touched the SA1100 DMA api :) Jul 05 12:22:37 do13_: Having as cause to look at the old 2.4 corgi audio drivers recently, I fully appreciate what ASoC does for us :) Jul 05 12:22:42 cbrake, I had a bootstrap-image error as well on my work computer but just ruled it out because I had not updated OE on it in a while. The builds at home (over the weekend) were all up to date OE builds. Jul 05 12:22:58 hrw|work: We should continue to try and enable things on a request basis IMO Jul 05 12:23:06 Gerrath_: ok, I'm looking into it now ... Jul 05 12:23:13 cbrake, the fresh download/build I'm running right now is on my work computer so we will see what happens. Jul 05 12:23:30 Gerrath_: ok, sounds good. Jul 05 12:23:32 RP: 'if you want some extra from kernel then build it' way? Jul 05 12:25:14 Gerrath_: I still need to sort the *ARCH settings in the maching conf as well. Jul 05 12:25:30 03hrw 07org.oe.oz354x * r9a9275ed... 10/packages/links/ (files/ac-prog-cxx.patch links-x11_2.1pre22.bb): links-x11: added 2.1pre22 from .dev Jul 05 12:25:35 03hrw 07org.oe.oz354x * r294c9caf... 10/packages/wlan-ng/ (wlan-ng-utils_0.2.1-pre26.bb wlan-ng-utils_0.2.3.bb): wlan-ng-utils: package man pages Jul 05 12:25:39 03hrw 07org.oe.dev * r938588db... 10/packages/wlan-ng/ (3 files): wlan-ng-utils: package man pages Jul 05 12:31:17 ooh scary, OZ users are getting links goodness :-) Jul 05 12:31:56 XorA: haha Jul 05 12:32:22 I still wish there was a qt gfx driver for links Jul 05 12:32:59 XorA: I even add newer abiword Jul 05 12:33:16 hrw|work: gah, those users are spoiled Jul 05 12:33:30 XorA: added some dictionaries? Jul 05 12:34:19 hrw|work: only for languages I speak Jul 05 12:34:48 * ixoth flashing his new mips device with his new oe-based distro Jul 05 12:38:12 cbrake, here is the version of OE I compiled over the weekend: 9e8da2379fb73583cd868e8429e611b0450b8c7f Jul 05 12:39:59 Gerrath_: aha, NOTE: multiple providers are available (task-bootstrap-unionroot, task-bootstrap); Jul 05 12:40:07 hrw|work: "File a bug and we'll add it to the defconfig" Jul 05 12:40:16 RP: sure Jul 05 12:40:46 1018M of packages... Jul 05 12:41:02 oz 3.5.4.1 is scary Jul 05 12:41:45 Gerrath_: so OE must be building the unionroot version, but something goes wrong at ipkg time when bootstrap-image is looking for task-bootstrap Jul 05 12:46:56 task-bootstrap-unionroot.bb defines PROVIDES=task-bootstrap, but at bootstrap-image time, ipkg can't find task-bootstrap. Jul 05 12:47:12 * cbrake takes a look at the task-bootstrap ipkg file ... Jul 05 12:48:47 * cbrake wonders if RPROVIDES should also be set to task-bootstrap??? Jul 05 12:50:06 koen: I'm suprised - we lack gnumeric 1.7.x... Jul 05 12:50:50 cbrake: It sounds like it should Jul 05 12:52:02 oh Jul 05 12:52:13 building tcpdump fails with "| CROSS COMPILE Badness: /usr/include in INCLUDEPATH: /usr/include" Jul 05 12:53:15 Gerrath_: can you take a look at your tmp/deploy/ipk dir and see if oe is building task-bootstrap or task-bootstrap-unionroot? Jul 05 12:53:22 hrw|work: we got the latest stable Jul 05 12:53:41 hrw|work: 1018M of packages an no regression testing? :) Jul 05 12:53:42 koen: yep - just checked that 1.7 is devel Jul 05 12:54:02 mithro: users are our testers Jul 05 12:54:22 cbrake, I have RPROVIDES set to task-bootstrap in my personal meta file. Jul 05 12:54:29 cbrake, I'll check.. Jul 05 12:56:57 cbrake, task-bootstrap. Jul 05 12:58:02 Gerrath_: ok, your builds should be fine then. Apparently, it is somewhat random what package OE builds if there are multiple providers. Jul 05 12:58:18 * cbrake should watch his build logs closer ... Jul 05 13:02:34 XorA: the mplayer.bb is wrong for the pxa270 clamshells Jul 05 13:02:48 XorA: it enables bvdd, but that's only present in 2.4.crappix Jul 05 13:03:33 mplayer-atty is obsoleted by 1.0pre8 iirc Jul 05 13:04:45 hrw|work: no-one reads .bb files or commit messages :-) Jul 05 13:08:13 if it's obsolete, why is it still present? Jul 05 13:08:28 we have an scm, remember? Jul 05 13:10:06 do most people set PREFERRED_PROVIDER_task-bootstrap in their local.conf? Jul 05 13:10:58 cbrake: No, there are DEPENDS in most of the image .bb files which have some influence over that Jul 05 13:12:12 koen: that can be said of a large chunk of OE metadata Jul 05 13:12:29 koen: and some people might still be using cxx00 2.4 models Jul 05 13:14:11 XorA: hopefully there are no official images for cxx00 with 2.4 kernels Jul 05 13:14:25 XorA: heh, if it were me that large chunk would be a lot less smaller Jul 05 13:14:27 Koen got a sec? Jul 05 13:15:00 Was it okey to push that opie-cvs..versions.inc file, using SRCDATE? Jul 05 13:15:25 Kristoffe: no idea, I don't use opie Jul 05 13:15:35 hehe Jul 05 13:15:37 gpe freak Jul 05 13:15:59 I wish gpe built for me Jul 05 13:16:04 Kristoffe: I do not think that this is needed - opie 1.2.2 should be ok Jul 05 13:16:04 then I could join in on the fun Jul 05 13:16:16 Opie 1.2.2 doesnt build for me at all Jul 05 13:16:22 its taken from cvs may right? Jul 05 13:16:27 Kristoffe: it is Jul 05 13:16:29 and my stuff started working in june Jul 05 13:17:09 Im aware of that most people here dont build much opie related stuff, but someone else with a new/obselete platform might have use of it Jul 05 13:17:37 Kristoffe: then I would use 1.2.2 + SRCDATE_libopie = 'workingdate' but not whole opie from one SRCDATE Jul 05 13:17:46 but do anything to get it working Jul 05 13:18:20 I had major hassle with it last night, and this was the best solution I could come up with at the time. Will probobly clean up in the future, but does the trick Jul 05 13:18:50 if it works then use it Jul 05 13:18:59 Btw, one question. When doing -c clean it doesnt clean out the .ipk Jul 05 13:19:09 yep Jul 05 13:19:38 package-index just grabs every ipk and makes index over them, so whatever I do (choose which packages I want to include) it shouldnt affect? Jul 05 13:19:55 Im asking cause I got different opie-xxxx versions and I just want one Jul 05 13:20:09 package-index will move old ones into morgue Jul 05 13:20:18 Aha, doh that explains it Jul 05 13:20:21 thx Jul 05 13:21:03 Btw, can one delete morgue without any future hassle? Jul 05 13:21:16 yes Jul 05 13:21:19 goodie Jul 05 13:24:04 03hrw 07org.oe.oz354x * r7af21694... 10/packages/tslib/ (5 files in 3 dirs): dropped tslib-maemo Jul 05 13:24:08 03hrw 07org.oe.oz354x * r087abd85... 10/packages/abiword/ (abiword-plugins_2.4.5.bb abiword_2.4.5.bb): abiword: added 2.4.5 from .dev Jul 05 13:24:13 03hrw 07org.oe.oz354x * rcc502f6d... 10/packages/gnome/ (goffice_0.2.0.bb goffice_0.2.1.bb): goffice: updated to 0.2.1 Jul 05 13:24:17 03hrw 07org.oe.oz354x * rf859021f... 10/packages/gnumeric/ (11 files in 5 dirs): gnumeric: updated to 1.6.3 and dropped older Jul 05 13:26:53 damn... thought I was nearly there with my x86 toolchain Jul 05 13:26:56 but no :( Jul 05 13:27:35 03koen 07org.oe.dev * r17886af9... 