**** BEGIN LOGGING AT Tue Jan 10 02:59:56 2006 Jan 10 03:04:26 ugh this can't be good Jan 10 03:05:10 just flashed an "-rjan2006prebreakage" image to my Akita and got _tons_ of "Empty flash at blahh ends at blahh" messages Jan 10 03:07:59 CoreDump|home: Only time I've had that happen is where I've dumped the .img to the CF card instead of the .tgz Jan 10 03:08:34 CosmicPenguin: Akita is known to print a lot of them. They should be harmless unless you're seeing lots and lots of them Jan 10 03:08:45 er, CoreDump|home even... Jan 10 03:08:59 i saw indeed lots and lots and LOTS =) Jan 10 03:09:05 ok harmless,thx Jan 10 03:09:28 might want to hide that for a release though Jan 10 03:10:33 NOTE: build 200601101010: completed Jan 10 03:10:37 glop glop Jan 10 03:10:44 but hey, "Attempted builds: 20 Jan 10 03:10:45 " Jan 10 03:10:57 =) Jan 10 03:10:59 stupid router/internet conection... Jan 10 03:15:26 meh opies konsole is badly messed up, leaving atrifacts all over the window Jan 10 03:15:56 hi Jan 10 03:16:01 hi hrw Jan 10 03:16:05 hi hrw Jan 10 03:16:16 CoreDump|home: you are alive ;) Jan 10 03:16:25 indeed I am=) Jan 10 03:16:39 CoreDump|home: altboot problem I have in 3.5.4-rc Jan 10 03:16:51 ah what is it? Jan 10 03:17:11 CoreDump|home: when I boot altboot powered collie then option 'copy rootfs to sd/cf' is fscked - no sd/cf support started Jan 10 03:17:33 yeah that Jan 10 03:17:41 shouldn't have added it =( Jan 10 03:17:51 it wasn't finished I'm afraid Jan 10 03:17:59 CoreDump|home: I tried to look at last version of your flavour of OZ and you have lots of extras which do it later iirc Jan 10 03:18:43 CoreDump|home: I would like to get altboot officially nominated as rboot hacks replacement Jan 10 03:18:54 heh =) Jan 10 03:19:09 hrw: altboot? Jan 10 03:19:32 NAbyss: oe/packages/altboot/ Jan 10 03:19:46 my problem right now is whether I should build the 3.5.4 branch or not. It's quite old and I don't know if it wasn't better to take a fresh snapshot for new images Jan 10 03:19:50 ah Jan 10 03:20:30 but getting kernel 2.6 to work smoothly _is_ going to need some work heh Jan 10 03:21:17 CoreDump|home: with 2.6 kernel you can prepare kernel which will boot right from sd Jan 10 03:21:28 (Oo) ??? i did a bitbake gpe-image and now I have a zImage but no initrd in my deploy/image !!??!! Jan 10 03:21:32 hrw: what's your point on the 3.5.4 branch? fix it or trash it? Jan 10 03:21:46 CoreDump|home: The release branch 2.6 support is marginal atm Jan 10 03:21:48 hrw: really? sounds neat Jan 10 03:21:48 CoreDump|home: .oz354fam083 branch? Jan 10 03:22:02 jep that one Jan 10 03:22:35 CoreDump|home: this is branch from which users will get stuff Jan 10 03:22:50 CoreDump|home: if users have to get something then it HAS to be in .oz354fam083 branch Jan 10 03:22:55 so it's still beeing worked on? Jan 10 03:23:02 CoreDump|home: yep Jan 10 03:23:07 ahh ok Jan 10 03:23:27 morning all Jan 10 03:23:33 morning Jan 10 03:23:51 hi lrg|home Jan 10 03:24:35 hey RP, I'm almost ready to send you an update :) Jan 10 03:25:18 lrg|home: Sounds good :). What's included in this one? Jan 10 03:25:19 what is theturtle IMAGE_FSTYPES for openzaurus ? Jan 10 03:25:24 morning all Jan 10 03:25:32 morning Dirk Jan 10 03:25:34 hi Dirk Jan 10 03:25:46 (why th hell did i write turtle ????) Jan 10 03:25:51 hi Richard, Liam Jan 10 03:26:24 RP: some new codecs, including some refactoring of the codec descriptors to support the new codecs. also intial tdm support Jan 10 03:26:39 lrg|home: changing rpu has no effect Jan 10 03:26:48 do13: :( Jan 10 03:27:16 lrg|home: tdm? Jan 10 03:27:35 do13: I should get my tosa at the weekend. I'll be able to fix this bug then. I've never seen it do this on Mainstone or NBP Jan 10 03:27:37 lrg|home: touch alone and battery alone works. but if I do both the touch isn't usable Jan 10 03:28:20 lrg|home: it seems the locking in the driver doesn't works correctly Jan 10 03:28:21 http://www.hrw.one.pl/2006/01/02/2005-timeline/ if someone interested Jan 10 03:28:25 do13: ah, I think they are clobbering each other on the wm9712 as it's fine on the 13. Jan 10 03:29:22 do13: ok, I had made a lot of changes to make things smaller, I've obviously missed a few things. :( Jan 10 03:30:39 lrg|home: What's the difference between 12 and 13? xx_poll_sample looks very similar. Jan 10 03:30:42 do13: I have a few audio things to do first, so my priority for next week is to get the tough and audio working well on tosa Jan 10 03:31:05 do13: the touch register maps are different Jan 10 03:31:27 hi, has anyone been successfull building perl for an arm920t with glibc 2.2.5 so far? nothing seem to work right :( Jan 10 03:31:56 do13: did you try accelerated touch ? Jan 10 03:32:07 lrg|home: Not yet Jan 10 03:32:20 RP: to get all deps built I have to add some magic to meta-opie*? Jan 10 03:32:33 lrg|home: Do you exept this works better? Jan 10 03:32:58 do13: It uses an irq for pen down instead of polling Jan 10 03:33:16 do13: although I don't know which irq on tosa Jan 10 03:33:31 do13: so it's set up for mainstone atm Jan 10 03:33:46 lrg|home: Ahh. Ok. Jan 10 03:34:10 do13: I'm now wondering if LCD noise is contributing to fake pen down signals..... Jan 10 03:34:20 so guess that means either no or fuck off, anyway thanks Jan 10 03:34:46 lrg|home: No if I disable battery reading, the touch works Jan 10 03:34:59 do13: including the callibration ? Jan 10 03:35:06 lrg|home: Yep. Jan 10 03:35:12 do13: ah. Jan 10 03:35:55 do13: ok, then I can fix this on mainstone today. I'll send a patch with the hooks for getting a structy wm97xx as well Jan 10 03:36:16 lrg|home: ok Jan 10 03:36:29 lrg|home: The values are a bit noisy, because the touch needs to be in snyc with the lcd. this can be done later Jan 10 03:37:17 do13: it's also on my long todo list :) Jan 10 03:37:28 lrg|home: I'll clean up my tree and send you the patches later this week. Jan 10 03:37:37 do13: thanks :) Jan 10 03:37:58 * lrg|home reckons his todo list is almost as long as RP's todo list Jan 10 03:38:25 lrg|home: I changed soc/pxa/tosa.c to detect the earphone. Jan 10 03:39:14 do13: excellent. I had basically left this file until I had a tosa. I also need to add dpm to tosa. Jan 10 03:39:35 lrg|home: yep. this is missing. Jan 10 03:40:57 lrg|home: I'll also send you info about how the WM9712 is connected Jan 10 03:41:29 do13: thanks. this really helps the dpm. :) Jan 10 03:42:57 hrw: The patch you have for bitbake needs tweaking Jan 10 03:43:18 hrw: I can't provide one at the moment as my machine with the known working patch is unavailable :-( Jan 10 03:43:47 ok Jan 10 03:43:58 lrg|home: We need to disucss the open points until sunday, because next week I'am on vacation and the week after I'am on a business in spain. Jan 10 03:44:25 hrw: Have you a copy of that patch I sent you last night lying around I can look at? Jan 10 03:44:34 RP: moment Jan 10 03:44:45 hrw: There are some lines to comment out that will make it work better Jan 10 03:44:47 do13: np. I may be going to ShangHai later this month on business and will be behind the great firewall :( Jan 10 03:45:05 RP: http://ewi546.ewi.utwente.nl/tmp/hrw/bitbake-rproviders-r0.patch Jan 10 03:45:52 + if virtual != package: Jan 10 03:45:52 + bb.note("%s: %s not equal to %s, skipping" % (item, virtual, package)) Jan 10 03:45:52 + continue Jan 10 03:46:02 hrw: comment out those lines Jan 10 03:46:38 hrw: Also, keep in mind the circular dependency warnings are a bug in bitbake, not a bug in the metadata Jan 10 03:46:50 ok Jan 10 03:47:02 RP: any way I can resolve most of them Jan 10 03:47:14 hrw: I had most of opie compiling last night - the only other circular dependency I found was opie-mediaplayer1 Jan 10 03:48:04 At the moment, bitbake it being a bit too zealous about building all RDEPENDS when