**** BEGIN LOGGING AT Mon Feb 05 02:59:57 2007 Feb 05 03:03:18 anyone has dealt with i/o port level programming here? Feb 05 03:03:33 i am trying acccess allocated ports on the card and can't seem top Feb 05 03:03:35 to Feb 05 03:03:59 the ports suppose to be read/write registers , but they all return 0xfff Feb 05 03:04:11 of variation of that depending on size of the read. Feb 05 03:08:09 is there a working qemu binary out there or a way to convince qemu to bitbake correctly on a fc6 system? Feb 05 03:08:30 qemu-arm segfaults when I try it with gcc4.1 Feb 05 03:08:38 trying to convince it to use gcc3 Feb 05 03:08:44 .... Feb 05 03:50:15 03koen 07org.oe.dev * r11f0ee99... 10/ (1 packages/eds/eds-dbus_svn.bb): eds-dbus: make it even more granular Feb 05 03:53:52 koen|away: how are OpenEmbedded-derived distros going to be affected by the stupid US DST changes? The timezones package seems to get some binary opie files from somewhere ... Feb 05 03:55:43 * rwhitby will watch logs for any answers ... Feb 05 07:13:51 03rwhitby 07org.oe.dev * rf13f3dd6... 10/ (1 conf/distro/debianslug.conf conf/distro/openslug.conf): slugos: Removed reiserfs from standard image. Feb 05 07:24:22 koen: is OE doing anything about the DST change in the US? The timezones package seems to download binary files from somewhere - will they be updated by someone? Feb 05 07:24:42 no idea Feb 05 07:24:51 but mwester's suggestion is unacceptable Feb 05 07:24:55 good morning Feb 05 07:30:33 koen: that's why I'm asking here first ;-) Feb 05 07:31:36 hi , I am coming to fosdm, is there any chance someone may bring an extra neo ? Feb 05 07:31:43 koen: seems that an existing package that downloads binary timezone data from some opie site is unacceptable too. Feb 05 07:32:06 * rwhitby wonders if timezone data is endian dependent. Feb 05 07:32:17 I need to know, as my cellular is beginin to die Feb 05 07:33:43 rwhitby: iirc it is :-/ Feb 05 07:34:15 guess it's time for those OE developers in the US to work out a better system then ;-) Feb 05 07:34:48 rwhitby: anything involving "sorry, your timezone is now 2MB" is unacceptable Feb 05 07:35:07 koen: nod Feb 05 07:35:27 that's my only concern Feb 05 07:44:34 koen: I've replied to the list. Feb 05 07:44:43 the OE list? Feb 05 07:44:59 I was a bit annoyed I had to read it on the nslu2 list instead of the OE list Feb 05 07:45:59 but then again, I get annoyed easily before I have my coffee :) Feb 05 07:46:03 I've replied to that message on the lists it was sent, and will send a very different message to the OE lists :-) Feb 05 07:49:29 sent. Feb 05 08:01:02 morning all Feb 05 08:01:10 hey RP Feb 05 08:01:21 morning RP. Do you live in the US and care about timezones? ;-) Feb 05 08:02:03 rwhitby: I live in the UK and don't have a clue about timezones ;-) Feb 05 08:02:23 are there *any* OE developers in the US? Feb 05 08:02:35 kergoth... Feb 05 08:03:09 * rwhitby guess kergoth would refuse to update timezones specifically cause the government changed them Feb 05 08:03:21 kergoth has his own tz :) Feb 05 08:06:51 hi RP Feb 05 08:15:39 morning Feb 05 08:24:30 morning Feb 05 08:24:53 RP: hi Feb 05 08:24:58 hey hrw & XorA Feb 05 08:27:25 hi XorA, hi hrw Feb 05 08:30:41 hi XorA , hrw Feb 05 08:31:15 hi everyone Feb 05 08:38:18 RP: I will do build with bitbake 1.6-svn today Feb 05 08:43:33 build started Feb 05 09:00:48 gah, my knowledge of spinlocks fails Feb 05 09:20:05 argh.. Feb 05 09:20:16 memory usage of usual apps kills me Feb 05 09:20:30 total used free shared buffers cached Feb 05 09:20:30 Mem: 962312 951112 11200 0 56280 317176 Feb 05 09:23:40 hrw: stop running VIsta in vmware :-) Feb 05 09:25:58 ;)) Feb 05 09:29:13 XorA: one month and I will add '1' to that 962312 Feb 05 09:32:30 hrw: I need to buy new machine for that sort of memory Feb 05 09:34:08 XorA: this machine has 4 memory slots Feb 05 09:34:41 XorA: so will sell 2x512M which I use and buy 2x1G ones Feb 05 09:50:00 hi Liam Feb 05 09:50:25 hi lrg Feb 05 09:50:43 hey Marcin, Florian Feb 05 09:50:47 morning all Feb 05 09:51:23 hi Liam Feb 05 09:52:20 hi Richard Feb 05 09:53:06 hmm... if someone need to pull updates hosted at linuxtogo... it might be a good idea to do that now. Feb 05 10:02:37 * rwhitby watches http://gpephone.linuxtogo.org/download/emulator/ for signs of a vmware image ... Feb 05 10:04:03 rwhitby: i'm on it - but uploading from here takes ages Feb 05 10:36:02 ~curse qemu BADLY Feb 05 10:36:14 May the fleas of a thousand camels infest your most sensitive regions, qemu BADLY ! Feb 05 10:44:33 bbl Feb 05 10:48:25 hrw: I take cursing qemu is a favorite pastime Feb 05 10:49:07 mtaht4: yeah. it breaks on amd64, it breaks when someone use gcc4 etc Feb 05 10:49:28 It's broken in FC6 by default. Feb 05 10:50:16 I installed the gcc34 compat package to give that a go, then fell asleep Feb 05 10:52:48 22G www/mirror/ Feb 05 10:53:53 hrw: u sayin that gcc34 on x86_64 is not a viable strategy? Feb 05 10:55:46 gcc34 is ok for qemu Feb 05 10:55:54 tryin now Feb 05 11:02:34 [hrw@misc1 angstrom]$ ll ~/www/sources/|wc 37220 389387 5907817 Feb 05 11:02:38 o fsck.. Feb 05 11:03:25 37k files is too much Feb 05 11:04:37 hrw: how do I convince bitbake to trash the qemu directory and start over? Feb 05 11:04:39 (build dir) Feb 05 11:05:13 -c rebuild Feb 05 11:05:16 bitbake -crebuild qemu-native Feb 05 11:07:13 thx. tryin Feb 05 11:12:17 nope. segfault. This aint enough of a mod to qemu? Feb 05 11:12:22 EXTRA_OECONF = "CC=gcc34 --target-list=arm-user,arm-softmmu --disable-gfx-check" Feb 05 11:12:44 go through configure without complainin Feb 05 11:15:39 realized that it was already trying that Feb 05 11:15:57 trying again Feb 05 11:19:13 I'm trying to do some dynamic packaging through the use of PACKAGES_DYNAMIC Feb 05 11:19:22 and? Feb 05 11:19:26 I have set PACKAGES_DYNAMIC = "settopcore-env-*" Feb 05 11:20:03 and I want to use /usr/share/stv/environments to search for the available environments Feb 05 11:20:08 (in image/) Feb 05 11:20:26 I'm following an example from drm-kernel Feb 05 11:20:37 and I'm doing: Feb 05 11:20:37 do_split_packages(d, '/usr/share/stv/environments', '^[a-zA-Z](.*)$', Feb 05 11:21:02 (I want a seperate package for each subdir of usr/share/stv/environments) Feb 05 11:21:17 but I don't seem to end up with any new packages Feb 05 11:21:49 FILES_${PN} contain "${datadir}/stv" - right? Feb 05 11:22:14 hrw: no Feb 05 11:22:21 huess thats my problem then Feb 05 11:22:23 tkp: I think you miss a / Feb 05 11:22:23 guess* Feb 05 11:22:29 * tkp ammends Feb 05 11:22:30 '/usr/share/stv/environments' Feb 05 11:23:25 XorA: you mean the / at the start? That was a typo here... it's in the bb file Feb 05 11:24:05 tkp: ok Feb 05 11:24:38 XorA: would you mind... http://rafb.net/p/FQj76p16.html ? Feb 05 11:24:59 added .../environments to FILES_${PN}, but still no new packages Feb 05 11:26:10 tkp: looks right to me Feb 05 11:26:25 all the environments are now in the main package Feb 05 11:27:22 only thing I can think of is python is funny about tabs/spaces Feb 05 11:27:51 I'm thinking... ^[a-zA-Z](.