10/conf/machine/include/ (tune-ep9312.conf tune-iwmmxt.conf): Jul 05 13:27:35 tune-ep9312, tune-iwmmxt: use -mcpu, -march and -mtune Jul 05 13:27:35 : koen: aaargh, what idiots made the options do totally different things between x86 and arm? Jul 05 13:27:51 ;) Jul 05 13:28:17 koen: can you build esound-gpe-20020817-r4 ? Jul 05 13:28:25 dot past glibc-initial Jul 05 13:28:31 even got past intermediate Jul 05 13:28:34 RP: ping Jul 05 13:28:46 but now it's: add-on directory "nptl" does not exist Jul 05 13:29:06 hrw|work: pong Jul 05 13:29:29 RP: why eds-dbus need libiconv? iconv from glibc is not enough? Jul 05 13:29:59 funky Jul 05 13:30:21 I always thought glib-2.0 was supposed to do all the iconv stuff for gtk apps Jul 05 13:30:37 koen: :-D Jul 05 13:30:54 koen: so was wrong Jul 05 13:31:00 koen: so you was wrong Jul 05 13:31:03 hrw|work: Its supposed to be virtual/libiconv Jul 05 13:31:09 hrw|work: apparently Jul 05 13:31:19 Its fixed in poky but not in OE :-/ Jul 05 13:31:23 do13_: building it now Jul 05 13:34:10 do13_: Packaged contents of esound-gpe into /data/build/koen/OE/build/tmp/angstrom/deploy/ipk/libesd0_20020817-r4_iwmmxt.ipk Jul 05 13:34:13 grrrr Jul 05 13:34:18 ~lart glibc do_stage Jul 05 13:34:19 * ibot shoves a crumpet down glibc do_stage's throat, happy now?! Huh? Want some JAM with that? Jul 05 13:34:31 RP: users wanted tinymail Jul 05 13:34:45 koen: still fighting with glibc? Jul 05 13:34:52 koen: SoC not going well? Jul 05 13:34:59 koen: is the lib in your staging dir? Jul 05 13:35:02 hrw|work: I'll change it to use virtual/libnv Jul 05 13:35:05 hrw|work: yes Jul 05 13:35:13 XorA: no progress, just anguish Jul 05 13:35:14 RP: ok. I changed it in .oz354x already Jul 05 13:35:44 do13_: yes Jul 05 13:35:50 koen: How's the SoC project going? Jul 05 13:35:53 Hmm. http://pastebin.ca/79513 Jul 05 13:36:05 I have only symlinks Jul 05 13:36:16 RP: no real progress, but I suspect to best glibc this week Jul 05 13:37:37 RP: | make[3]: dbus-binding-tool: Command not found Jul 05 13:37:46 RP: eds-dbus need dbus>0.23? Jul 05 13:38:03 hrw|work: yes Jul 05 13:38:16 ok so no tinymail for oz 354x Jul 05 13:38:47 koen: wil your SoC be glibc/uclibc/libc agnostic ? Jul 05 13:38:58 ade|desk: yes Jul 05 13:39:02 groovy Jul 05 13:39:44 do13_: -rwxr-xr-x 1 koen koen 44019 2006-07-05 15:51 libesd.so.0.2.28 Jul 05 13:40:45 03rpurdie 07org.oe.dev * rf1c9afde... 10/packages/eds/eds-dbus_svn.bb: eds-dbus: DEPEND on virtual/libiconv Jul 05 13:51:11 koen: thx, found my prob. locale settings are wrong Jul 05 14:02:14 lol, I like how mirrors are all in one file "bitbake.conf" The default sourceforge mirrow was down so I just changed the link in that file and everything works. I remember using buildroot and having to change the URL in multiple Make files (each package) :-) Jul 05 14:03:21 Gerrath_: that's probably wrong way to do it, you should add mirrors into local.conf and leave bitbake.conf alone... Jul 05 14:04:07 Gerrath_: I set SOURCEFORGE_MIRROR in local.conf Jul 05 14:04:23 03kristoffer 07org.oe.dev * r8c16cdbf... 10/ (4 files in 3 dirs): Jul 05 14:04:23 linux-jlime-sh3/LinuxSH-2.6.17.patch : Patch from Linux-SH for 2.6.17 Jul 05 14:04:23 * Addition of patch from Linux-sh tree Jul 05 14:04:23 linux-jlime-sh3-2.6.17/defconfig_jlime : Addition of defconfig Jul 05 14:04:23 linux-jlime-sh3_2.6.17.bb : New jlime kernel based on 2.6.17+linuxSH Jul 05 14:04:40 hrw|work, ixoth: thats even better :-) Jul 05 14:05:12 :) Jul 05 14:06:11 m orning Jul 05 14:09:20 CIA-9: ping Jul 05 14:10:00 03hrw 07org.oe.oz354x * r9aa284d5... 10/packages/db/ (db_4.3.27.bb db_4.3.29.bb): db4: updated to 4.3.29 Jul 05 14:10:04 03hrw 07org.oe.oz354x * r4cb783b3... 10/packages/linux/ (2 files in 2 dirs): linux-openzaurus: applied patch from #1084 to get squashfs built Jul 05 14:10:09 03hrw 07org.oe.oz354x * r8bb17d41... 10/packages/linux/ (4 files in 2 dirs): linux-openzaurus: enabled ALSA USB Audio Jul 05 14:10:13 03hrw 07org.oe.dev * ref6b7850... 10/packages/linux/ (6 files in 4 dirs): linux-openzaurus: applied patch from #1084 to get squashfs built Jul 05 14:10:33 RP: I will sync defconfigs later today Jul 05 14:12:58 RP: I tried to use the corgi audio setup on poodle to see how broken it is...:) Jul 05 14:14:01 oh goody, the uni switches to OWA for mail Jul 05 14:14:37 I can see that it is much better as the old system, since people can't login nor receive mail anymore Jul 05 14:14:44 ~lart microsoft crap Jul 05 14:14:44 * ibot whips microsoft crap with a wet and grimy noodle just because Jul 05 14:15:23 looks like snd-soc-core has some unresolved symbols Jul 05 14:15:51 OWA? Jul 05 14:16:38 Outlook Web Access Jul 05 14:16:49 O_utlook W_ith A_ttitude??? Jul 05 14:18:00 koen: exchange is fine as long as they start the IMAP service and allow relay SMTP Jul 05 14:18:50 ade|desk: don't know, all I want is that fetchmail gets the mail and dumps it in my maildir Jul 05 14:19:01 fairplay Jul 05 14:19:14 when testing the new .fetchmailrc I noticed that it isn't working *at all* Jul 05 14:19:19 when does fetchmail get MAPI working ;) Jul 05 14:20:20 OWA is funny on non-IE, half the menus are missing, right click menus won't work with non-IE Jul 05 14:20:48 and logout doesn't logout if you use firefox Jul 05 14:21:29 I'm getting the error: gnu-configize: command not found when trying to bitbake any of the glibc_2.3.* bb files Jul 05 14:21:40 keeps the session alive so people can read your mail if you leave the computer without killing session and removing cookies .. fun Jul 05 14:21:45 any ideas what that might be hinting at? Jul 05 14:21:52 tkp: you didn't build gnu-config-native? Jul 05 14:22:31 tkp: you used '-b' bitbake option? Jul 05 14:22:39 yes I did Jul 05 14:22:40 bitbake --help Jul 05 14:22:54 -b --broke-build break my build badly Jul 05 14:23:21 hmm. basically I had run make task-bootstrap, and it died when it got to glibc_2.3.2 Jul 05 14:23:47 so I thought I would try some other versions... can I do that without rebuilding theentire toolchain? Jul 05 14:24:45 hmm.. one day I have to check random MACHINE (!arm one) and random DISTRO and do build Jul 05 14:27:43 03hrw 07org.oe.oz354x * r56979359... 10/packages/t1lib/t1lib_5.0.2.bb: t1lib: ship static lib in -dev package - close #1133 Jul 05 14:27:47 03hrw 07org.oe.dev * rcaa51a01... 10/packages/t1lib/t1lib_5.0.2.bb: t1lib: ship static lib in -dev package - close #1133 Jul 05 14:28:43 hey guys Jul 05 14:29:05 i just managed to finally get task-bootstrap to work Jul 05 14:29:08 hey CosmicPenguin Jul 05 14:29:08 great Jul 05 14:29:11 so now i need a quick pointer or two Jul 05 14:29:19 mornintg Jul 05 14:29:23 i now have a .ipk of the unionroot Jul 05 14:29:36 i take it that is basically a system ready to go? Jul 05 14:31:44 should be Jul 05 14:32:26 how do i image that onto a disk? Jul 05 14:34:11 build bootstrap-image Jul 05 14:34:28 ah - a different target Jul 05 14:36:25 cbrake, everything compiled fine on my work computer with the latest OE (updated this morning). Jul 05 14:41:44 when a package is marked as native - is that native to the target or the host? Jul 05 14:42:27 host Jul 05 14:43:08 right Jul 05 14:43:42 so i take it if i installed some of this stuff normally it would use that? or does it always assume its not there? Jul 05 14:44:46 wow Jul 05 14:44:59 openzautus.org running at a crawl Jul 05 14:45:04 *openzaurus Jul 05 14:45:20 Gerrath_: that's not the latest ;) Jul 05 14:45:25 jkp: duhh, sf.net Jul 05 14:45:47 koen: is there any way to change the mirror its using? Jul 05 14:45:53 its going at 200b/s Jul 05 14:46:26 what is going on with OE..... Jul 05 14:46:31 so many revs in the past few days Jul 05 14:46:48 JustinP: oz 3541 goes into release Jul 05 14:46:59 JustinP: summertime, holiday time Jul 05 14:47:48 Gerrath_: great, things should work here