**** BEGIN LOGGING AT Mon Mar 29 02:59:57 2010 Mar 29 05:00:19 * grg watches a tumbleweed roll past Mar 29 06:37:36 03Denys Dmytriyenko  07org.openembedded.dev * rf1ca258e45 10openembedded.git/recipes/util-linux-ng/util-linux-ng_2.17.bb: Mar 29 06:37:36 util-linux-ng: disable building fallocate and chgrp-ing wall for -native Mar 29 06:37:36 Fixes following build/install errors on Ubuntu 8.04: Mar 29 06:37:36 1. Cannot find linux/falloc.h header Mar 29 06:37:36 2. Cannot chgrp tty wall on install Mar 29 06:37:37 Signed-off-by: Denys Dmytriyenko Mar 29 06:37:42 03Denys Dmytriyenko  07org.openembedded.dev * r8f605c607e 10openembedded.git/recipes/alsa/alsa-utils_1.0.20.bb: Mar 29 06:37:42 alsa-utils: add gettext-native dependency Mar 29 06:37:42 Fixes following error on Ubuntu 8.04: Mar 29 06:37:42 mv: cannot stat `t-ja.gmo': No such file or directory Mar 29 06:37:43 Signed-off-by: Denys Dmytriyenko Mar 29 08:16:20 good morning Mar 29 08:16:32 good moaning Mar 29 08:22:23 gm Mar 29 08:22:58 gm MWelchUK_work Mar 29 08:36:38 03Martin Jansa  07org.openembedded.dev * r23bbeea07b 10openembedded.git/conf/distro/include/sane-srcrevs-fso.inc: Mar 29 08:36:38 fso: bump SRCREVs to fix fso-specs build issue Mar 29 08:36:38 Signed-off-by: Martin Jansa Mar 29 08:46:27 how well-supported is uclibc in oe? i'm currently using buildroot for my custom linux/uclibc system, but am thinking of trying oe. Mar 29 08:49:18 Ulfalizer: it is always a bit behind glibc/eglibc, but it is still a first citizen. Mar 29 08:50:00 iirc just lacks nptl atm Mar 29 08:51:17 I was more thinking in terms of build breakage in terms of building stuff like gtk and higher upts. Mar 29 08:52:37 zecke: behind in the sense that it is not as well-tested? Mar 29 08:53:37 i couldn't get nptl to work with buildroot either Mar 29 08:54:38 Ulfalizer: yes, not well tested when stuff like autoconf gets upgraded. or a new version of a recipe is added. Mar 29 08:54:51 okay Mar 29 09:02:24 Ulfalizer: I also tried with atngw100 without success Mar 29 09:02:56 mckoan: buildroot + nptl? Mar 29 09:03:09 Ulfalizer: but would be great if OE would replace buildroot for linux/uclibc systems Mar 29 09:03:39 Ulfalizer: I mean atngw100 with OE Mar 29 09:05:28 what advantages does it have over buildroot except explicit support for different boards (haven't seen that in buildroot)? Mar 29 09:05:55 in what way did it fail? Mar 29 09:12:38 good morning Mar 29 09:14:18 ah..now that I remember, khem is working on uclibc-nptl. Possibly is working now Mar 29 09:17:06 is the kernel configuration used when building angstrom for a particular MACHINE the same that gets used when building just the kernel? the vanilla kernel-omap seems to have crappy defaults with lots of things off that needs to be on in order to boot from NAND, so i thought i'd use the angstrom config as a base. Mar 29 09:17:40 among other things Mar 29 09:17:43 *need Mar 29 09:17:55 ant_work: ok :) Mar 29 09:19:27 GNUtoo: about glib and zlib dependency, I could find this http://osdir.com/ml/svn-commits-list/2009-11/msg05629.html Mar 29 09:19:31 heh, should have said that the target is a beagleboard as well Mar 29 09:19:54 GNUtoo: I'm absolutely *not* a GTK/glib expert Mar 29 09:20:33 but AC_CHECK_LIB(z, inflate, [AC_CHECK_HEADER(zlib.h, found_zlib=yes)]) is clear Mar 29 09:20:36 ant_work, and,it's just a check added Mar 29 09:20:58 ah ok Mar 29 09:21:02 you had the header? Mar 29 09:21:32 some other packaged staged that header Mar 29 09:22:27 ah? Mar 29 09:22:41 which one? Mar 29 09:22:54 and why? Mar 29 09:23:14 because they bundle the lib in their source? Mar 29 09:23:17 which is bad Mar 29 09:23:44 oh..can't say... many are depending from zlib... Mar 29 09:27:22 so was the fix ok? Mar 29 09:28:55 if the dependency was added in 2.23.6, yes Mar 29 09:30:29 previous version did not list zlib as dependency, and as I said Gentoo glib-2.22.5 doesn't list it too Mar 29 09:32:13 ok Mar 29 09:33:03 btw there is a pb_ patch floating in patchwork which is supposed to catch these cases Mar 29 10:30:30 do i need to specify a DISTRO, or is TARGET_OS sufficient if i just want to build a kernel and a few packages, and don't care for a prebuilt rootfs? Mar 29 10:30:45 it's for a custom, non-distro-based system Mar 29 10:32:11 yes just use the mini dsitro or angstrom in this case Mar 29 10:32:17 Ulfalizer: well, for fast success stay with a distro. Mar 29 10:33:20 ok Mar 29 10:33:24 what's the mini distro? Mar 29 10:34:12 Ulfalizer: a distro is a set of policies, E.g. which package manager to use, which configs for like busybox to use. Mar 29 10:35:05 so the mini distro is just a reasonable "minimal" configuration for those things? Mar 29 10:35:13 Ulfalizer: the best supported one is angstrom, then there are two other ones with various support and success. mini is applying minimal set of policies, micro is getting rid of the /usr dear completely.. Mar 29 10:35:29 ok :) Mar 29 10:36:03 mini worked for me much better than angstrom building an image for the Sim One Mar 29 10:36:33 all i really need at the moment is a toolchain, kernel, busybox and directfb, preferably with a custom configuration for the latter two. Mar 29 10:36:46 for beagleboard Mar 29 10:37:40 i'll try mini and see how it goes.. Mar 29 10:39:34 what language are the .conf files in by the way? Mar 29 10:39:52 custom? Mar 29 10:40:04 Ulfalizer: I don't think it has a name, yes it is custom. Mar 29 10:59:57 Is tinderbox down? I don't seem to be able to access it. Mar 29 11:01:12 Oh right, same with cgit btw. :-} Mar 29 11:02:39 florian, Not just me then :-) Mar 29 11:05:32 ~/oe/openembedded> git pull -v fatal: read error (Connection reset by peer) Mar 29 11:05:44 anyone else having problem with git pull? Mar 29 11:07:33 That'a all the same machine... Mar 29 11:14:09 Bugger - m4 and libpcap aren't building for me. Was going to look at tinderbox to see if I was the only one having the problem. Failing that, when the build has finished all it can, I was going to do a git pull to see if it had been fixed :-s Mar 29 11:17:24 florian, Is the mailing list also on that machine? Mar 29 11:17:43 MWelchUK_work: no Mar 29 11:19:06 florian, Does some one with