**** BEGIN LOGGING AT Fri May 30 02:59:57 2008 May 30 04:07:45 Getting build failures with the latest .dev: "sed: -e expression #1, char 10: unknown option to `s'" May 30 04:08:30 I've seen this discussed before but haven't seen any resolution. Any suggestions? May 30 04:56:15 bitbake usage question. May 30 04:57:36 i'm trying to build mpd (a music player daemon) under oe. the way its autoconf configure script checks for libOggFLAC is broken: it somehow ends up with -I/usr/include in the command line to the cross compiler. May 30 04:58:10 after making a change, i'd like to run autoconf and ./configure again, to see what the config.log will say. May 30 04:58:45 but i haven't so far managed to construct manually the intricate environment needed for cross-compilation, under which the autoconf and configure script run when i use bitbake. May 30 04:59:03 so i have to repeatedly bitbake -cclean mpd and bitbake mpd to regenerate everything. May 30 04:59:31 my question is, is there an easy way to make the loop of cleaning and rebuilding smaller or faster? May 30 05:03:52 jaredj: does bitbake -c rebuild not clean it? May 30 05:04:56 try bitbake -c clean May 30 05:05:07 then rebuild May 30 05:05:31 ah - rebuild cleans first May 30 05:07:20 .. wth?! May 30 05:07:36 mpd's configure script is getting run with --includedir=/usr/include May 30 05:08:26 this surpises you why exactly? those autoconf paths are the runtime locations, not build time May 30 05:08:51 run time locations, not build time May 30 05:08:56 brain fart May 30 05:09:36 so it seems the libFLAC aclocal bit is lacking in cross-compile goodness? May 30 05:09:48 some buildsystems do a -I@includedir@ in their makefile.in's, those usually need to be patched out. then our actual build time locations get passed in via the cflags, etc May 30 05:09:53 because it ends up passing -I/usr/include to the compiler when compiling conftest May 30 05:12:14 hmm, i already passed --with-libFLAC=${STAGING_DIR} May 30 05:14:22 no -I@includedir@ in makefile.in; in fact this one uses automake May 30 05:14:51 and we're not even getting to that step: the cross compile badness happens when compiling a conftest as part of the do_configure step May 30 05:15:03 (by we're, i mean i'm) May 30 05:16:21 i have, in the config.log, the command line used to compile the conftest. it contains a -I/usr/include May 30 05:16:46 i have, in libFLAC.m4, the place where the command line is originally written May 30 05:17:07 what i do not have is a way to figure out through which of the variable substitutions the -I/usr/include gets into the commandline May 30 05:17:49 if i could edit the configure script, put in an "env" command, and re-run configure without deleting everything May 30 05:18:02 i've had that happen on lots of things ( cross compile badness ). just have to edit the makefile and change -I/usr/include to your OE staging include directory May 30 05:19:31 the badness is not happening during the make May 30 05:19:39 it's happening during the configure May 30 05:20:34 does it have a config.in? edit that May 30 05:20:58 wicknix, he already is. the m4 macros are how configure.in is processed to produce configure. May 30 05:21:27 ahh, right. May 30 05:23:08 i think the -I/usr/include is in the CFLAGS, by the time line 26438 of configure happens May 30 05:23:18 that is almost no lead May 30 05:23:56 well, we know thats not in the cflags in bitbake, unless its inheriting native May 30 05:23:59 so check aclocal.m4 for macros that mess with the cflags May 30 05:24:03 bahah! May 30 05:24:09 LIBOGGFLAC_CFLAGS='-I/usr/include' May 30 05:24:36 and the other -I/usr/include is from MPD_CFLAGS May 30 05:25:08 er, the other -I/usr/include -I/usr/include -I/usr/include, which I haven't hit yet. May 30 05:25:36 gah, and OGG_CFLAGS, and VORBIS_CFLAGS May 30 05:27:13 ok i have leads. May 30 05:31:30 (those were in the config.log. i was missing --with-everything-and-its-brother=${STAGING_DIR}) May 30 05:38:14 thanks everybodee. now i'm missing a header file from the OggFLAC library; sleuthing May 30 05:41:01 oho. /home/jaredj/src/gumstix-oe/tmp/staging/x86_64-linux/bin/install -c -m 644 'seekable_stream_decoder.h' '/home/jaredj/src/gumstix-oe/tmp/work/arm May 30 05:41:05 v5te-angstrom-linux-gnueabi/flac-1.1.2-r5/image/usr/include/FLAC/seekable_stream_decoder.h' May 30 05:41:19 in my log.do_install. May 30 05:41:39 notice that the h file gets installed to the image directory May 30 05:41:57 but i need it to get installed into the staging directory so mpd can have that file to include May 30 05:43:53 the flac bb file says PACKAGES += "libflac libflac++ liboggflac liboggflac++" May 30 05:44:15 i tried to say DEPENDS = "libflac liboggflac" in my mpd_0.13.1.bb and it couldn't find those dependencies May 30 05:44:26 i had to say DEPENDS = "flac [otherstuff]" May 30 05:46:11 just found do_stage in the flac.bb May 30 05:53:40 wewt! it appears my mpd built May 30 06:45:05 morning May 30 07:15:26 Morning all... May 30 07:17:19 Got an annoying problem with my OE build that I am not sure how to handle. As of today 08:15 GMT 30-05-08 when building gpe-image (or any image I assume) it is tying to pull down fakeroot_1.9.4.tar.gz from the debian repositries. only Debian have pulled that from their mirrors and replaces it with fakeroot_1.9.5.tar.gz. May 30 07:17:47 Any ideas how to get around this problem. I tried adding a new .bb file but it just complained it was not part of the database.. May 30 07:20:01 For the record. mtn: updating along branch 'org.openembedded.dev' May 30 07:20:01 mtn: already up to date at cff1bfc5bbe28ffa46dfd581865e6537bb2220cb May 30 07:20:58 Is the version of the tree I am using. and DISTRO = "generic-uclibc" Is the distro I am building to May 30 07:23:23 any suggestions? How can I edit the DB to reflect the changed in the Debian Mirrors? May 30 07:47:03 DonAlex: you could probably just update the .bb file containing fakeroot and bump it to the new version.... or you could fetch the old version from http://snapshot.debian.net/archive/2008/04/15/debian/pool/main/f/fakeroot/fakeroot_1.9.4.tar.gz May 30 07:49:17 fatal__: Well that is what I tried to do.. Seems there is no version information inside the bb file so I assumed it used the file name to determine what was