as well once I get the bootstrap stuff straightened out. Jul 05 14:48:24 ahhh Jul 05 14:48:26 nice Jul 05 14:48:31 ctl-c :) Jul 05 14:48:44 ade|desk: what's the state of uclibc_svn in OE? Jul 05 14:48:56 in OE , crap Jul 05 14:49:05 koen, thats why I specified this morning since I know OE changes by the minute ;-) Jul 05 14:49:25 in my so f**ked up tree i can't push .. not too bad Jul 05 14:49:25 hrw|work: did you fix the oz-bootsplash problem? Jul 05 14:49:38 hrw|work: I just saw it this morning, but there have been >50 revs since then... Jul 05 14:50:10 hrw|work: oz-bootsplash wasn't being built for images so the do_rootfs died. I had to manually build it Jul 05 14:50:25 cd .. Jul 05 14:50:29 lol Jul 05 14:50:31 heh Jul 05 14:50:43 ~ewi Jul 05 14:50:44 JustinP: I create images from feed - feed contain oz-bootsplash now Jul 05 14:50:44 rumour has it, ewi is ewi546.ewi.utwente.nl - main backup server for OE monotone. Jul 05 14:50:57 shit.. some new stuff built... Jul 05 14:51:12 hrw|work: well, I don't really feel like rebuilding from scratch yet again to test this.... Jul 05 14:51:51 JustinP: I can probably test it but not now Jul 05 14:52:10 ibot: no ewi is ewi546.ewi.utwente.nl - secondary backup server for OE monotone. Jul 05 14:52:12 okay, koen Jul 05 14:52:37 which is the OE snapshot on there koen ? Jul 05 14:52:54 the .db or tar'ed stuff Jul 05 14:53:16 How do I go about tracking the failure of a patch during do_patch? Apparently, bitbake cleans up after itself so the sources and the patch are gone. Trying to fix bug 410 and build kaffe-qt. Jul 05 14:54:35 Laibsch: bitbake something. if it fail on do_patch then cd into WORKDIR and use quilt, editor etc to fix it, refresh patches, put them in metadata and rebuild Jul 05 14:54:38 ade|desk: see GettingStarted ;) Jul 05 14:54:40 ade|desk: http://www.openembedded.org/snapshots/ Jul 05 14:55:03 ~OE.db Jul 05 14:55:23 ibot: OE.db is http://www.openembedded.org/snapshots/OE.db.bz2 Jul 05 14:55:43 okay, hrw|work Jul 05 14:55:47 ~OE.db Jul 05 14:55:54 i heard oe.db is http://www.openembedded.org/snapshots/OE.db.bz2 Jul 05 14:55:57 feel free to edit ;D Jul 05 14:57:33 so let me get this right Jul 05 14:57:54 everything i might want to do, basically needs to be done by writing bitbake recipies Jul 05 14:58:26 im looking at some of them, for example the bootstrap one, and it just leaves me asking where everything actually happens! Jul 05 15:00:50 koen: hrw|work: cheers Jul 05 15:01:29 jkp: err for the time beeing you should ignore bootstrap as best as you can. It's overy complicated Jul 05 15:02:02 CoreDump|home: my goal i to build one image / ext2 file that represents everything thohjg Jul 05 15:02:10 bootstrap is evil Jul 05 15:02:22 jkp: then I guess you have no choice =) Jul 05 15:02:24 how should i do what i want then? Jul 05 15:02:38 CoreDump|home: how do you go about building your embedded system if not like that? Jul 05 15:02:49 i come from buildroot and thats what im used too.... Jul 05 15:03:02 maybe there is a better way Jul 05 15:03:35 jkp: you want everything in one image? That woul dend up huge.... Jul 05 15:03:42 jkp: I was under the impression you wanted to learn what .bb's are all about. Bootstrap would be the worst choice for that IMO. But since you indeed want to create complete images you'll have not much choice Jul 05 15:03:46 everything for the target Jul 05 15:04:34 "for the target"? Jul 05 15:04:44 goodbye everyone Jul 05 15:04:50 if you just want an image with your own set of packages installed to it see, say, e-image Jul 05 15:04:56 it's pretty simple Jul 05 15:05:07 ok , im getting the impressions this isnt the OE way of doing things. In BR we had a makefile that build the toolchain. then then target binaries and libs, then put them into the fakeroot, then built an image for the final system Jul 05 15:05:20 JustinP: how is that different from bootstrap? Jul 05 15:05:31 jkp: that's basically what OE does, only wayy more flexible Jul 05 15:05:54 heh - nice. but i still dont have a picture of the overall workflow Jul 05 15:06:08 jkp: bootstrap is just a cimplicated little image...I'm suggetsing you look at a simpler example first Jul 05 15:06:09 jkp: If you want to look at something simpler, you could have a look at poky. Its like OE but is a bit leaner/cleaner. It might illustrate how things work better Jul 05 15:06:13 ~poky Jul 05 15:06:18 poky is, like, http://projects.o-hand.com/poky Jul 05 15:06:20 jkp: you use bitbake+OE to build packages. then you have packages in tmp/deploy/ipk Jul 05 15:06:45 jkp: then you can build image which contain some of packages - it can be cramfs, jffs2, tarball, ext2fs Jul 05 15:06:50 RP: ill take a look at that Jul 05 15:07:05 jkp: you install image on target device and can add packages to it with ipkg Jul 05 15:07:07 * JustinP goes back to sleep Jul 05 15:07:08 hrw|work: ok, its the last step im hazy on Jul 05 15:07:21 RP: mallum gives you a euro every time you mention that? :p Jul 05 15:07:46 how do you take a load of ipkgs and turn that into an image of a fs? Jul 05 15:07:51 jkp: jkp-image.bb can contain 'inherit image_ipk' + IPKG_INSTALL = "my own set of packages which I want in image" + RDEPENDS = "${IPKG_INSTALL}" Jul 05 15:08:13 hrw|work: ah...its that simple? Jul 05 15:08:24 jkp: yep Jul 05 15:08:29 how about choosing the format of the image? Jul 05 15:08:31 jkp: Basically with OE, each piece of software is a .bb file, be them native pacjkage that get used as build tools (autoconf-native), bits of the toolchain (binutils-cross) or target device programs (busybox). We then have .bb files which combine the packages (meta/task-*.bb) and then image .bb files which create images from groups of packages Jul 05 15:08:37 jkp: look at IMAGE_FSTYPES Jul 05 15:08:54 right Jul 05 15:08:57 so the steps are Jul 05 15:09:03 jkp: IMAGE_FSTYPES="ext2 jffs2 tar.gz" for example Jul 05 15:09:06 packages/angstrom/ has some examples for images and tasks Jul 05 15:09:37 koen: Why are they not in meta? Jul 05 15:09:45 * RP was totally unaware of them :-/ Jul 05 15:09:55 build some packages for the target, then combine them all using the kind of .bb file you outlined Jul 05 15:10:03 exactly Jul 05 15:10:16 RP: I'm affraid that some people will revert them because they 'duplicate' behaviour Jul 05 15:10:17 image_ipk is native to bitbake? Jul 05 15:10:24 jkp: Except that the image .bb is clever enough to know to build the packages it needs Jul 05 15:10:37 RP sure Jul 05 15:10:40 dependencies Jul 05 15:10:46 one more question Jul 05 15:11:19 how do you define whether something is building for the native system, or for the target system? Jul 05 15:11:26 do you assume it is building for the target system? Jul 05 15:11:48 jkp: it's always for the target system, unless it's called foo-native Jul 05 15:11:49 jkp: At present, native packages are named *-native.bb Jul 05 15:12:07 if i named something native, would it automatically build targetted native? Jul 05 15:12:20 ie - is that the criteria Jul 05 15:12:25 jkp: no, you'd need 'inherit native' Jul 05 15:12:28 jkp: No, you'd "inherit native" for that Jul 05 15:12:30 right Jul 05 15:12:49 where are things such as native and image_ipk actually defined? Jul 05 15:12:58 are they part of bitbake? or specific to OE? Jul 05 15:12:59 classes/ Jul 05 15:13:07 jkp: inherit native means include classes/native.bbclass