*)$ does tis miss out the first character from the captured backreference? Feb 05 11:28:06 (because of thge position of the brackets) Feb 05 11:28:23 tkp: yeah it will do Feb 05 11:28:32 right... /me tries again Feb 05 11:28:52 hm.. qemu recent cvs boots arm kernel/image on my amd64 Feb 05 11:29:35 '^([a-zA-Z].*)$' = no change :( Feb 05 11:30:09 trying to catch all subdirs which start with a normal character (to exclude invisible dirs etc) Feb 05 11:33:51 tkp: ah, I missed that you were trying to catch directories, maybe to code doesnt work with directories, Ive only seen it used with libXXX.so style constructs Feb 05 11:34:17 I've just seen that do_split_packages defaults to allow_dirs=False Feb 05 11:35:09 tkp: sounds like that could be the problem Feb 05 11:36:09 ok... well It sees them now, but I just get 'Not creating empty archive for settopcore-env-...' Feb 05 11:37:30 does someone has static busybox/arm? Feb 05 11:38:12 * koen searches Feb 05 11:38:28 drat, I don't have my old initrds anymore Feb 05 11:38:40 hrw: sorry, no :( Feb 05 11:39:01 ok Feb 05 11:39:12 I will look how locales are generated Feb 05 11:40:35 hrw: well, qemu at least thinks it built with gcc34 from FC6 compat for x86_64 but it still segfaults. try -m32 or something like that? beg someone for a workin binary? Feb 05 11:42:27 mtaht4: you familiar with building RPM? if so, build the FC6 qemu 32bit rpm with the OE patches Feb 05 11:42:40 mtaht4: then ASSUME_PROVIDED += "qemu-native" Feb 05 11:43:11 XorA Feb 05 11:43:32 the 64 bit rpm from fc6 goes boom Feb 05 11:43:41 so I will try building 32 bit Feb 05 11:43:42 thx Feb 05 11:43:49 'with OE patches' is the keyword Feb 05 11:43:53 you need the nptl patch Feb 05 11:43:56 I grok. EABI Feb 05 11:44:15 http://www.openembedded.org/repo/org.openembedded.dev/packages/qemu/files/arm_nptl.patch <- that one Feb 05 11:44:18 re Feb 05 11:44:29 Jin^eLD: wb Feb 05 11:44:49 still easier to beg someone for a workin binary or rpm. :) Feb 05 11:44:53 .... patching... Feb 05 11:45:09 another mips fight for me today ) let's see how far I get this time Feb 05 11:45:22 qemu is a POS concerning its build Feb 05 11:45:57 assuming I get a working one, can someone toss the rpm somewhere up on the site Feb 05 11:46:00 * XorA just doesnt use binary locales Feb 05 11:46:36 gcc34 32 bit I assume? Feb 05 11:47:03 mickey|tv: you mentioned that someone was looking at http://bugzilla.gnome.org/show_bug.cgi?id=90621, any news on that? Feb 05 11:50:26 hrw: scrolling back in the irc log implies that more recent cvs of qemu is working on x86_64 or is that non-eabi? Feb 05 11:50:39 also suckin down latest cvs of qemu Feb 05 11:50:44 non nptl Feb 05 12:12:11 latest src rpm seems much later/greater, guess that's just for the kernel version. http://fedoranews.org/tchung/qemu/0.8.2/fc6/2.6.19-1.2895.fc6/ Feb 05 12:16:17 patching, building... patched rpms for ntpl are in http://www.transconf.net/~mtaht/qemu - testing now Feb 05 12:16:49 is there something similar to base_conditional which simply lets me test if an variable is set? Feb 05 12:17:51 hmm... that doesn't appear to build a qemu-arm binary Feb 05 12:19:19 gah. target-list Feb 05 12:19:24 re-re-re-rebuilding Feb 05 12:26:22 rebuilt... testing... Feb 05 12:26:37 does pkg_postinst_ get done when the rootfs is built for an image? or does it happen when the mchine first boots? Feb 05 12:29:36 both Feb 05 12:30:26 tkp: anything what can be done during rootfs creation should be done in that phase Feb 05 12:30:50 hrw: wb Feb 05 12:31:11 * hrw is testing cup warmer hub Feb 05 12:33:16 hrw, koen: Yes! Yes! a workin qemu. (well, it's generating locales now... any way to check for correctness?) I will keep those rpms up on transconf if someone else wants them.... Feb 05 12:33:27 hrw: I get (offline root mode: not running settopcore.postinst) when it tries to run my pkg_postinst_ durin rootfs creation Feb 05 12:34:08 mtaht4: 0.8.2+oe patches or other ver? Feb 05 12:34:36 all I did was add the ntpl patch to the latest/greatest fc6 rpm. I didn' Feb 05 12:34:47 t have it generate anything but arm and x86 emulators tho Feb 05 12:35:31 shall I go full tilt boogie and try to put in all the oe patches? Feb 05 12:35:41 and generate all the emulators? Feb 05 12:41:16 greetings Feb 05 12:41:22 hi Ifaistos Feb 05 12:45:23 very quiet :) Feb 05 12:45:32 hey Ifaistos ) Feb 05 12:46:46 hrw, Jin^eLD : hi guys Feb 05 12:50:07 guys.. any ideas regarding this problem: http://www.deadlock.dhs.org/jin/glibclog.txt Feb 05 12:50:36 I want my distro to use glibc-2.5, the .bb file is there, but its getting ignored for some reason Feb 05 12:50:52 I can't figure out who is selecting 2.3.5 Feb 05 12:53:07 Jin^eLD : i had something similar with the ppc on Anstrom. You need to define glibc intermediate Feb 05 12:53:30 Jin^eLD: lack of nptl in your config Feb 05 12:53:37 # nptl needs unwind support in gcc, which can't be built without glibc. Feb 05 12:53:37 PROVIDES = "virtual/libc ${@['virtual/${TARGET_PREFIX}libc-for-gcc', '']['nptl' in '${GLIBC_ADDONS}']}" Feb 05 12:54:18 hrw: but what about this in glibc_2.5.bb file? GLIBC_ADDONS ?= "ports,nptl,libidn" Feb 05 12:54:26 I am not setting the GLIBC_ADDONS elsewhere Feb 05 12:54:36 I even tried replacing ?= with = in the .bb file - it did not help Feb 05 12:54:50 Ifaistos: I am setting glibc and glibc_intermediate preferred versions to 2.5 Feb 05 12:54:52 angstrom defines it for mipsel Feb 05 12:55:58 koen: where exactly? I looked at the conf file, there was stuff like PREFERRED_VERSION_glibc ?= "2.5" Feb 05 12:56:01 I used the same Feb 05 12:56:09 but I did not see anything mipsel specific, did I miss it? Feb 05 12:56:21 yes :) Feb 05 12:56:27 =) Feb 05 12:56:36 argos-wlan:~/Projects/OpenEmbedded/org.openembedded.dev koen$ grep mips conf/distro/angstrom-2007.1.conf Feb 05 12:56:37 PREFERRED_PROVIDER_virtual/mipsel-linux-libc-for-gcc = "glibc-intermediate" Feb 05 12:57:15 boy, locale generation takes forever. Feb 05 12:57:29 oh.. crap.. but that must be new? I do update each couple of days, I guess I should update everyday :) Feb 05 12:57:31 thanks, I will try that Feb 05 13:10:18 ok that brought at least some progress Feb 05 13:10:20 it now says: Feb 05 13:10:23 NOTE: preferred version 2.5 of glibc not available Feb 05 13:10:23 NOTE: package glibc-intermediate-2.5: started Feb 05 13:12:21 Jin^eLD: what architecture are you building for? Feb 05 13:12:22 mipsel Feb 05 13:12:31 cool Feb 05 13:12:49 you also have some mipsel hardware? Feb 05 13:12:54 I'm keeping an eye on everything glibc2.5 as I'm hoping to run it on my slug one day. Feb 05 13:13:23 but by then I'll have outgrown my slug so... Feb 05 13:13:39 well, I am trying 2.5 out of desperation