pulled. May 30 07:49:38 fatal__: Trouble is copying the bb file and renaming it threw up an error in the DB May 30 07:49:49 fatal__: could not account for the strange BB file.. May 30 07:50:45 fatal__: Where does it determine which mirrors to look at anyway.. because I notice it completely by passes my UK ones and that cannot be good for speed. May 30 08:06:06 * * OE Bug 4337 has been created by linuxrulez(AT)land.ru May 30 08:06:07 * * Compiling "gpe-image" target has linux-handhelds problem May 30 08:06:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4337 May 30 08:47:54 morning May 30 08:50:06 * * OE Bug 4338 has been created by udovdh(AT)xs4all.nl May 30 08:50:08 * * reboot doesn't reboot May 30 08:50:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4338 May 30 09:13:23 BBFILES = "/stuff/org.openembedded.dev/packages/*/*.bb /home/markus/oe/mystuff/*.bb" works but BBFILES = "/home/markus/oe/mystuff/*.bb /stuff/org.openembedded.dev/packages/*/*.bb" fails. May 30 09:14:02 it seems like you can't have your own path first in BBFILES, then bitbake will crash. May 30 09:14:08 is this a bug? May 30 09:15:28 return path_to_bbfiles[0][:path_to_bbfiles[0].rindex( "packages" )] <-- this line in base_get_scmbasepath(d): in base.bbclass will throw an exception. May 30 09:16:50 it assumes that oe-bbfiles is listed first and that dir contains a subdir called "packages" May 30 09:17:11 ..the split() on the line before also assumes there is no paths with whitespaces '\\ ' May 30 09:17:29 yes. May 30 09:17:54 can somebody tell me if this is suitable for a bug report? May 30 09:28:26 gm all May 30 09:35:40 markus: it certainly is, because even if its valid, a bug answer will tell later comers why this is done this way May 30 09:40:47 03koen 07org.oe.dev * rd76a97e8... 10/ (1 conf/machine/neuros-osd2.conf): neuros-osd2: start machine file for the neuros osd v2.0 machine May 30 09:51:06 * * OE Bug 4339 has been created by emil(AT)myhrman.name May 30 09:51:08 * * Value Error when "org.openembedded.dev/packages" is not listed first in BBFILES May 30 09:51:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4339 May 30 09:57:26 RP: ping May 30 10:01:56 johncylee: pong May 30 10:03:07 RP: I'm looking at the meta toolchain stuff... I think I need native opkg-cl in toolchain, but I don't know the proper way to add it. is it "opkg-sdk" in poky ? but opkg-sdk seems to build target binaries in OE.dev. May 30 10:03:34 johncylee: Yes, its opkg-sdk in poky May 30 10:04:13 johncylee: It sounds like something isn't in sync between poky and OE.dev May 30 10:04:54 RP: okay. I'll just check poky and port the necessary changes. is that okay? May 30 10:05:36 johncylee: yes, thats fine May 30 10:05:49 johncylee: Feel free to run anything by me if you want to confirm anything May 30 10:06:08 johncylee: I'm surprised the sdk stuff is out of sync :/ May 30 10:11:18 RP: hi May 30 10:11:27 RP: was looking at http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=history;f=packages/udev/udev_118.bb;h=20fef83d7474fc4c669b34d6c5636d73ebbdae84;hb=HEAD May 30 10:11:44 do you maintain udev anymore ? May 30 10:16:28 ant|work: maintain would be to strong a word. I help with it occasionally ;-)( May 30 10:18:03 hm .. http://rafb.net/p/SLLNNJ98.html. I'm not sure to understand what's wrong because the expat library do/should not depend on gcc, right ? May 30 10:24:24 I can't find where 'GCC_3.5' comes from. There is no host or OE generated gcc3.5, nor reference to it in the expat working directory :-/ May 30 10:24:36 hi all May 30 10:24:41 hi flo_lap May 30 10:25:02 flo_lap, how is linuxtag? May 30 10:28:21 RP: about udev, there is some mount problem (root: mount: mount point /dev/shm does not exist) May 30 10:28:55 but reverting to 0.92 fixes all (devices are created) May 30 10:28:56 Crofton|work: pretty nice May 30 10:28:59 ant|work: hmm :/ May 30 10:29:09 so should be a pure udev regression May 30 10:29:24 ant|work: Is there a later version that fixes it? May 30 10:29:25 Crofton|work: just answered your mail :) May 30 10:29:43 I'd lik to test with 115 but is not in tree May 30 10:30:05 Did you guys have a chance to see the talk? May 30 10:30:51 I'm guessing it is running Angstrom May 30 10:32:18 Crofton|work: unluckily not May 30 10:32:51 If you would like, I suspect I can get you the slides May 30 10:33:14 How has interest in OE been? May 30 10:35:12 Crofton|work: quite good - and even better knowing the fact that OE is interesting for a few people only May 30 10:35:54 that sentnece reads funny May 30 10:36:04 oe is only interesting to a few people? May 30 10:39:16 Laibsch: I see you uses "do_rm_work" for building. I remeber it was considered hazardous...seems ok now May 30 10:43:54 ah, koen is using it too May 30 10:51:58 hello. May 30 10:54:16 i just tried to play with oe. i'm using org.openembedded.stable as the base and MACHINE = "alix", TARGET_OS = "linux-uclibc" and DISTRO = "generic-uclibc". a bitbake helloworld stop after a while with: May 30 10:54:44 | /home/babel/oe/build/conf/tmp/cross/lib/gcc/i586-linux-uclibc/4.1.1/../../../../i586-linux-uclibc/bin/ld: crt1.o: No such file: No such file or directory May 30 10:54:45 | collect2: ld returned 1 exit status May 30 10:56:28 http://oe.linuxtogo.org/wiki/OeFaq is a dead link and google show no real help with the first matches. May 30 10:56:54 cjhbabel, use DISTRO = "angstrom-2008.1" May 30 10:57:08 the generic distros are not well maintained May 30 10:57:43 * Crofton|work needs to talk to mickey_away about either fixing generic, convincing someone to fix generic, or removing it ... May 30 10:57:47 uih, i thougt that generic and stable should just "work" :) May 30 10:58:03 right May 30 10:58:16 but, most people are using angstrom May 30 10:58:27 and we want generic to work :) May 30 10:58:41 actually, we want to rename it to minimal May 30 10:59:07 and build, but not have "features" like package manament etc May 30 11:00:57 our project so far uses a heavily modified toolchain from uclibc. i visited OE talk on LT and just want to try oe out. since i didn't have fancy hardware generic would match may needs. but i know this kind of problems, no time left to fix the "obvious" things for experts ;) May 30 11:02:11 and we know the website needs work May 30 11:02:38 