the ability to look at it know the machine is not responding, or shall I send an email? Mar 29 11:20:11 MWelchUK_work: I don't think its necessary... I think the admins of this macheine read this channel too. Mar 29 11:23:32 * MWelchUK_work finds the "Infrastructure" wiki page and agrees. Mar 29 11:26:58 hi,is git down? Mar 29 11:28:33 seem down for me Mar 29 11:28:50 MWelchUK_work, osul has been maving a bunch of stuff around Mar 29 11:29:23 Crofton, ah, ok. Mar 29 11:29:28 MWelchUK_work, I've a fix for libpcap Mar 29 11:29:45 GNUtoo, Yes git is down Mar 29 11:29:52 osuosl Mar 29 11:29:59 MWelchUK_work, ok thanks,do you want the fix via pastebin? Mar 29 11:30:12 don't see anything on their webpage, but remember seeing something somewhere Mar 29 11:30:13 GNUtoo, that would be great, cheers :-) Mar 29 11:30:26 Crofton, cheers for the info Mar 29 11:30:55 http://pastebin.com/iRjCwknp Mar 29 11:31:11 git am it Mar 29 11:31:29 and if it doesn't work I can host a file in an apache server Mar 29 11:31:32 GNUtoo, I'd kinda expected that looking at the failure :-) Mar 29 11:31:33 which would be better Mar 29 11:31:38 ok Mar 29 11:31:58 * MWelchUK_work is going to lunch, will try it when he gets back. Mar 29 11:32:40 MWelchUK_work, it was because of the commit where it was converted to new staging Mar 29 11:32:46 I'll eat too Mar 29 11:39:34 i am running into a issue where i cant mount nfs stuff in a oe image Mar 29 11:40:00 but i do have nfs-utils, portmap and some codepages in kernel .. what do i miss ? Mar 29 11:51:36 rob_w, have you tried running it? Mar 29 11:51:51 what u mean Mar 29 11:52:00 Do you have the output from a serial console you can put in a pastebin? Mar 29 11:52:20 ount: wrong fs type, bad option, bad superblock on 10.0.0.24:/home/bob, Mar 29 11:52:33 and so on .. that hatefull generic nfs error message Mar 29 11:52:51 rob_w, At a guess the kernel support isn't compiled into the kernel. Mar 29 11:53:53 or the boot args are wrong. Mar 29 11:55:04 all nfs support is compiled in .. and i am not booting onto nfs .. just want to mount something Mar 29 11:58:02 rob_w, Ah, sorry got the wrong end of the stick Mar 29 11:59:07 I'm afraid I don't know then. Mar 29 11:59:39 np Mar 29 12:05:44 MWelchUK_work, about NSF I've a big issue: I can mount directories but I can't boot : error -13 permission denied Mar 29 12:05:53 machine : bug 1.x Mar 29 12:05:54 nfs :) Mar 29 12:06:11 distro: angstrom Mar 29 12:06:42 maybe that's a bug in the 2.6.27 kernel Mar 29 12:07:58 long time ago I had some success with a wrt54gs, I think it was 2.6.3x Mar 29 12:08:35 GNUtoo, are you sure you're trying to mount the same directory each time? Mar 29 12:08:39 yes Mar 29 12:08:57 it was some time ago tough, I tried multiple times Mar 29 12:08:59 That looks like it's failing to mount. Mar 29 12:09:04 it is Mar 29 12:09:15 the reason is error -13: permission denied Mar 29 12:09:20 s/is/was Mar 29 12:09:24 morning Mar 29 12:09:34 It worked the last time I tried it, wasn't that long ago. Mar 29 12:09:41 hrw, morning. Mar 29 12:11:17 ok Mar 29 12:12:42 MWelchUK_work, if you're interested here are the details: Mar 29 12:12:47 GNUtoo, although that is on a PPC board. Mar 29 12:12:47 http://lists.buglabs.net/pipermail/bug-dev/2010-March/000324.html Mar 29 12:12:50 ok Mar 29 12:12:58 when it worked for me it was mips Mar 29 12:13:03 here it's arm Mar 29 12:15:16 Are you sure that IP address has the correct permissions for that directory? Mar 29 12:16:25 it worked when mounting it normally Mar 29 12:16:31 from NFS userspace Mar 29 12:16:53 How do you assign the IP address to the machine when mounting in userspace? Mar 29 12:17:47 ifconfig Mar 29 12:18:02 and you give it the same IP? Mar 29 12:18:15 yes Mar 29 12:18:29 192.168.20.2 Mar 29 12:18:44 in the first case(else it wouldn't have mounted it) Mar 29 12:19:13 *192.168.20.20 Mar 29 12:20:20 showmount -e gave that: /home/gnutoo/embedded/bug-device/nfs-angstrom-02_01_2009 192.168.20.20 Mar 29 12:20:39 and in the second case: ip=192.168.20.20:192.168.20.2:192.168.20.2:255.255.255.0:bug:eth0:off Mar 29 12:21:35 The only 2 things I can think of are removing the gateway entry and possibly the "eth0" entry. Mar 29 12:21:45 ok Mar 29 12:22:03 I'll try Mar 29 12:22:07 hi thebohemian Mar 29 12:22:09 I tend to use ip=dhcp Mar 29 12:22:45 GNUtoo: hi Mar 29 12:23:11 thebohemian, if I've java fixes(for now only hacks(not yet commitable)) for x86,for jalimo,where should I send them? Mar 29 12:24:39 GNUtoo: nice. if it for jalimo send them to its dev list - i cant promise to have a look at them quickly (due to heavy schedule) but so they are not forgotten Mar 29 12:25:16 thebohemian, ok I'll try to find time to make them something presentable(not only hacks) Mar 29 12:25:33 by dev list you mean the dev list of jalimo? Mar 29 12:25:42 GNUtoo, the libpcap patch seems to do the trick Mar 29 12:25:52 MWelchUK_work, ok Mar 29 12:26:03 ah its Mar 29 12:26:07 ok Mar 29 12:26:14 read too fast as usal Mar 29 12:26:18 thebohemian, thanks a lot Mar 29 12:36:03 GNUtoo: yes, I mean jalimo's devlist Mar 29 12:37:02 thebohemian, ok I read too fast sorry (you said its) Mar 29 12:37:16 GNUtoo: np :) Mar 29 12:40:04 is the kernel config used independent of the DISTRO? Mar 29 12:41:30 usually Mar 29 12:41:44 ok Mar 29 12:45:56 Good day. It appears the git repository is down at git://git.openembedded.org. Anyone confirm that? Mar 29 12:47:02 really? lemme check Mar 29 12:47:03 k1mgy: confirmed Mar 29 12:47:14 hmmm Mar 29 12:47:38 Thanks. Will try later. Mar 29 12:48:09 python-sip is unbuildable the source is gone :( Mar 29 12:48:14 mirrored anywhere? Mar 29 12:48:22 k1mgy, yup, Crofton thinks that osuosl are moving servers arround Mar 29 12:51:22 k1mgy, JaMa, please test git again? Mar 29 12:51:56 OK. It's back and working. Mar 29 12:52:08 Thanks a million. Mar 29 12:52:12 kernel "bug" Mar 29 12:52:26 ka6sox, thanks Mar 29 12:52:54 I'll try to analzye why it did that...hopefully it doesn't come back. Mar 29 12:53:57 "hopefully it doesn't come back"... what I said after this morning's breakfast. Mar 29 13:18:45 03Martin Jansa  07org.openembedded.dev * rcfcd292c81 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Mar 