now :) I managed to get 2.3.5 to build based on the dreambox branch, merged the changes over to dev Feb 05 13:13:41 Some nice embedded cpu coming out: PASEMI's 1682M looks good. Feb 05 13:13:55 intel also has a new pentium M variant. Feb 05 13:13:59 * koen wants that 3dlabs cpu Feb 05 13:14:03 but then my hadrware refused to boot, tells "cannot set up thread-local storage: unknown error", could not find out what it means Feb 05 13:14:20 so I am trying 2.5 with cross-lfs patches Feb 05 13:14:22 koen: what's the specs? Feb 05 13:14:58 Jin^eLD: that's why I'm just waiting... for somebody to do it for me as I know my limitations. Feb 05 13:15:21 ) Feb 05 13:15:53 My aim is robotics not kernel hacking (although It would be nice). Feb 05 13:16:03 btw, anyone seen this thing: http://www.intel.com/design/servers/storage/ss4000-E/ Feb 05 13:16:06 what CPU is that? Feb 05 13:16:22 VoodooZ: kernel hacking would be interesting, but me experience in that equals to zero :) Feb 05 13:16:39 koen: This one? http://www.3dlabs.com/content/productOverview.asp Feb 05 13:16:52 I am just trying to create a usable OE based distro for the dev board I hve Feb 05 13:16:55 VoodooZ: that one Feb 05 13:16:56 Jin^eLD: yeah, exactly. me too. Feb 05 13:17:01 koen: very nice!!! Feb 05 13:21:54 ~curse git Feb 05 13:22:02 May you be reincarnated as a Windows XP administrator, git ! Feb 05 13:22:05 cg-seek v2.6.20 just fscked my kernel tree Feb 05 13:22:49 wow that PASEMI looks excellent. does it have real floating point? want ardour on one of these: http://www.linuxguitar.org/ Feb 05 13:23:43 and sooperlooper Feb 05 13:25:34 mtaht4: Each chip is a single FP piped PowerPC that can issue three instructions per clock Feb 05 13:25:44 mtaht4: http://www.theinquirer.net/default.aspx?article=37426 Feb 05 13:31:04 hrw: ping Feb 05 13:36:06 CM: still strikes me as a rather warm guitar. Feb 05 13:37:03 pöng Laibsch Feb 05 13:54:37 doh.. glibc_2.5 failed with stuff like undefined reference to `_Unwind_Resume' and so on.. any ideas what to look for? cross lfs suggested libc_cv_forced_unwind=yes which is set (as I can see from config.log) Feb 05 13:57:13 mornig Feb 05 13:57:20 hi Feb 05 13:59:13 yea, I finally got through generating locale data. Not for the first time do I wish that everyone spoke english. Feb 05 13:59:32 :) Feb 05 14:26:58 how can I get ipkg to install a package from a local file? without it attempting to download anything from it's feed urls? Feb 05 14:27:26 it just keeps telling me 'Package xyx is not available from any configured src' Feb 05 14:27:26 rm /var/lib/ipkg/* ; ipkg install foo.ipk Feb 05 14:27:57 koen: that dir is already empty Feb 05 14:35:31 serenity:~# netstat | wc -l Feb 05 14:35:31 357 Feb 05 14:35:38 all hail slashdot Feb 05 14:36:12 koen: I did add the things exactly as in the angstrom conf file, it does indeed try to build the glibc-2.5 version now, but still get the not available message Feb 05 14:36:18 is that ok? http://www.deadlock.dhs.org/jin/glibc2.txt Feb 05 14:36:29 yes, that's expected Feb 05 14:36:37 ok thanks Feb 05 14:43:16 interesting, seems philips stb810 support made it into 2.6.20 Feb 05 14:48:45 03koen 07org.oe.dev * r5c48574c... 10/ (3 files in 2 dirs): libjingle: update to 0.3.10 release Feb 05 14:54:16 good morning Feb 05 14:54:56 hey Feb 05 14:55:35 building my first bootstrap image with my new edb9302 kernel this morning Feb 05 14:56:02 Is there any automated target for making a ramdisk from the bootstrap image? Feb 05 14:57:23 I think there is some way to produce various images, I've seen it in some conf files Feb 05 14:57:29 but I did not try it yet Feb 05 14:57:46 kk, grepping begins Feb 05 14:58:19 stuff like IMAGE_FSTYPES = "cramfs" Feb 05 15:08:36 crap.. glibc 2.5 refuses to build.. complains about undefined reference to `_Unwind_Resume' Feb 05 15:08:45 any ideas? Feb 05 15:09:04 anyone had a similar problem? Feb 05 15:09:51 Jin^eLD: read comments in blibc_2.5.bb? Feb 05 15:10:21 Jin^eLD: which gcc/binutils you use? Feb 05 15:10:25 about nptl and unwind support? Feb 05 15:10:29 yes Feb 05 15:10:43 gcc 4.1.1, binutils 2.17 Feb 05 15:11:02 maybe mipsel lack unwind support? Feb 05 15:11:14 hmm Feb 05 15:11:34 cross lfs suggest to do that: Feb 05 15:11:35 Jin^eLD: http://packages.debian.org/experimental/libs/libc6 - check debian diff Feb 05 15:11:36 echo "libc_cv_forced_unwind=yes" > config.cache Feb 05 15:11:52 I checked my config.log of glibc and that var is set Feb 05 15:12:00 ok thanks, let me see Feb 05 15:14:54 03koen 07org.oe.dev * rabecc468... 10/ (7 files in 2 dirs): tapioca: update to something more recent Feb 05 15:21:22 hmm, there are some mips patches there of course, but I do not see anything unwind related Feb 05 15:22:35 so unwind support is actually in gcc? Feb 05 15:22:47 and glibc is needed to build gcc Feb 05 15:22:51 I guess thats the glibc-intermediate thing Feb 05 15:23:06 and after that is done, then the "real" glibc is built, do I understand that correctly? Feb 05 15:25:30 Jin^eLD: yes Feb 05 15:26:00 so how do I ensure that there is unwind support in gcc? Feb 05 15:26:53 both glibc-intermediate and glibc do have libc_cv_forced_unwind=yes in config.log Feb 05 15:27:02 but thats glibc.. now gcc hm Feb 05 15:27:17 Jin^eLD: Sorry, I think unwind support is needed in glibc. Feb 05 15:27:28 aha.. Feb 05 15:27:35 * Jin^eLD is pretty confused :) Feb 05 15:27:49 this comment confused me a little Feb 05 15:27:50 # nptl needs unwind support in gcc, which can't be built without glibc. Feb 05 15:28:18 A cut down version of gcc is built that doesn't need unwind. This is just used to build an intermediate glibc with unwind support. A full gcc can then be built which can then be used to build a full glibc Feb 05 15:28:46 Jin^eLD: I suspect I'm probably wrong ;-) Feb 05 15:28:49 ok I see.. Feb 05 15:28:50 hehe Feb 05 15:28:51 :) Feb 05 15:28:53 * RP can't remember the specifics... Feb 05 15:29:03 well, I am at the point where I have the glibc-intermediate and gcc Feb 05 15:29:32 both, glibc-intermediate and glibc have libc_cv_forced_unwind=yes and libc_cv_c_cleanup=yes in their config.log files Feb 05 15:29:43 those are the only settings that cross-lfs is mentioning regarding unwind Feb 05 15:30:22 I wonder why this is not working out then.. hmm Feb 05 15:31:51 what is "libunwind" Feb 05 15:31:55 a library that does the same? Feb 05 15:32:28 the sconfigure script says something like Feb 05 15:32:30 checking for libunwind-support in compiler... no Feb 05 15:32:34 checking for forced unwind support... yes Feb 05 15:32:42 but I guess that should be ok.. Feb 05 15:34:40 Perhaps compare it with an x86 or ARM build? Feb 05 15:35:06 05 17:06 < Jin^eLD> checking for libunwind-support in compiler... no Feb 05 15:35:22 does not look good Feb 05 15:35:50 ok.. so that is needed after all? that means my problem is actually with gcc and not with glibc? Feb 05 15:36:36 