Jul 05 15:13:16 ish :) Jul 05 15:13:16 right Jul 05 15:15:33 RP: what do you think about changing 'inherit' from 'include' to 'require'? Jul 05 15:16:23 Personally, I like that Jul 05 15:16:39 I don't like the idea of inherit being optional Jul 05 15:16:42 right, so now i see how it all fits together :) Jul 05 15:16:45 koen: You mean the behaviour? Jul 05 15:17:06 I can't see why someone would want an inherit to fail. Its not like the conf file arguments... Jul 05 15:17:13 ie - im a coder, and just seeing these sparse files with no guts makes me nervous :) Jul 05 15:17:27 jkp: The guts of OE is in the classes Jul 05 15:17:33 RP: i noticed :) Jul 05 15:17:39 jkp: Its one of its powerful features :) Jul 05 15:17:59 RP: it would catch 'inherit naive' type of bugs Jul 05 15:18:02 RP: you suggested poky...do you really think I ought to do that? Jul 05 15:18:07 or look at it at least Jul 05 15:18:52 jkp: Its a simple cut down version of OE. If you want something simpler to look at, it might be interesting. If you're now happy with OE, you probably don't need to. Jul 05 15:19:09 jkp: At present poky mainly targets arm devices (zaurus and emulation) Jul 05 15:19:25 I'd like to see it gain x86 support though Jul 05 15:19:34 hrw|work: The question is where do I find the stuff to fix? bitbake does a "bitbake $PACKAGE clean" if it fails, so what I need is gone, isn't it? Jul 05 15:19:37 i think now i understand how it works im happier with it Jul 05 15:19:54 Laibsch: it does not do it Jul 05 15:19:55 Laibsch: it isn't Jul 05 15:20:21 Laibsch: Are you using tinderbox? Jul 05 15:20:25 yes Jul 05 15:20:32 Laibsch: Disable that ;-) Jul 05 15:20:40 OK Jul 05 15:20:41 I've figured out the glibc 2.4 problem. The "rm -rf ${S}/bits" that's a workaround for the CVS problems breaks anything not using ports. I presume this is left over from being copied from a cvs .bb file... Jul 05 15:21:24 v8jlene: It sounds likely - I added that for 2.3.5 Jul 05 15:25:21 RP: Yep. And I think angstrom is ARM and it's the only thing that uses glibc 2.4 so far. x86 and sh were both broken and neither have been moved to ports. Just doing a clean build now to verify x86. Well test sh3 tommorow.. Jul 05 15:25:27 RP: I would like to help add x86 to poky, but I think we're probably too closely tied into OE at this point to switch over Jul 05 15:25:54 And poor Ray is busy enough as it is Jul 05 15:26:30 v8jlene: angstrom isn't tied to arm Jul 05 15:26:47 CosmicPenguin: I suspect it will happen soon enough anyway :) Jul 05 15:27:06 CosmicPenguin: I'm hoping to see a qemux86 :) Jul 05 15:27:07 We really, really, really need to make our tree available too Jul 05 15:27:13 koen: yeah, ok - but arm is what you have tested against? Jul 05 15:27:21 v8jlene: yes Jul 05 15:27:32 CosmicPenguin: How different is it from OE itself? Jul 05 15:27:59 RP: We've overloaded some of the key classes and added many others Jul 05 15:28:22 nother quicki question: does stuff build against glibc by default? i mean, im trying to get away from uclibc at this point Jul 05 15:28:32 We're mainly concerned with building our releases - we've done some pretty cool things with quilt Jul 05 15:28:49 jkp: it depends on distro Jul 05 15:28:56 CosmicPenguin: I like quilt - proper intergration with OE would be cool :) Jul 05 15:28:58 jkp: switching glibc<>uclibc is easy Jul 05 15:28:58 ahhh yeah Jul 05 15:29:00 i remember Jul 05 15:29:08 do most things compile either way? Jul 05 15:29:16 we had a nightmare with mplayer and uclibc Jul 05 15:29:22 jkp: We have people using both Jul 05 15:29:26 RP: we store our kernels in RCS as quilt trees - so we have an OE class that can grok that Jul 05 15:29:56 CosmicPenguin: You've seen my nightmare kernel .bbs ? :) Jul 05 15:30:04 RP: heh - no Jul 05 15:30:20 CosmicPenguin: Look at linux-openzaurus_2.6.16.bb ;-) Jul 05 15:30:28 RP: what about using a custom kernel config or a custom busybox config? Jul 05 15:30:54 jkp: we have overrides for it Jul 05 15:31:05 RP: heh - yeah, I know the feeling Jul 05 15:31:19 busybox/files/YOURDISTRO/ or busybox/files/YOURMACHINE can contain distro/machine specific files Jul 05 15:31:26 RP: right now, my biggest push is to make OE better able to build and understand initrds Jul 05 15:31:26 CosmicPenguin: It has about 90 patches. Thankfully most are now upstream :) Jul 05 15:31:45 hrw|work: and those will always override everything else? Jul 05 15:31:53 jkp: yes Jul 05 15:32:07 jkp: Basically there are lots of ways you can handle it, depending on what you want to achieve Jul 05 15:32:08 hrw|work: excellend Jul 05 15:32:12 OE++ :D Jul 05 15:32:27 looks like hopefully we are going to be able to get up and running fairly fast Jul 05 15:33:52 CosmicPenguin: We probably need a task-initrd.bb ... Jul 05 15:34:01 so say there are options in EXTRA_OECONF that i want to remove Jul 05 15:34:01 CosmicPenguin: I've pushed a cset that should make building initrds easier: http://www.openembedded.org/viewmtn/revision.psp?id=f813b9fe0054ffebcf34d6ab5ec92091b1627194 Jul 05 15:34:19 CosmicPenguin: (and an asssociated initrd-image.bb) Jul 05 15:34:19 what is the best way to do that? Jul 05 15:34:39 CosmicPenguin: that way to know the important modules without using bootstrap* Jul 05 15:34:46 RP: yeah - we've been building the initrd with the same glibc and kernel as the main image, and thats probably not smart Jul 05 15:34:55 jkp: EXTRA_OECONF_yourmachine = "" Jul 05 15:35:11 CosmicPenguin: btw, aren't the alchemy boards in OE? Jul 05 15:35:19 food time, bbl Jul 05 15:35:31 so in the package folder for a package called foo Jul 05 15:35:34 jkp: You need to learn about overrides in OE/bitbake Jul 05 15:35:47 RP: is it in the bitbake manual? Jul 05 15:35:52 RP: I've been playing with OLPC, and we use the the -nostdlib trick to compile in uclibc for the LinuxBIOS payload Jul 05 15:36:19 I think we can do the same thing to build uclibc for the initrd - though the current uclibc.bb assumes its the standard c library for the image Jul 05 15:36:25 We would need to teach glibc and uclibc to live well together Jul 05 15:37:34 koen: the xxs1500 is an alchemy board Jul 05 15:37:35 CosmicPenguin: The 770 uses a uclibc initrd and a glibc rootfs Jul 05 15:37:49 RP: Oh, I didn't know that Jul 05 15:37:57 RP: so somebody already solved my problems.. as usual Jul 05 15:38:00 * CosmicPenguin hugs OSS Jul 05 15:38:10 CosmicPenguin: We have an OLPC board but I'm not going to have anything to do with it and won't be doing anything OE related with it. There was a lot of high feelings at GUADEC and its too political for us to involve OE :-( Jul 05 15:38:21 anyone know what this means when building the toolchain? Jul 05 15:38:22 "stop trying to boot 2.4 kernels on libc's you build against linux-2.6 headers" Jul 05 15:38:24 oops Jul 05 15:38:35 "fatal: kernel too old" Jul 05 15:38:47 RP: I'm not seeing how DEPENDS in the image .bb files are influencing which task-bootstrap is being built? Jul 05 15:38:49 RP: yeah, I get that feeling too - but I've been mirroring the Fedora image with OE Jul 05 15:39:13 RP: both task-bootstrap and task-bootstrap-unionroot provide task-bootstrap ... Jul 05 15:39:15 CosmicPenguin: I simply can't get involved :-/ Jul 05 15:39:16 RP: and, for what its worth, the first kernel/rootfs that booted with LinuxBIOS was OE generated... :) Jul 05 15:39:36 CosmicPenguin: That gives me some hope :) Jul 05 15:39:45 RP: for a while there, I had hoped that they would go with Matchbox, - Jim Gettys had nothing but good things to say about Matthew at the Linux PM summit Jul 05 15:40:07 http://www.infradead.org/~jcrouse/olpc-screenshot.jpg Jul 05 15:40:18 CosmicPenguin: Some thought they were going with matchbox and others not at GUADEC. Its total confusion as far as I can see... Jul 05 15:40:30 Thats from my camera phone, so its fuzzy - but if you squint, you'll see OpenEmbedded on the banner.. :) Jul 05 15:40:51 CosmicPenguin: I saw it - Even the shape of the boot messages is familar :) Jul 05 15:41:11 cbrake: Does task-bootstrap RPROVIDE task-bootstrap? Jul 05 15:41:34 RP: I just got X up - maybe I'll toss on my matchbox images for shits and giggles.. :) Jul 05 15:41:44 cbrake: If not, ipkg won't see it. How ipkg chooses between two task-bootstraps is a more difficult quesiton though Jul 05 15:41:53 /nick XorA|gone Jul 05 15:42:15 CosmicPenguin: If you want a low overhead X+mb image, poky generates them but I didn't say that ;-) Jul 05 15:42:32 Whos talking? Jul 05 15:42:34 :) Jul 05 15:42:54 cosmicpenguin: sash! Jul 05 15:42:59 heh Jul 05 15:43:01 RP: no, task-bootstrap does not RPROVIDE task-bootstrap, but it should use the ipkg name be default. Jul 05 15:43:55 RP: nod, if you have two ipkg that RPROVIDE task-bootstrap, that would be a problem. At this point, I'd just like to know the correct way to ensure the right one gets built. Jul 05 15:44:35 cbrake: To ensure it builds the right one set PREFERRED_PROVIDER_task-boostrap Jul 05 15:45:22 RP: I suppose this should be done in local.conf vs. the distro? Jul 05 15:45:53 RP: I can imagine that the distro could support both task-bootstrap packages (in theory) Jul 05 15:46:23 cbrake: Right, so its a local.conf issue Jul 05 15:49:57 hrw|work, mickeyl: http://gnumonks.org/~laforge/weblog/2006/06/30#20060630-sharp-slc3200 Jul 05 15:50:30 koen: news@11 Jul 05 15:52:02 hmm, looks like DEFAULT_PREFERENCE is not working for me. task-bootstrap-unionroot is being built rather than task-bootstrap :-\ Jul 05 15:52:43 ~large opie-image/merge_feeds Jul 05 15:52:47 ~lart opie-image/merge_feeds Jul 05 15:52:47 * ibot eats opie-image/merge_feeds and falls over dead Jul 05 15:54:33 ibot: lart hrw backwards Jul 05 15:54:33 * ibot executes killall -HUP hrw backwards Jul 05 15:56:43 cbrake: Have a look at the output of bitbake -DDD imagename and see if it says why its doing that? Jul 05 15:56:56 RP: ok Jul 05 15:56:59 aaargh Jul 05 15:57:00 http://www.openembedded.biz/ Jul 05 15:57:16 note the new 'offline' message Jul 05 15:58:46 ~lart append <> += changes Jul 05 15:58:47 * ibot does a little 'renice 20 -u append <> += changes' Jul 05 16:01:53 RP: http://pastebin.ca/79636 Jul 05 16:02:48 cbrake: Is says you didn't set PREFERRED_PROVIDER_task-bootstrap ? Jul 05 16:03:14 BBMASK="(task-bootstrap-unionroot)" Jul 05 16:03:23 RP: nod, but should not DEFAULT_PREFERENCE kick in here? Jul 05 16:04:04 RP: task-bootstrap-unionroot:DEFAULT_PREFERENCE = "-1" Jul 05 16:04:20 hrw|work: so is this what you do in your local.conf files? Jul 05 16:05:32 cbrake: Set PREFERRED_PROVIDER_task-bootstrap = "task-bootstrap" in your local.conf file. Jul 05 16:05:47 cbrake: I'm not sure I've see that DEFAULT_PREFERENCE syntax before... Jul 05 16:05:50 cbrake: no - I never had problem which you have Jul 05 16:06:12 hrw|work: I suspect DEFAULT_PREFERENCE is working for you then. Jul 05 16:06:52 cbrake: looks like Jul 05 16:06:56 RP: yeah, I'm sure that will work -- just kind of curious why no one else is seeing this problem and DEFAULT_PREFERENCE is not working. Jul 05 16:07:07 cbrake: ah, right, its in the .bb file... Jul 05 16:07:19 strange... Jul 05 16:08:08 RP: may be some obscure BB bug -- other builds seem to be working fine for me. Jul 05 16:08:26 cbrake: Could be. You're using 1.4.2? Jul 05 16:08:46 RP: 1.4.3 Jul 05 16:10:02 koen: thats just typical of mvista Jul 05 16:10:31 cbrake: So if it is a bitbake bug, its in the current version. not good :-/ Jul 05 16:12:58 RP: i'll roll back to 1.4.2 and see what happens ... Jul 05 16:13:26 has anyone built the davinci-linux kernel using OE? Jul 05 16:13:44 jasuris: does it is in OE? Jul 05 16:13:58 cbrake: I doubt it will be any different as we haven't changed anything to do with that (that I'm aware of ) Jul 05 16:14:16 im not sure.... dont think so Jul 05 16:14:28 RP: ok, then maybe it is just quicker to debug it -- I'll look into it. Jul 05 16:14:52 03koen 07org.oe.dev * r384828c9... 10/packages/angstrom/angstrom-gpe-image.bb: angstrom-gpe-image: set some preferred_providers Jul 05 16:29:11 * hrw|work want BLEBLE_c7x0 += "stuff" be same as BLEBLE_append_c7x0 = "stuff" Jul 05 16:30:28 03cbrake 07org.oe.dev * rb3afec96... 10/packages/meta/task-bootstrap-unionroot.bb: Jul 05 16:30:28 task-bootstrap-uionroot: add RPROVIDES Jul 05 16:30:28 - package PROVIDES task-bootstrap, so it should also RPROVIDE so that Jul 05 16:30:28 ipkg can find it Jul 05 16:32:03 03hrw 07org.oe.oz354x * r26a7cdcb... 10/packages/ipkg/ (ipkg-collateral/c7x0/dest ipkg-collateral.bb): ipkg-collateral: added sd,cf,home dests for c7x0 Jul 05 16:32:07 03hrw 07org.oe.oz354x * re5417dda... 10/conf/distro/openzaurus-3.5.4.1.conf: openzaurus 3.5.4.1 is going into RELEASE Jul 05 16:32:14 03hrw 07org.oe.oz354x * r8bec59f6... 10/conf/distro/openzaurus-3.5.4.1.conf: oz 3.5.4.1: add usbutils in proper way Jul 05 16:37:04 cu Jul 05 16:37:25 http://ewi546.ewi.utwente.nl/tmp/hrw/3.5.4.1-test/final/ - OZ 3.5.4.1 images (if someone want to test) Jul 05 17:01:42 hello brace OE developers! Just to make note, I just did a monotone pull and update and tried a bitbake opie-imave (angstrom)... bitbake crashed: Traceback (most recent call last): Jul 05 17:03:19 OSError: [Errno 2] No such file or directory: '/mnt/ext0/poli/4700/oe/org.openembedded.dev/packages/binutils/binutils_2.15.91.0.2.bb' Jul 05 17:04:06 poli: I suspect rm tmp/cache/* will fix that Jul 05 17:06:03 or touch conf/local.conf Jul 05 17:08:33 koen: touch as in get it to mark the current date? Jul 05 17:08:45 yes Jul 05 17:13:34 * mrz80 is away: Metaphysically I'm still here; physically, I'm elsewhere Jul 05 17:15:54 hey kergoth Jul 05 17:16:36 hey Jul 05 17:19:11 poli, RP, koen, have the same problem as poli, just did an mt update Jul 05 17:19:53 touch conf/local.conf Jul 05 17:21:44 koen, doesn't help, actually I did not even have a local.conf , only a local.conf.sample Jul 05 17:21:52 have I lost local.conf during the upgrade Jul 05 17:22:18 no local.conf? Jul 05 17:22:25 not even in the builddir? Jul 05 17:23:21 nope Jul 05 17:23:32 i assume you mean openslug/conf Jul 05 17:25:07 koen, had a back tree, but that one does not have a local.conf either Jul 05 17:25:48 no local.conf on your system? Jul 05 17:26:01 how do you specify which machine to use for builds? Jul 05 17:27:32 koen, either make openslug-image or cd openslug; . setup-env; bb Jul 05 17:29:02 bb complains about .bb files that are missing, Jul 05 17:29:15 for me initially it complained about /home/frans/OpenSlugHead/openslug/openembedded/packages/binutils/binutils_2.15.91.0.1.bb Jul 05 17:29:34 as a test copied that one from the backup tree, then it compialed about the same file but 0.2.bb Jul 05 17:29:44 s/compialed/complained Jul 05 17:33:05 koen, is there a way to remove