**** BEGIN LOGGING AT Mon Aug 19 02:59:59 2013 Aug 19 07:26:01 I get "cronie_common.h: No such file or directory" since the upgrade to 1.4.11 Aug 19 08:00:03 morning all Aug 19 08:00:57 hi bluelightning Aug 19 08:01:03 morning bluelightning and #oe Aug 19 08:01:04 hi woglinde Aug 19 08:01:08 hi silviof Aug 19 08:07:20 can anyone else build the upgraded cronie? I get a missing include file (path) Aug 19 08:09:48 I do not need cronie Aug 19 08:13:36 Zagor: Yes i got the same failure "cronie_common.h: No such file or directory" Aug 19 08:25:12 Zagor: also confirmed - I've sent an email about it Aug 19 08:25:20 Zagor: thanks for the heads up Aug 19 08:27:14 ah, saul already submitted a patch even Aug 19 08:31:27 ah yes, indeed Aug 19 09:07:14 hi stefan Aug 19 10:16:48 morning all Aug 19 10:17:09 morning pb_ Aug 19 10:17:16 hi bluelightning Aug 19 10:24:03 hi pb_ Aug 19 10:25:35 ant_work: hi, have you checked *-klibc failures? Aug 19 10:26:00 hi ant_work Aug 19 10:26:49 JaMa: from sstate youmean, maybe yes, somehow... nailed down th INHIBIT_DEFAULT_DEPS in linux-libc-headers recipe Aug 19 10:28:42 JaMa: pb_: do you think in OE world linux-libc-headers recipe doesn't depend on any tc part ? Aug 19 10:29:44 I can't think why it would need to. It doesn't compile anything. Aug 19 10:30:25 hi pb Aug 19 10:30:33 ok, so who is the tipical user of it? the dev having a full tc on host Aug 19 10:30:37 hi woglinde Aug 19 10:30:49 ant_work: typical user of what, linux-libc-headers? Aug 19 10:30:53 yep Aug 19 10:31:10 well, eglibc depends on it. Aug 19 10:31:25 I guess maybe I don't understand the question. Aug 19 10:31:26 being a fact that some don't like kernel's make headers_install we still have it around Aug 19 10:31:53 oh, I see, you mean why do we have a separate recipe for linux-libc-headers rather than just using the kernel itself? Aug 19 10:32:00 pb_: it happens that klibc (which depends on linux-libc-headers) does fail to rebuild from sstate Aug 19 10:32:24 some basic headers are still missing (limits.h iirc) Aug 19 10:33:02 ant_work how many? Aug 19 10:33:17 isn't a kernel header, it's provided by either libc or gcc. I forget which exactly, but I think it's eglibc. Aug 19 10:33:18 on first run just a couple iirc Aug 19 10:33:46 pb_: I'm incurring in a race or in a missing dependency Aug 19 10:34:06 well, not I, JaMa's autobuilder really ;) Aug 19 10:34:26 can you show me the actual error? Aug 19 10:37:44 I'm sorry I see only huge logs .. Aug 19 10:39:14 it's not related to sstate, you'll get the same when you tried to build any -klibc recipe with empty tmp-eglibc Aug 19 10:39:26 s/ied/y/ Aug 19 10:39:37 I would copy over the missing headers Aug 19 10:40:30 pb_: e.g. this one http://logs.nslu2-linux.org/buildlogs/oe/oe-shr-core-branches/log.world.20130816_112209.log/min_failed/kexecboot-klibc Aug 19 10:41:50 JaMa: once klibc was depending on kernel source tree, now can be built against linux-libc headers and as bonus isn't anymore machine-specific Aug 19 10:41:54 http://logs.nslu2-linux.org/buildlogs/oe/oe-shr-core-branches/log.world.20130816_112209.log/min_failed/kexec-tools-klibc Aug 19 10:41:57 http://logs.nslu2-linux.org/buildlogs/oe/oe-shr-core-branches/log.world.20130816_112209.log/min_failed/ubi-utils-klibc Aug 19 10:42:28 JaMa: the point is now, linux-libc-headers alone isn't enough because it doesn't inherit anymore the rest of the dev stuff Aug 19 10:42:59 ah, right. so it's not klibc itself that fails to build, it's the packages that are trying to link against klibc. Aug 19 10:43:10 this apparently after Bruce # ccf3bf70a1a83ac55e47acfa4759838f7f4834ae Aug 19 10:43:11 and it's not actually you're missing, it's Aug 19 10:43:53 pb_ as well fatal error: asm/unistd.h Aug 19 10:44:11 yeah, it looks like you're basically missing klibc's copies of all the kernel headers. Aug 19 10:45:16 probably because when klibc does its copy he heders are not yet staged Aug 19 10:45:45 yousee, build from scratch does succeed Aug 19 10:45:46 yeah, possibly. I don't know enough about how klibc builds itself to say anything very intelligent about that. Aug 19 10:46:12 okay. so, do you know which package you have to remove from sstate to make it work again? Aug 19 