thats a clue.. let me see Feb 05 15:40:18 hrw: on the other hand, I looked at the glibc config.logs from my arm builds, libunwind support in compiler is also "no" there Feb 05 15:45:33 oh dear Feb 05 15:45:43 some opie luser is pm'ing me Feb 05 15:47:30 koen: poinson hatemail Feb 05 15:48:01 OMG MAEMO SUCKS!!!!!ONE!!! OMG OPIE!!!!!ONE!!! Feb 05 15:48:14 exciting Feb 05 15:48:27 koen ??? Feb 05 15:48:33 * koen mentions "dead" "undeveloped" "trolltech is evil" and some other stuff Feb 05 15:48:39 makes me want to by herbal enhancments Feb 05 15:48:54 chouimat: the executive summary of the pm Feb 05 15:49:05 * XorA sits back and waits for the fireworks to begin Feb 05 15:49:14 koen: maybe you could offer some gentle explanation of how the shift key works? Feb 05 15:49:29 lol Feb 05 15:49:57 :) Feb 05 15:49:57 koen I'm still enjoy qtopia 4.2 ... better than the 2.x series ... now I can have a nicer code base ... no more #ifdef for qt4/3/2 :) Feb 05 15:50:20 * hvontres|poodle wonders if there is a way to re-braodcast a pm session for our collective amusement :) Feb 05 15:50:35 hvontres|poodle: yes it's called pasting ;) Feb 05 15:50:39 o...lets not Feb 05 15:50:45 back to work cybermen Feb 05 15:50:59 chouimat: qt4 is pretty neat Feb 05 15:51:07 natezg: party pooper :) Feb 05 15:51:21 natezg: I want to but my todo list is empty and working (or trying to work) with a broken knee is fun Feb 05 15:51:55 hi CosmicPenguin Feb 05 15:52:39 chouimat: Ouch...did you forget Guido's payment last month :) Feb 05 15:53:51 hvontres|poodle: ice related ... Feb 05 15:54:59 chouimat: ahhh...ice...hmm that's that frozen water thingy, right ? Feb 05 15:55:06 chouimat: morning - hows the leg? Feb 05 15:55:16 * hvontres|poodle smiles at so cal "winters" Feb 05 16:01:41 CosmicPenguin: better but painful ... I have a nice apointment tomorrow morning with the orthopedic doctor ... Feb 05 16:04:33 * chouimat goes back reading BLFS book Feb 05 16:25:52 pretty old busybox in oe Feb 05 16:27:57 working on creating a ramdisk for my edb9302A clone: Feb 05 16:28:02 mtaht4: 1.3.0 is old? Feb 05 16:28:39 I'm looking for diffenences between creating and genextfs ramdisk image for flash and ramdisk and don't really see any Feb 05 16:28:58 So I tried to boot it directly, just to see what it does Feb 05 16:29:21 Detected it, but then just stopped and did nothing else Feb 05 16:29:41 should the bootstrap-image target boot to a shell? Feb 05 16:29:46 yes Feb 05 16:29:53 kk Feb 05 16:30:22 I suppose that question could have been shorter ;-) Feb 05 16:31:19 hi mickeyl Feb 05 16:31:48 hey Feb 05 16:32:00 hello everyone Feb 05 16:32:02 more specifically, will it boot to a shell with the boot arg console=ttyAM,57600 Feb 05 16:32:04 natezg: images will not differ. their format can differ (cramfs, jffs2, ext2, tarball) but not content Feb 05 16:32:23 natezg: bootstrap boot to login prompt Feb 05 16:32:35 serially? Feb 05 16:33:00 I'll decompress and image and look at it Feb 05 16:33:30 I have a little probléme to compile gpe-imagewith OE tools : http://rafb.net/p/FwnZT045.html Feb 05 16:34:52 http://rafb.net/p/FwnZTO45.html sorry :p Feb 05 16:35:45 I don't know what to do after that Feb 05 16:35:49 hrw: busybox 1.4.1 is out. Also, my angstrom build for the ep9302 just built 1.2.1 Feb 05 16:36:16 I mean, 1.3.1 is a whole month out of date! intolerable. :) Feb 05 16:36:18 hey mickeyl Feb 05 16:36:27 mtaht4: busybox project release more often then we can test Feb 05 16:36:36 Didn't used to be the case. Feb 05 16:36:45 Used to be a bunch of patches flying in loose formation. Feb 05 16:36:56 busybox add regressions faster than we can fix is more to the truth Feb 05 16:37:00 mtaht4: and some of us feel that busybox adds new bugs each time so we stick at 1.2.1 now Feb 05 16:37:49 well, I'm pleased to say that aside from the qemu madness, that everything I might use for my current ep9302 project just finished building on the x86_64. Good work guys! Feb 05 16:38:01 Now... will any of it be bootable... that's another question. Feb 05 16:38:56 is there an easy way to build a basic (busybox based) bootable image for qemu these days in bitbake? Feb 05 16:39:32 mtaht4: MACHINE=qemuarm Feb 05 16:39:40 so...my ramdisk silently fails to boot Feb 05 16:39:46 Aggh. You mean a rebuild from scratch? Feb 05 16:40:02 Any good debug options out there for OE? Feb 05 16:40:21 mtaht4: which distro you used? Feb 05 16:40:34 angstrom Feb 05 16:40:48 or do you mean FC6? Feb 05 16:41:36 mtaht4: if angstrom then change machine and it will build only some packages Feb 05 16:41:38 hi goxboxlive Feb 05 16:42:03 hi hrw Feb 05 16:43:07 thx Feb 05 16:43:07 hrw: Did you receive my PM I sent to you today around 1PM? Feb 05 16:44:21 Laibsch: nope Feb 05 16:46:01 does anyone solved the problem with base-files-3.0.14-r8 ? Feb 05 16:46:59 Delvin_fr: what problem? I build bootstrap-image Feb 05 16:47:03 i don't there is one Feb 05 16:47:04 update Feb 05 16:47:09 and rebuild base-files Feb 05 16:47:24 http://rafb.net/p/FwnZTO45.html Feb 05 16:47:34 hrw, ther's the error message Feb 05 16:47:42 that's been fixed since a couple of days Feb 05 16:49:42 ok i didn't make the mtn update in org.openembedded.dev Feb 05 16:50:25 I'm just starting with OE , mtn and bitbake, sorry :p Feb 05 16:51:00 Delvin_fr: OE is changing often so it is good to do mtn pull/update after hiting bug Feb 05 16:51:21 ok thanks Feb 05 16:56:17 trying to expand IMAGE_CMD options Feb 05 16:56:27 added IMAGE_CMD_ramd = "genext2fs -b 16384 -d ${IMAGE_ROOTFS} ${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.ramrootfs.ext2" Feb 05 16:56:39 to my distro configuration Feb 05 16:56:54 this fails with ERROR: IMAGE_CMD not defined Feb 05 16:57:23 Does the IMAGE_CMD addition depend on something else? I'm just following the format it bitbake.conf Feb 05 16:58:56 natezg: IMAGE_FSTYPES +="ext2" is better ;D Feb 05 17:01:42 crap.. I thought maybe there was some problems with the patches from cross-lfs, but that was not it; can't get past my unwind problemo Feb 05 17:02:28 hrw:yea I tried that, but I get that silent failure. I need to be able to move the image type around and change things, try to get some sort of response from the system Feb 05 17:03:05 I have a cirrus ramdisk that works with the oe generated kernel, so I know the kernel is ok Feb 05 17:03:08 05 18:14 < natezg> so...my ramdisk silently fails to boot Feb 05 17:03:11 that one? Feb 05 17:03:15 thats the one Feb 05 17:03:25 I would compare both rootfs content then Feb 05 17:03:33 hrw, just made an update and I still have the bug, what can I do ? Feb 05 17:03:38 natezg: try to boot with 'init=/bin/sh sh' Feb 05 17:03:38 k Feb 05 17:03:47 Delvin_fr: bitbake -crebuild base-files Feb 05 17:03:51 ok, good plan Feb 05 17:04:27 console=ttyAM0,57600, right? Feb 05 17:04:33 console=ttyAM,57600 won't work Feb 05 17:04:43 hrw, thx, it works now Feb 05 