some aren't needed but this is probably better than the other way around... Jan 10 03:48:08 RP: ok - will look into this too Jan 10 03:48:34 hrw: It actually looked correct to me so I just need to fix bitbake :) Jan 10 03:49:42 k Jan 10 03:50:12 RP: what I have to add into meta-opie* to finally get it build? BUILD_ALL_DEPS=1 into recipe? Jan 10 03:54:39 hrw: meta-opie shouldn't build all - it should just produce a set of package s with the appropriate RDEPENDS Jan 10 03:55:09 hrw: but for testing run "BUILD_ALL_DEPS=1 bitbake meta-opie" Jan 10 03:56:33 RP: so in future I would have to remember of that var?? for me 'meta-opie' HAS to build all opie stuff Jan 10 03:58:21 hrw: That is a misconception - it doesn't have to build it, that's just the behaviour you expect. Jan 10 03:58:54 RP: agreed about expecting. Jan 10 03:59:34 RP: but that also mean that to build all opie stuff I would have to run 'bitbake opie* meta-opie' or something like that? Jan 10 04:00:07 hrw: no, you'd need to trigger the BUILD_ALL_DEPS functionality Jan 10 04:00:13 hrw: You could create a .bb file that did 'include meta-opie.bb BUILD_ALL_DEPS="1"' Jan 10 04:00:19 ok Jan 10 04:00:36 hrw: not include, DEPENDS and RDEPENDS, sorry Jan 10 04:01:00 ok. I added b_a_d into meta-opie.bb and started build Jan 10 04:01:42 and now I would see... Jan 10 04:05:04 RP: http://ewi546.ewi.utwente.nl/tmp/hrw/bitbake.log-1.txt - why it selects each kernel when machine=c7x0? Jan 10 04:06:14 hrw: monotone pull Jan 10 04:06:33 ERROR: Nothing provides kernel-image-2.6.15-rc7.. Jan 10 04:06:34 o Jan 10 04:06:35 k Jan 10 04:06:38 The key line is "ERROR: Nothing provides kernel-image-" Jan 10 04:06:50 I added a line to kernel.bbclass which sorts that out Jan 10 04:08:24 ok. now opie-advancedfm Jan 10 04:09:08 same file Jan 10 04:09:51 ERROR: /home/hrw/devel/oe/org.openembedded/packages/opie-advancedfm/opie-advancedfm_cvs.bb depends on itself (eventually) Jan 10 04:09:59 its bitbake fault now Jan 10 04:15:36 hi Jan 10 04:15:39 hi zecke Jan 10 04:16:01 hey zecke Jan 10 04:16:17 hey zecke Jan 10 04:16:36 lrg|home: the package has been brought to the post office... well I hope my mother brought it there Jan 10 04:17:12 zecke: :) Jan 10 04:18:25 lrg|home: I didn't find the USB mini cable for the dockingstation, but I think you have one already... Jan 10 04:18:58 zecke: yes thanks, i've got one with borzoi Jan 10 04:22:28 how to mark recipe as few machines/kernels only? opie-camera need sharp-camera-support which mean 2.4.18/2.4.20 for zaurus machines only Jan 10 04:24:23 hrw: I think opie needs to add a variable, something like BOOTSTRAP_EXTRA_RDEPENDS, which would get set by the machine files to include opie-camera Jan 10 04:24:45 OPIE_EXTRA_RDEPENDS Jan 10 04:25:05 ok Jan 10 04:25:17 I just removed opie-camera from meta-opie Jan 10 04:25:26 ERROR: /home/hrw/devel/oe/org.openembedded/packages/opie-button-settings/opie-button-settings_cvs.bb depends on itself (eventually) Jan 10 04:25:32 another self RDEPENDS Jan 10 04:25:56 I didn't see that one for some reason... Jan 10 04:26:29 RP: add b_a_d into meta-opie.bb and try to '-nv' build it Jan 10 04:26:44 lrg|home: http://oe.handhelds.org/cgi-bin/moin.cgi/ProjectDevices?action=show Jan 10 04:26:53 lrg|home: please update the page once it arrived ;) Jan 10 04:26:55 hrw: I've lost access to my build box for now :-( Jan 10 04:27:32 ok Jan 10 04:27:35 I did add something to my patched bitbake to cope with RDEPENDS of the form "dep (version)" so just ignore errors caused by them (sharp-camera-support uses them) Jan 10 04:28:06 zecke: np Jan 10 04:29:40 RP: that kind of depends 'problem' should be resolved by bitbake Jan 10 04:33:10 hrw: It is, in the version I don't have access to... Jan 10 04:33:59 ok Jan 10 04:37:02 RP: I have to build sqlite to let bitbake know that libsqlite-bin is part of it? Jan 10 04:37:36 ignore me Jan 10 04:42:19 * CoreDump|home heads to work Jan 10 04:42:22 later guys Jan 10 04:42:26 cu CoreDump|home Jan 10 04:55:51 re Jan 10 05:27:36 morning Jan 10 05:36:05 hrw|afk: Would it be ok to do something like this with opie cvs versioing?: http://www.rpsys.net/openzaurus/temp/opie-pv.patch Jan 10 05:37:42 hi Jan 10 05:37:46 hi Mardy Jan 10 05:40:13 s/OPIECVSPV/OPIE_VERSION_CVS maybe? Jan 10 05:40:17 brb Jan 10 05:40:38 RP: did you have a look at tskeys? Jan 10 05:40:49 hrw|afk: It is a PV string, not a VERSION though Jan 10 05:41:22 Mardy: No, and I really want to get that done... Jan 10 05:41:59 Mardy: The time I set aside for that got taken over by other things but I will get it looked at Jan 10 05:42:28 :-) Jan 10 05:43:21 * france is away: Away Jan 10 05:43:28 Mardy: Its not just tskeys that's suffering - there's something called chkhinge26 in ohand svn which I also want to fix up and get into OE... Jan 10 05:43:56 Both are very related and should probably become one program Jan 10 05:44:36 I have visions of this device-manager deamon doing nice things with the hardware... Jan 10 05:52:07 RP: but it contain current opie version + cvs date Jan 10 05:54:29 hrw: in the form of a PV string :) Jan 10 05:54:51 hrw: Would OPIE_CVS_PV look better? Jan 10 05:54:59 OPIE_VERSION_CVS_PV? Jan 10 05:55:33 OPIE_CVS_PV looks better anyway Jan 10 05:56:05 hmm. oz 3.5.4 todolist need to be created... Jan 10 05:56:20 We probably want something with just one VERSION in it. OPIE_CVS_PV does look better, I must admit. Jan 10 06:00:38 ok Jan 10 06:07:21 fortunes-mod need recode-native.. Jan 10 06:09:02 is it normal that "bitbake nano" doesn't build packages (.ipk) ? Jan 10 06:10:05 theturtle: you use .dev and have wrong DISTRO Jan 10 06:12:33 i use .dev and distro is set to DISTRO = "familiar-0.8.3" Jan 10 06:12:58 can i force it to build packages in the local.conf or somewhere ? Jan 10 06:13:00 there is no such distro in .dev Jan 10 06:13:33 ho it has changed... Jan 10 06:14:11 which correspond to the old -0.8.3 ? familiar or familiar-unstable ? Jan 10 06:15:26 unstable Jan 10 06:15:37 ok thanks a lot hrw Jan 10 06:15:41 I propose suscription to oe ML Jan 10 06:25:52 fortune-mod fixed Jan 10 06:26:14 time to build meta-opie again Jan 10 06:27:04 Hm.. does anyone else have issues using -rjan2006prebreakage? Jan 10 06:29:37 NAbyss: notice breakage in it Jan 10 06:29:48 Hi all! anyone can help me? I'm trying bitbake for VIA Epia but it complains about glibc-intermediate, it seems to be related to gcc and i586-oe-linux-gcc.... I'm usinc fedora core 4 Jan 10 06:31:14 hrw: $ monotone update -rjan2006prebreakage Jan 10 06:31:15 monotone: expanding selection 'jan2006prebreakage' Jan 10 06:31:15 monotone: misuse: no match for selection 'jan2006prebreakage' Jan 10 06:31:31 NAbyss: when last time you pulled? Jan 10 06:32:01 NAbyss : a pull solved that problem for me Jan 10 06:32:11 hrw: Just pulled 5 min ago.. 0 updates from everything Jan 10 06:32:23 Can't connect to the vanille.de monotone server though.. the mirrors have no updates Jan 10 06:32:29 NAbyss: monotone list tags Jan 10 06:32:52 jan2006prebreakage 0b9a2dfead1136f996280e9827949f83cb032537 rpurdie@openembedded.org Jan 10 06:33:11 http://pastebin.com/499174 Jan 10 06:33:52 NAbyss: monotone head Jan 10 06:33:57 monotone: branch 'org.openembedded.dev' is currently merged: Jan 10 06:33:57 459f44cba1ccadda754cda0955f1ac1e4877443c hrw@openembedded.org 2006-01-10T13:28:32 Jan 10 06:34:20 $ monotone head Jan 10 06:34:20 monotone: branch 'org.openembedded.dev' is currently merged: Jan 10 06:34:20 a3a85ad4928b7808b987844dcb1af5f05f81dfba koen@openembedded.org 2006-01-04T11:58:20 Jan 10 06:34:22 http://ewi546.ewi.utwente.nl/tmp/viewmtn/branch.psp?branch=org.openembedded.dev Jan 10 06:34:40 NAbyss: 6 days... I would not call it 'fresh'... Jan 10 06:34:50 hrw: Nor I.. but a pull's not grabbing a single thing Jan 10 06:35:10 alias mtpulle='mt