I'd be wealthy if I had a euro for everytime I've had to say use Angstrom instead of generic May 30 11:02:47 Thansk for giving OE a try :) May 30 11:02:51 no problem i someone here is around to point a newbie to the right direction. May 30 11:02:57 if someone ... May 30 11:09:20 hi everybody May 30 11:23:25 i would like to test so minimal distros May 30 11:23:32 thesing: hi, how's it going with kernel dev? May 30 11:23:44 hhm ok I messed up with the moko makefile, hence my problem May 30 11:24:11 cyrilRomain: slap ;-) May 30 11:24:54 ant|work: quite slowly I'm mostly busy fixing things that upstream broke. May 30 11:25:09 thesing: the initramfs thing seems having lost the "momentum" it had, unfortunately... May 30 11:25:47 are you btw experimenting 2.6.25 kernels? May 30 11:25:57 markus, ping May 30 11:26:11 just curiois about size growth/shrink May 30 11:26:23 vs. 2.6.24 May 30 11:29:07 ant|work: no I work with bleeding edge git. May 30 11:29:32 I see May 30 11:29:56 everything else makes no sense. You end up fixing bugs that are already fixed upstream, your patches don't apply upstream May 30 11:30:31 it did just happen for WPA and USB_net fixed May 30 11:30:44 in .25 and backported to .24 May 30 11:34:25 Crofton|work: hello, i couldn't boot using that 2.6.25-kernel May 30 11:34:49 what happened? May 30 11:38:06 http://rafb.net/p/CauF1991.html May 30 11:38:17 Crofton|work: look at noobeans rafb-paste. May 30 11:38:35 but there are a lot of other errors. May 30 11:40:16 one thing about nfsroot May 30 11:40:43 look in /etc/network/interfaces May 30 11:40:57 # Wired or wireless interfaces May 30 11:40:57 #auto eth0 May 30 11:41:09 find these lines and make sure auto eth0 is commented out May 30 11:41:27 i think dhcp is disabled. May 30 11:41:49 for nfsroot, the kenel brings up theinterface May 30 11:41:58 but the auto line break the nfs server connection May 30 11:42:10 I do not think that s the probelm, but you should do it now May 30 11:42:13 before we forget May 30 11:42:20 okay... fixing... May 30 11:42:35 i have to do that every time we make a new root May 30 11:43:46 yeah May 30 11:43:53 sorry :( May 30 11:44:01 at least when you do nfsroot May 30 11:45:16 ok, fixed eth0 auto, still the same problem. May 30 11:45:52 see line 69 in the pastebin May 30 11:46:02 is that "normal" May 30 11:48:05 i have no idea :) May 30 11:48:53 but it seems like ethernet is mounting root ok May 30 11:48:54 question: it seems if I build foobar and its dependency is helloworld, there won't be helloworld.ipk, only foobar.ipk. is that the correct behavior? how can i make both ipkgs ? May 30 11:48:58 03john_lee 07org.oe.dev * rfcac20a3... 10/ (7 files in 3 dirs): May 30 11:48:58 pkgconfig-sdk_0.23: add pkgconfig-sdk from poky. May 30 11:48:58 * modify pkgconfig.inc to work with different pkgconfig versions. May 30 11:49:01 03john_lee 07org.oe.dev * r068af7c0... 10/ (1 packages/curl/curl-sdk_7.18.0.bb): curl-sdk: add from poky May 30 11:49:10 03john_lee 07org.oe.dev * r52b77d35... 10/ (1 packages/zlib/zlib-sdk.inc packages/zlib/zlib-sdk_1.2.3.bb): zlib-sdk: add from poky May 30 11:49:15 03john_lee 07org.oe.dev * r578cd827... 10/ (1 packages/tasks/task-sdk-host.bb): May 30 11:49:15 task-sdk-host: add pkgconfig-sdk and opkg-sdk May 30 11:49:15 * remove unnecessary RDEPENDS May 30 11:49:15 * port from poky May 30 11:49:35 basically, I am trying to identify the first error and see if we can fix it May 30 11:50:28 Crofton|work: if another kernel is used, it boots and a login shows May 30 11:50:42 can you compare the logs May 30 11:50:53 ok May 30 11:50:57 thanks May 30 11:51:01 comparing with 2.6.22 then? May 30 11:51:13 whatever is easiet May 30 11:51:22 the PHY message is the first strange one May 30 11:51:36 johncylee: bitbake -c buildall xyz ? May 30 11:51:50 I used the provided defconfig with only a few mods for the .25 kernel May 30 11:51:50 does it matter if i use uclibc or not? May 30 11:52:31 ant|work: thanks :D is it always like this? May 30 11:52:50 there are some magic BB_ variables you could edit May 30 11:52:51 did you? May 30 11:53:06 instead of giving the arguments May 30 11:53:48 also, does the fs match the kernel? May 30 11:53:59 it feels like you are loaing non-matching modules May 30 11:54:07 the vfat erros are confusing me May 30 11:54:26 johncylee: ask RP or Laibsch for details :-D May 30 11:54:43 Crofton|work: i probably am May 30 11:54:49 ant|work: thanks again :) May 30 11:56:03 I suppose I need to get the OSK out of the closet and try this locallay :) May 30 11:56:42 OSK? what is that? May 30 11:56:49 ~osk May 30 11:56:50 somebody said osk was the OMAP5912 Starter Kit, a $295 development board with 32/32, serial, ethernet, usb host/client, CF, jtag, and expansion connectors available from http://www.spectrumdigital.com/cgi/catalog.cgi?show_product=701875 or http://tree.celinuxforum.org/pubwiki/moin.cgi/OSK, or http://www.ti.com/omap5912osk May 30 11:57:00 thank u May 30 11:57:01 :) May 30 11:57:40 old references May 30 11:57:48 yeah May 30 11:58:11 but thanks anyway May 30 11:58:15 i could check them May 30 11:58:32 alot of OSK stuff needs updating May 30 11:58:37 like the OE build :) May 30 11:59:09 hmmm... i'm compiling... May 30 11:59:17 looks nice the board May 30 11:59:44 i catch the new links if you want :) May 30 12:00:21 thanks May 30 12:00:42 the first one: http://www.spectrumdigital.com/product_info.php?cPath=27_77&products_id=39&osCsid=5a12ba6ca1405ead24545d568a636921 May 30 12:00:57 and the second one: http://tree.celinuxforum.org/CelfPubWiki/OSK May 30 12:01:11 ibot, forget osk May 30 12:01:12 Crofton|work: i forgot osk May 30 12:01:49 osk is the OMAP5912 Starter Kit, a $295 development board with 32/32, serial, ethernet, usb host/client, CF, jtag, and expansion connectors available from http://www.spectrumdigital.com/product_info.php?cPath=27_77&products_id=39&osCsid=5a12ba6ca1405ead24545d568a636921 May 30 12:02:04 osk is http://tree.celinuxforum.org/CelfPubWiki/OSK May 30 12:02:08 ~osk May 30 12:02:30 ibot, osk is the OMAP5912 Starter Kit, a $295 development board with 32/32, serial, ethernet, usb host/client, CF, jtag, and expansion connectors available from http://www.spectrumdigital.com/product_info.php?cPath=27_77&products_id=39&osCsid=5a12ba6ca1405ead24545d568a636921 