17:06:19 I was using ttyAM, changed that Feb 05 17:06:42 that alone doesn't fix it Feb 05 17:06:48 trying hrw init idea Feb 05 17:08:32 nope, still jus stops at RAMDISK: Compressed image found at block 0 Feb 05 17:08:52 command line = root=/dev/ram console=ttyAM0,57600 init=/bin/sh sh Feb 05 17:09:30 arm? Feb 05 17:09:32 natezg: arm? Feb 05 17:09:43 hrw:yep ep93xx Feb 05 17:10:06 natezg: kernel without EABI + EABI rootfs? Feb 05 17:10:37 hrw:mmm lost me, think I need to read something Feb 05 17:10:59 ep93xx kernel from OE? Feb 05 17:11:13 hrw:nope, too old Feb 05 17:11:25 hrw:took it from mainstream 2.6.20-rc7 Feb 05 17:11:44 hrw:had to too support my board platform Feb 05 17:11:51 natezg: pester koen to update OE one Feb 05 17:12:01 natezg: enable EABI in kernel config Feb 05 17:12:06 hrw:kk, thx Feb 05 17:12:10 CONFIG_AEABI=y Feb 05 17:12:10 CONFIG_OABI_COMPAT=y Feb 05 17:15:18 hrw:they were off, changed and rebuilding. Feb 05 17:15:50 ;) Feb 05 17:21:02 hrw:koen fyi if oe does get an update for 2.6.20-rc7 it will need the discontig mem patch from Buytenhek to work Feb 05 17:21:36 hrw:koen has a slight confict, posted fix to cirrux-linux list will probably make it into mainstream soon enough\ Feb 05 17:26:43 03koen 07org.oe.dev * rf18d11c9... 10/ (3 files in 3 dirs): dbus: add patch by Rob Taylor, inspired by Rob McQueen to fix waitpid problems Feb 05 17:27:49 natezg: freelist mangled your patch, could you put it online somewhere? Feb 05 17:31:53 yep, gimme a few Feb 05 17:32:13 koen,RP: are there any examples of building two versions of a kernel, the only dfference being the command-line ? Feb 05 17:32:27 hvontres|poodle: for one machine? Feb 05 17:32:34 koen: yes Feb 05 17:32:46 hvontres|poodle: collie-kernels iirc Feb 05 17:32:48 hvontres|poodle: not really, but linux-rp contains some magic for CMDLINE Feb 05 17:32:58 brb Feb 05 17:33:01 koen: dbus fixed then? Feb 05 17:33:39 XorA|gone: yes Feb 05 17:33:46 koen: woowoo Feb 05 17:33:49 ha! it was the glibc_eh patch from cross-lfs that broke my build! I should not have added it :> Feb 05 17:34:22 koen: ok, I'll take a look there some more. I am still trying to figure out a clean way to build one image that boots a machine from flash and one that can boot from SD :) Feb 05 17:34:48 XorA|gone: now hal and NM should work a bit better Feb 05 17:37:29 koen: cool Feb 05 17:45:20 koen:ftp://ftp.emacinc.com/dynamic-phys-offset-2.6.20-rc7.diff Feb 05 17:45:46 03koen 07org.oe.dev * r8cfe6d3e... 10/ (1 packages/linux/ep93xx-kernel_2.6.19+2.6.20-rc7.bb): ep93xx-kernel: update to 2.6.20rc7, needs a new defconfig and discontigmem patch Feb 05 17:46:34 * koen adds patch Feb 05 17:47:01 that patch applies to 2.6.20-rc7, not 2.6.19 is that ok? Feb 05 17:47:29 yes Feb 05 17:47:36 kewl Feb 05 17:47:47 2.6.19+2.6.20-rc7 = 2.6.19 + .20rc7 patch Feb 05 17:48:00 ah, thx Feb 05 17:48:51 cu Feb 05 17:51:52 mickeyl: http://projects.linuxtogo.org/plugins/scmsvn/viewcvs.php/trunk/marketing/GPEPE-announce-en.txt?rev=560&root=gpephone&view=markup Feb 05 17:55:01 03koen 07org.oe.dev * r24a6bbd3... 10/ (3 files in 3 dirs): ep93xx kernel: add patch to make some more boards work Feb 05 17:59:45 koen, any idea if any wireless usb dongles will work on EFIKA? Feb 05 18:01:09 hi Feb 05 18:02:13 i've made some change to one of the .bb files and i've added a custom configure flag that should be passed to ./configure; but when bitbake runs i think it skips the configure step and doesn't run ./configure with the new flag. how can i force it to reconfigure the application i'm trying to build ? Feb 05 18:03:04 i've used this to specify a custom flag: EXTRA_OECONF = "--disable-mcs-build" Feb 05 18:03:33 Ebli1: I do not know if that is the correct approach, but I usually clean a package if I want it to be rebuild completly, I use bitbake -c clean mypackage Feb 05 18:03:40 in config.log that flag is not present and the timestamp is a bit older than my last attempt. Feb 05 18:03:43 aha, i'll try that Feb 05 18:03:51 i tried -c configure but it didn't work :) Feb 05 18:04:17 no you can -c a whole package :) if you build it again after that it will go through all the steps, unpacking, configuring, etc.. Feb 05 18:08:45 so we are talking about GPE phone edition today? Feb 05 18:08:56 I want screenshots :) Feb 05 18:11:23 koen: if I do a svn update will that pull in the new ep93xx patches you just put in? Feb 05 18:13:18 no Feb 05 18:13:26 mtn update, however will Feb 05 18:13:32 k, thx Feb 05 18:13:46 mtn pull && mtn up to be exact Feb 05 18:16:29 does bitbake use the command line util tar to extract archives ? i've come across some archives that MC for one can read properly, it loses some files and displays the subfolders wrong Feb 05 18:16:55 and now i've noticed that one of the archives unpacked by bitbake is missing some files Feb 05 18:17:10 Ebli1: yes bitbake and base.bbclass of OE use the normal tar command line application Feb 05 18:17:51 aha, weird then ... i'll check a bit more :) Feb 05 18:18:56 Crofton: they should work, if they have drivers Feb 05 18:20:11 ok Feb 05 18:20:36 http://www.maniacchallenge.org/ Feb 05 18:20:48 hi, all! Feb 05 18:20:51 being difficult, I would like to use an efika, rather than a laptopo :) Feb 05 18:21:08 please could somebody look at bug 1846, it is Zire72 machine update Feb 05 18:22:25 how much is this board? Feb 05 18:22:29 (efika) Feb 05 18:22:34 $200 Feb 05 18:22:48 wow ... Feb 05 18:22:51 http://www.genesippc.com/efika.php Feb 05 18:22:56 efika is http://www.genesippc.com/efika.php Feb 05 18:23:03 Crofton found the page ;) Feb 05 18:23:06 ok Feb 05 18:23:21 figured I should make sure the bot knows for future reference Feb 05 18:23:38 Crofton could be usefull ... Feb 05 18:23:53 yeah Feb 05 18:23:54 good OE support :) Feb 05 18:24:22 ibot: efika is http://www.genesippc.com/efika.php Feb 05 18:24:24 koen: okay Feb 05 18:24:46 ~koen Feb 05 18:24:48 i heard koen is our regular image monkey Feb 05 18:24:58 ~Crofton Feb 05 18:57:53 mickeyl: what would you think of a browser-engine layer for GPE-mini-browser? Feb 05 18:58:38 SO I can abstract out the parts of the engine I need. So projects can use their html render (gtk-webcore, webkit, minimo, mozilla) without having to change the ui. Feb 05 18:59:08 webkit is good enough for everyone (I'm biased though) Feb 05 19:00:27 zecke: if there would be a nice GTK+ interface I would be using it :-) Feb 05 19:01:51 webkit is good enough for everyone (I'm biased though) Feb 05 19:11:22 koen: which machine builds well with angstrom? And is available in enough numbers to people to test? Feb 05 19:11:38 * philippe wants to start the GPE svn build images tonight... Feb 05 19:12:08 ep93xx, h2200, hx4700, c7x0, spitz, borzoi, h3900, efika Feb 05 19:14:45 03koen 07org.oe.dev * r5e37839a... 10/ (1 packages/libsndfile/libsndfile1_1.0.16.bb): libsbdfile: fix packaging Feb 05 19:16:20 03slapinid 07org.oe.dev * r30bf52d6... 