**** BEGIN LOGGING AT Wed Aug 05 02:59:56 2009 Aug 05 03:03:03 03Angus Ainslie  07fso/milestone5.5 * rdeee949270 10openembedded.git/recipes/linux/ (8 files in 2 dirs): linux-openmoko-2.6.31: update openwrt patches Aug 05 03:58:35 whoa, ld is using over 600mb ram to link libwebkit... Aug 05 04:25:57 and can someone explain why libwebkit-1.0.so.2 is relinked in both do_install and do_populate_staging? Aug 05 04:26:27 egads... it takes an eternity to change desktops with that crap in the background :) Aug 05 07:12:36 Fetching omap-kernel fails for some reason. The command "git fetch git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6.git master" fails ... Aug 05 07:40:30 good morning Aug 05 08:06:27 all linux-omap_* fails to build. repository cannot be fetched - what might be wrong ? Aug 05 08:22:22 good morning Aug 05 08:42:22 helllo.....did a OE-build for smdk2440 and want symply to patch the kernel-machine-name ...... could someone give me a hint? Aug 05 08:42:58 had to patch this # git://opensource.wolfsonmicro.com/linux-2.6-asoc-ggdev;protocol=git;tag=asoc-merge-0002 #ORIG - not in WEB! Aug 05 08:42:58 SRC_URI = "git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git;protocol=git;tag=v2.6.21-rc5 \ Aug 05 08:46:56 in oe you can append SRC_URI with a patch Aug 05 08:48:22 tahnks. like file://0001-Enable-cs8900A-network-device-for-smdk2440-board.patch;patch=1 \ ? Aug 05 08:48:58 right Aug 05 08:50:58 this patch seems to be the git-patch? Aug 05 08:51:07 where is the git source? Aug 05 08:51:26 my problem is ...how do I have to start.... Aug 05 08:52:38 just add file:// entry for your patch created against main git:// kernel sources Aug 05 08:54:14 booxter: where can I find the mach-smdk2410.c file to patch this? Aug 05 08:54:26 morning Aug 05 08:54:56 I can not find it... Aug 05 08:55:36 schasch: arch/arm/mach-s3c2410/mach-smdk2410.c Aug 05 08:55:43 hi hrw Aug 05 08:57:01 florian:#find .OE_SMDK2440/ -name mach-smdk2410.c Aug 05 08:57:01 # (nothing) Aug 05 08:57:58 schasch: that's because oe cleans up your kernel tree after it has finished building Aug 05 08:58:22 schasch: if you don't have a patch known to work do the following Aug 05 08:58:47 schasch: bitbake -b virtual/kernel -c configure Aug 05 08:59:12 schasch: then you get the kerel tree with all config and patches from oe Aug 05 08:59:45 schasch: copy this to a location outside oe and prepare the patches you need Aug 05 09:00:50 could someone please verify that this command "git fetch git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6.git master Aug 05 09:01:15 I think it is just plain rubbish. Aug 05 09:01:28 florian:TAHNKS a lot ...i will try... Aug 05 09:01:57 sgh: "fatal: Not a git repository" Aug 05 09:02:05 schasch: yw Aug 05 09:04:24 florian: all linux-omap kernels end that way. What has changed in OE ? Aug 05 09:41:16 florian: I copied the sources and chaged one file...but can not make diff Aug 05 09:41:25 florian:bucket@packet-linux-entwicklung:~/moko/OE_A9M24x0/linux-smdk2440-2.6.20+git-r1$ git log Aug 05 09:41:25 fatal: Not a git repository Aug 05 09:41:25 bucket@packet-linux-entwicklung:~/moko/OE_A9M24x0/linux-smdk2440-2.6.20+git-r1$ Aug 05 09:42:10 florian:bucket@packet-linux-entwicklung:~/moko/OE_A9M24x0/linux-smdk2440-2.6.20+git-r1$ cd git/ Aug 05 09:42:10 bucket@packet-linux-entwicklung:~/moko/OE_A9M24x0/linux-smdk2440-2.6.20+git-r1/git$ git diff Aug 05 09:42:27 # (nothng) Aug 05 10:09:54 can I use SRC_URI to get entire directories recursively with a file:// Aug 05 10:13:15 koobe: so far yes, but do not base on it Aug 05 10:14:04 Is that considered evil and removed later on? Aug 05 10:15:54 hi all anybody here to talk on toolchain ?? Aug 05 10:16:23 I would like to have package source in sync with the metadata overlay. And still keep the package source out of the metadata tree. Aug 05 10:20:45 Is it possible to distribute toolchain obtained from me on my PC ?? Aug 05 10:21:11 I want to provide my team with toolchain Aug 05 10:24:36 dash_, I think I asked something similar in the past. I understood that it was not a good idea to package the toolchain. I was adviced to let the OE build it for everyone. Aug 05 10:25:48 dash_, koobe: We have the meta-toolchain packages in OE for this purpose. Aug 05 10:26:38 These are for distributable toolchains... some people who develop software migt not want to use OE all the time. Aug 05 10:38:30 florian: do you have any ideas about fetching linux-omap? Who else might know something? Aug 05 10:39:40 sgh: linux omap... koen or Crofton Aug 05 10:40:38 florian: thanks Aug 05 10:46:32 Crofton: Hi! Do you know why all linux-omap kernels fails to fetched ? Aug 05 10:51:35 florian: Sorry for late reply Aug 