May 30 12:02:31 Crofton|work: okay May 30 12:02:43 ibot, osk is also http://tree.celinuxforum.org/CelfPubWiki/OSK May 30 12:02:44 Crofton|work: okay May 30 12:02:47 ~osk May 30 12:02:48 from memory, osk is the OMAP5912 Starter Kit, a $295 development board with 32/32, serial, ethernet, usb host/client, CF, jtag, and expansion connectors available from http://www.spectrumdigital.com/product_info.php?cPath=27_77&products_id=39&osCsid=5a12ba6ca1405ead24545d568a636921, or http://tree.celinuxforum.org/CelfPubWiki/OSK May 30 12:03:44 ibot, osk is also http://elinux.org/OSK May 30 12:03:44 Crofton|work: okay May 30 12:13:53 hello, May 30 12:13:57 any ideas about this? http://rafb.net/p/V5hWYk58.html May 30 12:14:59 udovdh: ideas yes, solution not yet May 30 12:15:15 it's failing during initramfs creation May 30 12:15:25 johncylee: BB_DEAULT_TASK = "buildall" probably does what you want May 30 12:16:21 RP: proably this is what most people wants... May 30 12:20:14 ant|work: So encourage your distro of choice to set it ;-) May 30 12:20:18 * RP shrugs May 30 12:21:00 are there any other configurations besides angstrom-2008.1.conf known to work? May 30 12:21:28 cjhbabel: yes but they're normally more niche setups May 30 12:22:04 cjhbabel: You could try poky May 30 12:22:40 (pokylinux.org) May 30 12:23:03 RP: buts easier to create a working one with more working examples. i need to create a x86 uclibc/uclibc++ system with the smallest possible disk footprint. May 30 12:23:24 ant|work, if you have things I can try, just post them here; May 30 12:23:36 cjhbabel: poky has a minimal image and supports uclibc May 30 12:23:38 I am about to leave for home so I can read them back when I get there. May 30 12:24:05 initramfs is the uclibc issue for angstrom? May 30 12:24:22 so if I build the ramdisk (for kernel) manually stuff could proceed? May 30 12:26:04 RP: i'm just looking at their website. our "old" toolchain was able to produce a really small system with kernel 2.6 + uclibc + dsl pppoe + openvpn + dropbear that fits on a 1.44mb disk. and i'm looking for something similar. May 30 12:26:33 udovdh: I think the initramfs stuff is commented out in .dev. you could try to uncomment it. May 30 12:26:43 classes/kernel.bbclass May 30 12:27:05 * * OE Bug 3435 has been marked as DUPLICATE of bug 3740 by spam(AT)campana.vi.it May 30 12:27:07 * * hal: Missing /var/cache/hald May 30 12:27:09 * * http://bugs.openembedded.net/show_bug.cgi?id=3435 May 30 12:36:20 03koen 07org.oe.dev * r726109e6... 10/ (5 files in 3 dirs): linux-omap2 git: add LED and NAND drivers May 30 12:38:25 udovdh: thesing is right: http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=commit;h=28dcab7459161d3451801d4d86c61bf9c9c73615 May 30 12:47:11 thesing: How usable is the initramfs stuff? May 30 12:47:56 * Laibsch would like to generally switch my Zs over to some kind of initramfs + kexec in flash and the "real" OS on HD/SD/CF May 30 12:50:09 Laibsch: initramfs-kexec is working for c7x0 and akita May 30 12:50:32 Laibsch: well it worked. but I lost my patches. I even got the initramfs-bootmenu stuff small enough. May 30 12:50:43 you lost your patches? May 30 12:50:48 what does that mean? May 30 12:51:00 you lost them locally before you could commit them? May 30 12:51:24 how does one use this? May 30 12:51:31 * ant|work is sure thesing posted this on the ML and on IRC -> look hentges logs May 30 12:51:32 "bitbake $whatimage"? May 30 12:51:42 well, that is no use May 30 12:51:57 the full logs are too unspecific May 30 12:52:02 I had to revert them for some reason and had a patch laying around to apply again later. But I don't find the patch any more May 30 12:52:42 Maybe " sudo rgrep $nameofthefile2patch /" will eventually find it? May 30 12:52:55 thesing: I'm pretty sure I have a copy at home May 30 12:53:03 or that May 30 12:53:35 I found it again. But I think it won't probably apply to current .dev May 30 12:53:37 thesing: can you give me an executive summary on the usage? or point me to a reference if it is already online somewhere? May 30 12:53:49 well, at least you got it May 30 12:53:57 It can probably be beaten into shape May 30 12:54:54 the usage of my stuff would be: bitbake linux-rp May 30 12:55:17 you would get a kernel with initramfs an one without and a packaged version of each of them. May 30 12:55:45 all you have to set is one variable. May 30 12:55:49 wasn't ANGSTROM_MODE = "uclibc" necessary? Wasn't a two step process? May 30 12:55:59 that was pauls solution May 30 12:56:19 I stopped my tests then... May 30 12:56:43 thesing: What variable is that? May 30 12:56:58 set where? local.conf? distro.conf? May 30 12:57:05 the second thing I have laying around is a hack on the initramfs images to make them use klibc utils. It's nearly finished. May 30 12:58:03 In some conf file. I doesn't really matter. May 30 12:58:12 Can you commit whatever you have that is subjectable to some kind of testing? May 30 12:58:36 Or at least provide a patch to pastebin? May 30 12:59:39 I can upload them to my side. May 30 12:59:51 But first mtn pull has to finish. May 30 13:00:03 ( I didn't pull for some months) May 30 13:00:25 IIRC the last klibc fix was the addition of the wc command May 30 13:00:31 thesing: hehe switch to git then because git pull is way faster ;) May 30 13:01:38 it doesn't really matter to me and as long as oe doesn't switch officially I won't too. ;) May 30 13:02:05 Laibsch: this was the starting thread http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-February/004278.html May 30 13:02:18 You know: never change a running workflow. May 30 13:03:18 ant|work: I remember wc is needed for initramfs-bootmenu but I didn't test it after I added wc to klibc utils. May 30 13:04:15 I confess I did'n test too...other breakages overwhelmed me... :D May 30 13:04:29 btw. the simple initramfs-kexec images should work fine. May 30 13:05:44 Laibsch: RP follow-up (got bored from pfalcon :D) http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-February/004323.html May 30 13:06:52 ant|work: Thank you for the links May 30 13:07:09 pfalcon striking again: http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-February/004467.html May 30 13:07:21 ;-} May 30 13:07:33 But there are no replies from thesing to any of the