pull ewi546.ewi.utwente.nl;mt merge;mt update' Jan 10 06:35:17 run that Jan 10 06:36:15 Crap, that's going to pull a lot.. I've only got .dev on this box.. should I run it like that, or add the branch? Jan 10 06:36:52 depends.. if you use branch then only pull branch Jan 10 06:36:58 if dev then dev Jan 10 06:37:04 same if both (like I do) Jan 10 06:37:29 .oz354fam083 head: ab345d02993e406515d397441e80a52e5be89074 hrw@openembedded.org 2006-01-10T13:16:10 Jan 10 06:38:12 Weird.. it's pulling now Jan 10 06:46:25 hrw, if you want ro work with oz or familiar now dont you need both? Jan 10 06:46:31 to* Jan 10 06:47:07 any failure reports with glibc building ? can't get it compile : http://pastebin.com/499191 Jan 10 06:49:10 theturtle, what gcc are you using? Jan 10 06:49:31 emte: depends - if I want oz 3.5.4-rc then I use branch only Jan 10 06:50:03 hrw, so the distro meta were added back into dev then? Jan 10 06:50:29 distro meta? Jan 10 06:51:28 conf/distro Jan 10 06:52:01 emte: no - conf/distro/ are renamed.. fam-0.8.3 and oz-3.5.4 are in branch only. Jan 10 06:52:04 when they first split the branches it was missing for familiar in .dev Jan 10 06:52:15 .dev has oz-unstable/fam-unstable Jan 10 06:52:23 ah Jan 10 06:53:17 good good Jan 10 06:54:47 <[lala]> hi guys Jan 10 06:55:27 hi lala Jan 10 06:57:48 <[lala]> does anyone know which pkg is responsible for setting the hotplug binary in /proc/sys/kernel/hotplug? Jan 10 06:58:04 <[lala]> because it's empty in my image Jan 10 07:01:28 emte : arm-linux-gcc-3.4.4 Jan 10 07:02:15 i'm trying with gcc-4.0.2-r1 Jan 10 07:03:05 [lala]: udev disables that Jan 10 07:04:36 <[lala]> RP: whats the point of that? afaik /sbin/hotplug is still essential for udev Jan 10 07:07:08 RP: it would be possible to send the c3100 from lrg to you. for seeing the white screen problem Jan 10 07:07:52 <[lala]> hmm ... in my ubuntu /sbin/udevsend is configured as hotplug binary Jan 10 07:09:20 [lala]: I don't think udev needs hotplug now (in the latest versions) Jan 10 07:09:54 greentux: If lrg can confirm he sees those problems on the device, perhaps yes Jan 10 07:10:10 lrg|home: ? Jan 10 07:10:48 greentux: how do I reprodce the problem ? Jan 10 07:10:57 greentux: I might have a look through the code in pxafb and see if I can find the timing issue that causes the problem before we do that Jan 10 07:11:12 <[lala]> RP: i thought that at least a hotplug agent has to configured in /proc/sys/kernel/hotplug Jan 10 07:11:17 lrg|home: nearly after every flash we had a whit screen afte rreboot Jan 10 07:11:31 [lala]: not anymore as far as I understand it Jan 10 07:11:39 lrg|home: but also while switching to console anf back to gpe Jan 10 07:12:35 greentux: I've reflashed acouple of times and can't remeber the screnn color. I'll try the console change Jan 10 07:13:04 lrg|home: white screen means you have to reboot. nothing to see there Jan 10 07:13:34 greentux: so it hangs :( Jan 10 07:13:47 you sure its not just generationg it's keys? Jan 10 07:14:34 * emte doesnt know if oz generates keys at the start like familiar does Jan 10 07:14:38 <[lala]> RP: hmm. do u have any docu handy for the latest udev? Jan 10 07:15:48 greentux: Does a suspend/resume cycle fix the screen? Jan 10 07:16:02 RP: no suspend doesnt worked Jan 10 07:16:57 greentux: Can you still see the device over the network or anything? I'm trying to work out if its a hardware lockup or if the screen just stops working Jan 10 07:17:15 [lala]: Just what's available from google and kernel.org I'm afraid - there isn't a good udev document out there yet Jan 10 07:17:24 <[lala]> RP: yes, the device is still accessible while in white screen mode Jan 10 07:17:52 [lala]: What does dmesg say about the suspend/resume request? Jan 10 07:18:27 <[lala]> haven't looked yet - sorry. Jan 10 07:22:15 [lala]: It might reveal the problem a bit more Jan 10 07:23:31 <[lala]> yep, i guess so :-) problem is that i hardly had network configured when it happened - only recently it happend while i had a network connection :/ Jan 10 07:24:09 [lala]: ok. If it happens again, you know what to do! Jan 10 07:24:19 <[lala]> hehe Jan 10 07:32:56 lol git conversion is done Jan 10 07:37:31 zecke: excellent :) Jan 10 07:37:56 RP: will convert the new 74 revs since the 6th of January now Jan 10 08:02:45 hi i hope today somebody can help me with my error message for compiling php5 : http://pastebin.ca/36293 Jan 10 08:03:21 kami22: same answer as yesterday from my site ;) Jan 10 08:03:46 ah new error ;) Jan 10 08:04:31 yes new error i finished compiling Jan 10 08:04:37 with now problems :) Jan 10 08:04:47 kami22: a) -L/usr/lib is wrong ;) Jan 10 08:05:05 kami22: (if you find real bugs file bug reports with patches attached!) Jan 10 08:05:39 kami22: and you have a libcrypt.so in staging/armeb-linux/lib? Jan 10 08:07:19 zecke: everything is standard and i have got 6 libcrypt files in ../openslug/tmp/staging/armeb-linux/lib Jan 10 08:08:36 libcrypt-2.3.90.so Jan 10 08:08:36 libcrypt.a Jan 10 08:08:36 libcrypto.a Jan 10 08:08:37 libcrypto.so Jan 10 08:08:37 libcrypto.so.0.9.7 Jan 10 08:08:38 libcrypt.so Jan 10 08:08:39 libcrypt.so.1 Jan 10 08:09:06 morning Jan 10 08:09:14 Did somebody ping me over the night? Jan 10 08:09:15 morning Jan 10 08:09:21 kami22: then find out why you have -L/usr/lib in there Jan 10 08:10:00 sorry i don't know what -L/usr/lib means? Jan 10 08:10:09 kami22: man ld Jan 10 08:10:11 kami22: mand gcc Jan 10 08:10:27 Dear linker please get the libraries from /usr/lib Jan 10 08:10:51 but as /usr/lib has libraries for your system and it is not a armeb-linux Jan 10 08:10:56 you will see errors Jan 10 08:11:12 so first step. Make sure the -L/usr/lib is not in the link line Jan 10 08:11:18 2. step compile again Jan 10 08:11:24 3. look for the next error Jan 10 08:13:01 re Jan 10 08:13:10 okay i read the man for ld and the meaning of library-path=searchdir but i dont understand what you mean with this : so first step. Make sure the -L/usr/lib is not in the link line Jan 10 08:13:37 kami22: look at line two of your error Jan 10 08:13:48 kami22: you see a -L/usr/lib there. this is an error Jan 10 08:14:02 CosmicPenguin: I think I managed to type your name instead of CoreDump's with some unintended tabcompletion, sorry Jan 10 08:14:02 kami22: teach php to not add /usr/lib per default Jan 10 08:14:22 Ahhh.... Jan 10 08:14:23 no worries Jan 10 08:14:28 not that I have anything to teach anybody anyway Jan 10 08:17:34 okay i think i understand what you mean. php is looking in /usr/lib and this is not okay but how can i configure/teach php to stop doing this? Jan 10 08:17:49 kami22: find the place where it add /usr/lib Jan 10 08:17:55 RP: so far meta-opie* works nice - only some self-depends exist Jan 10 08:18:23 you mean find it in the config.log? Jan 10 08:18:48 kami22: this could be either due a autoconf test (yes you would find it on in config.log) or some Makefile.am is add /usr/lib Jan 10 08:24:12 zecke: i see in the config.log that the ./configure command got this option : --libdir=/usr/lib is this the problem? Jan 10 08:29:23 hrw|afk: That's good the hear. I'll sort out self depends handling in the next revision of the bitbake patch. Jan 10 08:29:49 My build machine is back online now - the ADSL router needed resetting :-/ Jan 10 08:36:10 kami22: no, I don't know, it shouldn't be wrong Jan 10 08:38:49 RP: good to hear too Jan 10 08:40:04 RP: looks like both meta-opie* are building now (with b_a_d added) Jan 10 08:41:49 can't you do an easy bb php on your cross development system? Jan 10 08:42:41 kami22: I have switched from X86 to PPC/OS X and I'm in the process of getting glibc to build Jan 10 08:42:48 kami22: so sadly I can not try it myself now Jan 10 08:43:07 kami22: even with OE cross compiling remains a tough