05 10:51:44 R U still there ? Aug 05 10:52:13 a little bit :) Aug 05 10:52:50 florian: Can u suggest me what do I need to do for this Aug 05 10:53:21 I have cloned OE git and built kernel and some packages with bitbake Aug 05 10:53:30 And all are working fine Aug 05 10:54:07 florian: Now I want to share same toolcahin with my team. So can u suggest me somehing ? Aug 05 10:54:16 dash_: bitbake meta-toolchain Aug 05 10:54:27 OK Aug 05 10:54:36 dash_: it builds a tar archive. Aug 05 10:54:55 but it won't clear up my any other code right ? Aug 05 10:55:33 or otherwise is it necessary to give -c option with -f ? Aug 05 10:59:34 florian: will it deploy the toolchain ? Aug 05 10:59:42 where will it keep ? Aug 05 11:00:46 Crofton|work: Hi! Do you know why all linux-omap kernels fails to fetched ? Aug 05 11:00:49 daq4th: no its just a target like other ones Aug 05 11:01:09 dash_: deploy/glibc/sdk whatever Aug 05 11:01:32 ok thanks Aug 05 11:08:17 sgh, no Aug 05 11:08:42 what machine are you building? Aug 05 11:14:52 XorA: ping Aug 05 11:17:17 shit.. Aug 05 11:17:30 replacing Poky busybox with OE busybox was not good idea Aug 05 11:27:41 florian: pong Aug 05 11:28:01 XorA: linux-smdk2440 is not fetchable any more Aug 05 11:28:22 florian: I dont have a board anymore so feel free to fix/mark as broken Aug 05 11:28:27 * florian needs to kick mickey|patio and zecke again Aug 05 11:28:41 XorA: oops Aug 05 11:29:00 florian: maybe broonie has an interest, he took over from me Aug 05 11:29:09 hello, i'm using OE with the angstrom distribution on an ARM processor and I notice that it takes a long time (20 - 30 sec) from when it says "Freeing init memory" statement to when I see "starting udev version" statement. What's going on during that time? Are there scripts running that I can look at to try and take out things I know are not in my system? Aug 05 11:30:24 leonardo__: mounting the jffs2 filesystem Aug 05 11:30:37 XorA: ok thanks Aug 05 11:30:41 broonie: ping :) Aug 05 11:31:25 florian: is there anyway to speed up that process? Is it the size of the filesystem that determines that time? Aug 05 11:33:08 leonardo__: switch to 2.6.29 or newer and use UBIFS instead of JFFS2 Aug 05 11:34:03 hrw: i'm using 2.6.30 but haven't looked at UBIFS Aug 05 11:34:09 then look Aug 05 11:34:12 it is worth Aug 05 11:36:40 you can even 100% fill ubifs without it breaking Aug 05 11:38:50 vendor kernels suxx Aug 05 11:42:28 the most important button on developer board: RESET Aug 05 11:43:40 yeah... there are way too many devices without reset button :) Aug 05 11:44:12 some have it in wrong places Aug 05 11:44:21 at91sam9263-ek has reset in center of board Aug 05 11:46:31 uf.. booted Aug 05 11:56:29 florian: hrm? Aug 05 11:57:19 broonie: XorA you mentioned your name in one sentence with the smdk2440 he can't support any longer. its kernel bb isn't fetachable any longer Aug 05 11:57:38 I don't have a SMDK24xx that I can boot. Aug 05 11:58:03 The ones we have all seem to be flashed for WinCE. Aug 05 11:58:13 broonie: hmm.. i guess that's not because of the kernel bb, right? Aug 05 11:58:23 doh Aug 05 11:58:37 florian: Yeah, but since I don't have any hardware I'm not even likely to try bulding it. Aug 05 11:58:47 ah too bad Aug 05 11:58:49 broonie: you broke my old one? Aug 05 11:58:57 brb, food Aug 05 11:59:00 XorA: It had a WinCE bootloader on it AFAICT. Aug 05 11:59:12 broonie: you can boot it from usb Aug 05 11:59:18 broonie: Ben Dooks wrote a util Aug 05 11:59:43 Ah, right. I just gave it to ian. Aug 05 11:59:52 broonie: hehe, best thing for it Aug 05 11:59:58 broonie: nasty platform Aug 05 12:00:42 Yeah, I don't really have time to do too much with it anyway. Aug 05 12:00:58 smdk2440 probably has some support in kernel anyway so maybe switch to linux_2.6.30 and mail ML with status? Aug 05 12:01:00 broonie: you have a gta0X anyway, much better devkit Aug 05 12:01:18 Yes, it's definately got mainline support. Aug 05 12:02:15 Possibly worth asking on #s3cxxxx Aug 05 12:02:21 ~hail in-kernel dhcp client Aug 05 12:02:22 * ibot bows down to in-kernel dhcp client and chants, "I'M NOT WORTHY!!" Aug 05 12:02:41 its had mainline support for 2 years now Aug 05 12:03:21 so no need for linux-smdk2440 recipe Aug 05 12:03:53 hrw: no, it was needed when I wrote the recipe as mainline support was lacking, but it was fixed long ago AFAIK Aug 05 12:04:01 Hrm, yeah. That looks like it's just pulling in random audio support. Aug 05 12:04:09 the only issue in mainline is probably just the network driver Aug 05 12:04:34 which didnt use the platform device model back then and was flaky as hell Aug 05 12:05:25 anyway anyone with the hardware feel free to get rid of that junk Aug 