29 13:18:45 EFL: bump SRCREV, mostly bug fixes Mar 29 13:18:45 Signed-off-by: Martin Jansa Mar 29 13:34:11 eFfeM: you was working on mythtv - what hardware you run it on? Mar 29 13:37:23 * mwester is exploring the idea of running mythtv on the WDTV Live hardware -- would be perfect as a MythTV client. Mar 29 13:39:04 why does the *.so without .0 at the end go to -dev package? Mar 29 13:39:43 * chouimat is hunting a segfault on a not so fun platform to work with :( Mar 29 13:39:59 * MWelchUK_work runs mythtv on a via C7, but that's with Mythbuntu Mar 29 13:40:17 and it's a little underpowered as a client and server Mar 29 13:46:41 hi, all! Mar 29 13:47:02 is it possible to build profiling libc with OE? Mar 29 13:48:08 I need to have info about executing libc functions in context of application, using gprof. Mar 29 13:50:52 03Thomas Zimmermann  07org.openembedded.dev * r95d884b788 10openembedded.git/recipes/alsa/alsa-oss_1.0.17.bb: Mar 29 13:50:52 alsa-oss: fix aoss and convert to new staging Mar 29 13:50:52 * as libaoss.so is in -dev it has to be renamed in aoss Mar 29 13:50:52 Signed-off-by: Thomas Zimmermann Mar 29 13:57:12 hrw, i did most of the work for beagleboard, and I mostly used it as a backend for my hauppauge wintv pvr usb2 card which has a hw mpeg2 encoder (analog usb card) Mar 29 13:57:45 but I had that hauppauge card also working on nlsu2 several years back Mar 29 13:58:13 ok Mar 29 13:58:29 still looking for an inexpensive usb dvb-s card, guess recording of transport streams would not be too expensive cpu-wise Mar 29 13:58:38 eFfeM: I am planning to use beagleboard to play videos from samba share over wifi Mar 29 13:58:55 (dvb-t coverage in my area is not good enough to try that with a simple indoor antenna) Mar 29 13:59:25 hrw, i mostly used the backend part Mar 29 14:00:24 hrw, guess bandwidth might be the problem, not sure if there is a good 11n setup, otherwise you'll need a good 11g card and not too much distance between AP and beagle Mar 29 14:01:13 hrw, also instead of samba you could also consider nfs or iscsi. I've been using tgt with success (and the recipe is in oe and up-to-date) Mar 29 14:03:17 is git still up? Mar 29 14:03:34 eFfeM: nfs is option of course Mar 29 14:03:43 Ka6sox-mobile: yes (at least web if is) Mar 29 14:03:57 eFfeM: my 11g setup is fine here to play divx on my wife laptop Mar 29 14:04:08 hrw never really bothered to measure what gives the best througput Mar 29 14:04:10 Thanks Mar 29 14:05:07 hrw, did some wifi assessment for a previous employer, wifi throughput greatly depends on the chipset used Mar 29 14:05:34 el cheapo chipsets often give el crappo results Mar 29 14:06:08 I know and this makes me scary Mar 29 14:06:17 laptop has atheros, for BB I have ralink Mar 29 14:06:47 how does atheros stack up? Mar 29 14:07:22 enough for divx movies (8MB cache in mplayer), too slow for 720p Mar 29 14:07:29 did not tested with iperf Mar 29 14:07:29 don't recall exact details, and this was 4 years ago iirc, but at that point atheros was one of the best (if not the best) Mar 29 14:07:53 and i think ralink was so-so Mar 29 14:08:06 03Koen Kooi  07org.openembedded.dev * r324f773851 10openembedded.git/recipes/openssl/openssl-native_0.9.8m.bb: openssl-native 0.9.8m: fix staging Mar 29 14:08:08 but your mileage may vary and things might have improved over time Mar 29 14:08:16 now I am waiting for 11n router + 11n usb wifi (ath9k_htc driver iirc) to test them vefore it will go outside Mar 29 14:09:28 i heard rumours that the latest marvell 11n offering is quite nice but not sure if there are already products where it is in Mar 29 14:11:09 hrw, there is also a very nice belkin gigabit powerline box Mar 29 14:11:24 wired ofc but guess you have power at both sides anyway Mar 29 14:13:47 | oe_libinstall: cd /media/e79a5359-8155-4799-b39c-3c9394e679f3/stuff/tmp/work/armv7a-angstrom-linux-gnueabi/libpcap-1.0.0-r6/libpcap-1.0.0 Mar 29 14:13:50 | install: cannot create regular file `/media/e79a5359-8155-4799-b39c-3c9394e679f3/stuff/tmp/work/armv7a-angstrom-linux-gnueabi/libpcap-1.0.0-r6/image/usr/share/aclocal/libpcap.m4': No such file or directory Mar 29 14:13:54 wtf Mar 29 14:16:05 MWelchUK_work, still the same error,and I bet I need eth0 Mar 29 14:16:18 wroberts2, I've a fix Mar 29 14:16:47 eFfeM: no wired ethernet in salon - would have to put ~20m of cat5e and replace strips in one room (few hours of work) Mar 29 14:16:52 wroberts2, http://pastebin.com/iRjCwknp Mar 29 14:16:54 eFfeM: so prefer wifi Mar 29 14:17:17 git still down I bet Mar 29 14:17:30 ah no Mar 29 14:17:32 I'll push Mar 29 14:17:32 GNUtoo: git is up Mar 29 14:17:36 ok thanks Mar 29 14:17:36 This one: http://pastebin.com/MbHTWa7L Mar 29 14:17:47 Should be UP Mar 29 14:18:09 MWelchUK_work, no this one: [42949387.050000] Root-NFS: Server returned error -13 while mounting /home/gnutoo/embedded/bug-device/nfs-angstrom-02_01_2009 Mar 29 14:18:11 hrw you misunderstood me, the belkin thing is powerline, so ethernet over 230 (or whatever you have) Mar 29 14:18:30 GNUtoo, Ah, yes sorry - need a coffee Mar 29 14:18:50 you just have a power brick that you put in an outlet and that has an rj45 jack out of it, ofc you need one at each side Mar 29 14:19:30 hrw: this one:http://www.engadget.com/2009/06/22/belkins-gigabit-powerline-adapter-ups-the-ante-for-electrical-o/ Mar 29 14:19:41 collegue has one and is happy with it Mar 29 14:19:42 eFfeM: ah. Mar 29 14:22:10 gnutoo: got it Mar 29 14:22:58 wroberts2, I'll push very soon the fix Mar 29 14:23:00 * mwester notes that the powerline adaptors don't work so well when there's a sodium-vapor yard light on the powerline between the two end-points. :) Mar 29 14:23:55 i also noticed that openssh-native wanted to install man pages onto my system. had to make those directories writable to continue Mar 29 14:23:55 mwester: no idea, i did do some tests several years ago with 12 Mbit devices and that sucked. Mar 29 14:25:44 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r2fc69ae0d5 10openembedded.git/recipes/libpcap/libpcap.inc: Mar 29 14:25:44 libpcap.inc : fix staging of libpcap.m4 Mar 29 14:25:44 Without that fix it fails like this: Mar 29 