business Jan 10 08:43:08 okay Jan 10 08:44:15 kami22: did your grep for /usr/lib yet? Jan 10 08:48:22 RP: NOTE: package opie-image-1.0-r19: task do_build: completed Jan 10 08:49:04 hrw|afk: excellent. So I just need to work all this into a proper patch for bitbake :) Jan 10 08:49:42 zecke: i looked in the config.log and found out that really often the -L/usr/lib is used and i think its an autoconf command Jan 10 08:49:51 but i dont know what to do now? Jan 10 08:50:02 RP: too much kernel modules Jan 10 08:50:31 hrw|afk: How do you mean? Jan 10 08:50:52 kernel-modules pacakge is in image Jan 10 08:51:13 http://ewi546.ewi.utwente.nl/tmp/hrw/ipkg-status.txt Jan 10 08:51:37 kami22: the php_5.0.5.bb looks correct Jan 10 08:52:12 back to the movie.. Jan 10 08:52:23 hrw|afk: Ah, thats a bug then. I'm not sure why that would happen as I shouldn't have changes the ipks ipkg was being told to install Jan 10 08:56:43 funny thing - nothing depend on kernel-module Jan 10 08:56:44 s Jan 10 09:07:38 zecke: okay but where can be the problem? Jan 10 09:09:10 kami22: well take a look at the source (sorry for that answer) Jan 10 09:09:27 kami22: check config.log which pulls /usr/lib, check Makefile*, use grep Jan 10 09:20:32 hrw|tv: Hmm. Something is very wrong then :-/ Jan 10 09:22:04 i find only something like this in Makefile : Jan 10 09:22:07 EXTRA_LDFLAGS = -L/usr/lib Jan 10 09:22:08 EXTRA_LDFLAGS_PROGRAM = -L/usr/lib Jan 10 10:07:49 What's the best way to remove samba from my images? Jan 10 10:13:53 Luke-Jr: is it in your image ? I think it is a dep from abiword... Jan 10 10:14:33 alan|laptop: pretty sure it's in the image... :\ Jan 10 10:14:43 arf... Jan 10 10:26:09 hi .. Jan 10 10:27:10 i am doing a bb for btsco and i fail as the package extracts simply to btsco but oe makes a btsco-VERSION tmp/work/btsco-Version/ Jan 10 10:28:52 so to clarify the bb work dir is btsco-0.41-r0 but inside the actual data sources go into simply tmp/btsco-0.41-r0/btsco and oe try`s to build in mp/btsco-0.41-r0/btsco-0.41 .. so i either miss something in the bb or do it wrong ? Jan 10 10:29:01 * Philippe is back (gone 19:01:12) Jan 10 10:30:10 oh han on .. maybe it is vice versa then i now what todo maybe Jan 10 10:31:39 ah got it , thanks you all :-) Jan 10 10:36:45 evening Jan 10 10:39:25 reenoo, if hrw|gone was here, he would say Jan 10 10:39:29 ~ugt Jan 10 10:39:33 methinks ugt is Universial Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. Jan 10 10:39:43 but hrw|gone is not here... ^^ Jan 10 10:41:02 guys, what is the difference detween the DEPENDS line and the RDEPENDS line in a bb file ? Jan 10 10:41:19 ugt is the surpeme law Jan 10 10:41:25 amen Jan 10 10:47:07 heh Jan 10 10:47:25 I guess I couldn't care less Jan 10 11:43:29 Hi, my OE and monotone are up to date. When I try to build, bitbake eventually aborts because some file apparently cannot be downloaded. -> http://rafb.net/paste/results/kzNI2w77.html Jan 10 11:43:43 Any comments as to the problem and remedy? Jan 10 11:59:13 need a little oe help, having difficulty Jan 10 12:20:41 hey pb_ mithro Jan 10 12:22:52 RP: it looks like packages/meta/slugos-image.bb needs updating, but at present I have a simple failure which you may know about already: Jan 10 12:22:55 NOTE: dependencies for ixp-osal are: quilt-native virtual/armeb-linux-uclibc-gcc virtual/libc virtual/kernel kernel (2.6.15) Jan 10 12:22:55 ERROR: Nothing provides (2.6.15) Jan 10 12:23:27 I.e. the dependency is "kernel (2.6.15)" but it looks like something has done a split and got "kernel" "(2.6.15)" Jan 10 12:23:28 anyone here using nylon a meshcube with an atheros card ? Jan 10 12:23:36 hi lrg|home Jan 10 12:24:03 * pb_ just got home from visiting a customer Jan 10 12:24:11 I hadn't realised Cirencester was so far away Jan 10 12:24:14 heh Jan 10 12:24:32 pb_: where do you live ? Jan 10 12:24:37 Cambridge Jan 10 12:24:40 ah Jan 10 12:26:01 jbowler: Yes, I know about this. http://sam.rpsys.net/bitbake-rproviders-r1.patch will solve that issue Jan 10 12:26:35 jbowler: This should let you move to more subtle ones. I'm busy totally rewriting the bitbake patch, properly this time :) Jan 10 12:27:22 Yes, I know :) Jan 10 12:27:50 Is that patch checked in to SVN? (I've got 332 at present) Jan 10 12:28:08 Stupid question, I can find that out myself. Jan 10 12:28:18 jbowler: No. Whilst its better, it isn't right enough to check in Jan 10 12:29:25 Ok, I'll patch my 332 and see what happens - will I need to change packages/meta/slugos-image (NSLU2 is not using task-bootstrap)? Jan 10 12:31:21 jbowler: In theory, you can drop the DEPENDS and add the line RDEPENDS = "${IPKG_INSTALL}" Jan 10 12:31:55 The idea of these patches is to remove the duplication and requirement to sync DEPENDS and RDEPENDS Jan 10 12:33:43 RP: right, so far as I am aware NSLU2 does not have this problem - I've checked all the images very carefully for this (because we are at the limit of the flash capacity for openslug). Jan 10 12:34:52 I'll try the change you suggest - at present the bitbake fix seems to have solved the problems, but slugos-image hasn't quite built yet. Jan 10 12:35:20 jbowler: The issue I'm aiming to fix is a metadata problem, not one that will change the image at all. Jan 10 12:35:32 The RDEPENDS line in slugos-image is already correct, it should be sufficient just to drop DEPENDS. Jan 10 12:35:58 Right, I understand - if bitbake builds slugos-image it will be ok, unless it builds the wrong version of something, which it hasn't so far. Jan 10 12:35:59 jbowler: If the RDEPENDS is right in theory you can do just that :) Jan 10 12:36:26 hey all, I keep running into do_install problems because coreutils-native is not built (as a dependency of gpe.bbclass) Jan 10 12:36:43 In fact it will still have DEPENDS, because 'upslug2-native' must be built and nothing other than the image requires it (it is used to flash the image). Jan 10 12:37:28 jbowler: That should be a depends of the machine file - see how the zaurus handles zaurus-updater Jan 10 12:37:50 jbowler: I put that badly :-/ Jan 10 12:38:02 slugos-image now builds Jan 10 12:38:16 Which file should I look at? Jan 10 12:38:41 Ok: EXTRA_IMAGEDEPENDS += "zaurus-updater" Jan 10 12:38:46 jbowler: EXTRA_IMAGEDEPENDS - see zaurus-clamshell.conf Jan 10 12:39:15 Guess I need to rebuild from scratch to test this... Jan 10 12:40:13 pH5: that's a weird problem Jan 10 12:40:34 it might be a task for Coremaster Zecke (or, if he were not on vacation, Coremaster Mickey) Jan 10 12:42:08 pH5: Is this related to recent changes or not? I'm hoping its not... Jan 10 12:43:24 RP: no, this started to happen earlier Jan 10 12:44:08 pH5: With debug on, does coreutils appear in the depends list? Jan 10 12:44:11 NOTE: dependencies for libxsettings are: coreutils-native virtual/libintl intltool-native ipkg-utils-native quilt-native virtual/arm-linux-gcc virtual/libc gettext-native x11 Jan 10 12:44:38 Looks like some problem in the core then :-/ Jan 10 12:48:38 RP: neat, openslug.conf is becoming much simpler (I no longer need to know which package the ext2 programs come from for example). Jan 10 12:53:17 coreutils-native is in status.ignored_dependencies here for some reason Jan 10 12:54:03 ouch Jan 10 12:54:04 :-) Jan 10 12:54:12 ~lart pH5 Jan 10 12:54:12 * ibot blasts pH5 to oblivion with a kamehameha wave Jan 10 12:54:19 what might be the reason for loads of such messages when pullin updates? monotone: warning: discarding revision data packet ea1b1fd1c744adb0b6a3d4bfd0d80654e097fee8 with unmet dependencies Jan 10 12:54:25 mhh still problem with glibc ... and i cut off all my .bb files and wipe out tmp dir :( :( :( ... Jan 10 12:54:27 i put it into assume_provided myself... Jan 10 12:54:41 pH5: ! Jan 10 12:55:05 ... --> | Applying patch arm-audit.patch Jan 10 12:56:11 