05 12:05:28 * XorA no longer cares :-) Aug 05 12:06:40 * broonie recently flipped the i.MX31ADS over to use mainline for the same reason. Aug 05 12:06:50 Should really test that the OE built kernel boots. Aug 05 12:09:15 fsck... Aug 05 12:21:40 florian: Thanks for your help. I am done with toolchain and tested it on other machine. Works fine Aug 05 12:41:46 Crofton|work: Any hint on where to start looking? Aug 05 12:44:07 what is the error? Aug 05 12:47:22 Crofton: http://pastebin.ca/1519144 Aug 05 12:49:28 does git work from the command line? Aug 05 12:49:36 can you access the repository that way Aug 05 12:50:33 Crofton: No, execution "git fetch ......" just gives "fatal: Not a git repository" Aug 05 12:50:56 well, if it does not work from the command line Aug 05 12:51:01 the oe fetcher will not work Aug 05 12:51:15 do you need to clone first (from the command line) Aug 05 12:51:22 maybe you have firewall issues? Aug 05 12:51:33 Crofton: replacing "fetch" with "clone" works. Aug 05 12:52:10 hmmm Aug 05 12:52:15 Crofton: But I don't know how to change that in the recipe, linux.inc or kernel.bbclass Aug 05 12:52:32 maybe you have a corrupt git repo in the oe_source dir Aug 05 12:53:08 Crofton: in what way? Aug 05 12:53:15 not sure Aug 05 12:53:28 anyone know how to clean up things for the git fetcher? Aug 05 12:55:38 I'm not certain how the git fetcher works Aug 05 12:56:04 can anyone maybe try to do a "fetch linux-omap" within an interactive bitbake session? Aug 05 12:56:10 I am sort of guessing it thinks it already has cloned the repo, so it tries to fetch, but the repo is corrupt, so the fetch fails Aug 05 12:56:32 I think there is an easy solution, just do not know what it is Aug 05 12:56:38 XorA, ping Aug 05 12:56:45 Crofton: I have tried to remove tmp several times, so it is a clean build. Aug 05 12:56:50 yeah Aug 05 12:56:58 but there is a seperate sources dir Aug 05 12:57:02 outside of tmp Aug 05 12:57:11 Crofton: aaahhhh ...... Aug 05 12:57:56 inside the diredctory you download sources to Aug 05 12:58:05 or more correrctly that oe downloads to Aug 05 12:58:11 cd into the git dir Aug 05 12:58:25 and remove the directory for the omap kernel Aug 05 12:58:27 and try again Aug 05 12:58:50 Crofton: will do.... Aug 05 13:07:05 Crofton|work: gnop Aug 05 13:08:10 did you see the discussion about the git fetcher dying? Aug 05 13:09:25 Crofton: no Aug 05 13:09:32 Crofton: comes as no real surprise :-) Aug 05 13:09:47 bother Aug 05 13:09:51 er, why? Aug 05 13:10:02 Crofton|work: its never worked that reliably Aug 05 13:10:16 it works on my pc ..... Aug 05 13:10:23 well, last time I tried .... Aug 05 13:10:44 doing stuff like changing the SRC_URI gets it knickers in a twist Aug 05 13:10:51 as does git push -f Aug 05 13:11:55 although must admit havent seen it dissapear up its own ass for a few months now Aug 05 13:12:40 Crofton: brilliant!!!! it ort of worked. Turned out to be an invalid tar.gz snapshot laying around. It must have been confused. Anyway "rm -r git ; rm git*.tar.gz" resolved it. Thank you very much. Aug 05 13:18:53 great Aug 05 13:23:10 sgh, glad to help Aug 05 13:25:22 hi, I'm adding g1 support...where should I post initial recipes? Aug 05 13:25:24 in a branch? Aug 05 13:28:50 anyone ncountered any truly odd issues with setting up an oe buildmachine under ubuntu? Aug 05 13:30:11 xcasex: plenty ;) Aug 05 13:30:28 xcasex: symlinks and DASH/BASH are popular ones ;-) Aug 05 13:31:02 well i've worked around some, but now there's an issue with python that has me slamming my head against the wall Aug 05 13:31:25 Oh, what Ubuntu and what is the error? Aug 05 13:33:08 hrw: or anyone else, i switched to ubifs from jffs2 to speed up mounting and everything works except that I still have the same 30 second delay between the "freeing init memory" statement and "starting udev" so is it something other than the mounting that's taking so long? Aug 05 13:33:23 DJWillis: jaunty and the error is this: http://pastie.org/572487 Aug 05 13:40:53 xcasex: hmmm, what bitbake version are you using? Aug 05 13:41:08 DJWillis: latest stable from the website Aug 05 13:53:55 03Elena Grandi  07org.openembedded.dev * r7de8c2a6cd 10openembedded.git/ (conf/checksums.ini recipes/ipython/ipython_0.10.bb): ipython: new recipe for version 0.10 Aug 05 14:04:53 hi all. I've got a problem with staging-* packages. Is there a reason why they have an architecture? Would it be possible to make them noarch? Aug 05 14:15:31 hi all Aug 05 14:16:33 in my angstrom desktop, text size is sometimes is big, sometimes is normal. when I reset my board, sometimes I see big texts, sometimes normal.. how can I get the default size? Aug 05 14:25:49 sx: this is a bit