14:25:44 | oe_libinstall: cd ${WORKDIR}/armv6-angstrom-linux-gnueabi/libpcap-1.0.0-r6/libpcap-1.0.0 Mar 29 14:25:44 | install: cannot create regular file `${WORKDIR}/armv6-angstrom-linux-gnueabi/libpcap-1.0.0-r6/image/usr/share/aclocal/libpcap.m4': No such file or directory Mar 29 14:25:45 NOTE: Task failed: ${WORKDIR}/armv6-angstrom-linux-gnueabi/libpcap-1.0.0-r6/temp/log.do_install.32266 Mar 29 14:27:53 eFfeM, I bought 54Mbit devices, and had to set them down to 11Mbit to get it to work reliably - so no video across that link Mar 29 14:34:06 mmm gst-plugins-bad => | make[3]: /home/embedded/oetmp_bug/staging/i686-linux/home/embedded/oetmp_bug/staging/i686-linux/usr/bin/orcc: Command not found Mar 29 14:34:28 I saw an irc thread about orcc and cross-compiling on the net Mar 29 14:35:39 http://lists.mplayerhq.hu/pipermail/ffmpeg-devel-irc/2010-March/000049.html Mar 29 14:35:41 mmm Mar 29 14:35:47 mwester: thougth the next gen powerline was 84 or so Mar 29 14:36:22 these ones seem to be a lot better Mar 29 14:36:35 with the 14 mbit one i could get 0.5 mbit in my garden :-( Mar 29 14:36:44 ok there is a native recipe for orc Mar 29 14:36:56 (BBCLASSEXTEND = "native") Mar 29 14:36:57 morning all Mar 29 14:37:08 GNUtoo: yes, made that a whiel ago Mar 29 14:37:28 * mwester can't check right now, maybe it was 84 Mb Mar 29 14:37:36 ok Mar 29 14:38:28 strange...DEPENDS += "orc-native gst-plugins-base openssl directfb" is already in gst-plugins-bad_0.10.18.bb Mar 29 14:38:37 I'll check Mar 29 14:38:48 yeah koen has been digging into that Mar 29 14:40:06 ahh ok Mar 29 14:40:15 /home/embedded/oetmp_bug/staging/i686-linux/home/embedded/oetmp_bug/staging/i686-linux/usr/bin/orcc: Mar 29 14:40:21 there is the path twice Mar 29 14:40:55 there is somethign odd there, don't know details Mar 29 14:41:53 * eFfeM heads home :-) Mar 29 14:42:01 ok I'll try to fix it Mar 29 14:42:59 maybe: Mar 29 14:43:01 sed -i -e s:/usr/bin/orcc:${STAGING_BINDIR_NATIVE}/orcc:g $i Mar 29 14:43:15 maybe it's already ${STAGING_BINDIR_NATIVE} Mar 29 14:43:22 so it sed it again Mar 29 14:47:13 have anybody done profiling libc for target? any patches? Mar 29 14:48:55 slapin: profiling as in gprof or oprofile? Mar 29 14:49:00 slapin: patches in which way? Mar 29 14:50:11 zecke: gprof Mar 29 14:50:25 zecke: glibc_*.bb Mar 29 14:51:52 zecke: I just need to be able to profile libc functions to optimize usage of them, not for libc hacking (at the moment at least). Mar 29 14:52:42 I just need for apps to produce gprof.out with all needed information Mar 29 14:53:28 having -pg in addiion to -dbg could be nice, but that seems to be a long shot for now. Mar 29 14:54:16 03Thomas Zimmermann  07org.openembedded.dev * r263d5be381 10openembedded.git/recipes/findutils/findutils.inc: Mar 29 14:54:16 findutils: fix path of sort in updatedb Mar 29 14:54:16 * sort pointed to staging in updatedb script Mar 29 14:54:16 Signed-off-by: Thomas Zimmermann Mar 29 15:09:30 GNUtoo, just realised I can't NFS boot on the board I'm working on. Mar 29 15:09:52 MWelchUK_work, ah why? no real ethernet? Mar 29 15:10:02 no drivers in the bootloader? Mar 29 15:10:21 Though that appears to be because the kernel isn't doing the DHCP request it should be. Mar 29 15:10:34 ah you meant that it's possible but you didn't succeed Mar 29 15:10:42 No, it's found the Eth Mar 29 15:11:05 I understood that it was not possible,sorry Mar 29 15:11:52 GNUtoo, I've definitely had it working on boards not that long ago. Mar 29 15:12:36 ok Mar 29 15:12:59 I meant that I didn't understood " GNUtoo, just realised I can't NFS boot on the board I'm working on. " well Mar 29 15:13:55 It failed to NFS boot. Mar 29 15:15:32 ok Mar 29 15:32:53 khem: around? Mar 29 15:58:08 03Thomas Zimmermann  07org.openembedded.dev * r21c2b8d910 10openembedded.git/recipes/tasks/task-shr-feed.bb: Mar 29 15:58:08 task-shr-feed: add libyaml-perl and libnet-dbus-perl as requested in SHR Bug: #998 Mar 29 15:58:08 Signed-off-by: Thomas Zimmermann Mar 29 16:05:40 ouch... Mar 29 16:05:44 do_configure_append(){ Mar 29 16:05:57 in inc file get into run.do_configure Mar 29 16:06:00 followed by Mar 29 16:06:21 : of do_configure_append(){ in recipe Mar 29 16:06:51 how can I override do_configure_append in the recipe when it's already defined in inc file? Mar 29 16:09:00 you don't. would have to put those bits into a separate shell function and call it from do_configure in the .inc, so the .bb can override the do_configure, or similar. Mar 29 16:09:50 ouch thanks a lot Mar 29 16:09:57 or a separate task? Mar 29 16:10:23 and override the task in the .bb Mar 29 16:32:02 how much disk space do i need for angstrom? Mar 29 16:33:22 etrunko, on target? Mar 29 16:33:23 etrunko, to build or run? Mar 29 16:33:39 to build Mar 29 16:33:41 etrunko, and which image Mar 29 16:33:46 x-image Mar 29 16:34:15 I'd guess at about 20GB Mar 29 16:34:22 But that really is a guess. Mar 29 16:34:34 weird Mar 29 16:34:39 Why? Mar 29 16:35:12 MWelchUK_work: i guess i'm using much more than that Mar 29 16:35:25 just waiting for du to finish] Mar 29 16:35:46 etrunko, I did say it was a guess - I'm not close to the machine I've built x11 images on. Mar 29 16:36:02 i generated a console image for minimal distro which used around 16GB Mar 29 16:36:08 console-image can take about 14GB I think Mar 29 16:36:25 MWelchUK_work: yep, something like that Mar 29 16:36:36 etrunko, my estimates are obviously a little low. :-) Mar 29 16:36:41 ;) Mar 29 16:36:59 Probably depends a little on arch as well. Mar 29 16:38:11 14G console-image? Mar 29 16:38:35 oh disk space for build :) Mar 29 16:38:40 boatloads ... Mar 29 16:38:41 Crofton: the build dir is about that Mar 29 16:39:10 Crofton, I wouldn't like to ramdisk boot it if the filesystem was that big ;-) Mar 29 16:39:29 Clearly - I'd run out of RAM though :-) Mar 29 16:40:24 * MWelchUK_work decides to rest his brain. S'later. Mar 29 16:40:48 have a nice rest of day Mar 29 16:44:11 cheers Mar 29 16:44:54 gn Mar 29 16:54:27 etrunko, rm_work can help for disk-space but debugging is worse with it Mar 29 16:55:43 GNUtoo: still waiting du Mar 29 16:55:56 will get the number sson Mar 29 16:55:59 *soon Mar 29 16:56:05 