bitbake's cache? Jul 05 17:34:41 interesting, tried bb -v -D spca5xx Jul 05 17:34:57 DEBUG: Removing /home/frans/OpenSlugHead/openslug/openembedded/packages/binutils/binutils-cross_2.15.94.0.1.bb from cache Jul 05 17:35:49 oops, that was only the cross one Jul 05 17:40:01 the odd thing is that if I go to packages/binutils I see that some Jul 05 17:41:31 hmm. koen, i think i found it. Jul 05 17:41:44 binutils_2.15.91.0.1.bb apparently was dropped and is removed from the tree after an update Jul 05 17:42:04 I know Jul 05 17:42:07 however binutils-cross-sdk_2.15.91.0.1.bb is still there Jul 05 17:42:10 touching local.conf solves that Jul 05 17:42:29 and that one says: include binutils_${PV}.bb Jul 05 17:43:12 yeah, and if I do not have a local.conf ? apart from ./openembedded/packages/fontconfig/files/local.conf Jul 05 17:43:50 shouldnt the binutils-cross bb files be dropped ? Jul 05 17:44:18 of course only the ones where the sibling package is not there any more Jul 05 17:47:37 I just ran bitbake essential-machine-image and it got most of the way through, and then finished off with: Jul 05 17:47:40 ERROR: Nothing provides dependency ${MACHINE_DEPENDS} Jul 05 17:48:46 I'm not sure what that means, and In not sure if it actually finished off the build Jul 05 17:49:11 if it did, where would my shiny new ext2 file be located? Jul 05 17:50:06 did you initialize teh MACHINE_DEPENDS var? Jul 05 17:50:22 like MACHINE_DEPENDS ?= "" Jul 05 17:50:29 apparently not Jul 05 17:51:03 is that in local.conf or in my distro.conf? Jul 05 17:51:18 ${MACHINE}.conf Jul 05 17:51:55 essential-machine-image relies heavily on the machine setting all the vars (yes, that's a bug) Jul 05 17:52:18 ok, so I just add it to epia.conf Jul 05 17:53:11 where can I find out a list of 'all the vas', maybe there are some other ones missing too Jul 05 17:54:12 probably in task-essential-to-boot.bb Jul 05 17:55:28 has anyone with a gumstix tried plugging a icuiti M920-CF into the CF slot of the gumstix netcf? are the drivers there for that, or do they need to be written? Jul 05 18:01:41 eFfeM: rm tmp/cache/* Jul 05 18:02:09 its a bug in bitbake's cache handling :-/ Jul 05 18:04:31 koen: thanks, seems to have done it... it's continuing with the build now :) Jul 05 18:08:31 hrm, klibc 1.4 doesn't compile Jul 05 18:09:12 RP, i fixed it temporarily by creating the missing bb files Jul 05 18:09:26 will undo that and try the cache solution Jul 05 18:10:08 ~lart PIERRE! Jul 05 18:10:08 * ibot resizes PIERRE!'s terminal to 40x24 Jul 05 18:12:05 RP: DEFAULT_PREFERENCE is only considered when deciding between two versions of the same package -- does not apply to different packages with the same provides. Jul 05 18:14:26 RP: thanks, that did it Jul 05 18:14:32 03eFfeM 07org.oe.dev * r9e783e78... 10/packages/pvrusb2-mci/ (pvrusb2-mci-20060702/Makefile.patch pvrusb2-mci_20060702.bb): pvrusb2-mci: added 20060702 snapshot Jul 05 18:17:47 cbrake: That sounds like something we should consider then... Jul 05 18:18:19 RP: nod -- seems like it would be useful. Jul 05 18:41:00 * mrz80 is back (gone 01:27:27) Jul 05 18:53:00 hi Jul 05 18:53:03 kergoth: ping Jul 05 18:54:21 hi hrw Jul 05 18:54:40 I saw kergoth alive so appeared Jul 05 18:55:32 kergoth: did you got my mail about feeds.openzaurus.org CNAME? Jul 05 18:59:12 hrw: congrats on the final OZ build Jul 05 18:59:25 thx koen Jul 05 18:59:37 koen: just rebuilt it Jul 05 19:03:25 hmm.. altboot faild to build... Jul 05 19:03:40 seems to be looking for altboot*.cfg, which doesn't exist Jul 05 19:05:35 tkp: this is rather zaurus related stuff.. Jul 05 19:05:47 CoreDump|home: add COMPATIBLE_MACHINE into it maybe? Jul 05 19:06:11 hrw: I called bitbake essential-machine-image Jul 05 19:06:29 and it tried to do altboot along the way Jul 05 19:08:29 updated OZ 3.5.4.1 images available - now they should not warn about wrong kernel Jul 05 19:09:15 hello Jul 05 19:09:25 hi sirfred Jul 05 19:09:47 hrw: Hello. The boot image is showing great now on my c760 Jul 05 19:10:06 sirfred: grabbed from my c760 Jul 05 19:10:29 sirfred: have to add same for tosa Jul 05 19:10:32 hrw: opie still doesn't start in mine. Still the sigsegv error. Jul 05 19:10:49 sirfred: fsck.. will flash mine later maybe Jul 05 19:10:59 have to prepare press release Jul 05 19:11:35 And the same strange effect in the setup wizard. Windows only showing in the very left side. Jul 05 19:12:02 http://linuxdevices.com/news/NS4392038828.html will be base of my press release - I like how they changed my previous note Jul 05 19:12:12 hrw: Some days ago, I started opie from a gdb session, and the segfault seemed to be in some clipping related function. Jul 05 19:12:29 bad luck Jul 05 19:12:42 CoreDump told that opie and gpe images works ok on his akita Jul 05 19:13:00 so I'll look on c760 - maybe w100 stuff break something Jul 05 19:13:16 http://www-128.ibm.com/developerworks/linux/library/l-scheduler/?ca=dgr-lnxw09LinuxScheduler Jul 05 19:13:31 hrw: It's disabled by default. Anyway, I tried enabling w100 qte driver with the same result. Jul 05 19:13:40 hrw: Well, perhaps you mean at kernel level. Jul 05 19:14:09 hrw: i'm already reading ir :) Jul 05 19:14:34 hrw: where does the root file system come from when you build an image? Jul 05 19:14:44 is there a package? Jul 05 19:15:38 hrw: how did edb-native kill ewi? Jul 05 19:15:46 jkp_: no Jul 05 19:15:52 jkp_: they're in tmp/deploy/images Jul 05 19:16:03 JustinP: do_configure started lot of gtk-config (gtk 1.2 stuff) Jul 05 19:16:15 JustinP: by lot I mean A LOG Jul 05 19:16:21 JustinP: by lot I mean A LOT Jul 05 19:16:23 i mean, what part of the system actually generates stuff like /dev/* and the directory structure? Jul 05 19:16:24 hrw: :-P We can disable the gtk interface... Jul 05 19:16:34 hrw: FYI , /etc/init.d/checkversion is complaining about a bad kernel version (I'm sure I downloaded initrd.bin and zImage.bin) Jul 05 19:16:42 sirfred: already fixed Jul 05 19:16:53 hrw: Great Jul 05 19:17:30 hrw: I've killed my tmp again, re-copied local.conf, and cleared my ccache and I still have rxvt issues in e..... Jul 05 19:17:56 sirfred: I noticed it when was walking home from work Jul 05 19:18:04 :) Jul 05 19:18:08 sirfred: and then got mail from CoreDump|home with that info too Jul 05 19:18:17 Program received signal SIGSEGV, Segmentation fault. Jul 05 19:18:17 0x4038ea1c in QPainter::setClipping () from /opt/QtPalmtop/lib/libqte.so.2 Jul 05 19:18:22 hrw: That is what gdb says. Jul 05 19:18:51 sirfred: I was debugging qt/e over years ago... in OZ 3.5.0 times Jul 05 19:19:07 '3.5.0' == OZ from OE before 3.5.1 release Jul 05 19:19:26 The old and good times. Jul 05 19:19:31 Stunning observation :p Jul 05 19:19:39 qte debug + qpe debug = 40M.. had to use cf.. Jul 05 19:19:50 I'd be worried if OZ 3.5.0 came out after 3.5.1 :p Jul 05 19:20:00 openzaurus.org Jul 05 19:20:40 CoreDump|home: your mirror has feed synced Jul 05 19:20:51 nice Jul 05 19:20:58 * CoreDump|home will bbl Jul 05 19:21:02 ~hail rsync Jul 05 19:21:03 * ibot bows down to rsync and chants, "I'M NOT WORTHY!!" Jul 05 19:22:34 makedevs...any ideas what provides that? Jul 05 19:23:32 here's a shot in the dark - makedevs? Jul 05 19:23:39 busybox has a makedevs too Jul 05 19:23:43 might be using that Jul 05 19:23:52 CosmicPenguin: heh - its not installed by something that required it and