discussion, so it was not obvious how his work would be related to that of "don't use recursive bitbake" May 30 13:07:58 be patient, erst im Maerz May 30 13:08:30 http://lists.linuxtogo.org/pipermail/openembedded-devel/2008-March/004613.html May 30 13:10:27 this way why I had to revert my changes locally. I had to separate them from the do_deploy stuff. May 30 13:11:22 ant|work: But I suppose that should already be in .dev May 30 13:11:37 * Laibsch is utterly about the status May 30 13:11:57 thesing: what is left in left your local stuff that has not yet been committed? May 30 13:12:12 s/utterly/utterly confused/ May 30 13:13:09 wait a moment. May 30 13:13:26 Laibsch: http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=commit;h=28dcab7459161d3451801d4d86c61bf9c9c73615 just needs to be reenabled, failed intermittently May 30 13:13:35 locks problem? May 30 13:13:51 we should retry now on oestats May 30 13:14:31 RP: do we still have those lock problems? May 30 13:15:44 http://www.pastebin.ca/1034062 is the mechanism of building kernels with buildin initramfs May 30 13:16:19 Laibsch: Which lock problems? May 30 13:16:30 http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=commit;h=342be92e862f499a0cb0622a4a5abc96720c9b1f May 30 13:16:42 could have fixed these ? May 30 13:18:05 RP: sorry to ping you, I was just about to write a different message to you. We were talking about the lock problems that led to http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=commit;h=28dcab7459161d3451801d4d86c61bf9c9c73615 May 30 13:18:06 * * OE Bug 4340 has been created by  May 30 13:18:08 * * commit otavio's patches May 30 13:18:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4340 May 30 13:18:44 ant|work: I doubt the two are related May 30 13:18:56 Laibsch: Its likely the image rootfs class changes could have fixed that locking issue May 30 13:19:07 http://www.pastebin.ca/1034066 is my work on making initramfs-bootmenu use klibc May 30 13:19:08 OK May 30 13:19:11 but I'm not 100% sure since I lack context about the exact problem May 30 13:19:12 RP: Thanks May 30 13:19:14 hi May 30 13:19:22 RP: me too, a bit May 30 13:19:55 RP: was about failure building in parallel before koen deactivated the initramfs stuff May 30 13:20:20 bitbake was not finding updater.sh and /or angstrom-version May 30 13:20:39 ant|work, I mean this thing: http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2008-February/001920.html May 30 13:20:49 is that it? May 30 13:21:09 RP: and was "missing lock" or smthg May 30 13:21:19 then still: I am building an app, why need to build a kernel/ramdisk when unpacking the kernel would giev access to the same files? May 30 13:21:51 udovdh: sorry but you should ask pfalcon for linux-hh May 30 13:22:04 we are discussing linux-rp May 30 13:22:13 ok May 30 13:22:19 some of ...;-) May 30 13:24:19 udovdh: I think the kernel is build before some apps because of some dependencies and because the hh kernel wants initramfs (which doesn't work atm) your build fails. May 30 13:24:41 hmmm. May 30 13:24:49 so then I can't build apps on hh/oe/angstrom. May 30 13:24:56 I tried navit, gpsd and flite May 30 13:25:01 navit fails on this one May 30 13:25:06 * * OE Bug 4341 has been created by  May 30 13:25:08 * * provide a working initramfs and corresponding image definition May 30 13:25:08 flite and gpsd on something else May 30 13:25:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4341 May 30 13:25:21 (but that was in stable, now it is dev) May 30 13:25:42 in stable dbus gives me an error (bug is reported) May 30 13:25:50 so so far it is unusable May 30 13:25:52 udovdh: If I where you I would try to revert this 28dcab commit. May 30 13:26:00 Laibsch: nice to see you taking the hot potato in your hand and giving a new impulse ! May 30 13:26:07 when a few days ago using a `generic` distro i could build navit. May 30 13:26:21 thesing, how can I do that? May 30 13:26:28 commit to the kernel? May 30 13:26:45 I have experience with compiling an dpatching etc May 30 13:26:49 ant|work: Is this controversial? I know that pfalcon's stuff caused a bit of a commotion. Let's do it right this time and everybody will be happy May 30 13:26:51 but this specific environment is kinda new May 30 13:26:53 no change you copy of oe metadata. May 30 13:27:05 * * OE Bug 4342 has been created by  May 30 13:27:07 * * Default kernel for HTC Universal does not start May 30 13:27:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4342 May 30 13:27:46 look at the changes this commit does and revert them (urls above) May 30 13:27:59 so basically there is two issues: May 30 13:28:02 the reporting is unclear May 30 13:28:08 and the process doesn't help itself May 30 13:28:17 which is what bitbake is supposed to do May 30 13:28:19 right? May 30 13:28:25 at least one of them is true May 30 13:28:33 maybe both of them May 30 13:30:49 RP: what I wanted to ask you earlier is if you can think of a way to use oestats, but build against .dev and .stable (IOW, inherit oestats for .dev only since it is not in .stable) May 30 13:31:41 Laibsch: for now, use an overlay with oestats in it ? May 30 13:31:48 inherit in the distro config for example instead of local.conf where I suppose base.bbclass is already available May 30 13:32:18 Laibsch: An anonymous python function which adds oestats to the INHERIT variable when using .dev? May 30 13:32:52 RP: sounds cool ;-) Can that be used in local.conf? May 30 13:33:17 RP: do you have an example of a similar anonymous python function? May 30 13:37:56 udovdh: I built the 3 a.m. GPS programs for angstrom-2008.1-test. Pls look my last oestats May 30 13:38:07 Laibsch: I'm not sure it can be used in local.conf, it may need to be a class file May 30 13:38:26 ant|work, interesting. oestats? will have a look May 30 13:38:33 Laibsch: You can put class files into a class directory in the same way as local.conf is in a conf directory though May 30 13:38:53 grep OE for anonymous and you'll find examples of anonymous functions May 30 13:38:59 OK, thanks May 30 13:39:07 udovdh: http://tinderbox.openembedded.net/ May 30 13:39:21 thanks ant|work May 30 13:40:23 Laibsch: Do we really need a packaged kernel with build-in initramfs or is a zImage to flash enough? May 30 13:40:48 i686? need