ok Mar 29 17:02:44 ouch Mar 29 17:02:47 64GB Mar 29 17:03:00 and i was still on task ~4500 Mar 29 17:14:07 03Roman I Khimov  07org.openembedded.dev * rb064810bbb 10openembedded.git/recipes/glib-2.0/ (2 files in 2 dirs): Mar 29 17:14:07 glib-2.0_2.23.6: add gatomic-proper-pointer-get-cast.patch Mar 29 17:14:07 Fixes "cast discards qualifiers from pointer target type" warning on Mar 29 17:14:07 gthread.h:348 which is especially annoying when package is being built Mar 29 17:14:07 with -Werror. Mar 29 17:14:08 2.22 glib branch is not affected. Mar 29 17:14:08 Signed-off-by: Roman I Khimov Mar 29 17:14:09 03Roman I Khimov  07org.openembedded.dev * r5115150567 10openembedded.git/recipes/openais/ (files/fix-lcrso-linkage.patch openais_1.1.2.bb): (log message trimmed) Mar 29 17:14:10 openais: add new recipe Mar 29 17:14:10 The OpenAIS Standards Based Cluster Framework is an OSI Certified Mar 29 17:14:11 implementation of the Service Availability Forum Application Mar 29 17:14:11 Interface Specification (AIS). The Application Interface Specification Mar 29 17:14:12 is a software API and policies which are used to develop applications Mar 29 17:14:12 that maintain service during faults. Restarting and failover of Mar 29 17:14:13 03Roman I Khimov  07org.openembedded.dev * r08533fd2a8 10openembedded.git/recipes/linux-ha/ (3 files in 2 dirs): Mar 29 17:14:13 cluster-glue: add new recipe Mar 29 17:14:55 of a complete cluster engine. Mar 29 17:14:55 Signed-off-by: Roman I Khimov Mar 29 17:14:55 03Roman I Khimov  07org.openembedded.dev * r43ef9ce201 10openembedded.git/: Merge branch 'pacemaker' into org.openembedded.dev Mar 29 17:14:55 03Roman I Khimov  07org.openembedded.dev * rc5497549b9 10openembedded.git/recipes/linux-ha/cluster-resource-agents_1.0.2.bb: Mar 29 17:14:56 cluster-resource-agents: add new recipe Mar 29 17:14:56 A resource agent is a standardized interface for a cluster resource. Mar 29 17:14:57 In translates a standard set of operations into steps specific to the Mar 29 17:14:57 resource or application, and interprets their results as success or Mar 29 17:14:58 failure. Mar 29 17:14:58 Signed-off-by: Roman I Khimov Mar 29 17:27:27 03Denys Dmytriyenko  07org.openembedded.dev * r1dd39944c8 10openembedded.git/recipes/util-linux-ng/util-linux-ng.inc: Mar 29 17:27:27 util-linux-ng-native: add dependency on lzo-native Mar 29 17:27:27 Signed-off-by: Denys Dmytriyenko Mar 29 18:08:10 hi mickeyl, kergoth_ Mar 29 18:08:59 evening pb_ Mar 29 18:17:03 03Roman I Khimov  07org.openembedded.dev * r99ac56dac1 10openembedded.git/recipes/quagga/ (files/volatiles.03_quagga quagga.inc): Mar 29 18:17:03 quagga: fix access rights, add more postinst Mar 29 18:17:03 Add null conf files for protocols (allows tighter permissions Mar 29 18:17:03 setting and sets some starting point for configuration in general), Mar 29 18:17:03 fix permissions for log directory. Mar 29 18:17:04 Signed-off-by: Roman I Khimov Mar 29 18:17:05 03Roman I Khimov  07org.openembedded.dev * r5803d4e2d5 10openembedded.git/recipes/quagga/ (quagga.inc quagga_0.99.16.bb): Mar 29 18:17:05 quagga: add version 0.99.16 Mar 29 18:17:06 Signed-off-by: Roman I Khimov Mar 29 18:17:10 03Roman I Khimov  07org.openembedded.dev * r2d72e73306 10openembedded.git/recipes/quagga/ (5 files): Mar 29 18:17:10 quagga: switch to INC_PR Mar 29 18:17:10 Signed-off-by: Roman I Khimov Mar 29 18:19:28 hmm... pkgconfig fails building in do_install Mar 29 18:19:37 with the following patch it builds Mar 29 18:19:39 - install -d -m 0755 ${libdir}/pkgconfig Mar 29 18:19:40 + install -d -m 0755 ${D}${libdir}/pkgconfig Mar 29 18:19:49 not sure if that is the correct thing to do though Mar 29 18:26:14 is this the makefile you're patching, or the recipe? Mar 29 18:26:24 if it's the makefile, and it's an autotools package, you should use DESTDIR not ${D} Mar 29 18:27:18 the recipe Mar 29 18:27:40 pb_: do_install_append Mar 29 18:28:32 in that case it should be fine Mar 29 18:29:17 ok, shall I commit that? or better go via the list? Mar 29 18:29:20 Hi guys, I am getting `/oe/openembedded/recipes/gcc/gcc-cross-intermediate_4.3.3.bb do_compile failed`. I would like to prevent rebuilding the whole tree. What do I have to run with `clean` to get this build? Mar 29 18:30:10 It complains about `/oe/angstrom-dev/work/armv7a-angstrom-linux-gnueabi/gcc-cross-intermediate-4.3.3-r11.1/gcc-4.3.3/libgcc/../gcc/tsystem.h:111:18: error: time.h: No such file or directory`. Mar 29 18:30:19 It complains about `make[2]: *** [_negdi2.o] Error 1 Mar 29 18:43:59 03Denis 'GNUtoo' Carikli  07org.openembedded.dev * r50e27651d4 10openembedded.git/recipes/linux/ (linux-leviathan/defconfig linux-leviathan_git.bb): linux-leviathan: defconfig: enable earlysuspend for console to fix suspend/resume w/ console on a VT Mar 29 18:46:32 KeyError: 'virtual:native:/data/pkg/oe/openembedded/recipes/util-linux-ng/util-linux-ng_2.17.bb' Mar 29 18:46:35 *sigh* Mar 29 19:08:06 03Klaus Kurzmann  07org.openembedded.dev * r97a32c6428 10openembedded.git/conf/distro/include/sane-srcrevs.inc: Mar 29 19:08:06 sane-srcrevs.inc: bump revs for libgee and libgee-native Mar 29 19:08:06 to get a fix to make it build with current vala Mar 29 19:08:06 Signed-off-by: Klaus Kurzmann Mar 29 19:36:21 sigh.. mondays Mar 29 19:37:05 Indeed Mar 29 19:38:53 Could you please test my Enna recipe in his build environment. Mar 29 19:39:16 I am trying to build the tree with GCC 4.3.3 but it takes very long. Mar 29 19:47:04 03Koen Kooi  07org.openembedded.dev * r6a00236c41 10openembedded.git/recipes/efl1/ecore_svn.bb: Mar 29 19:47:04 ecore-native: disable curl again Mar 29 19:47:04 * this avoids an error with curl internal checks that get triggered for some reason on 64 bit buildhosts Mar 29 20:01:32 woglinde: Just for your interest. Mar 29 20:06:21 hehe okay Mar 29 20:06:37 I didnt even remember about this patch Mar 29 20:13:26 03Koen Kooi  07org.openembedded.dev * r4172d7593e 10openembedded.git/recipes/geoclue/geoclue_0.12.0.bb: geoclue: add 0.12.0 release Mar 29 20:26:05 hi, does anyone has | configure.ac:25: require Automake 1.11.1, but have 1.10.3 in org.openembedded.dev/recipes/m4/m4_1.4.14.bb ? Mar 29 20:26:15 I recompiled from scratch(rm -rf tmpdir) Mar 29 20:26:20 and still this issue Mar 29 20:28:32 I really don't have all the time required to dive into theses autotools issues Mar 29 20:40:57 PaulePanter: hmmm gcc build it failing on you. because your libc is removed from staging Mar 29 20:41:05 PaulePanter: you could do this Mar 29 20:41:33 bitbake -c clean gcc-cross gcc-cross-intermediate gcc-cross-initial glibc glibc-initial Mar 29 20:41:48 then bitbake cross-gcc Mar 29 20:41:54 khem: Thanks! Mar 29 20:42:20 PaulePanter: if you are not using glibc then whatever libc you use Mar 29 20:42:25 eglibc or uclibc Mar 29 20:42:28 khem: I will try it later. I already started from scratch again. Mar 29 20:42:37 PaulePanter: ok, Mar 29 20:43:12 Or is there a way to stop BitBake in such a way, that all tasks get finished and no new ones are started? Mar 29 20:43:41 Ctrl+C Mar 29 20:43:51 I guess Mar 29 20:43:56 but dont quote me on that Mar 29 20:44:03 khem: I forgot `glibc` and `glibc-initial` when I tried `bitbake -c clean …`. Mar 29 20:44:23 khem: I think from Ctrc + c I got my problems. Mar 29 20:44:30 yeah there is some catch 22 situation when it comes too toolchain Mar 29 20:44:36 I have been thinking on improving it Mar 29 20:46:01 khem: I guess there are more important things to do as always. Thank you anyway! Mar 29 20:46:34 PaulePanter: np Mar 29 20:54:16 khem, should I install automake 1.11.1? Mar 29 20:55:51 I bet there is no wrapper script no? Mar 29 20:56:09 hm. I have funny erors with (GNU automake) 1.10.3 Mar 29 20:56:15 ant__, me too Mar 29 20:56:26 I recompiled tmpdir from scratch because of that Mar 29 20:56:33 I had no time to dig into it Mar 29 20:56:36 /oe/build/tmp/work/armv5te-angstrom-linux-gnueabi/kexec-tools-klibc-static-2.0.1-r4/kexec-tools-2.0.1/configure: line 5880: syntax error: unexpected end of file Mar 29 20:56:40 file is fine Mar 29 20:57:31 ant__: it might be spitting out a file and cant find EOF token Mar 29 20:57:43 ant__: paste the configure script somewhere Mar 29 20:58:21 last 2 bytres are 0x00A, EOF Mar 29 20:58:23 ok Mar 29 20:58:23 GNUtoo: its better to recompile everything when you have a autotools update Mar 29 20:58:31 has anyone else run into staging issues with libnet after a clean build? Mar 29 20:58:41 khem, I just did that Mar 29 20:58:48 GNUtoo: good Mar 29 20:58:55 I don't want to do it again Mar 29 20:59:03 seems that both the headers and lib aren't staged now Mar 29 20:59:09 sakoman: hmmm Mar 29 20:59:19 basically I rmed my tmpdir and rebuilt hoping it fixed it but it didn't Mar 29 20:59:40 GNUtoo: which distro are u using Mar 29 20:59:51 angstrom Mar 29 20:59:55 machine: bug 1.x Mar 29 20:59:57 fun..pastebin is down Mar 29 21:00:04 khem: the recipe has an explicit do_stage Mar 29 21:00:12 sakoman: may be thats the problem Mar 29 21:00:21 sakoman: convert it to new staging Mar 29 21:00:35 I'm not sure how to do that :-) Mar 29 21:00:50 I tried removing the do_stage and it then stages the headers Mar 29 21:01:07 sakoman, how to reproduce with angstrom and .dev? Mar 29 21:01:09 sakoman: whatever staging is doing put into install Mar 29 21:01:11 but libnet seems to be a static build and the .la is not staged Mar 29 21:01:14 khem: http://filebin.ca/ecomye/configure Mar 29 21:01:48 ant__, not same error Mar 29 21:02:13 khem: I tried moving the do stage stuff to install, but it seemed to not be necessary. the proper stuff was being installed into the image directory Mar 29 21:02:35 sakoman: then you are set Mar 29 21:02:36 GNUtoo: the recipe was building few days ago Mar 29 21:02:41 just remove do_stage Mar 29 21:03:10 ant__: Generated by GNU Autoconf 2.65 for kexec-tools 2.0.1 Mar 29 21:03:12 hmmm Mar 29 21:03:38 is it compatible with that new autoconf ? Mar 29 21:05:38 khem: just wondering... Mar 29 21:06:18 ant__: what version is there unmodified version have Mar 29 21:06:51 ant__: just do bitbake -c patch Mar 29 21:06:59 and then check the configure script Mar 29 21:10:11 GNUtoo: for your error you might want to regenerate the autoconf stuff for geoclue_0.12.0.bb Mar 29 21:10:28 ah Mar 29 21:10:40 GNUtoo: btw how come you chose so new autotools Mar 29 21:10:48 I didn't Mar 29 21:10:50 I did: Mar 29 21:10:56 rm -rf tmpdir Mar 29 21:11:01 bitbake illume-image Mar 29 21:11:03 GNUtoo: oh Mar 29 21:11:04 and I had that error Mar 29 21:11:25 ok that recipe is newly added like few hours back Mar 29 21:11:26 http://pastebin.com/8RBzt8qE Mar 29 21:12:07 khem: interestingly now the configure is longer... Mar 29 21:12:10 ok so your problem is opposite of ant__ Mar 29 21:12:35 GNUtoo: you seem to have older version of autotools than what the recipe wants Mar 29 21:12:47 I know but what should I do to fix that Mar 29 21:13:01 khem: http://filebin.ca/prcjsp/configure Mar 29 21:13:10 I bet autotools aren't compatible between verisons Mar 29 21:13:18 so other things will break Mar 29 21:13:19 no? Mar 29 21:13:21 sakoman, what package can I build that depends on libnet being staged? Mar 29 21:14:36 hmmm . . . for me I think it was tcptraceroute that failed Mar 29 21:14:52 ideally we should have multiple versions of autotools live together in native staging Mar 29 21:15:08 and then let the packages use whatever they are configured with Mar 29 21:15:18 khem: argh..some step of bitbake did overwrite the configure..now is back to 5880 lines... Mar 29 21:15:22 was 6416 Mar 29 21:15:47 ant__: what version of autoconf did it use Mar 29 21:16:01 2.63 Mar 29 21:16:04 Crofton_|work: another issue I noticed is that some of my images have tripled in size after the pull! Mar 29 21:16:21 did you look at the testlab reports? Mar 29 21:16:39 sakoman: interesting, Mar 29 21:17:00 sakoman: Do you mean the individual binaries or the .ipks Mar 29 21:17:13 Crofton_|work: no, I didn't Mar 29 21:17:31 khem: I'm referring to the size of the image tar.bz2 Mar 29 21:18:01 sakoman: hmmm so its pullin in more deps I guess Mar 29 21:18:02 looking at the untarred old & new images it seems that the /usr/lib has grown dramatically Mar 29 21:18:06 which images do you build Mar 29 21:18:37 sakoman: one wrong dependency can double the size :) that my theory Mar 29 21:19:57 khem: it seems to be a number of individual