gremlin[it]: What date glibc is it building? Jan 10 12:56:25 jbowler: This was the general idea :) Jan 10 12:56:54 jbowler: The entire removal for BOOTSTRAP_EXTRA_DEPENDS was possible which was nice for the other devices :) Jan 10 12:57:17 really unsettling is that I don't remember that I did this, let alone why Jan 10 12:57:24 pH5: oops :) Jan 10 12:58:01 package glibc-2.3.5+cvs20050627-r1 Jan 10 12:58:50 but bitbake did a cvs update today (less then an hour ago) Jan 10 12:59:08 gremlin[it]: The last time I saw this error, the person hadn't updated to the latest bitbake svn... Jan 10 12:59:18 ok i try ! Jan 10 12:59:29 which meant the CVSDATE -> SRCDATE conversion caught them out Jan 10 13:00:54 RP: do I need to define IPKG_INSTALL at all now? (Can I just define RDEPENDS?) Jan 10 13:01:38 jbowler: no, you need IPKG_INSTALL for now. Depending on how things work out, that might become unneccessary, we'll see... Jan 10 13:02:35 mhhh seem also fast parfing all .bb files :) Jan 10 13:02:51 Ok, and slugos-image apparently works at present without a line of the form RDEPENDS = "${IPKG_INSTALL}", but I assume it is required now that I have deleted DEPENDS. Jan 10 13:03:18 i got an exception parsing org.openembedded.dev/packages/linux/handhelds-sa-2.6_cvs.old.bb Jan 10 13:03:34 and after the same error ... i have to wipe out tmp dir ? Jan 10 13:03:41 jbowler: For all I know, image_ipk is clever and does that for us :) Jan 10 13:03:59 I looked and I don't think it is, I think it is a side effect of the DEPENDS Jan 10 13:04:26 glurp... is there no sound for c7x0 ? Jan 10 13:04:39 on 2.6 kernels ? Jan 10 13:04:56 (fresh installed onw build...) Jan 10 13:05:27 alan|laptop: sound works, you need change your mixer settings. Jan 10 13:05:54 lrg : i saw that on oz website, but how do you change that ? Jan 10 13:06:36 s/lrg/lrg|home Jan 10 13:06:39 alan|laptop: use alsamixer and enable the output mixer Jan 10 13:06:58 jbowler: Yes, you probably need the RDEPENDS then. Jan 10 13:07:33 lrg|home: function snd_ctl_open failed for default : No such device Jan 10 13:07:44 alan|laptop: Which kernel is this? Jan 10 13:08:03 2.6.15-rc7 Jan 10 13:08:22 is that normal ? Jan 10 13:08:48 yes, that's fine. Did you install a new image when you flashed it? Jan 10 13:08:54 yep Jan 10 13:09:12 alan|laptop: what is the output of cat /proc/asound/pcm ? Jan 10 13:10:14 00-00: WM8731-i2s-0 : : playback 1 : capture 1 Jan 10 13:10:43 alan|laptop: looks like the driver is loaded ok. Jan 10 13:10:51 iirc, i should turn off something, but how/what ? Jan 10 13:11:09 * RP wonders if all of alsa is present in a default oz image? Jan 10 13:11:24 RP: ah Jan 10 13:11:34 i just used a bitbake gpe-image Jan 10 13:11:55 can i use oss/install alsa ? Jan 10 13:12:47 alan|laptop: I suspect we're missing some key alsa package, perhaps alsa-conf. Try installing that and see if it helps Jan 10 13:12:57 If so, tell us so we can fix OE :) Jan 10 13:14:31 alsa-conf is installed and up to date... Jan 10 13:14:54 1.0.10-r0 Jan 10 13:15:17 it was already installed Jan 10 13:15:30 alan|laptop: i assume aplay fails as well ? Jan 10 13:15:43 alan|laptop: perhaps one of the other alsa packages then...? Jan 10 13:17:11 lrg|home: i don't know what aplay is, but this command is not on my zaurus... Jan 10 13:18:09 alan|laptop: Its in alsa-tools-aplay I think Jan 10 13:18:26 alan|laptop: aplay is the alsa audio player. please try installing the other packages as RP suggests Jan 10 13:18:40 alsa-utils-aplay Jan 10 13:18:53 RP: here is what compiled by default with gpe-image Jan 10 13:18:54 http://pastebin.ca/36385 Jan 10 13:19:14 lrg|home: fresh gpe sound worked for me Jan 10 13:19:28 and me on borzio Jan 10 13:19:46 alsa-utils-aplay is not default... but doent need it for setup mixer Jan 10 13:20:28 ha... Jan 10 13:20:31 iirc it worked last night with Luke-Jr Jan 10 13:21:03 oz ljr is too big for my c750 Jan 10 13:21:19 * alan|laptop has no luck today :'( Jan 10 13:22:38 is there no way to solve this problem ? Jan 10 13:23:56 on oz website, i see : c7×0 users need to unmute ?Output Mixer HiFi Playback Switch? Jan 10 13:24:14 how can i do this ? it is not explained... Jan 10 13:24:22 alan|laptop: you need to be able to load alsamixer. If you can't load that there is some problem Jan 10 13:24:34 RP: ha. Ok. Jan 10 13:27:07 the very strange thing is if i do a "mplayer -ac mad audio.mp3", mplayer doesn't crash and acts as if it was playing the file... Jan 10 13:27:28 alan|laptop: The sound is just muted and you need to unmute it Jan 10 13:27:45 haaaa Jan 10 13:27:50 to unmute it you need alsamixer. Jan 10 13:27:57 got it ^^ Jan 10 13:28:15 (maybe the only thing i got ;-( ) Jan 10 13:29:21 ? Jan 10 13:30:22 greentux: do you not agree ? Jan 10 13:33:11 alan|laptop: no i dont know what you mean with "got it" Jan 10 13:38:17 gremlin[it]: I think he meant he understood what we were saying Jan 10 13:38:37 greentux: I meant to say that to you... Jan 10 13:39:04 Too many similar nicknames for the first two letters to work :) Jan 10 13:39:35 hiyas Jan 10 13:39:41 can someone explain to me how the busybox route command works? Jan 10 13:39:56 route add 192.168.0.21 192.168.1.200 Jan 10 13:40:04 route: SIOC[ADD|DEL]RT: No such device Jan 10 13:40:21 root@collie:~# route add 192.168.0.21 192.168.1.200 usbd0 : usage info Jan 10 13:40:27 I think we need to start adhering the topic a bit... Jan 10 13:40:46 ah yeah sry failed to notice that Jan 10 13:41:04 unomi: you need "gw" in there somewhere Jan 10 13:41:14 its just that the docs at http://www.busybox.net/downloads/BusyBox.html dont mention much Jan 10 13:41:30 gw? Jan 10 13:41:36 yeah Jan 10 13:41:56 gateway Jan 10 13:41:58 again, but like im a total noob Jan 10 13:42:10 "route add 192.168.0.21 gw 192.168.1.200" Jan 10 13:42:19 lovely! Jan 10 13:42:34 ! thanks guys Jan 10 13:42:35 unomi: Dunno? Jan 10 13:42:48 and now it is my internet connection that give me up... Jan 10 13:42:57 i deserve it... i am stupid Jan 10 13:43:19 ~lart alan|laptop Jan 10 13:43:20 * ibot takes out a cattle prod and gives alan|laptop a good jolt Jan 10 13:53:43 heh, "RPDEPENDS" Jan 10 13:53:50 that's kind of cute Jan 10 13:54:04 glibc now compiling !!! :) Jan 10 13:54:59 greentux, lrg|home RP : thanks a lot for your help, and sorry i've been so stupid Jan 10 13:55:51 gremlin[it]: your bitbake was out of date? Jan 10 13:56:38 pb_: Did I type that somewhere? Jan 10 13:56:39 RP : yes after svn co of bitbake and a tmp wipe out all worked :) Jan 10 13:57:07 gremlin[it]: excellent Jan 10 13:57:08 RP: yeah, in one of your emails to the oe list Jan 10 13:57:39 "This means the existing preferred provider scheme can't be used with Jan 10 13:57:39 RPDEPENDS :-(." Jan 10 13:58:00 pb_: hehe Jan 10 13:58:23 That email was just plain wrong as far as I can work out now... Jan 10 14:07:31 RP: Synchronization Framework Jan 10 14:07:32 To back-up their device or share data with desktop computers, users can synchronize their contacts, calendar events, to do items, multimedia files and documents with either Qtopia Desktop, a cross-platform desktop PIM suite from Trolltech or Microsoft® Outlook. Jan 10 14:07:42 RP: i have to test this :) Jan 10 14:08:18 greentux: Sounds like fun :) Jan 10 14:10:45 RP: will check tomorrow. you know what the default browser in opie is? Jan 10 14:11:37 greentux: konqueror I think Jan 10 14:12:11 RP: mmmh but its unmaintained. i guess minimo will run only with x environment...? Jan 10 14:12:43 RP: i read on openedhand.com: Ipaq 2750 Linux port... are you also familiar with 4700hx? Jan 10 14:13:25 greentux: The hx2750 and 4700 are very similar so I know a bit about it, yes Jan 10 14:13:46 well very similar in some respects and totally different in others :) Jan 10 14:14:25 RP: ok... Jan 10 14:20:47 so i will go to