10:46:13 or even s/not staged/missing Aug 19 10:46:16 klibc, klcc, ...? Aug 19 10:47:10 http://cgit.openembedded.org/meta-openembedded/tree/meta-initramfs/recipes-devtools/klibc/klibc-2.0.2/klibc-linux-libc-dev.patch Aug 19 10:47:46 eh, it makes symlinks? Aug 19 10:47:48 I think things go wrong here Aug 19 10:47:55 what's $(KLIBCKERNELSRC) in this context? Aug 19 10:48:38 KLIBCKERNELSRC=${STAGING_DIR_TARGET}${exec_prefix}' \ Aug 19 10:49:50 hm, right. I don't quite know how sstate will react to symlinks with absolute paths in them. Aug 19 10:50:10 do those symlinks appear when the package is unpackaged from sstate? Aug 19 10:50:18 and, if so, do they point to the right place? Aug 19 10:55:25 pb_: I'll do deep tests later at home. many thx Aug 19 11:34:29 if a software uses the even-odd convention (even version numbers: stable releases, odd numbers: unstable), and I want to give the user the option to use an unstable release, what is the recommended way to do that in OE? Aug 19 11:35:01 I thought of adding recipes for the unstable version and putting DEFAULT_PREFERENCE = "-1" at the beginning Aug 19 11:35:08 that would be fine Aug 19 11:35:15 release managing is no part of oe Aug 19 11:36:46 dv_ btw. user in compile all myself or user in using distribution Aug 19 11:44:57 um, not sure where the difference is Aug 19 11:45:07 whoever modifies the local.conf ? Aug 19 11:46:08 dv_ the second you could control via opkg Aug 19 11:46:19 ah, then the first one Aug 19 11:46:29 and opkg conf Aug 19 11:46:52 first with PREFFRED_VERSION Aug 19 11:50:52 okay. suits me well. "regular" users get the stable version, bleeding edge and experimental people can pick the newest Aug 19 13:49:49 How do i add the depmod binary to my core-image-minimal? I thought it would've been fairly fundamental Aug 19 13:50:42 why? Aug 19 13:50:49 busybox has insmod Aug 19 13:51:06 but no depmod for modprobey goodness? Aug 19 13:51:42 why? you do not need depmod this often in embedded stuff Aug 19 13:51:55 it boots than its run Aug 19 13:51:57 for development I do Aug 19 13:52:26 than use depmod recipe Aug 19 13:52:37 instead of busybox module handling Aug 19 13:52:47 bitbake depmod? Didn't yield any joy Aug 19 13:54:15 how about git grep depmod in oe-core Aug 19 13:54:26 than you find the right recipe Aug 19 13:55:33 meta/recipes-kernel/kmod/kmod? Aug 19 13:55:43 yes Aug 19 13:55:57 woglinde, I tried a bitbake kmod and it didn't do any work. Nothing needed to rerun... still no depmod in the rootfs Aug 19 13:56:45 exosyst what does opkg files kmod tell you on target? Aug 19 13:58:05 XorA_, are there any "major" airports that are convenient to Edinburgh via train, besides Glasgow? Aug 19 13:58:40 crfoton why not fly to berlin and than with easyjet to Edinburgh? Aug 19 13:58:54 rofl Aug 19 13:59:06 package kmod not installed Aug 19 13:59:08 Crofton: no Aug 19 13:59:10 to get to Berlin I'd have to change in Frankfurt anyway Aug 19 13:59:28 Crofton: newark->Edin is pretty much only route :-( Aug 19 13:59:33 exosyst opkg install kmod Aug 19 14:00:05 frakking newark Aug 19 14:00:14 Glasgow is not convenient to anywhere :-) Aug 19 14:00:35 and nothing via an "easy" train ride? Aug 19 14:00:36 * mwester thinks newark is still preferable to Chicago O'Hare Aug 19 14:00:47 heh Aug 19 14:00:59 Crofton|work: Manchest Aug 19 14:01:00 I can easily avoid O'Hare :) Aug 19 14:01:03 Crofton|work: Manchester Aug 19 14:01:07 yeah Aug 19 14:01:10 that may be the route Aug 19 14:01:14 how long is the train? Aug 19 14:01:35 Crofton|work: I have no idea how you get Manchester airport to mainline train, but mainline train is 3 hours I think Aug 19 14:01:54 Crofton|work: if it works from abroad, www.nationalrail.co.uk Aug 19 14:05:13 Crofton|work: why obsession with train? Aug 19 14:09:07 trying to find a sane routing Aug 19 14:09:19 I think Machester is easier for me to get to Aug 19 14:10:12 if you live here pretty much resigned to catching short flight to Amsterdam or Heathrow Aug 19 14:10:47 yeah Aug 19 14:10:52 I already have short flight here Aug 19 14:11:09 and I am thinking short flight will be about same amount of time as train Aug 19 14:12:24 go to Dublin on your USA passport then to UK on your