offtopic, it depends on your desktop environment? what do you use? Aug 05 14:26:19 zecke, I think gpe Aug 05 14:26:24 (x11-image) Aug 05 14:26:52 hi is it off topic if Xfbdev is not starting and that I see a _ on top-left of the screen? Aug 05 14:26:52 but logs says X is started Aug 05 14:26:59 by the way should I request a branch? Aug 05 14:27:38 sx: isn't there a gpe-conf or gpe-settings app? in general GPE/GTK+ honors XSETTINGS, you could google for these :) Aug 05 14:30:24 Hi! Aug 05 14:31:10 I'm facing some problems with glibc do_package, when it comes to generate locale Aug 05 14:31:29 qemu rejects me with a nice "Kernel too old" Aug 05 14:32:12 update metadata Aug 05 14:32:18 I saw that this could be avoided by setting OLDEST_KERNEL to "None", but I'm not sure this is a good choice Aug 05 14:33:05 I'm up to date (stable/2009 branch) Aug 05 14:33:38 strange... I updated stable/2009 to cover that Aug 05 14:33:43 which DISTRO you use? Aug 05 14:33:51 Angstrom Aug 05 14:34:10 hm. my builds works Aug 05 14:35:01 I'm building for ARM by the way Aug 05 14:35:01 is there a problem with the git server ? Aug 05 14:35:19 while trying to clone I get an error Aug 05 14:35:35 .git/refs/heads/master: No such file or directory Aug 05 14:35:47 error: there are still refs under 'refs/heads/origin' Aug 05 14:35:50 morning Aug 05 14:37:04 hrw: I have to check my distro configuration, I'm using a copy on a separate tree to build a customized distro based on Angstrom, and it seems that updates were done, you were right Aug 05 14:38:38 Gnutoo: I have seen that a few times since the XServer upgrade, X starts, flashing _ at the top of the screen then it gives up after a while and goes to the console login screen. Aug 05 14:38:59 DJWillis, do you have a solution? Aug 05 14:40:49 Gnutoo: I was working on one but it 'just went away' when I did a GIT PULL a few days ago without a rebuild and I can't see ANY X server activity that would have fixed it :(. I did prove that it was not the XVideo driver (tried several, fbdev, OMAP etc. on my boards, Pandora in this case). Aug 05 14:41:46 I still have an issue where all the recent work on X seems to have shown up stability issues in gpe-login on the boards I am trying it with, not sure whats borked there. Aug 05 14:42:16 Gnutoo: this is all on oe.dev not stable2009 Aug 05 14:42:24 DJWillis, shr branch Aug 05 14:42:33 DJWillis, and G1's kernel Aug 05 14:42:36 2.6.29 Aug 05 14:43:33 i am trying to wirte a big file ( > 2 gig ) to a flash memory on an omap running oe Aug 05 14:43:44 i use open64 and lseek64 functions Aug 05 14:43:59 however those seem to fail when exceeding the 2 gig limit Aug 05 14:44:07 any idea what to consider? Aug 05 14:45:36 change filesystem? Aug 05 14:45:42 Gnutoo: I spent ages thinking it was linked to some kernel changes we did for the Pandora (DSS2 additions, if you an OMAP man) but I really never got to the bottom of it, i'll try and nail the commit that fixed it but as you can understand git bisecting OE builds is one of the most fruitless tasks known to man when there are a lot of changesets :( Aug 05 14:46:24 DJWillis, ok I'll ask the one who get Xfbdev working then Aug 05 14:46:25 boto|w: you fail to mention what the FS and NAND layer is ;-) (or is it SD?) Aug 05 14:46:33 hrw: the file system is yaffs2 Aug 05 14:46:34 DJWillis, yes SD Aug 05 14:46:37 oops Aug 05 14:46:51 it is a nand flash Aug 05 14:46:58 Gnutoo: I tried that also, flash to SD, NAND etc. tried just about everything ;-) Aug 05 14:46:59 boto|w: I do not know does yaffs support >2GB files Aug 05 14:47:11 boto|w: btw - how big flash you have there? Aug 05 14:47:15 ok Aug 05 14:47:18 boto|w: is YAFFS2 even 2GiB > OK? Aug 05 14:47:21 the flash ist 8 gig Aug 05 14:47:40 i check if yaffs2 supports > 2gig ... Aug 05 14:47:55 I know old YAFFS(2) used to bork with anything near that size. YAFFS1 did for sure. Aug 05 14:48:43 boto|w: use 2.6.29+ kernel and ubifs instead? Aug 05 14:49:39 what i even googled ist that yaffs2 should support big files, at least the recent versions of yaffs2 Aug 05 14:49:40 Gnutoo: pretty sure it started working about the end of July (the images I auto-baked on the 30th worked with X again). Aug 05 14:49:54 ok Aug 05 14:50:01 I've no pandora Aug 05 14:50:05 but I've a G1 Aug 05 14:50:17 boto|w: so are you carrying the latest versions of YAFFS? Aug 05 14:50:18 hrw: i am using kernel 2.6.28 Aug 05 14:50:42 well, the kernel is synced with oe rep Aug 05 14:50:54 i have to check which version of yaffs2 i use ... Aug 05 14:51:56 Gnutoo: I realise that, just giving some dates that 'may' help you work out if there is a fix in there ;-). A sort of 'worked for me' answer. You mentioned you had a G1, so