sleep cu and gn8 Jan 10 14:22:31 ho no ! it still doesn't work... all i can hear is a high really dirty sound... Jan 10 14:23:04 and it doesn't come from my mp3 file... i double checked with two different files that work on desktop... Jan 10 14:23:48 sound test in GPE sound setup works, though... Jan 10 14:24:14 alan__: the mixer is to overloaded in the moment. very confusing. but lrg works on it. so weh sould wait a day or two :) Jan 10 14:25:07 greentux: do you mean this time it doesn't come from my stupidity ? haaaa !! that is a bit relieving ! ;) Jan 10 14:25:52 alan__: dont know, but i tried a voip call yesterday an setup the mixer around 30 minutes with many tests Jan 10 14:26:12 greentux: you mean there may ne Jan 10 14:26:17 excuse... Jan 10 14:26:34 greentux: you mean there may be a way to set this up anyway ? with alsamixer ? Jan 10 14:27:00 alan__: yes of course Jan 10 14:27:11 alan__: keep in mind you have a c7x0 and greentux has a c3000 Jan 10 14:27:27 the sound is different Jan 10 14:27:30 RP: good point... Jan 10 14:27:56 C3100! :) Jan 10 14:28:13 greentux: near enough ;-) Jan 10 14:28:29 plus i really don't know anything about alsamixer... i think i'm going to blind tweak and report here if i find anything interesting... Jan 10 14:28:47 ok folks. cu tom Jan 10 14:30:49 <[lala_]> RP: do u have already updated kernel to 2.6.15 in oe? Jan 10 14:31:29 [lala_]: I did add that, yes Jan 10 14:31:51 <[lala_]> RP: so its now the default kernel for borzoi? Jan 10 14:32:08 [lala_]: Yes. Not much changed from -rc7 though Jan 10 14:32:13 <[lala_]> i see Jan 10 14:33:13 <[lala_]> do u know if it's possible to update only some packages in the working directory with monotone update or do i always get the full bunch of changes? Jan 10 14:33:43 [lala_]: you should be able to go into the appropriate packages/foo directory and run a monotone update there. Not sure if it's wise though. Jan 10 14:34:25 NAbyss: That's not correct at all Jan 10 14:34:48 [lala_]: you can't I'm afraid. Best of checkout a separate tree and get the files that way Jan 10 14:34:59 RP: Hm, update does the whole tree? Jan 10 14:35:05 NAbyss: correct Jan 10 14:35:05 <[lala_]> RP: i was afraid of that :-/ Jan 10 14:35:44 <[lala_]> RP: problem is, that i have a working image now, but after the last commit mails i noticed that quite a lot has changed Jan 10 14:35:57 Anyways, off to work Jan 10 14:36:03 <[lala_]> so i'm a bit reluctant wiht updates atm Jan 10 14:36:24 [lala_]: I'd avoid them to be honest as things are a bit broken at the moment, as advised on oe@ Jan 10 14:36:55 <[lala_]> yep Jan 10 14:36:56 checkout a new tree and copy the kernel files accross to your good one Jan 10 14:37:11 <[lala_]> sounds as best solution for now Jan 10 14:37:52 <[lala_]> maybe i should even use a seperate .db file for that Jan 10 14:38:42 [lala_]: a separate db won't gain you anything Jan 10 14:39:56 <[lala_]> how so? Jan 10 14:40:34 <[lala_]> if i would delete some files by accident in my working copy i could check out it again if i preserve the .db Jan 10 14:40:52 You still can check out previous revisions Jan 10 14:41:26 <[lala_]> true, but i find the version numbers not very well adjusted for human beeings ;-) Jan 10 14:41:37 see the topic, the oe mailing list or the oe website for the infomation about what i tagged the repository as before making these changes Jan 10 14:41:43 That's why there's tags ;) Jan 10 14:41:59 <[lala_]> ah, i see Jan 10 14:42:21 <[lala_]> didn't noticed the topic cause it's always so crowded ;) Jan 10 14:47:00 mhhh for an ipaq (strong arm) package linux-jlime-sh3-2.6.11 have sense ???? Jan 10 14:47:37 gremlin[it]: current bitbake and the oe metadata have a few issues Jan 10 14:48:02 ah ok ... ;) Jan 10 14:48:46 mhh in fact ... before there was a suggestion --> NOTE: consider defining PREFERRED_PROVIDER_kernel <<- Jan 10 14:49:22 hey on the kernel provider is listed also handhelds-sa-2.6 is just a few usable ? Jan 10 14:50:16 gremlin[it]: Yes, its known about and we need to fix it... Jan 10 14:50:29 * RP is working on a fix Jan 10 15:00:45 RP: looks like there is an unresolved order dependency in building external kernel modules: Jan 10 15:00:49 NOTE: dependencies for ixp400-eth are: ipkg-utils-native quilt-native virtual/armeb-linux-gcc virtual/libc ixp4xx-csr virtual/kernel Jan 10 15:00:49 NOTE: building all deps Jan 10 15:00:49 NOTE: bb.build.FuncFailed:("can't read from file '%s' (%s)", ('/home/work-tmp/jbowler/nslu2/openslug/staging/armeb-linux/kernel/kernel-abiversion', )) while evaluating: Jan 10 15:00:50 ${@base_read_file('/home/work-tmp/jbowler/nslu2/openslug/staging/armeb-linux/kernel/kernel-abiversion')} Jan 10 15:00:50 NOTE: bb.build.FuncFailed:("can't read from file '%s' (%s)", ('/home/work-tmp/jbowler/nslu2/openslug/staging/armeb-linux/kernel/kernel-abiversion', )) while evaluating: Jan 10 15:00:52 ${SHLIBS}ixp4xx-csr kernel (${KERNEL_VERSION}) Jan 10 15:01:54 I.e. ixp4xx-csr (which inherits module) needs virtual/kernel, which is not yet built (because of the order in RDEPENDS I guess) and the bb file read evaluation fails because it is not yet built... Jan 10 15:03:16 jbowler-away: With any luck, order issues like that should get resolved when I splt up the depends and rdepends handling in bitbake Jan 10 15:04:12 I think I can hide it just by reversing the order of the two things (virtual/kernel vs modules) in slugos-image.bb Jan 10 15:04:53 Ah, virtual/kernel isn't in slugos-image.bb Jan 10 15:05:23 jbowler-away: The kernel module should be depending on virtual/kernel... Jan 10 15:05:39 The bug is in module.bbclass: Jan 10 15:05:39 It can't build without it being provided... Jan 10 15:05:45 kernel.bbclass:PROVIDES += "virtual/kernel" Jan 10 15:05:46 module.bbclass:DEPENDS += "virtual/kernel" Jan 10 15:06:11 Or the problem, rather. Jan 10 15:06:24 Why is that a problem? Jan 10 15:06:56 The module has a DEPENDS on virtual/kernel, the image has an RDEPENDS on virtual/kernel, the result is a bad ordering Jan 10 15:07:51 Hum... when you say "the kernel module" what do you mean? I.e. where should the DEPEND on virtual/kernel be? Jan 10 15:08:07 module.bblcass, as it is Jan 10 15:08:25 "bad ordering" shouldn't be possible and its happening because of the way bitbake combined depends and rdepends. I'm rewriting it to seperate this Jan 10 15:08:45 Ok. Bug in slugos-image maybe - maybe it never had an RDEPENDS on virtual/kernel. Jan 10 15:09:07 if has an RDEPENDS on a kernel module though? Jan 10 15:09:28 Right, I understand that there's a bitbake bug too, but I think slugos-image is maybe broken - yes, it has an RDEPENDS on the modules but not the kernel (which is wrong) Jan 10 15:09:36 That kernel module has a DEPENDS on virtual kernel so v/k should be build before attempting to build the module Jan 10 15:10:11 You'd find things end up in your image as the kernel modules should RDEPEND on virtual/kernel for you Jan 10 15:10:15 It will be built, but when bitbake attempts to work out what v/k to build it attempts to evaluate KERNEL_VERSION, which presupposes a built virtual/kernel... Jan 10 15:10:26 So including one module pulls in the whole kernel Jan 10 15:10:34 which is correct Jan 10 15:10:37 Right Jan 10 15:10:57 (But it is valid to build a slugos-image with no kernel modules too) Jan 10 15:11:28 ah, if that image should have a kernel in, then you rdepends are wrong :) Jan 10 15:11:49 Indeed they are. Jan 10 15:12:49 is the KERNEL_VERSION string in a DEPENDS or RDEPENDS? Jan 10 15:13:27 KERNEL_VERSION depends on something in staging right? Jan 10 15:13:31 It's from cooker.cook and it is evaluating: Jan 10 15:13:38 ${SHLIBS}ixp4xx-csr kernel (${KERNEL_VERSION}) Jan 10 15:14:14 I believe that's the result of bitbake changing virtual/kernel to the actual package Jan 10 