British :-D Aug 19 14:16:45 hi guys! Aug 19 14:17:23 is there an automated way to remove old package versions from the work directory? I have up to 15 versions of the same package lying around and wasting space... Aug 19 14:18:27 unfortunately, most of them differ only in git revision and not in version so "a -c cleanall package_ver.sion" does not work... Aug 19 14:21:28 i just wipe tmpdir regularly, personally, and keep sstate_dir, cache_dir outside of it. doesn't take terribly long to restore everything from sstate Aug 19 14:21:40 but i think there's a contrib script to clean up workdir, maybe. check oe-core/poky Aug 19 14:24:58 hsychla there is a script Aug 19 14:25:40 hsychla hm do you mean ipk or sstate stuff? Aug 19 14:26:30 for sstate it is in oe-core Aug 19 14:26:32 scripts/sstate-cache-management.sh Aug 19 14:29:21 I mean the tmp/work/*/ stuff that stays after building Aug 19 14:29:55 i want to keep the ipks, images and latest temporary data (in case I want to rebuild) but remove old packages Aug 19 14:30:05 or rather tmp stuff of old packages Aug 19 14:31:31 FYI, the ipks are all also stored in sstate, if you're not using oe-classic. so getting those back is a matter of tarfile extraction, as with restoring any other sstate archive Aug 19 14:31:44 and rebuilding an image is just a matter of installing those ipks, so not much time is lost at all Aug 19 14:31:48 * kergoth shrugs Aug 19 14:32:37 just found scripts/cleanup-workdir. it looks like it checks versions and deletes the old ones (my python is bad/non-existent) will try it Aug 19 14:34:28 ah Aug 19 14:34:29 okay Aug 19 14:35:01 did not notice it before Aug 19 14:35:06 thanks hsychla Aug 19 14:35:24 cool, thought i remembered seeing something in there for it Aug 19 14:36:45 hm but I wonder about this builddir = run_command('echo $BUILDDIR').strip() Aug 19 14:36:50 why not using getenv Aug 19 14:37:05 the script is running now, i will know shortly if it works Aug 19 14:38:11 hyschla you may want to use both scripts Aug 19 14:38:12 woglinde: hah, yikes Aug 19 14:38:20 someone new to python, i guess :) Aug 19 14:40:05 woglinde, will do, thx Aug 19 14:45:20 scripts/sstate-cache-management.sh did not change my free space but scripts/cleanup-workdir gave me 21G. nice! Aug 19 14:46:09 hsychla scripts/sstate-cache-management.sh -d Aug 19 14:46:35 for finding duplicates Aug 19 14:46:48 woglinde, yes, did that. it removed 900 something duplicates Aug 19 14:47:25 ah okay Aug 19 19:08:30 hi Aug 19 20:45:05 ericben: ping :) Aug 19 20:46:30 JaMa: sorry to correct you, build from scratch are ok Aug 19 20:46:31 http://paste.debian.net/27195/ Aug 19 20:46:51 your errors show during build from sstate (setcscene tasks) Aug 19 20:49:00 * kergoth thinks about making hte variable typing bits a python package which uses pkgutil to be a namespace package, so other layers could extend the available types Aug 19 20:56:13 ant_home: 145 didn't need to be rerun <- doesn't look like from scratch Aug 19 21:10:38 JaMa, http://paste.debian.net/27202/ Aug 19 21:11:10 oops too long Aug 19 21:11:40 I'm just having hard time reproducing the issue ;) Aug 19 21:14:38 JaMa: this is qemuarm http://sprunge.us/bCcA Aug 19 21:25:46 JaMa: it was from scratch because I built for 3 qemu so -native were shared Aug 19 21:26:14 JaMa: to be picky I've started a build from ground just for logging purposes ;) Aug 19 21:26:24 will take some time Aug 19 21:30:34 if I recall correctly, the mesa recipe was fixed in master, to be able to turn on/off its GLES/EGL features so it doesnt block with GPU libraries? Aug 19 22:03:02 JaMa: full qemuarm log http://sprunge.us/cLgT Aug 19 22:07:02 ~stab cmake Aug 19 22:07:03 * infobot runs at cmake with an origami Swiss Army knife, and inflicts a nasty paper cut. Aug 19 22:07:04 bluelightning: Error: "stab" is not a valid command. Aug 19 22:07:24 Crofton-bot: you don't even support stabbing? for shame... Aug 19 22:07:25 bluelightning: Error: "you" is not a valid command. Aug 19 22:07:38 bots these days... Aug 19 22:38:53 JaMa:hm.. as pb_ said the symlinks are not part of the libklibc-dev ipk Aug 19 22:39:57 er.. I must be blind Aug 19 22:39:59 they are **** ENDING LOGGING AT Tue Aug 20 02:59:58 2013