you have that running OE off SD ok? I had issues when I looked at it a while ago, not got the G1 any more :(. Aug 05 14:52:17 XorA: you still work on zoom2? Aug 05 14:52:29 DJWillis, are you nils? Aug 05 14:52:39 he got a g1 too Aug 05 14:52:47 Nils? Nope. Aug 05 14:52:50 and has a blog inside oe's feed Aug 05 14:52:54 hrw: on and off Aug 05 14:53:20 XorA: Mike Turquette works on zoom2 for Poky Aug 05 14:53:31 Gnutoo: that's nilsf... he doesn't use irc that frequently Aug 05 14:53:39 XorA: http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=mturquette/zoom2 Aug 05 14:53:41 yes I was told that Aug 05 14:53:55 (that he didn't use irc frequentely) Aug 05 14:53:56 hrw: cool Aug 05 14:54:05 XorA: have you ever given any thoughts to how one might fit a 'first run' wizard type of thing into OE (Angstrom based distro) without to much hacking, or for that matter, ever considered alts. to gpe-login? Aug 05 14:55:12 DJWillis: sakoman has full gnome running including gdm I beleive Aug 05 14:55:26 XorA: seems overkill ;) Aug 05 14:55:46 XorA: or maybe not ;-) Aug 05 14:55:59 DJWillis: openmoko just ran X directly, E has a display manager as well Aug 05 14:56:19 XorA: Entrance is not really supported is it? Aug 05 14:56:33 no idea Aug 05 14:56:37 ask raster Aug 05 14:56:44 will do Aug 05 14:58:04 DJWillis: gnome actually works better than I expected it to Aug 05 14:58:07 XorA: I did consider just running X, not sure if it's such a good idea. hmmm. Aug 05 14:58:25 DJWillis: I prefer my pda to look kinda like my desktop Aug 05 14:58:36 sakoman: really? Have you got a metadata tree anywhere? Really tempted to give it a spin. Aug 05 14:58:37 DJWillis: but I guess the emu boyz might not Aug 05 14:59:22 DJWillis: it required more than 100 recipe changes, so I still haven't gotten all the patches ready Aug 05 14:59:41 making oe behave under ubuntu is ;_; Aug 05 14:59:43 koen is pushing them for me - so far about a dozen are done Aug 05 14:59:46 XorA: that's my problem, I want it like a PDA that also has a game focus. The single user auto login, no concept of users etc. does not sit well with me :(. Aug 05 14:59:52 and that there's no other alternative for this sl-5600 is ;_; Aug 05 15:00:14 DJWillis: if you want to try a binary on Beagle or Overo: http://www.sakoman.com Aug 05 15:00:41 sakoman: ahhh, interesting. I guess I could *munge* that onto a Pandora, i'll try tonight. Aug 05 15:00:42 I'll keep doing a handfull or two of patches each day till they are done Aug 05 15:01:30 sakoman: what made you consider doing it? Sorry, been rather 'out of it' lately with my real job so seem to have missed out on a load of really cool stuff. Aug 05 15:01:37 but it feels like some days I generate more changes than I submit :-) Aug 05 15:02:04 DJWillis: just for the fun of it (and as a "demo" for my consulting business) Aug 05 15:03:15 sakoman: I see your site seems to have got updated a lot more (in favour of the consulting business) since I last looked. Good info there, i'll let you know if I can get a mashup going as that would be rather nice. Aug 05 15:05:05 DJWillis: I ordered a Touch Book and will try to get it going there too Aug 05 15:07:02 sakoman: trying to get one of those sent over to the UK (not easy). I still wish there where a lot more Pandora's to go out to devs, it seems such a shame etc. etc. Aug 05 15:09:43 DJWillis: agreed. any idea when they will be generally available? Aug 05 15:13:21 sakoman: really not sure, I would have hoped about 6 months ago :(. Not that close to that side of things as I am still just hacking on it 'for fun'. I know that several hundred as being built at the moment but not sure if they will be cased or 'dev board' versions. Aug 05 15:18:45 xcasex: you have problems with OE under ubuntu? Aug 05 15:20:06 hrw: yep Aug 05 15:20:18 xcasex: examples? Aug 05 15:20:34 hrw: http://pastie.org/572487 Aug 05 15:21:34 xcasex: you use old bitbake - update it first Aug 05 15:21:57 xcasex: DeprecationWarnings were fixed in all BB versions which are in use Aug 05 15:22:44 hrw: i'm using bitbake-1.8.12.tar.gz Aug 05 15:22:58 xcasex: with .dev tree? Aug 05 15:23:13 no Aug 05 15:23:19 or wait. .dev tree? Aug 05 15:23:31 xcasex: org.openembedded.dev you use? Aug 05 15:23:52 yeah Aug 05 15:24:20 ok Aug 05 15:24:21 moment Aug 05 15:25:12 np Aug 05 15:25:22 xcasex: show me your $BBPATH variable Aug 05 15:25:29 xcasex: it is 99% wrong Aug 05 15:25:40 BBPATH=/home/robert/oe/build/:/home/robert/oe/openembedded/ Aug 05 15:26:43 strange Aug 05 15:26:54 yes, i know. Aug 05 15:26:59 because check_app_exists() is defined in base.bbclass so it should just work Aug 05 15:27:02 this is not my first time using oe either. Aug 05 15:27:13 