15:14:52 So it is in the DEPENDS (from the DEPENDS += above) Jan 10 15:16:32 Right, this is from module.bbclass's RDEPENDS line Jan 10 15:17:02 The log from bitbake is here: http://pastebin.ca/36405 Jan 10 15:17:14 It's module.bbclass's DEPENDS line I think Jan 10 15:17:56 I think its the RDEPENDS line, hence the "kernel" Jan 10 15:17:58 Ok, I'm wrong - it's the RDEPENDS line Jan 10 15:18:12 (I was reading the file with egrep ;-) Jan 10 15:18:22 Basically we can't attempt to parse that until after virutal/kernel has staged. I'm not sure I like this... Jan 10 15:19:22 Lets see if the rewrite avoids the issue or not :) Jan 10 15:19:55 I don't think it should be needed. A normal kernel build automagically does the module deps. Jan 10 15:20:18 * Philippe is away: Time to dream :-) Jan 10 15:20:59 It might be there from a time when modules didn't depend properly... Jan 10 15:21:24 I believe there was, I think I fixed something at one point. Jan 10 15:21:36 Its certainly a rather nasty RDEPEND to have around... Jan 10 15:23:03 It's actually the thing the bitbake patch you gave me is fixing too. Jan 10 15:23:29 I wonder what happens if it is replaced by "virtual/kernel"... Jan 10 15:23:39 It fixes some things and breaks others :) Jan 10 15:23:56 kernel isn't the problem - its the KERNEL_VERSION that is Jan 10 15:24:04 have a look at what KERNEL_VERSION does... Jan 10 15:24:24 Right, and KERNEL_VERSION is not known until the kernel is built - it cannot be derived from PV, it depends on the kernel config Jan 10 15:25:42 That's because something like 2.6.15-rc2 can generate 2.6.15, 2.6.15-rc2, 2.6.15-rc2-private according to how the build is configed - these end up being the subdirectory name in /lib/modules/ Jan 10 15:31:40 jbowler-away: correct Jan 10 15:32:24 jbowler-away: Although perhaps the RDEPENDS should be using the kernel PV instead... Jan 10 15:33:00 RP: my gpe-image build just completed without further problems Jan 10 15:33:23 pH5: Excellent. At least things partially work :) Jan 10 15:45:23 hrw: are you there? Jan 10 15:53:44 RP: adding virtual/kernel to RDEPENDS is not a good idea... the rootfs step fails "Cannot find package virtual/kernel." Jan 10 15:54:38 I remember I had to use "${PREFERRED_PROVIDER_virtual/ixp-eth}" to work round this before Jan 10 15:55:01 er? virtual/kernel is a build time PROVIDES, not an RPROVIDES Jan 10 15:55:09 Right Jan 10 15:55:20 perhaps you meant "kernel" or "kernel-image"? Jan 10 15:55:23 or some such Jan 10 15:55:31 Yes, probably... Jan 10 15:56:01 And the same applies to virtual/ixp-eth - i.e. I should probably be using RPROVIDES for that. Jan 10 15:56:24 So why doesn't the module.bbclass do RDEPENDS = "kernel-image"? Jan 10 15:56:36 because modules dont provide kernels? Jan 10 15:56:44 err Jan 10 15:56:45 kergoth: depends, not provides Jan 10 15:56:51 No, but they RDEPEND on kernels Jan 10 15:56:52 yeah, saw that Jan 10 15:56:53 hehe Jan 10 15:57:02 jbowler-away: because modules depend on a particular kernel, not just any kernel Jan 10 15:57:25 actually, I'm not terribly sure why they do have that dependency at all; it should maybe just be a recommends or suggests Jan 10 16:00:03 that's the problem - see above - the DEPENDS in virtual/kernel, the module in question doesn't care which kernel version it gets Jan 10 16:00:17 that's a build time dependency Jan 10 16:00:39 the module indeed doesn't care what kernel it builds against, but it does care that the kernel it runs with matches the one it was built with Jan 10 16:01:01 virtual/kernel is correct in DEPENDS, because any kernel will do at that point. Jan 10 16:04:32 The only issue I see is that the RDEPENDS is using ${KERNEL_VERSION} and K_V != to the kernel package's ${PV} Jan 10 16:06:47 It is my imagination or does parsing somehow seem faster with my bitbake patches? Jan 10 16:07:26 let it be true Jan 10 16:08:25 There's no rational reason for it to be faster. If anything we chuck more data around and do more computation... Jan 10 16:27:02 computing RDEPENDS after building a package seemed like a good idea until you consider opie-image... Jan 10 16:30:59 good night Jan 10 17:00:10 * RP sets the new bitbake patch away building. It passes a dry run test... Jan 10 17:17:31 someone care to add --enable-ipv6 to curl's bb? Jan 10 18:41:26 RP: it looks like the changes I made to slugos-image are doing the correct thing, however I have a question about PREFERRED_PROVIDER; Jan 10 18:42:01 At present I have PREFERRED_PROVIDER_virtual/kernel, but (apart from the module.bbclass stuff) nothing DEPENDS on virtual/kernel - the image now RDEPENDS on "kernel" Jan 10 18:42:13 Can I change (should I change) to PREFERRED_PROVIDER_kernel? Jan 10 18:42:28 Or PREFERRED_RPROVIDER_kernel (if that works) Jan 10 20:06:07 Would OE support a Zaurus *host* system? Jan 10 20:07:11 if you had a drive capable of doing i would think it could Jan 10 20:08:35 emte: C3100 should be, I think Jan 10 20:08:40 But I'm thinking NFS Jan 10 20:09:59 there are some issues with NSF i belive Jan 10 20:12:17 I've used it before, I know... Jan 10 20:12:20 Yeah.. NFS = Never f. secure.. Jan 10 20:12:29 NAbyss: not even on a LAN? Jan 10 20:12:50 actually you "can" secure an nfs .. Jan 10 20:13:03 to a certain extent Jan 10 20:13:05 Luke-Jr: I wouldn't trust it without IPSec or a VPN between the two wndpoints Jan 10 20:13:12 shrug Jan 10 20:13:25 emte: Yeah, there's extensions that do secure it.. but not secure by itself.. Jan 10 20:13:30 ro,async,all_squash,anonuid=65534,anongid=35 Jan 10 20:13:34 secure Jan 10 20:13:48 tho of course that wouldn't work for compiling ;) Jan 10 20:14:03 Luke-Jr: True.. you'd have a bit of a problem there ;) Jan 10 20:14:23 its gonna take you a week atlest to compile a basic image i think ... using the Z Jan 10 20:14:28 atleast* Jan 10 20:14:37 emte: I'm not thinking about an image, honestly =p Jan 10 20:14:42 lol Jan 10 20:14:44 I still dream of KDE on my Z Jan 10 20:14:45 thats good Jan 10 20:14:49 eww Jan 10 20:15:03 i think you have no hope of that anymore Jan 10 20:15:10 ...? Jan 10 20:15:17 KDE seems to have some pretty heavy memory reqs now Jan 10 20:15:23 not really Jan 10 20:15:26 Not sure how usable it'd be at 640x480 either.. Jan 10 20:15:32 it's only been improving over the last two .y releases Jan 10 20:15:42 NAbyss: that might be an issue Jan 10 20:15:44 Although... you could just X -indirect in the meantime.. Jan 10 20:15:52 I wonder if the Z can software-scale it Jan 10 20:17:15 bug: Xfbdev isn't suid root Jan 10 20:19:18 * Luke-Jr runs KDE remotely to check resolution issues Jan 10 20:19:39 1. KPersonalizer is too tall Jan 10 20:20:04 Hmm.. would DPI fix that? Jan 10 20:20:15 Not sure if KDE scales by point sizes or pixel. Jan 10 20:20:17 doubt it, but maybe Jan 10 20:20:24 DPI would probably make it worse, actually Jan 10 20:20:44 xserver-kdrive lacks proper DPI support for the Z anyhow Jan 10 20:21:06 KDE looks nice on the Z :) Jan 10 20:21:18 only possible problem I can see right now is mouse activity Jan 10 20:21:37 which is easily fixed by modifying the mouse drivers to use touchscreen for movement and OK/Cancel for left/right buttons Jan 10 20:22:59 can always bind a button too Jan 10 20:23:17 ? Jan 10 20:23:48 some button you hardly use Jan 10 20:24:05 reassign it as your second mouse button Jan 10 20:24:16 that doesn't permit movement w/o clicking Jan 10 20:24:28 it could Jan 10 20:24:34 ? Jan 10 20:24:39 think ctrl or alt Jan 10 20:24:51 ... Jan 10 20:24:55 to toggle to second Jan 10 20:25:06 but how would I move the mouse w/o clicking? Jan 10 20:25:31 ah i see what your saying now Jan 10 20:25:58 i was thinking the other way Jan 10 20:26:59 another possibility would be to use the ts as one of those touchpads on a laptop Jan 10 20:27:03 but that'd be plain ugly Jan 