amr5vte or something May 30 13:40:59 Laibsch: about oestats: wouldn't it be good if the insane QA logs were logged too? May 30 13:41:07 udovdh: it's my host May 30 13:41:25 aha. I need for h2200 ipaq -> arm May 30 13:41:31 look at the right side of the window ;-) May 30 13:41:48 udovdh: is this a StrongArm ipaq? May 30 13:42:00 pxa255 or 250 or so May 30 13:42:08 hp ipaq h2210 May 30 13:42:24 udovdh: c7x0 == pxa255 May 30 13:42:25 polyonymous_, hi, long time no see ;-) May 30 13:42:31 Processor : XScale-PXA255 rev 6 (v5l) May 30 13:42:46 Features : swp half thumb fastmult edsp May 30 13:42:50 thesing: I guess a zImage is enough if that is easier. Feel free to make any adjustments and comment to the bug report May 30 13:42:59 udovdh: if its a pxa and you only want the package use corgie as machine. May 30 13:44:08 I guess I am missing something; this is a build-farm? May 30 13:46:22 no just some build status. May 30 13:48:10 ok, can select package and status May 30 13:48:13 so how get to that machine? May 30 13:49:50 Crofton|work: sorry i can't test it right now. glibc fails ( glibc_2.6.1.bb do_install failed ) and i dont know why. it's friday and stuff. May 30 13:50:08 so i can't compare the two right now, uclibc still works though May 30 13:50:10 ok May 30 13:50:18 I got the OSK out May 30 13:50:38 my arm is messed up atm, so I am slow crawling under the desk looking for a plug May 30 13:51:05 * * OE Bug 4343 has been created by mdarland(AT)pager.net May 30 13:51:07 * * syslogd init script fails May 30 13:51:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4343 May 30 13:51:16 ok, did anyone else here thing for a second he was talking about an ARM machine there? =) May 30 13:51:21 *think May 30 13:53:43 CoreDump: me :-D May 30 13:53:47 udovdh: it was just to show you which revision got built by me using .dev branch for a pxa255 cpu (c7x0 = my target) May 30 13:54:01 Laibsch: if we only want a zImage its simple. May 30 13:54:15 ah, didn't recognise c7x0 as pxa255 May 30 13:54:59 http://www.angstrom-distribution.org/repo/?action=details&pnm=gpsd shows a slightly older version May 30 13:55:16 * thesing doesn't remember why I wanted packaged initramfs-kernels. May 30 13:55:39 thesing, comment out the check... May 30 14:01:50 Crofton|work: i might have fooled you before. i'm not sure what i did test. don't worry about that rafb-paste. May 30 14:02:06 * * OE Bug 4344 has been created by udovdh(AT)xs4all.nl May 30 14:02:07 * * kernel (?) build as part of `bitbake navit` fails May 30 14:02:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4344 May 30 14:02:16 how so? May 30 14:03:47 03koen 07org.oe.dev * r545cc6bc... 10/ (1 packages/linux/linux-omap2-git/beagleboard/defconfig): linux-omap2 git: enable jffs2 filesystem May 30 14:04:04 i'm not sure what build the log shows. May 30 14:04:11 what build of the kernel May 30 14:04:19 and what rootfs May 30 14:09:57 markus, not sure what you are asking May 30 14:10:54 i'm not asking, i'm telling you not to trying to spend time understanding what i did May 30 14:11:11 ok :) May 30 14:16:02 RP: for now I am happy with http://rafb.net/p/K14xxB78.html for oestats May 30 14:16:19 anyone is getting opkg segfaulting? May 30 14:16:40 Laibsch: looks good May 30 14:16:55 and it seems to be working as expected, too May 30 14:19:36 always an added bonus May 30 14:20:55 ;-) May 30 14:22:14 anyone getting the following errors, ERROR: '/home/sudharsh/Projects/openmoko/openembedded/packages/linux/linux-openmoko_2.6.24+git.bb' failed ? May 30 14:22:57 oops...wrong channel..sorry :) May 30 14:32:03 What SRCREV_pn-opkg people are using? 4247 looks to segfault here May 30 14:35:33 03likewise 07org.oe.dev * r99e27c51... 10/ (1 packages/linux/linux.inc): May 30 14:35:33 linux.inc: Add device tree location for MPC8515E-RDB. May 30 14:35:33 Note that in 2.6.26, the device trees have been merged/changed. May 30 14:35:38 03likewise 07org.oe.dev * r3b41e4e7... 10/ (1 packages/linux/linux-rt_2.6.25.bb): linux-rt-2.6.25.bb: Updated to -rt4. May 30 14:36:50 otavio: Look in sane-srcrevs.inc May 30 14:36:59 that is what most people will be using May 30 14:37:05 it's 4247 May 30 14:37:51 Laibsch: I have a cleaned up version of the initramfs stuff now. I will test and commit if it works. May 30 14:38:08 thesing: GREAT! May 30 14:38:15 Are you talking, like "today"? May 30 14:39:07 depends on how much has changed in oe. I rebuild from scratch (with deploy/pstage not removed). May 30 14:39:12 morning May 30 14:39:17 But chances are good. May 30 14:41:20 thesing: Give a current patch and I can try and build an image as well May 30 14:41:40 re (netsplit victim) May 30 14:41:41 Laibsch: http://rafb.net/p/QHVxjF27.html if you want to test yourself. just set INITRAMFS_IMAGE to sth. like initramfs-kexec-image May 30 14:42:02 thesing: with deploy/pstage not removed the hanches of a clean build are LOW May 30 14:42:24 s/hanches/chances May 30 14:42:42 lately it spits out some nasty sed errors May 30 15:16:56 Laibsch: referring to oestats, how would you treat the QA logs? May 30 15:17:06 ignore like now? May 30 15:19:19 cyrilRomain: ping May 30 15:20:24 ant|work: pong May 30 15:20:48 bjour, do you happen to have some script for parsing the QA logs? May 30 15:21:26 ant|work: no May 30 15:21:59 I think the logs are too long to insepct May 30 15:22:07 ant|work: hmm what are you looking for? publishing QA result to oestats ? May 30 15:22:07 *inspect May 30 15:22:24 some sort of trigger, like elogs in gentoo May 30 15:23:03 ant|work: heh :) May 30 15:23:15 nobady seems to read them... May 30 15:23:53 I found a couple yesterday building gpsd May 30 15:24:18 with single packages it is easy to see them scroll May 30 15:24:39 ant|work: probably because nobody activate 'QA_LOG = "1"' in their conf file ? May 30 15:24:56 first reason is this :-D May 30 15:25:19 ant|work: but indeed if you don't look at the scrolling text, it's easy to miss them :-/ May 30 15:25:31 too easy May 30 15:27:33 hm the qa log file is generated only on QA issue May 30 15:28:04 so I'll think about a elog-like script that looks into temp directories for such qa log files May 30 15:28:13 ant|work: file an RFE against the tinderbox component to upload qa logs as well May 30 15:28:34 ok May 30 15:28:36 Laibsch: yeah