xcasex: try top of bitbake 1.8 branch maybe Aug 05 15:27:17 but it is my first time it has been this much of trouble Aug 05 15:27:38 hrw: the svn branch? Aug 05 15:28:41 or git branch as bitbake is now in git Aug 05 15:29:30 ah Aug 05 15:29:34 will try :) Aug 05 15:39:23 hrw: same error :D Aug 05 15:40:57 strange Aug 05 15:42:00 hrw: very strange as i did it by the oe guide Aug 05 15:43:09 my ubuntu9.04 VM is going to out of space Aug 05 15:44:15 :/ Aug 05 15:44:28 i'm going to set it up on my work laptop as well. Aug 05 15:44:44 if that doesnt work, i'm regressing to debian lenny and see if that'll work out better for hte build box Aug 05 15:47:39 I use debian sid Aug 05 15:59:42 bb in few Aug 05 16:03:16 Hi, when I run bitbake in my OE environment I have this error "COW.py:29:DeprecationWarning: the sets module is deprecated". I add python-native to the list of ASSUME_PROVIDED packages in build/conf/site.conf, but I have alaways this error. I need help Aug 05 16:03:46 albacore: it is no error, it is a warning Aug 05 16:04:13 albacore: that is why the message "Warning", it is because you are using Python2.6 which will warn about things that will be different in python3.0 Aug 05 16:04:21 albacore: if you have an error, it has a different cause Aug 05 16:07:50 zecke: yes, I thought that warning caused error, my error is "ConfHandler.py", line 109, in include raise ParseError("Could not %(error_out)s file %(fn)s" % vars() ) Aug 05 16:07:52 bb.parse.ParseError: Could not inherit file classes/distutils.bbclass" Aug 05 16:08:52 albacore: do you have a file called distutils.bbclass somewhere? did you set BBPATH? Aug 05 16:21:09 zecke: yes I have distutils.bbclass in openembedded/classes but BBPATH is not set Aug 05 16:23:33 sorry BBPATH=export BBPATH="${TOPDIR}:${TOPDIR}/openembedded:${TOPDIR}/shr/openembedded:${TOPDIR}/bitbake" Aug 05 16:26:35 sorry, need to sleep Aug 05 16:26:36 bbl Aug 05 16:27:16 zecke: ok, my script that Aug 05 16:27:19 hrw|gone: progress. now its just sitting there, afk making dinner Aug 05 16:28:12 zecke: thank, it's resoved, my script that set BBPATH failed, I correct, it's now ok Aug 05 16:56:57 03Angus Ainslie  07fso/milestone5.5 * rc310390656 10openembedded.git/ (34 files in 4 dirs): linux-openmoko-2.6.30 : update to 2.6.30.4 Aug 05 16:57:07 03Angus Ainslie  07fso/milestone5.5 * r2c024938ce 10openembedded.git/recipes/openmoko-projects/paroli_git.bb: paroli_git : remove paroli-elementary and depend on python-elementary Aug 05 16:57:07 03Angus Ainslie  07fso/milestone5.5 * ra112a006ff 10openembedded.git/recipes/efl1/ecore.inc: ecore.inc : update for svn 41040 Aug 05 17:12:09 03Klaus Kurzmann  07shr/import * r36b1844d20 10openembedded.git/recipes/tasks/task-shr-feed.bb: Aug 05 17:12:09 task-shr-feed: add eject Aug 05 17:12:09 Signed-off-by: Klaus Kurzmann Aug 05 18:27:57 anyone ever get "Could not obtain administrative lock" when building meta-toolchain? Aug 05 19:00:35 If I have a recipe with one .so file that I want to be included in the target and the -dev package what do I have to do. Adding it to each FILES_ macro puts the .so in only one of the packages. Aug 05 19:05:04 03Angus Ainslie  07fso/milestone5.5 * rc35aada710 10openembedded.git/recipes/python/python-elementary_svn.bb: python-elementary : update for efl 41040 Aug 05 19:14:19 oneshel, yes, that means that one of the dirs it's trying to use doesn't exist, at least sometimes Aug 05 19:17:27 why do we use 0777 umask in bitbake ? Aug 05 19:17:38 all build tree files have this Aug 05 19:24:29 Tartarus: thanks for the info, actually I found the problem, I needed a newer meta-toolchain.bb that I hadn't pulled down yet :) Aug 05 19:25:04 ah, yes, that too :) Aug 05 19:31:25 if I want my Qt related stuff (qmake, etc.) packaged up as meta-toolchain does, does a recipe exist for that? Aug 05 19:31:53 Hm Aug 05 19:32:04 Maybe? Check out the meta-toolchain-*.bb files Aug 05 19:32:07 I think there's a qt one Aug 05 19:32:12 I'm reading through the other meta-toolchain-*.bb files for ideas... Aug 05 19:32:28 you're right! meta-toolchain-qte.bb Aug 05 19:33:51 * oneshel <3's OE :) Aug 05 19:37:25 How do I get rid of the message "Couldn't find shared library provider for foo" when I build libfoo.so? I've tried using PROVIDES and adding a libfoo package to my recipe Aug 05 19:38:17 Is LEAD_SONAME mentioned around there too? Aug 05 19:38:24 If so, I think just ignore it Aug 05 20:04:47 Tartarus, LEAD_SONAME does not seem to remove the warning. The main problem I have is I need the .so I am creating in the target and the -dev package and that is not happening Aug 05 20:05:25 Er, you want the same file in 2 packages? Aug 05 20:05:54 That's not allowed. But libfoo.so.1 