10 20:29:00 setting correct DPI kills any hope of KDE on Z =p Jan 10 20:29:03 I like incorrect DPI Jan 11 01:08:02 RP: oe-git is current again Jan 11 01:12:04 pb_: god of binutils, can I force objdump to dissassemble everything? Currently it emits ... for one whole symbol Jan 11 01:12:36 does this mean it is not there (which could be as nm names it a weak object V) Jan 11 01:12:53 if it says "...", that means the space is full of zeros Jan 11 01:13:15 I think "--disassemble-zeros" will force it to disassemble those as well Jan 11 01:13:16 okay Jan 11 01:14:20 lol... stupid me ;) Jan 11 01:14:30 sure it is zero... it is a dso... Jan 11 01:14:34 thanks Jan 11 01:14:50 * zecke fights a bug in the old OE CSL toolchain Jan 11 01:23:02 pb_: as usual, thanks a lot Jan 11 01:35:09 plop Jan 11 01:47:13 morning all Jan 11 01:49:26 jbowler-zzz: Leave PREFERRED_PROVIDER as it is (with virtual/kernel). Having just kernel in there would be illegal. I think bitbake will end up using it, even if you don't specifically mention it in the image RDEPENDS Jan 11 01:51:03 jbowler-zzz: PREFERRED_RPROVIDER doesn't do anything and also isn't as much use as you'd first suspect Jan 11 01:53:51 RP: I have 2.6.15-rc7+GPE installed on an Akita. I was wondering if there was any specific testing you'd like me to do Jan 11 01:54:01 I asked on the oesf.org forums and lardman told me to ask here Jan 11 01:56:54 johnX: Basically does everything appear to work? :) Jan 11 01:57:12 uhm...alsa causes an Oops Jan 11 01:57:19 really that's about it though Jan 11 01:57:29 can you pastebin the oops? Jan 11 01:57:44 It shouldn't do that :) Jan 11 01:57:50 is monotone.vanille.de down ? Jan 11 01:57:55 its bug #584 on bugs.openembedded.org Jan 11 02:00:04 johnX: Ah, thanks. I should look at the bugzilla more often :) Jan 11 02:00:17 morning Jan 11 02:00:28 I only put it up yesterday I think Jan 11 02:00:51 but I have an enlightened self interest in 2.6 on Akita, so I thought I'd bug you here :D Jan 11 02:01:08 At a quick glance, I think I can guess the problem - its probably using a spitz only scoop interface and should be using the akita io expander... Jan 11 02:01:27 I thought I'd worked around this though :-/ Jan 11 02:01:39 I could be doing something wrong... Jan 11 02:01:49 johnX: This is the kernel's fault ;-) Jan 11 02:02:20 morning all Jan 11 02:02:31 RP: just as long as you're sure it's not the person who compiled the kernel's fault ;) Jan 11 02:03:29 johnX: I am. This is actually Liam's fault :) Jan 11 02:03:48 RP: ok fair enough Jan 11 02:04:00 johnX: I can see the problem in the code. Liam just hasn't realised the differences between akita and borzoi/spitz when he changed something Jan 11 02:04:34 johnX: I can't complain too much as Liam partly gave us alsa sound :) Jan 11 02:05:08 I'm not complaining at all Jan 11 02:05:20 I really appreciate all the work you guys do Jan 11 02:05:32 2.6 is down right "snappy" compared to 2.4 Jan 11 02:05:56 USB works, wireless works, SD, suspend by the front power button Jan 11 02:07:13 johnX: I have a fix - Basically delete the line with the second reference to reset_scoop_gpio(&spitzscoop2_device* in sound/soc/pxa/spitz.c Jan 11 02:07:39 johnX: If you're compiling your own kernel, could you test that for me? Jan 11 02:07:50 absolutely Jan 11 02:08:23 I'll do that and set it to rebuild before I go to sleep Jan 11 02:08:30 johnX: It looks like a simple cut and paste error Jan 11 02:08:56 RP: that's great Jan 11 02:12:49 I had to clean out my oe tree a little because it took up the whole 8G drive it was on. Jan 11 02:13:19 johnX: I know the problem, believe me :) Jan 11 02:13:41 use disk quota on your tmp dir ;) Jan 11 02:13:50 heh Jan 11 02:14:25 that machine that I use for builds is setup like this: / on an 8G drive and /home/bitbake on it's own 8G drive Jan 11 02:15:00 In case anyone is interested, http://sam.rpsys.net/bitbake-rproviders-r4.patch is the latest patch to bitbake. I suspect zecke is going to turn a funny colour... Jan 11 02:15:16 color you mean? Jan 11 02:16:23 zecke: That would depend whether you're american or british ;-) Jan 11 02:17:06 zecke: farbe in your case :) Jan 11 02:17:54 hehe Jan 11 02:18:22 RP: I love explosions, so explode must be good Jan 11 02:21:01 zecke: If my dictionary is that bad, its no wonder I was never any good at German :) Jan 11 02:21:03 RP: I would love if you would use """ and """ in methods ;) Jan 11 02:21:26 RP: no farbe matches perfectly with colour Jan 11 02:22:07 zecke: Ah, explode_deps? :) Jan 11 02:23:20 zecke: Using "and" would probably help the code a lot. I just never dare combine expressions as things might break Jan 11 02:23:25 Is there anyone well-experienced with OE on VIA Epia machine? Jan 11 02:23:46 zecke: It can be cleaned up a lot - the duplicate code in BuildProvider and BuildRProvider for example... Jan 11 02:24:14 RP: generally. Use documentation ;) then you can do print explodeDep.__doc__ ;) Jan 11 02:24:26 RP: and you can put unit tests inside this doc as well Jan 11 02:25:23 zecke: ok. I wanted to make sure it worked first (which its appearing to for an opie-image build so far) :) Jan 11 02:25:56 RP: how can it ever work without specification? Jan 11 02:29:31 RP: it is tough stuff, I need some time to review Jan 11 02:33:01 RP: did you try to unite buildProvider and buildRProvider? Jan 11 02:35:47 RP: we should put that delayed initialisation of the dict into a) a method or b) a class Jan 11 02:40:15 zecke: I did have them united at first but them split them up to aid development. Having finished, I think the middle parts of both functions should be factored into a common function Jan 11 02:41:21 zecke: The code isn't ready for bitbake yet but does give an idea of what's needed - I'll try and tidy it up/document it tonight. Jan 11 02:41:34 zecke: It is rather tough code :) Jan 11 02:43:21 morning all Jan 11 02:43:40 RP: thanks you said it (code not ready for bitbake yet) Jan 11 02:44:24 hi Jan 11 02:44:36 hey hrw Jan 11 02:45:42 hi Jan 11 02:46:10 I have not seen yet where you pass itemtype Jan 11 02:49:02 morning Liam Jan 11 02:49:14 hi Marcin Jan 11 02:49:32 do13: morning Dirk, I'm working on the touch stuff right now Jan 11 02:50:42 RP: do you need help in cleaning up? Jan 11 02:51:04 someone remember how pcmcia/bt/wifi tasks will be handled? Jan 11 02:51:35 lrg|home: thanks. any ideas why the aux reading generates pendown events? Jan 11 02:51:36 approx. at 14:30 I will be in the office and could start building/hacking myself Jan 11 02:52:44 I would like to finish meta-opie cleaning in a way for such things Jan 11 02:52:45 do13: not yet. Jan 11 02:53:22 lrg|home: do you have seen this on mainstone? Jan 11 02:55:00 do13: no atm, as I was previously concentrating on the wm9713 (customer issue). I've got the 12 working now and i'm adding some code now to test the aux Jan 11 02:55:28 hi hrw, lrg|home Jan 11 02:55:38 hey Richard Jan 11 02:56:08 hi RP Jan 11 02:56:44 RP: do you have patch for selfdepends? Jan 11 02:56:54 meta/meta-opie.bb | 20 +++++++++----------- Jan 11 02:56:59 nice change.. Jan 11 02:57:08 zecke: I'm always open to help. Its probably easiest if I do the things I'm aware of like the unused function and the buildProvider refactoring, then open it up to other help from people who know python better than me :) Jan 11 02:57:16 hrw: http://sam.rpsys.net/bitbake-rproviders-r4.patch Jan 11 02:57:20 RP: thx Jan 11 02:58:19 lrg|home: I have a small fix for spitz.c in ASoC - we've had a bug reported on akita Jan 11 02:58:44 RP: thanks Jan 11 02:58:45 lrg|home: Basically delete the line with the second reference to reset_scoop_gpio(&spitzscoop2_device* in sound/soc/pxa/spitz.c Jan 11 02:59:02 Its covered by the set_bias() call underneath **** ENDING LOGGING AT Wed Jan 11 02:59:58 2006