libs that have grown Mar 29 21:20:46 for example libabiword was 5.5MB , now it is 161.6MB! Mar 29 21:20:49 sakoman: hmmm, did your toolchain versions change ? Mar 29 21:21:02 khem: OMG Mar 29 21:21:04 sakoman: oh could it be that now it is not stripped Mar 29 21:21:11 many are the same size Mar 29 21:21:16 diff -wB old configure Mar 29 21:21:33 < configured by $0, generated by GNU Autoconf 2.63, Mar 29 21:21:33 < with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\" Mar 29 21:21:33 --- Mar 29 21:21:33 > configured by $0, generated by GNU Autoconf 2.65, Mar 29 21:21:37 hi khem Mar 29 21:21:45 RP: hi there Mar 29 21:21:46 2 autoconf? Mar 29 21:21:58 khem: How do you make libgcc build without gcc itself? Mar 29 21:22:02 old is the configure after do_patch Mar 29 21:22:08 khem: libwebkit went from 13.2 MB to 773.8MB! Mar 29 21:22:13 configure is well..configure :) Mar 29 21:22:29 khem: I'm entering libgcc and running configure fine but when I run make, it wants headers and a makefile include from the gcc build directory? Mar 29 21:22:31 RP: in newer gcc libgcc is out of gcc tree Mar 29 21:22:38 RP: older versions I have not tried Mar 29 21:22:42 khem, yes that's the ideal,but I saw only that : Mar 29 21:22:46 khem: This is 4.3.3 Mar 29 21:23:03 org.openembedded.dev/recipes/automake $ ls Mar 29 21:23:04 automake_1.10.2.bb automake_1.10.3.bb automake_1.11.1.bb automake_1.9.6.bb automake.inc Mar 29 21:23:11 sakoman: do a file on new and old and see if they are stripped Mar 29 21:23:13 khem: Its in its own directory but needs a gcc builddir to work :/ Mar 29 21:23:20 khem: I too suspect they haven't been stripped Mar 29 21:23:24 will check Mar 29 21:24:11 RP: hmmm, even with gcc 4.3 and newer Mar 29 21:24:19 khem: yes indeed. old was stripped, new is not Mar 29 21:24:33 sakoman: so that bloat is all debug info Mar 29 21:24:46 khem: right Mar 29 21:24:55 what might have changed to cause this? Mar 29 21:25:07 it happened after a pull from oe.dev last week Mar 29 21:25:21 khem: I've been used in Gentoo having more autotools contemporarly..but each packages just used one set.. Mar 29 21:25:51 even in OE you cannot depend on a specific version... Mar 29 21:25:57 khem: I will post this recipe I'm playing with and you can take a look if you like :) Mar 29 21:26:23 RP: sure Mar 29 21:26:34 RP: that will give me ideas too :) Mar 29 21:27:03 How do you build the OE tree and develop. The tree is in a Git repository and if you switch branches BitBake gets confused, does not it? Do you copy the tree you want to build to a seperate folder and point BitBake to it? Mar 29 21:28:25 PaulePanter: ? Mar 29 21:28:27 khem: I just did a clean build and libabiword is unstripped there too Mar 29 21:29:07 PaulePanter: you can just checkout the branch you want to work on Mar 29 21:29:30 khem: But if the files in this branch differ from the branch I want to get build? Mar 29 21:29:49 khem: http://git.pokylinux.org/cgit.cgi/poky/log/?h=master-gcc-runtime-testing Mar 29 21:29:54 khem: The last two commits Mar 29 21:32:08 RP: have you an idea how this can happen? Mar 29 21:32:09 < generated by GNU Autoconf 2.63. Invocation command line was Mar 29 21:32:09 --- Mar 29 21:32:09 > generated by GNU Autoconf 2.65. Invocation command line was Mar 29 21:32:26 on the same packages, on do_patch and do_configure Mar 29 21:33:13 RP: nice. its on the lines I was thing libstdc++-precompile-dev may be called libstdc++-pch-dev Mar 29 21:33:20 And to get on everybody’s nerves another. It’d be great if you tested my recipe for Enna I posted , try to execute `enna` and report back. Mar 29 21:33:24 ant__: No idea Mar 29 21:33:47 khem: quite a few libs have become unstripped, among them libwebkit, libabiword, libcrypto, libgtk-x11, libssl, libavcodec, libgdk-x11, libpython, libevas, . . . Mar 29 21:33:48 khem: Right, I wanted to package them separately rather than delete which was what the previous code did Mar 29 21:34:00 khem: but I don't like having to stash that magic headers list :/ Mar 29 21:34:09 khem: if I don't do that it doesn't build Mar 29 21:36:02 RP, you saw my email with the bitbake stuff in pw? Mar 29 21:36:15 sakoman: it sounds a problem in general Mar 29 21:36:19 Crofton_|work: Yes, how do you mark a pw patch as archived? Mar 29 21:36:22 how old is your old snapshot Mar 29 21:36:30 Crofton_|work: Do I need some kind of privs or something? Mar 29 21:36:33 do you have an account on pw? Mar 29 21:36:35 yeah Mar 29 21:36:44 Crofton_|work: I added one Mar 29 21:37:12 Crofton_|work: Those patches were either applied or deliberately disregarded Mar 29 21:37:21 I think was one was applied, then reverted Mar 29 21:37:29 ok Mar 29 21:37:49 unless you think they should stay archive them Mar 29 21:37:53 RP: Doing a configure should generate them Mar 29 21:38:18 khem: configure at which level? Mar 29 21:38:26 I should go through and send an email with links to the class related patches also Mar 29 21:38:34 yeah it has to go into gcc/gcc Mar 29 21:38:50 RP: sometimes make also does some configure work Mar 29 21:38:51 khem: We end up in a dependency sinkhole then Mar 29 21:38:59 and some headers are generated during make Mar 29 21:39:15 khem: The resaon is that I want to build these the same as any other target lib so no cross class Mar 29 21:39:28 khem: I'm realy confused now - I rebuilt an offending package and the size is back to normal! Mar 29 21:39:34 khem: so target==host which is not what you want when building gcc Mar 29 21:39:42 RP: could we use same staged tree ? Mar 29 21:39:59 RP: for gcc-runtime recipe Mar 29 21:40:14 khem: Thats effectively what I'm doing by copying the headers... Mar 29 21:40:23 yeah Mar 29 21:41:01 hmmm, http://tinderbox.openembedded.net/public/logs/task/5469166.txt Mar 29 21:41:51 RP: I think it needs quite a surgery to separate gcc libs so I think for now staging tree is good Mar 29 21:42:16 actually you can generate a package for it Mar 29 21:42:34 but then you wont be able to use external toolchain to rebuild them Mar 29 21:43:07 but I think we should have boundaries how far we can go Mar 29 21:43:18 khem: Just staging the header as I've done or something more substancial? Mar 29 21:43:24 We