should end up in libfoo.ipk and libfoo.so symlink should be in libfoo-dev Aug 05 20:06:55 I do not see the symlink. in the listing of dpkg-deb --contents. Does it occur in a post inst script? Aug 05 20:07:14 Well, do you have the symlink being made, in your makefiles? Aug 05 20:07:33 The packaging rules should split things up correctly Aug 05 20:07:40 But the build rules need to make it :) Aug 05 20:09:30 OK let me backup , and sorry for my ignorance. I am building a meta-toolchain for devs to use outside of OE. I have a package foo that installs some apps plus an libfoo.so. Finally, I need to populate the meta-toolchain with the .h files and the .so. Aug 05 20:09:44 Right Aug 05 20:10:11 So meta-toolchain will pull in libfoo-dev.deb, and in turn that brings in libfoo.deb Aug 05 20:10:19 Once you fill out the hook to have libfoo-dev pulled in, of course Aug 05 20:10:35 It sounds like the problem you have is that libfoo-dev.deb doesn't have the contents you want Aug 05 20:11:24 Now, if foo.bb builds and installs libfoo.so.1 and libfoo.so symlink to libfoo.so.1 into the right place ($D/usr/lib/ or so), it should get put into the right debs Aug 05 20:11:41 Right now libfoo-dev has the .h files but no reference to the .so and not depends. Does the meta-toolchain magically install foo and foo-dev? Am I just insane? Aug 05 20:11:59 You've got two problems Aug 05 20:12:01 First problem Aug 05 20:12:19 Does foo actually make the symlink, when built outside of OE? Aug 05 20:12:36 It sounds like no, and that's a problem that needs fixing, inside of foos autofun Aug 05 20:13:06 I do all install stuff within do_install. So I should make that happen now Aug 05 20:13:09 Second problem, are you sure libfoo-dev doesn't have a dep on libfoo itself? ar -x and verify the control file Aug 05 20:14:34 libfoo-dev has no depends. Aug 05 20:15:21 by default, ${PN}-dev rdepends on ${PN}, unless the recipe is overriding it... see the default values in bitbake.conf Aug 05 20:17:59 let me do some hunting here... Aug 05 20:29:15 03Khem Raj  07org.openembedded.dev * r3ea400c9bf 10openembedded.git/recipes/uclibc/uclibc_nptl.bb: uclibc_nptl: Rev up to a buildable and bootable revision. Aug 05 20:29:20 03Khem Raj  07org.openembedded.dev * r4b3eefe53f 10openembedded.git/recipes/util-linux-ng/ (2 files in 2 dirs): Aug 05 20:29:20 util-linux-ng-2.16: Replace siginterrupt with sigaction. Aug 05 20:29:20 siginterrupt is obsoleted and newer versions of uclibc Aug 05 20:29:20 has moved these functions to a different category which is Aug 05 20:29:22 not enabled by default. Best is to replace it with Aug 05 20:29:24 sigaction. Aug 05 20:29:26 Signed-off-by: Khem Raj Aug 05 20:30:04 kergoth, Tartarus, hx for the help. I was missing the symlink part Aug 05 20:30:42 eve waite Aug 05 20:31:12 he timtimred Aug 05 21:41:24 03Khem Raj  07org.openembedded.dev * rc24a415882 10openembedded.git/recipes/python/python-native_2.6.1.bb: (log message trimmed) Aug 05 21:41:24 python-native_2.6.1.bb: Unset LIBC Aug 05 21:41:24 LIBC is used internally by python configure if it gets it from Aug 05 21:41:24 enviromment it honors it. Coincidently we also use LIBC in OE Aug 05 21:41:24 to denote one of uclibc eglibc or glibc to be system C library Aug 05 21:41:26 It conflicts if somehow LIBC= is set in the envronment Aug 05 21:41:30 so we pass empty LIBC for native recipe. Aug 05 21:42:46 khem: you can also do "unexport LIBC" at toplevel to keep it from being emitted into the shell tasks Aug 05 21:44:08 kergoth: you mean in the shell where bitbake will be called Aug 05 21:44:19 no, i mean in the recipe Aug 05 21:44:25 oh Aug 05 21:44:55 yeah sure Aug 05 21:44:59 * kergoth doesn't know which is best, just pointing out that it exists Aug 05 21:45:19 I think if I pass it to configure thats all what matters Aug 05 21:47:14 LIBC[unexport] = "1" is this analogous Aug 05 21:47:56 kergoth: none of our recipes use unexport so far :) Aug 05 21:48:03 base.bbclass does Aug 05 21:48:18 it unexports a couple vars that can break certain things, like binutils. imo that should really have been done in the individual recipes Aug 05 21:48:56 I bet yes Aug 05 21:50:02 course, i think we export a lot of things we shouldnt be to begin with Aug 05 21:50:16 the only things we should be exporting by default are cc, cflags, etc, for configure & make to pick up :) Aug 05 21:50:24 yeah I think many of those are prolly legacy Aug 05 21:50:30 yep Aug 05 21:50:53 nobody wants to touch it, inc ase it breaks something they don't know about :) Aug 05 21:56:13 right Aug 05 22:13:45 * kergoth mulls over ways to integrate scratchbox style functionality into OpenEmbedded Aug 05 22:16:10 kergoth: like devshell Aug 05 