errored thats all Jul 05 19:24:01 probably a broken dependency Jul 05 19:25:28 yeah it is Jul 05 19:27:45 hrw: I would change the second paragraph Jul 05 19:28:05 guys - bit of oddness here Jul 05 19:28:13 hrw: remove the "it is based on OPIE bit" Jul 05 19:28:13 building essential-image Jul 05 19:28:34 which depends task-essential-to-boot Jul 05 19:28:50 but it has no dependencies for other tools needed to create the image Jul 05 19:29:07 makedevs-native genext2fs-native Jul 05 19:29:11 for example Jul 05 19:29:36 we added them manually to a copy of the file, but is there an umbrella .bb file that includes the tools needed to create the image? Jul 05 19:30:22 seems odd that inherit image_ipk doesnt give you that stuff Jul 05 19:30:38 koen: me too Jul 05 19:33:12 cu Jul 05 19:36:18 koen: help! Jul 05 19:36:22 im extremely puzzled Jul 05 19:37:00 what do we need to put in the .bb file to make all the tools needed to build images (prerequisutes) part of the dependencies? Jul 05 19:39:42 heh Jul 05 19:39:45 ok - silly error Jul 05 19:39:54 we had DEPENDENCIES = foo Jul 05 19:40:03 not DEPENDENCIES += foo Jul 05 20:05:41 does anyone have a system I could do some bitbaking on? Jul 05 20:05:50 I have one myself but I can't figure out what this issue is..... Jul 05 20:06:00 or if someone could just run a spitz e-image-core build for me that would be enough Jul 05 20:20:58 fluxbox makes usage of gpe better Jul 05 20:23:21 hrw-tosa: you use it with perlpanel and no desktop icons? Jul 05 20:24:03 marcos: with matchbox panel and matchbox desktop Jul 05 20:24:21 ah Jul 05 20:25:25 i'll scap Jul 05 20:26:59 thejapa: http://handhelds.org/scap/ Jul 05 20:27:28 ah, i was about to ask. :) Jul 05 20:28:19 g Jul 05 20:29:20 re Jul 05 20:29:40 bah, not fair. tosa got keyboard and vga screen and wifi... Jul 05 20:30:06 it even got recent distro builds Jul 05 20:30:28 wifi which break after suspend, which has shitty driver... Jul 05 20:31:42 mickeyl: i need your help Jul 05 20:32:06 mickeyl: 3541 does not work on c7x0/opie Jul 05 20:34:00 guys - managed to build our first ext2 image file but we are having major issues with it Jul 05 20:34:14 a: the kernel wouldnt boot, but we can sort that later Jul 05 20:34:18 details please Jul 05 20:34:25 b: we cant get the filesystem to mount Jul 05 20:34:39 c: things seem to be linked to non-existent libraries Jul 05 20:34:46 we used bitbake essential-image Jul 05 20:34:51 tried to reconfigure kernel? Jul 05 20:35:04 jkp_: i suggest bootstrap-image Jul 05 20:35:10 really??? Jul 05 20:35:15 i was told *not* to use that one Jul 05 20:35:36 * thejapa announces - final cup game - france x italy Jul 05 20:35:45 * jkp_ woots :) Jul 05 20:35:54 we are looking for a base system to work from hrw-tosa Jul 05 20:35:58 heh.. something changed in .dev then... Jul 05 20:36:15 right Jul 05 20:36:18 before bootstrapimage was default tot try Jul 05 20:36:29 hrw-tosa: ok - will do that Jul 05 20:36:37 is that a good one to base our distro on? Jul 05 20:37:23 what was odd, was that busybox was dynamically linked to libgcc_s Jul 05 20:37:28 which didnt get installed on the box Jul 05 20:37:34 o.. Jul 05 20:37:49 try bootstrap-image Jul 05 20:37:56 yeah Jul 05 20:38:01 i will, one thing with that though Jul 05 20:38:26 when you run task-bootstrap...it says its ambigous, and that it doesnt know whether to use task-bootstrap, or task-union-bootstrap Jul 05 20:38:31 and it uses the later Jul 05 20:38:46 then when it comes to make the image it cant find the dependency Jul 05 20:38:58 seems like that ambiguity needs fixing maybe? Jul 05 20:39:16 use BBMASK="(task-bootstrap-unionroot)" in local.conf Jul 05 20:39:25 its workaround Jul 05 20:39:47 ok - so it will use the unionroot version? Jul 05 20:39:57 * jkp_ knows nothing about unionroot Jul 05 20:40:12 it will ignore that recipe Jul 05 20:40:24 ahhhh Jul 05 20:42:16 mickeyl: qpe segfault after scannig docs... Jul 05 20:42:40 : LauncherView::removeAllItems() Jul 05 20:42:40 : WARNING: XXXXXXXXXXX rescan Jul 05 20:42:40 Segmentation fault. Jul 05 20:45:21 mickeyl: after that it does not start at all: W100 driver cannot connect Jul 05 20:45:36 when bootstrap-image runs udev, I get lots of: NOTE: Couldn't find shared library provider for libgcc_s.so.1 Jul 05 20:45:57 tkp: looks like packages are fcsked Jul 05 20:45:57 I'm a little worried as that was the lib busybox tried to link to before Jul 05 20:46:16 tkp: which MACHINE you used? Jul 05 20:46:25 epia Jul 05 20:46:57 im wondering if there isnt a revision of the repository we could revert to that is in a known good state for what we are trying to do? Jul 05 20:47:09 looks like it used host libs.. Jul 05 20:47:19 aggghhh :/ Jul 05 20:47:27 thats not good Jul 05 20:47:29 libgcc1 should be built when gcc-cross was built Jul 05 20:47:36 * hrw-tosa goes sleep Jul 05 20:47:39 cu Jul 05 20:47:46 cya hrw-tosa Jul 05 20:47:51 thanks for all the help Jul 05 20:52:10 JustinP: i can build e-image-core for you, it's going to take some time thp Jul 05 20:52:51 i think this is all linked to an error we both had when building gcc-cross Jul 05 20:53:05 it said FATAL kernel to old Jul 05 20:53:14 but then we both hit rebuild and it all worked Jul 05 20:53:36 but im guessing that it didnt properly build gcc-cross, and things got linked to other libraries on the system Jul 05 20:53:51 would that make any sense? Jul 05 20:54:41 we saw some errors when building something else later (an mplayer dep) errors about multiple definitions in gcc-final stuff Jul 05 21:03:22 CoreDump|home: I'd appreciate it. I'd like to see if rxvt's font is ok on another image Jul 05 21:03:51 CoreDump|home: from oz354x, please, if you can. .dev is ok if you must, but I've been testing oz354x lately Jul 05 21:04:29 * CoreDump|home doesn't have a .dev build env anyway ;) Jul 05 21:04:48 CoreDump|home: :-) Jul 05 21:05:39 it's going to take a few hours, I'll have it ready by tomorrow Jul 05 21:05:48 CoreDump|home: could you please make a note of the rev of the workspace you're building from? If yours works I'll want to build the same rev on my machine and look for differences. Jul 05 21:06:15 CoreDump|home: thanks much...hrw keeps having problems with e on ewi.... Jul 05 21:06:24 sure, but "most-recent" pretty much matches my rev Jul 05 21:06:30 ok Jul 05 21:06:40 I just want to make sure I'm building the same rev for comparison purposes Jul 05 21:06:55 I'll basically do a diff -r of the image :-] Jul 05 21:07:09 this problem has been plauging me.....forever Jul 05 21:07:34 JustinP: interesting problem, do you have a screenshot? Jul 05 21:08:30 CoreDump|home: ummm...I'd do an scap now...hold on, let me see if I can get it up Jul 05 21:09:13 anyone know how to enter a WEP "password"? Jul 05 21:09:17 not a key, a password.... Jul 05 21:09:43 s:ASCII_TXT Jul 05 21:10:31 * JustinP reboots Jul 05 21:15:39 bitbake does x-compiling? Jul 05 21:16:22 JustinP: not sure which distro... but i remember wep wasn't working with someone here... Jul 05 21:16:28 duh Jul 05 21:17:53 CoreDump|home: thanks :-) http://handhelds.org/scap/port.9533.png Jul 05 21:18:13 REdOG: no, it's magic.....of course it cross compiles.... Jul 05 21:18:29 JustinP: e17 running where? Jul 05 21:18:30 thejapa: WEP works fine for me. I just didn't know how to do a password instead of a key Jul 05 21:18:30 JustinP: LOL, that's one