better even :) May 30 15:29:07 that will need a bit of work though to add a way to sift through them, too May 30 15:29:24 so it will on jeremy having time for this and deeming it important enough May 30 15:29:32 BTW the right table, the one with machine, revision,..., is not well displayed on IE6, IE7, Firefox (too much right-shifted) May 30 15:30:35 this both on 1024x768 and 1280x1024 May 30 15:32:32 ant|work: jeremy is not around, so there is not much use to discuss that here now May 30 15:32:51 ok, let's open a second bug May 30 15:32:54 file a bug against the tinderbox component for things you think could be improved May 30 15:32:59 k May 30 15:34:21 Laibsch: sorry to bother you again - last question: can you recommend do_rm_work ? May 30 15:34:27 hmm the problem is that qa log file can be created even if a package does'nt fail. So maybe a 'Succeed but...' (or a better name) need to be added in case the package build but with some warnings May 30 15:35:11 ant|work: I'm not having any issues with it May 30 15:35:33 cyrilRomain: LEt's leave that up to jeremy May 30 15:35:54 My favourite would be a somewhat separate interface to those logs May 30 15:36:20 qa logs need not be fatal, but they are only created *if* a problem is present May 30 15:36:39 so, IMHO just a long list of qa logs uploaded would be all that is necessary May 30 15:38:58 Laibsch: you mean separating QA issue from build log? Hmm I would prefer seing QA issue within the current oestat interface, although I agree with you a separate interface to see QA issue only is a plus :) May 30 15:40:05 * * OE Bug 4345 has been created by andrea.adami(AT)gmail.com May 30 15:40:07 * * RFE: display QA_LOGS May 30 15:40:09 * * http://bugs.openembedded.net/show_bug.cgi?id=4345 May 30 15:46:06 * * OE Bug 4346 has been created by andrea.adami(AT)gmail.com May 30 15:46:07 * * oestats pagination of Build Information May 30 15:46:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4346 May 30 15:57:48 cyrilRomain: a small field of one char would be enough on screen (green - yellow - red) for oestats May 30 15:57:52 cyrilRomain: I also meant current oestats interface, but possibly a separate page May 30 15:58:11 but let's leave that to Jeremy to consider May 30 15:59:02 ant|work, Laibsch: ok May 30 16:05:46 I use a SQL statement like this CASE .status WHEN 0 THEN 'GREEN' WHEN 1 THEN IF (.samples>1,'YELLOW','MAGENTA') WHEN 2 THEN 'ORANGE' WHEN 3 THEN 'RED' END as colorstatus May 30 16:10:59 bbl May 30 17:28:58 sounds like the beagle at LinuxTag is running Angstrom :) May 30 17:39:24 ok May 30 17:39:29 let me call berensens May 30 17:46:24 Did someone got this segfault? http://pastebin.com/f186a9e30 May 30 18:04:09 http://www.flickr.com/photos/jadon/2535894759/in/pool-beagleboard May 30 18:06:09 very cool Crofton May 30 18:07:51 hey Khem ? May 30 18:08:24 The beagle is great stuff May 30 18:08:56 how does one add German locale data to a OE image? May 30 18:09:11 wading through all the provides stuff, and its not clicking yet ... May 30 18:37:01 otavio: No, opkg is working fine here May 30 18:37:08 What distro, what machine? May 30 18:37:27 hello all, is this the place to ask questions about bitbake, et. al? May 30 18:37:57 Laibsch: found May 30 18:38:33 jadam: yes May 30 18:39:41 okay, great. well I am trying to make a recipe for a package called soundmodem, which is normally a standard 'configure; make; make install' type package May 30 18:40:16 I am trying to use autotools to build the package in my recipe but it seems to do nothing in the configure and compile step, and errors in the install step because there's nothing there May 30 18:40:47 jadam pastebin the recipe you have May 30 18:40:56 * Laibsch suspects S May 30 18:41:38 m76287d8a May 30 18:41:53 http://pastebin.com/m76287d8a May 30 18:42:28 (I should mention that I'm obviously very new to creating recipes, etc.) May 30 18:43:09 don't worry May 30 18:43:21 What version of the software are you trying to build May 30 18:43:22 ? May 30 18:43:42 this is an .inc file and the soundmodem-0.10 recipe just defines PV="0.10" and PR="r0" (soundmodem's version is 0.10) May 30 18:46:16 does your log look like http://tinderbox.openembedded.net/public/logs/140928.txt May 30 18:46:18 ? May 30 18:47:37 no, I don't see any mention of automake, autoreconf, or anything... May 30 18:47:58 actually I noticed something very strange just now in the tmp/work dir. May 30 18:48:38 previously I had named the tarball something else (soundmodem_no_gtk) and it seems to be unpacking that tarball somehow instead of the new one that is properly named May 30 18:48:55 there is a soundmodem_0.10 dir but it only has a patches dir May 30 18:49:40 sounds like your S= var isnt set correctly May 30 18:49:48 oe needs to know where the sources unpack -to- May 30 18:49:51 I use the -crebuild option whenever I try to make the package, and I have rm'ed the tmp/work/soundmodem* dir to clean it out May 30 18:50:25 is there a way to completely clear out any cached information, tarballs, etc for a package? maybe I should start there May 30 18:50:47 kergoth: I thought at first, too, it was a problem with S, but there certainly is more May 30 18:50:53 nod May 30 18:50:54 kergoth: good to see you around May 30 18:51:01 for some values of "around" May 30 18:51:04 havent had my caffeine yet May 30 18:51:19 jadam: "bitbake -c clean soundmodem" May 30 18:52:54 You may also want to remove the source tarball in case that changed May 30 18:53:26 laibsch, yes it has... where is that located? or do you mean the one hosted on google code? May 30 18:53:41 no May 30 18:53:46 the one you downloade May 30 18:53:47 d May 30 18:54:12 a responsible upstream project does not change released tarballs May 30 18:54:26 but it looks like you are your own upstream, right? May 30 18:54:39 kergoth, gm May 30 18:54:52 hey crofton May 30 18:54:53 how's work? May 30 18:55:00 paid work is slow May 30 18:55:06 fun work is overwhelming May 30 18:55:13 ahh, I should explain this as well... soundmodem has a configuration program written in gtk+ that we don't want on the gumstix so I stripped out the gtk references in the configuration files and made my own tarball May 30 18:55:15 hehe May 30 18:55:18 need to turn some fun stuff into paid :) May 30 18:55:33 I've