22:18:18 khem: hmm, perhaps.. or a flag like fakeroot for a task that spawns it in that environment, for lazy people who don't want to properly crosscompile, or just need to get up and running and want to make it nice later Aug 05 22:19:08 I think devshell can be powerful but its under developed Aug 05 22:26:34 03Michael 'Mickey' Lauer  07org.openembedded.dev * r77d48ff9a9 10openembedded.git/ (13 files in 3 dirs): python: add 2.6.2 Aug 05 22:27:33 hmmm Aug 05 22:29:36 Has anyone looked into using OE to produce qemu/vmware/livecd to do OpenEmbedded development in? Aug 05 22:30:37 people have run it in vmware images Aug 05 22:31:21 we really need to get the image creation to handle full disk image creation rather than just fs/partition generation, at some point.. i think poky does that, has syslinux stuff Aug 05 22:33:13 hm what the hell texasi is doing here Aug 05 22:33:15 iirc, that stuff is still there Aug 05 22:33:21 I have full rights Aug 05 22:33:22 dunno when it was last used / tried Aug 05 22:33:31 and the error is I have no rights Aug 05 22:33:36 when deleting Aug 05 22:33:36 but there's stuff, iirc, from amd, for making bootable isos Aug 05 22:34:36 It might, at least for arm, be easier to say "here's your flash" and pass it a jffs2, or whatever Aug 05 22:34:53 That could work on x86 too, even, it's just not as common there Aug 05 22:36:35 yeah, x86 is just a pain since the individual pieces are rarely as useful, need to take them and make a cohesive disk image out of it Aug 05 22:37:01 Well, depends on how flexiable you want it to be Aug 05 22:37:08 It's not that hard to make a livecd Aug 05 22:37:29 Heck, take a look at the ubi stuff, making a syslinux conf file isn't any more complex than that Aug 05 22:37:41 good nite Aug 05 22:42:20 03Angus Ainslie  07fso/milestone5.5 * r0547b37262 10openembedded.git/recipes/linux/ (4 files in 2 dirs): linux-openmoko-2.6.30.1 : update openwrt patches Aug 05 22:42:22 03Angus Ainslie  07fso/milestone5.5 * r12fec90c58 10openembedded.git/ (34 files in 4 dirs): Aug 05 22:42:22 Revert "linux-openmoko-2.6.30 : update to 2.6.30.4" Aug 05 22:42:22 This reverts commit c31039065610137137ad6f5e445bc4ba298de28e. Aug 05 22:43:03 hmm Aug 05 22:49:58 * kergoth watches some of the elc2009 videos Aug 05 22:51:12 i wonder if we'd be better off developing multiple tools to meet various peoples needs, rather than trying to cram everything into the bitbake model. it seems like some people really are happier using scratchbox or buildroot for certain things.. itd be nice to be able to meet all those needs but still have something cohesive and well integrated amongst them, you know, but without trying to be all things to all people in a single tool Aug 05 22:51:16 * kergoth ponders Aug 05 22:51:52 heh Aug 05 22:51:57 It depends really, I think Aug 05 22:52:21 kergoth: yeah it depends, scratchbox is great, but doesn't have any way to package things by default Aug 05 22:52:36 kergoth: so it is cool for a developer to mess around, but not build a rootfs Aug 05 22:53:16 kergoth: and I'm pissed off by /scratchbox/login every time... Aug 05 22:53:20 hehe Aug 05 22:55:03 hmm, it seems like a lot of the complaints about OpenEmbedded end up being about the image production aspects. having to use packages, the way we pull things in, not having a good way to jump in and kill the sysvinit stuff in favor of custom scripts, etc. i wonder if we should make a separate tool for that instead of wedging it in as recipes Aug 05 22:55:50 kergoth, some people won't believe you that having a package manager on the target isn't desired :) Aug 05 22:56:07 well, a lot of people don't seem to know about ONLINE_PACKAGE_MANAGEMENT Aug 05 22:56:22 hmm Aug 05 22:56:40 That too, but it's not exactly correct Aug 05 22:56:48 yeah, the postinst stuff is a *mess* Aug 05 22:56:50 kergoth: I a lot of people here don't know about strace, gdb, oprofile... and reset their device because the serial has no job contral Aug 05 22:56:53 Yeap :| Aug 05 22:56:55 if you dont get opkg in the image, half your shit doesn't work Aug 05 22:57:02 Can't use that hook unless you know there's no postinit stuff Aug 05 23:01:09 would be nice to abstract out aspects of the startup script mess at some point. put the command to start a service in this variable, stop it in that variable, etc, and make it generate lines for a busybox inittab, or emit an upstart script or sysvinit startup script or whatever Aug 05 23:02:14 kergoth: init scripts, yes, certainly Aug 05 23:53:32 03Andrea Adami  07org.openembedded.dev * r6211c54804 10openembedded.git/ (13 files in 4 dirs): klibc, klibc-utils-static: bump to Testing/1.5.15 **** ENDING LOGGING AT Thu Aug 06 02:59:57 2009