10/ (4 files in 4 dirs): palmz72: Update machine config, and linux-hackndev-2.6_svn defconfig. Feb 05 19:19:23 koen: which device are you willing to reflash ? Feb 05 19:19:40 reflash? Feb 05 19:19:45 ipkg update ; ipkg upgrade Feb 05 19:20:05 hmm, ok true.... Feb 05 19:20:19 * philippe needs some more time to make h38xx work again. Feb 05 19:35:43 koen: lennert's freescale mail was nice to read Feb 05 19:35:57 did you read the follow up? Feb 05 19:36:15 yes Feb 05 19:36:17 "Let me state clearly that I recognize the folly and the stupidity Feb 05 19:36:18 of waiting this long to engage the community. I am utterly Feb 05 19:36:18 embarrassed to approach it this way. It is a travesty to let Feb 05 19:36:18 all that architecture and design and implementation churn Feb 05 19:36:18 and simmer so long in captivity, and then to expect the Feb 05 19:36:19 community to accept it." Feb 05 19:36:28 hmm Feb 05 19:36:31 where was this? Feb 05 19:36:52 Crofton: arm linux kernel mailinglist Feb 05 19:37:02 Lennert seems to rock Feb 05 19:37:04 ah Feb 05 19:37:32 zecke: lennert rocks Feb 05 19:37:40 zecke: it's a shame he isn't attending fosdem Feb 05 19:37:53 (he's boycotting mvista) Feb 05 19:38:36 koen: aeh, and FOSDEM is a MVista event? Feb 05 19:38:43 which architecture he refer to ? Feb 05 19:38:44 nope. Feb 05 19:38:51 zecke: the embedded room is Feb 05 19:38:59 I think we could hijack the MVista talk and show our blank ass Feb 05 19:39:03 koen: it is not an mvista event... Feb 05 19:39:09 it's a travesty to allow the mv dude to be in the panel after last years presentation Feb 05 19:39:27 I'm surprised harald didn't sue him for libel and slander Feb 05 19:39:33 koen: he has interesting contacts and he did not overdo the mvista stuff... Feb 05 19:40:15 if we go that way I cannot let OpenMoko and QTopia give a talk either. Feb 05 19:40:40 koen: it is a shame lennert is not coming, but we are free to not to visit mvista talks Feb 05 19:41:47 philippe: having TT talk about "opensource development" after keeping it closed for over a year is a travesty as well Feb 05 19:42:26 sorry to move the discussion, i have booked all for FOSDEM ... i came with another italian guy ... does you (OE) need some othe attendant ? can I help in some way ? :) Feb 05 19:43:19 koen: not all talks are of the same level... hard to predict beforehand. Feb 05 19:43:25 koen: "Now Trolltech has (soon) released Qtopia 4 Open Source Edition" Feb 05 19:43:40 gremlin[it]: hey, you could help us with getting our flyers together :} Feb 05 19:43:49 gremlin[it]: start a wiki site with booth duties... Feb 05 19:43:59 start? Feb 05 19:44:09 koen: do we have one already? Feb 05 19:44:11 :} Feb 05 19:44:16 http://www.openembedded.org/wiki/Fosdem2007 Feb 05 19:44:20 hopelessly lost in the Windows CE world Feb 05 19:44:53 zecke: the first 10 minutes of my lecture on signal was about putting (remove) in your mail address to fight spam Feb 05 19:45:33 zecke, perfect :) Feb 05 19:45:36 koen: you lost me :) Feb 05 19:45:39 * koen was underwhelmed Feb 05 19:45:41 do we know what hours we nee to staff the booth? Feb 05 19:45:50 zecke: I, again, wasted 10 minutes of my life Feb 05 19:45:57 Crofton: yes, we do Feb 05 19:46:19 Crofton: and we can ask phillipe. On saturday is a normal lecture day. So we can staff our booth at about 14h Feb 05 19:46:28 maybe we should put them on the wiki Feb 05 19:46:34 :) Feb 05 19:46:51 what we'll do on booth ? Feb 05 19:47:03 does anyone an arm device handy and can build on it ? Feb 05 19:47:26 Eblis: OE cross-compiles just to avoid that :-) Feb 05 19:47:43 philippe: i need to build something on arm so i can get a header for arm :) Feb 05 19:47:47 philippe: pssst: monos buildsystem sucks Feb 05 19:47:55 Eblis: sure Feb 05 19:48:08 maybe there's a way to run make with qemu or something ? Feb 05 19:48:17 koen: do you have an arm system ? Feb 05 19:48:20 * philippe runs away screaming.... Java 2.0 (otherwise called mono or C#) Feb 05 19:48:34 Eblis: multiple ones Feb 05 19:48:42 could you build something for me ? :D Feb 05 19:48:43 embedded track looks like some interesting stuff Feb 05 19:48:54 Eblis: sure Feb 05 19:49:16 can you download the mono sources ? :) are they too big ? Feb 05 19:49:25 Crofton: and did you find the mvista bias??? Feb 05 19:49:42 Crofton yes but is really a shame it's at the same time as X11 talks :( grrrr Feb 05 19:49:44 hehe Feb 05 19:50:31 crap Feb 05 19:50:37 I just found out about exmap Feb 05 19:50:53 gremlin[it]: it is pretty hard to match up everything for everyone... there were 300+ talks last year Feb 05 19:51:03 koen: inside mono/mono/mini you have to do a make cpu-arm.h :) Feb 05 19:51:14 for me it fails because it doesn't know an instruction on line 544 Feb 05 19:51:34 philippe: we all know MVista is shitty and that koen should be sometimes more quiet Feb 05 19:51:44 philippe the only main tracks i'm interested are kernel and X11 :( ... Feb 05 19:51:47 root@h2200:/data# wget http://go-mono.com/sources/mono/mono-1.2.2.1.tar.gz Feb 05 19:51:56 zecke, ROFL Feb 05 19:51:57 philippe: I like to watch bad movies, so I will bring popcorn and watch how MVista invented Linux Feb 05 19:52:14 gremlin[it]: then you will be glad I tried to be careful with Andrew Morton's talk. Feb 05 19:52:23 people say that about me also Feb 05 19:52:43 philippe sure !!! Feb 05 19:52:51 * philippe would like to talk to lennert about this Mvista bull Feb 05 19:53:24 philippe: you want to sign the petition? Feb 05 19:53:59 philippe, um the mm talk is the smae time as the one I want to see in embedded Feb 05 19:53:59 mvmd Feb 05 19:54:25 koen: which petition??? I would have liked somebody actually talked to me about those issues. Feb 05 19:56:04 well, a lot of big names aren't attending FOSDEM due to MV being on the panel *and* insulting people in his talk Feb 05 19:56:06 something to think about for next year Feb 05 19:58:28 Eblis: root@h2200:/data# tar zxf mono-1.2.2.1.tar.gz Feb 05 19:59:03 go inside mono-1.2.2.1/mono/mini Feb 05 19:59:04 make cpu-arm.h Feb 05 19:59:17 a file called cpu-arm.h should be created, that's the one i want :) Feb 05 19:59:25 my sd-card isn't that fast :) Feb 05 19:59:25 koen: those people are just childish if they cannot discuss this with me. Feb 05 20:00:17 back-stabbing -in the dark- politics don't help. They could easily have contacted me. Feb 05 20:01:48 afaik they did contact someone in the panel Feb 05 20:03:30 koen: the only thing that happend was that somebody made vague threats and did not even wanted to start a discussion. And this was only found out by the mvista guy through rumours.... I would not call that contacting. Feb 05 20:07:17 hmm, i just tried to build treecc for arm and i got an error. did anyone else have a problem with this ? Feb 05 20:07:41 * koen taps foot while mono is still extracting Feb 05 20:07:52 o_O Feb 05 20:09:34 