should assume that all is provided when someone is using external toolchain Mar 29 21:43:44 RP: just headers probably Mar 29 21:43:52 khem: Yes, for an external toolchain I'm happy to assume its provided Mar 29 21:44:12 some other folks I am taking to in gcc community never imagined that it could be built this way :) Mar 29 21:44:15 khem: So that recipe as is problably works... Mar 29 21:44:18 so you see the support Mar 29 21:44:44 khem: I can imagine the gcc people thinking we're nuts ;-) Mar 29 21:44:51 yeah Mar 29 21:45:04 khem: Building this stuff in "target" context makes a lot of sense though Mar 29 21:45:51 khem: My other thought was to separately "make install-target-libgcc" into a separate directory in gcc-cross, then have gcc-runtime package it Mar 29 21:47:11 RP: As I see this is only needed in staging prolly what you are doing is ok Mar 29 21:48:48 khem: Fancy lending a hand in making this work in OE? :) Mar 29 21:49:29 RP: I can however not until apr 20th Mar 29 21:50:05 khem: ok, fair enough. I will continue to experiment a bit and may or may not get it pushed into Poky by then Mar 29 21:50:08 I think its ok so give it that much time in poky to stabilize Mar 29 21:50:19 khem: Right, we can test it there :) Mar 29 21:50:20 thats fine. Mar 29 21:50:41 I guess what I really want to check is that OE isn't going to stongly object :) Mar 29 21:50:50 strongly Mar 29 21:51:02 khem: and I think you and pb_ would be the main people who'd dislike it ;-) Mar 29 21:51:09 (potentially) Mar 29 21:51:34 pb_ I think will like it though as it helps his cross.bbclass desuck work Mar 29 21:51:43 RP: yeah I still dont like it fully Mar 29 21:51:51 but there are problems at hand Mar 29 21:52:03 khem: which bits don't you like? Mar 29 21:52:04 RP: this incremental staging Mar 29 21:52:20 khem: The -initial and -intermediate? Mar 29 21:52:27 but then we dont have a way around Mar 29 21:52:48 khem: Could we use a different sysroot directory? Mar 29 21:52:51 RP: actually that too plus the one gcc-runtime has Mar 29 21:53:07 RP: yes for toolchain it would make sense Mar 29 21:53:19 khem: which incremantal staging does gcc-runtime have? Mar 29 21:53:26 * RP doesn't follow :/ Mar 29 21:53:35 RP: this headers information passing from gcc build to gcc-runtime Mar 29 21:53:44 khem: ah, right, yes Mar 29 21:54:01 * RP isn't happy with that either but can't see an alternative Mar 29 21:54:15 RP: I think we need bitbake to treat gcc and glibc specially Mar 29 21:54:38 The one thing in this cases favour is that the directory is isolated and not need by any other part of the build Mar 29 21:54:44 khem: treat specially now? Mar 29 21:54:45 how? Mar 29 21:54:46 RP: or have an attribute where we can say this recipe aliases to another recipe Mar 29 21:55:06 khem: Ah, right :/ Mar 29 21:55:14 so if gcc-cross is built then there is no need to rebuild gcc-cross-* unless asked to do so Mar 29 21:55:14 khem: I can't see this happening to be honest Mar 29 21:55:46 RP: if we stage toolchain in different sysroot problem still remains Mar 29 21:56:15 only thing is that glibc might be now not in same sysroot Mar 29 21:56:19 khem: Well, there is an efficency problem there certainly but its not a problem any more than that if we have separate sysroots? Mar 29 21:56:56 RP: so you mean have separate sysroot for earch stage ? Mar 29 21:57:15 khem: yes. build against one sysroot, install into the next stage sysroot Mar 29 21:57:16 hmm that might work Mar 29 21:57:35 and unroll it when uninstalling ? Mar 29 21:58:04 khem: no files would conflict so there would be no overwriting and it would be easy to cleanup Mar 29 21:58:40 It would work if and only if gcc/glibc can cope with building one location and installing to another Mar 29 21:58:50 but I can at least see it working in theory Mar 29 21:58:58 thats not a problem Mar 29 21:59:17 we could use DESTDIR Mar 29 21:59:26 and use buildtime-tools options Mar 29 21:59:41 so it picks the tools from intermediate sysroots Mar 29 21:59:49 khem: We already use destdir, its just the buildtime sysroot options that would be needed Mar 29 22:00:27 yes thats possible Mar 29 22:00:33 interesting Mar 29 22:00:40 makes me feel better Mar 29 22:01:38 RP: I will try to play with what you have this week as much as I can afford Mar 29 22:02:05 khem: Sounds good. I will keep that branch updated with any of my experiments with gcc-runtime Mar 29 22:02:23 yeah that wud be nice Mar 29 22:02:58 khem: 11 days autoconf_2.65.bb : add newest version of autoconf. Mar 29 22:03:10 I'll revert locally Mar 29 22:05:52 ant__: Please do not forget to report your error with the logs to the mailing list. Mar 29 22:06:00 ant__: So that it can be fixed later. Mar 29 22:06:15 strangely is +DEFAULT_PREFERENCE = "-1" Mar 29 22:06:24 but still kicking in Mar 29 22:09:12 I'll have to ping XorA|gone Mar 29 22:10:15 seems some m4 stuff is needed after autoconf >2.63 Mar 29 22:11:58 ah.. I see Mar 29 22:12:12 │/autoconf-native-2.65-r9.0 was built here Mar 29 22:14:11 hm..how does BBCLASSEXTEND has to do with this? Mar 29 22:15:57 ah.. ok.. PREFERRED_VERSION_autoconf = "2.65" Mar 29 22:16:00 beats all Mar 29 22:16:37 well, PREFERRED_VERSION_autoconf-native = "2.65" Mar 29 22:17:52 11 days angstrom-2008-preferred-versions.inc : update angstrom to autoconf 2.65 Graeme Gregory Mar 29 22:18:01 ok, I know enough Mar 29 22:18:04 :) Mar 29 22:25:11 :)) with 2.63 all built just fine Mar 29 22:25:33 khem: as you suspected kexec-tools is not autoconf-2.65 compatible Mar 29 22:26:16 atm is the only package failing..let's catch others( if any...) Mar 29 22:33:46 03Klaus Kurzmann  07org.openembedded.dev * r6a01455b80 10openembedded.git/recipes/pkgconfig/pkgconfig.inc: Mar 29 22:33:46 pkgconfig.inc: add missing ${D} when creating ${libdir}/pkgconfig in do_install_append Mar 29 22:33:46 Signed-off-by: Klaus Kurzmann Mar 29 22:33:46 Signed-off-by: Philip Balister Mar 30 02:50:14 anyone know what happened to this file? Mar 30 02:50:15 /home/balister/oe/tmp/staging/armv7a-angstrom-linux-gnueabi/usr/include/boost/detail/atomic_count_gcc.hpp:20:28: error: bits/atomicity.h: No such file or directory **** ENDING LOGGING AT Tue Mar 30 02:59:57 2010