bad-ass cursor Jul 05 21:18:40 thejapa: my spitz Jul 05 21:18:44 CoreDump|home: yes...yes, it is Jul 05 21:18:46 cool Jul 05 21:18:59 it's just a font size issue I assume. no font with that size so it displays nothing Jul 05 21:19:09 typing things works I typed scap into it) Jul 05 21:19:19 heh Jul 05 21:19:42 * thejapa tests ibot Jul 05 21:19:45 ~scap Jul 05 21:19:46 rumour has it, scap is a CGI application to easily make screenshots from ibot's iPaq and have them posted on the Web. Refer to http://handhelds.org/scap/ Jul 05 21:19:47 JustinP: does it also build the x-compiler? Jul 05 21:19:50 aha Jul 05 21:19:53 REdOG: yes Jul 05 21:19:53 nice Jul 05 21:20:23 JustinP: is it able to use uclibc? Jul 05 21:21:33 * tkp leave his laptop hoping to wake to a shiny new bootable fs Jul 05 21:24:39 REdOG: it's supposed to be able to. I've never used uclibc Jul 05 21:26:29 CoreDump|home: http://handhelds.org/scap/port.10610.png Jul 05 21:26:38 CoreDump|home: rxvt -fn mono-4 Jul 05 21:26:41 :-P Jul 05 21:27:10 well, that looks usable Jul 05 21:27:42 except no chars still... Jul 05 21:28:19 maybe fonts aren't being installed....? Jul 05 21:28:22 silly question I know but err did you try that with a newly created $USER? Jul 05 21:30:16 ? Jul 05 21:30:21 ummm...no.... Jul 05 21:30:28 I always just login as root Jul 05 21:30:44 e-image doesn't have the user creation widget ATM Jul 05 21:31:01 and this used to work fine....and the last few e-images that hrw made did work... Jul 05 21:31:03 IIRC Jul 05 21:32:00 worth a try then ;) Those nasty dot files can be a real PITA at times Jul 05 21:32:21 ie: ~/.Xdefaults Jul 05 21:32:51 fwiw, e-image-core for akita just finished Jul 05 21:32:57 * CoreDump|home boots Jul 05 21:34:25 nope, newly created user has same issue Jul 05 21:34:53 no ~/.Xdefaults, though Jul 05 21:34:58 for root or new user Jul 05 21:35:13 .etc/X11/Xdefaults of course Jul 05 21:35:19 / Jul 05 21:40:33 "I've noticed that in several places (most prominently, Help-About), there is the product version, build number, etc. Please have your team remove this information in our next release." Jul 05 21:40:42 * koen loves thedailywtf.com Jul 05 21:41:05 hahah Jul 05 21:41:41 "The reason is simple: when they see "Version 2.8 (build 448)," they will think that it took us 28 releases and over *four hundred* builds to get right. " Jul 05 21:41:59 I suspect I did over 400 rebuilds with OE by now Jul 05 21:42:48 oh noes, we had to build it lots! Jul 05 21:42:52 idiots Jul 05 21:43:51 haha Jul 05 21:44:18 kergoth: Installed openwrt over the weekend after my magnia died. Jul 05 21:44:23 kergoth: It's pretty slick Jul 05 21:45:54 nice. i havent played with it much, didnt have the nerve to flash it when i was in MN with only one wrt Jul 05 21:46:10 It pretty much worked out of the box for me Jul 05 21:46:34 a bit sluggish when doing a file transfer over openvpn, but it works Jul 05 21:47:41 didn't someone fork openwrt (again) last month? Jul 05 21:48:11 ah yes, freewrt Jul 05 21:48:40 * koen wonders when someone will start 'wrtux' Jul 05 21:49:17 koen: no, bsdwrt is the next one ;-) Jul 05 21:49:42 there is much bickering in the wrt camp :-( Jul 05 21:50:14 freewrt.org runs on cherokee and jaws Jul 05 21:50:18 that's pretty cute Jul 05 21:50:50 i wanted an oe based distro on the wrt, but adding the kenrel proved to be a real pain in the ass, so i gave up. has anyone else messed with it? Jul 05 21:51:39 kergoth: I tried to convince the openwrt people just last week. No good response. But maybe I got them to reconsider at least a little bit. Jul 05 21:51:53 yay, black screen of death Jul 05 21:52:40 kergoth: hrw did some work and br1 Jul 05 21:52:44 Laibsch: i spoke with one of the maintainers at one point, and he was open to it. i was going to get the kernel in to get the hard parts done, then hand them a local.conf as a starting point :) Jul 05 21:52:48 thats cool Jul 05 21:52:50 kergoth: Let me know if you get more things working on the wrt. I also own such a device and would rather have OE based stuff on it. Jul 05 21:52:58 but br1 couldn't continue it since he gets paid by 4g-systems Jul 05 21:53:04 ah Jul 05 21:53:23 kergoth: Did you solder your device? Jul 05 21:53:25 iirc the images work but need a serialport to be usefull Jul 05 21:53:50 not yet, no. my wrt is in active use back in MN. need to pick up a new one for my use Jul 05 21:53:57 but you can build packages that run on openwrt afaik Jul 05 21:54:27 (build them with OE, that is) Jul 05 21:56:57 kergoth: the user "webmind" hangs around here and seems to be active in #openwrt as well. Maybe you guys can join forces? Jul 05 21:57:27 itll be a bit before i can work on it, have other priorities. in a couple months i'll pick up a wrt Jul 05 21:57:35 JustinP: http://handhelds.org/scap/port.14123.png <- e-image on akita Jul 05 21:57:37 koen: That is also the rumours that I heard. I'll certainly try that out once I flashed the linksys. Jul 05 22:00:17 Why is OE building kaffeh-native-1.1.5 instead of the CVS version? I can't seem to be able to override it with PREFERRED_VERSION_kaffeh-native = "cvs" either. Still trying to fix bug 410. Jul 05 22:01:04 Laibsch: "cvs" isn't a version Jul 05 22:01:14 kergoth: maybe you can get a wrt from fon.com for free ;-) Jul 05 22:02:01 Laibsch: see kaffeh-native_cvs.bb: PV = "1.1.5+cvs${SRCDATE}" Jul 05 22:02:22 koen: I see. Jul 05 22:04:29 ~lart http://www.openembedded.biz/ Jul 05 22:04:29 * ibot DoSes http://www.openembedded.biz/ Jul 05 22:04:41 ~lart http://www.openembedded.biz/ again Jul 05 22:04:41 * ibot whips out a sword and chops http://www.openembedded.biz/ again in half Jul 05 22:05:47 heh - funny Jul 05 22:06:15 I heard mvista has a lot of job openings nowadays Jul 05 22:18:08 JustinP: I'm now trying hentges-e-image which I created ages ago but still builds Jul 05 22:21:28 CoreDump|home: ah....so it's not just me....whew Jul 05 22:21:37 CoreDump|home: still, though, this is a real PITA.... Jul 05 22:21:51 CoreDump|home: what kind of box/distro are you building on? Jul 05 22:22:06 Debian SID on a P4 2.8G Jul 05 22:24:21 hentges-e-image is b0rked, too Jul 05 22:24:22 ok, so it's not a Gentoo thing either Jul 05 22:24:38 I wonder what I'd need to bitbake in a chroot.... Jul 05 22:25:00 I do that all the time ;) Jul 05 22:26:27 time for bed, n8 all Jul 05 22:28:22 hrw|gone: what distro do you bitbake on? Jul 05 22:42:42 v8jlene: I wouldn't be surprised is ssmtp has my name in the maintainer field, no. Anything in particular? Jul 06 00:18:09 NAiL: Yeah, it has you as maintainer ;) Was just going to make sure your ok with me checking in a change to stop it passing -s to install (which fails when it tries to call the host strip).. Jul 06 00:19:45 hmm Jul 06 00:20:11 strange ... i was under the impression that UTF8 encompasses all languages in it's char set Jul 06 00:23:58 unicode seems to claim that... Jul 06 00:31:44 ah yeah that would have been what i was thinking Jul 06 00:34:46 UTF8 is just an encoding of unicode and so should encompass everthing as well... Jul 06 00:35:55 there is a utf8 unicode Jul 06 00:39:02 If your not already looking at http://www.unicode.org/ then it's a good place to look around. Lots of useful stuff. Jul 06 01:35:49 oh... just installed eclipse 3.2... never tried eclipse before... cool :) **** ENDING LOGGING AT Thu Jul 06 02:59:57 2006