ah, it's done extracting Feb 05 20:09:38 * koen ./configure Feb 05 20:10:01 hmm, you don't have to build mono altogether Feb 05 20:10:10 i only need the header file for arm cpu Feb 05 20:10:17 otherwise there is no makefile to make arm-cpu.h Feb 05 20:10:19 ah Feb 05 20:10:39 that's going to take forever :) Feb 05 20:22:04 Eblis: it needs some parts of mono, building that now Feb 05 20:22:13 ok Feb 05 20:22:43 man, this is cumbersome :-/ Feb 05 20:23:55 i've tried to build portable .net and it has 2 folders for libraries inside it: libffi and libgc. when the do_cofigure target of pnet starts it deletes the ./configure files inside those 2 folders. how could i stop that ? Feb 05 20:24:30 doh.. busybox does not build for me anymore... at least not for my mipsel target, thats odd Feb 05 20:24:36 * philippe wonders if there is something like a ./configure cache.... Feb 05 20:24:41 the question is "why do you want to stop that?" Feb 05 20:25:11 because the do_configure target fails when it reaches those lib folders; it recursively does ./configure in them apparently Feb 05 20:25:49 i copied the 2 files by hand while the target was busy processing and the target completed successfully. now i just need to create a .bb rule to do that :) Feb 05 20:26:04 staging/mipsel-linux/include/linux/ip.h:84: error: redefinition of 'struct iphdr' - thats part of kernel headers? Feb 05 20:26:08 isnt it? Feb 05 20:26:19 then i got to the arm part but treecc didn't build, so i did a mtn pull, mtn update and now i'm waiting ... Feb 05 20:27:27 hmm, how can that be broken? Feb 05 20:27:50 pasting log ... (maybe i'm doing something wrong) Feb 05 20:28:48 http://miranda.pastebin.ca/341729 Feb 05 20:29:10 why does it want to run the binary file ? Feb 05 20:29:20 because its buildsystem sucks Feb 05 20:29:40 :) Feb 05 20:30:00 no, seriously, that's the reason Feb 05 20:31:10 hmm thats really odd... if_tunnel.h includes if.h and then gcc complains about various redefinitions of some structs like ifmap and so on Feb 05 20:31:18 so treecc isn't fully ported ? Feb 05 20:31:33 but those headers are from the libc-headers pacakge I believe? so they should be ok? Feb 05 20:31:47 Eblis: it needs to be patched to not run an arm treecc, but the x86 one Feb 05 20:31:57 aha Feb 05 20:32:30 which needs a treecc-native, etc Feb 05 20:34:45 doh.. linux/if.h and net/if.h should not conflict, right? Feb 05 20:34:55 or is the problem with busybox 1.2.1? Feb 05 20:37:28 koen: what is the link to your web site? Feb 05 20:38:17 The utwente one Feb 05 20:40:29 goxboxlive: http://dominion.kabel.utwente.nl/koen/cms Feb 05 20:40:53 i found it, thx anyway Feb 05 20:41:03 google is a nice place Feb 05 20:41:23 does that make sense to anyone? http://www.deadlock.dhs.org/jin/busyboxfails.txt Feb 05 20:41:53 system headers bad? Feb 05 20:42:08 or is busybox doing something it should not? Feb 05 20:44:46 Jin^eLD: could you run it without -j8? Might give a more clear error output Feb 05 20:45:09 ok let me try... Feb 05 20:45:10 Jin^eLD: and is -j8 not a bit overkill? Unless you have an 8-way machine.... Feb 05 20:45:36 someone told me that that if one should use num cpus * 4 Feb 05 20:45:38 :) Feb 05 20:45:59 or did I read it somewhere? I do not remember to be honest Feb 05 20:46:06 I'm on a dual xeon Feb 05 20:46:29 I was told num cpu's + 1 Feb 05 20:46:45 * philippe also has dual Xeon + HT ==> -j5 Feb 05 20:46:52 I would have thought cpus * 2 as if a task is stuck on IO another can process Feb 05 20:47:26 actually it would be interesting to benchmark that.. I could try full OE build with various settings Feb 05 20:47:57 * philippe has the feeling that most of the time goes into ./configure and glibc locales Feb 05 20:48:13 :) Feb 05 20:48:31 thats probably where fast SCSI drivers could make a difference Feb 05 20:48:43 philippe: hehe, I have that too Feb 05 20:48:56 I used to have -j30 but I had a icecream cluster Feb 05 20:50:16 Jin^eLD: well I have SATAII HW raid5.... Seems for the locales qemu eats up one cpu Feb 05 20:50:25 * koen discovers a bug in bison Feb 05 20:50:37 well looks pretty much the same without -j8.. the busybox problem I mean Feb 05 20:51:09 philippe: I have a sata software raid 0 :) so nothing I could be happy about heh Feb 05 20:51:17 Jin^eLD: probably. But now we might have the defined first in... part Feb 05 20:51:56 maybe it has something to do with the libc headers I am using.. I took the version that cross-lfs suggested for mips Feb 05 20:52:06 in conjunction with glibc 2.5 Feb 05 20:52:26 that would be 2.6.19.2-01232007 Feb 05 20:52:59 Jin^eLD: the areca/tekram ARC-1210 is pretty good and *relatively* inexpensive Feb 05 20:53:16 And in-kernel drivers with GPL-ed utilities. Feb 05 20:53:21 thats nice.. Feb 05 20:54:11 thats with parallel make turned off btw: http://www.deadlock.dhs.org/jin/busyboxfails2.txt Feb 05 20:54:14 Jin^eLD: you are cross-compiling by hand now? Feb 05 20:54:46 no, OE Feb 05 20:55:18 I am fighting for a couple of days now to get a mipsel target to work Feb 05 20:55:26 normally there should be some kind of "defined first here..." kind of message. Feb 05 20:55:33 Unless the toolchain is bad. Feb 05 20:55:43 I would think so too, it usually tells where it is defined Feb 05 20:55:45 OZ 3.5.5 was announced on the www.openzaurus.org website, does anyone know if it's going to be derived from the .dev branch? Feb 05 20:55:53 oh.. toolchain is bad.. well I was fighting a lot to get going with glibc 2.5 Feb 05 20:56:24 it did build in the end Feb 05 20:56:31 not sure how to figure out the "badness" Feb 05 20:57:42 I'll try with -DDD maybe it will tell me more Feb 05 20:57:55 seems the toolchain was build with very different headers than the kernel has maybe??? Feb 05 20:58:02 * philippe is not a toolchain expert Feb 05 20:58:15 yes.. that is true.. the kernel is older Feb 05 20:58:23 but I thought that libc-headers matter, the package Feb 05 20:58:36 so kernel headers and the libc-headers (the sanitized ones) have to match? Feb 05 20:59:19 not really Feb 05 20:59:38 the libc-headers are there so you don't need kernel headers Feb 05 20:59:42 ok.. thats what I thought.. thats why I took the sanitized headers suggested by cross-lfs Feb 05 20:59:53 koen is right... Feb 05 21:00:42 well, so its not the different headrs vs kernel version.. Feb 05 21:00:45 what then? Feb 05 21:00:48 hmm Feb 05 21:01:11 if I change the headers package version, I will have to rebuild glibc and so on? Feb 05 21:02:17 is there some documentation on the "proper" way to add packages to an image with bootstrap-extra-redepends? Feb 05 21:02:41 I could try going back from the cross-lfs suggested headers to something that is similarly recent but already present in OE Feb 05 21:03:20 Eblis: I'm going to let the compile run overnight, ping me tomorrow for the results Feb 05 21:03:52 oooww Feb 05 21:03:58 it's going to take that long ? D: Feb 05 21:04:03 ok, thanks Feb 05 21:04:17 can anyone make anything of this error ? http://miranda.pastebin.ca/341806 Feb 05 21:04:17 dunno how long it's going to take Feb 05 21:04:28 but I'm going to read a book in bed and go to sleep :) Feb 05 21:04:37 later Feb 05 21:04:39 hheh, i should probably do that too Feb 05 21:04:56 koen: don't forget to forward me those mails.... I want to sort that mess out. Feb 05 21:05:08 I'll do Feb 05 21:05:11 'night all Feb 05 21:05:15 night koen Feb 05 21:05:40 night, thanks again Feb 05 21:06:17 Eblis: no clue about yout error Feb 05 21:06:44 you all go to bed too early over there :) Feb 05 21:06:55 :( Feb 05 21:07:08 Its 10:39PM over here Feb 05 21:07:12 still not time for bed Feb 05 21:07:31 Btw, anyone gonna videostream from FOSDEM? Feb 05 21:09:15 Kristoffer: they are supposed to stream some stuff... But I guess it will be announced on the website. Feb 05 21:12:26 philippe oki thx Feb 05 21:15:52 ep93xx ramdisk still not working after rebuilding with CONFIG_AEABI and CONFIG_OABI_COMPAT flags Feb 05 21:16:05 Cirrus ramdisk does work, comparing with it. Feb 05 21:16:20 (any suggestions welcome) Feb 05 21:19:36 heh.. it was indeed the libc headers version, I downgraded and now busybox builds Feb 05 21:21:12 Jin^eLD: good you found it... and that our guesses were right Feb 05 21:21:55 yes... Feb 05 21:22:03 I wonder if I need to recompile everything now Feb 05 21:22:05 just to be sure Feb 05 21:48:19 WOHOOO ... i think I've managed to port Portable .Net :) although it's use might be limited as one package fails to build Feb 05 21:48:32 so it's currently skipped Feb 05 21:49:31 night everyone Feb 05 21:59:36 I have a theory, someone tell me if I'm crazy Feb 05 21:59:58 comparing the cirrus ramdisk and the OE ramdisk Feb 05 22:00:16 the cirrus uses uclibc and my OE build is using glibc Feb 05 22:00:51 the builds are about the same size but only about 6M memory is left after the uClibc version boots Feb 05 22:01:19 could the glibc version be running out of memory loading it's libraries into memory and thus causinga silent failure? Feb 05 22:01:30 or is that not very likely Feb 05 22:01:53 (building with uClibc while theorizing) Feb 05 22:09:13 natezg: how much RAM do you have? Feb 05 22:09:23 * Eblis is very happy, preliminary tests with Portable .Net work :) Feb 05 22:10:27 night all Feb 05 22:11:48 cbrake:32M total Feb 05 22:12:57 uclibc is probably a good idea in any case, but I'm wondering if it's more than that and actually the failure point Feb 05 22:13:32 I'm testing right now Feb 05 22:14:00 but I was wondering how possible this was, what sort of memory usage comparisons are available, etc.... Feb 05 22:14:23 question: I've been trying to build lyx for an ARM akita on a i386 compiling system; but it attempts to link against the libraries installed in /usr/src instead of those in the staging area, and so I get a binary incompatability error. How can I force the recipe to look in the staging area for libs to link against? Feb 05 22:14:45 (this is using a recipe I got out of mtn, haven't tweaked it myself yet) Feb 05 22:16:51 elizabeththegrey: at dinner in dabney on Friday, it was suggested that people too heavily associated with Blacker be removed from comptroller positions. Feb 05 22:16:55 erm I'm sorry Feb 05 22:16:57 that was the wrong window Feb 05 22:17:02 very much so Feb 05 22:18:07 calmofthestorm:links against libraries or the kernel source in /usr/src? Feb 05 22:18:16 libraries Feb 05 22:18:31 on the commandline to gcc Feb 05 22:18:37 it gives /usr/src/libXt.so Feb 05 22:18:39 (among other things) Feb 05 22:19:42 calmofthestorm: look in your machine.conf and make sure PREFERRED_PROVIDERS are set to sane values Feb 05 22:20:20 where is machine.conf, same dir as local.conf? Feb 05 22:20:25 or in the org/config/etc dir? Feb 05 22:20:55 calmofthe storm your machine's .conf file is located in conf/machine and is located based upon you machine setting in local.conf Feb 05 22:20:56 org.oe.dev/conf/machines Feb 05 22:21:01 thx Feb 05 22:21:08 this'll take a sec, I need to pull Feb 05 22:21:08 lo Feb 05 22:21:47 calmofthestorm7:your name is very long Feb 05 22:22:06 I suppose it is Feb 05 22:23:18 Somebody dealing with fosdem there ? Feb 05 22:28:03 so what would a sane value for PREFERRED_PROVIDERS be? Feb 05 22:28:12 PREFERRED_PROVIDER_xserver = "xserver-kdrive" Feb 05 22:28:13 atm Feb 05 22:28:32 and PREFERRED_PROVIDER_virtual/kernel = "linux-openzaurus" Feb 05 22:28:38 if your having gcc probs I'd look at those, i.e. Feb 05 22:28:46 PREFERRED_PROVIDERS += " virtual/${TARGET_PREFIX}gcc-initial:gcc-cross-initial" Feb 05 22:28:59 well many other packages compile Feb 05 22:29:02 most do in fact Feb 05 22:29:07 gcc works Feb 05 22:29:19 it's just that gcc has something wrong passed to it on the command line Feb 05 22:29:33 o i c Feb 05 22:29:45 it's being told by oe_runmake to link against /usr/lib/<5 or so X/qt related things>.so Feb 05 22:30:30 but when I try passing QTDIR= to the configure script it fails Feb 05 22:30:40 because it can't gneerate a working qt executable, since it's in arm format Feb 05 22:30:45 at least that's one theory of mine Feb 05 22:31:33 getting out of my area... Feb 05 22:31:40 but a handy suggestion someone gave me Feb 05 22:31:53 was to use bitbake -i and "peek" at variables Feb 05 22:31:58 to see what there getting set to Feb 05 22:32:11 then focus down on the one thats wrong Feb 05 22:32:56 also I'd check your machines .conf for a tune-xxxx include at the bottom which may be adding gcc flags Feb 05 22:33:52 thaks Feb 05 22:33:59 good to have somehwere to start looking Feb 05 22:35:28 having trouble building uclibc Feb 05 22:35:45 says:Unable to determine version for kernel headers Feb 05 22:35:45 provided in directory /home/nathan/lab/IPAC-9302/tmp/cross/arm-linux-uclibc Feb 05 22:35:56 where is that coming from? Feb 05 22:49:12 ~snapshots Feb 05 22:49:40 hi all Feb 05 22:50:03 howdy Feb 05 22:50:12 how i do to rebegin an new projet with oe Feb 05 22:50:27 disaster: Follow the GettingStarted page in the wiki? Feb 05 22:50:31 ~gettingstarted Feb 05 22:50:33 from memory, gettingstarted is http://www.openembedded.org/wiki/GettingStarted Feb 05 22:50:41 my making system has break and now when i run bitbake angstrom-gpe-image Feb 05 22:51:06 the source are not downloading Feb 05 22:51:13 and do error Feb 05 22:51:40 disaster: In that case, you might just want to rm -rf tmp, and recreate conf/build/local.conf from the template Feb 05 22:53:02 i can use my old local.conf ? Feb 05 22:53:48 or i must also delete local.conf Feb 05 22:54:30 disaster: You can do either, I'd suggest starting afresh if things are seriously broken Feb 05 22:55:31 kk thx Feb 05 22:55:34 i test now Feb 06 00:24:10 I think beer (even Stella) is cheaper in the US than in Brussells .... **** ENDING LOGGING AT Tue Feb 06 02:59:57 2007