got an OMAP3 board to play with May 30 18:55:45 nice May 30 18:55:59 Have you heard of the Beagle Board project at TI? May 30 18:56:08 hmm dont think so May 30 18:56:24 http://beagleboard.org/ May 30 18:56:45 havent paid much attention to embedded lately, trying to get back into it. i should email steve johnson at ti and see how he's been, i always lose contact with folks May 30 18:56:46 heh May 30 18:57:08 wow, cute litlte board there May 30 18:57:27 only $150 eh, cant argue with that May 30 18:57:37 yeah May 30 18:58:06 laibsch: if I wget the updated tarball with the fixed path (but which still has the same name) it untars to the 'right' directory, i.e., soundmodem-0.10... but when I run bitbake it is still untarring to the old directory name of soundmodem_no_gtk_0.10 May 30 18:58:43 laibsch: is there a cached tarball somewhere? May 30 18:59:39 Crofton: what is the power consumption of the beagleboard? How much computing power does it have? May 30 18:59:53 ahh, I see it... /usr/share/sources May 30 18:59:59 Laibsch, rtfm :) May 30 19:00:06 I'm going to run some errands May 30 19:00:25 l8r May 30 19:03:11 laibsch: okay, now it's actually running through the configure step after deleting all the old tarballs from /usr/share/sources. since this is probably going to be an iterative process where I might have to fiddle with the source to get it to compile for the gumstix, should I actually start naming the tarballs with the revision number at the end and update the recipe, so there's no confusion as to what tarball I'm using? May 30 19:04:53 either that or apply patches on top of the downloaded source May 30 19:05:21 you can use OE to beat the source code into the shape you want May 30 19:05:33 rwhitby: Are you there? May 30 19:10:43 Crofton: It's not in the manual May 30 19:24:09 Anyone here? May 30 19:29:28 Laibsch: I just committed the initramfs-stuff. May 30 19:29:35 03thesing 07org.oe.dev * r1ebb8850... 10/ (1 classes/kernel.bbclass): May 30 19:29:35 kernel.bbclass: get initramfs stuff working May 30 19:29:35 To use this you have to INITRAMFS_IMAGE to the desired image. In addition May 30 19:29:35 the you must have cpio.gz in IMAGE_FSTYPES (at least for that image see May 30 19:29:35 packages/images/initramfs-kexec-image.bb for an example) May 30 19:29:52 thesing: GREAT May 30 19:29:54 Thanks a lot May 30 19:30:00 I'll try it out at once May 30 19:30:58 thesing: I can set INITRAMFS_IMAGE in local.conf, I suppose? May 30 19:31:05 what are possible values? May 30 19:31:21 initramfs-kexec-image, for example? May 30 19:31:31 "initramfs-kexec-image", for example? May 30 19:32:05 thesing: will this work on a poodle or is it a collie only feature right now? May 30 19:32:22 Laibsch: this is the one I tested. May 30 19:33:03 hvontres|home: it will work on all machines but you may have to tweak initramfs-kexec-image. May 30 19:33:50 all it does is to mount flash and kexec a kernel from it. May 30 19:33:51 thesing: cool. Sounds like a good project for this weekend May 30 19:34:17 thesing: Does it mount anything besides flash? May 30 19:34:28 IOW, can it boot an image on SD/CF/HD? May 30 19:34:54 no. that would be the job of initramfs-bootmenu-image (?) May 30 19:35:24 do we have that, yet? May 30 19:35:25 btw. current collie config + initramfs is to big to flash. We might need linux-rp-light or so. May 30 19:36:13 yes its there but its to big atm. I started to make it use klibc but I didn't finish. May 30 19:36:33 thesing: I guess we could strip out all the network stuff May 30 19:37:08 for initramfs-kexec-image thats enough. The image is only 100K May 30 19:43:53 thesing: bug 4341 is fixed, then, I guess? May 30 19:45:15 yes unless we want to package initramfsed kernels at some later time. May 30 19:45:32 but I don't see why we should want to. May 30 19:47:28 OK May 30 19:48:06 * * OE Bug 4341 has been RESOLVED (FIXED) by May 30 19:48:09 * * provide a working initramfs and corresponding image definition May 30 19:48:10 * * http://bugs.openembedded.net/show_bug.cgi?id=4341 May 30 19:50:23 hm. the klibc version of initramfs-bootmenu-image is 120K.. May 30 19:58:45 bye May 30 20:00:31 03Laibsch 07org.oe.dev * r8a6d2bb7... 10/ (3 files in 2 dirs): gtkmm: update to latest release 2.12.7 May 30 20:03:25 laibsch: after cleaning up tarballs and what-not, it's actually running through the configure step now :) May 30 20:03:42 great May 30 20:03:46 keep it up May 30 20:03:54 however, it fails when checking for the cross-compiler path: "cross compiler not found" May 30 20:04:25 any ideas? May 30 20:04:28 sounds like you have some errors in your makefile May 30 20:04:41 but I can't help you much with that now May 30 20:25:51 Laibsch, I was afraid of that, I'll see if koen has made a measuremnt May 30 20:26:07 I suspect it depends strongly on what you are doing May 30 20:27:02 of course May 30 20:27:12 but can you give me some ballpark figures? May 30 20:27:24 Is it at least as fast as the Xeon? May 30 20:27:26 ;-) May 30 20:27:51 Will it be 1W, 5W, 20 or 100W power consumption? May 30 20:28:07 well, 10W at worst May 30 20:28:13 since I have a 2A supply May 30 20:28:27 I suspect the video interface may consume a lot May 30 20:28:38 so, with the minimal 500mA supply it would be 2-3W? May 30 20:28:41 unbelievable May 30 20:28:44 I would guess much less when power control is used corefully May 30 20:29:00 that is why we need to make measurements :) May 30 20:29:13 sure May 30 20:29:17 since they target mobile devices, I susecpt you can run pretty low May 30 20:29:17 but sounds awesome May 30 20:29:31 but I am not sure how much power NEON intensive apps will use May 30 20:29:36 yeah May 30 20:29:43 how much CPU power would it be? May 30 20:29:56 pentium 2 300 MHz range? May 30 20:29:58 ask me in a few months May 30 20:30:00 or more? May 30 20:30:05 I'm not really sure May 30 20:30:09 ok May 30 20:30:25 I think it will be hard to make comparisons like that May 30 20:30:38 I've been looking for something like that to have a small LAN server May 30 20:30:45 not necessarily fast May 30 20:30:55 this the beagle? May 30 20:30:56 but full-featured as far as software goes May 30 20:31:05 low-energy May 30 20:31:27 and fast enough to serve the odd mediawiki or wordpress page May 30 20:31:43 right now I use a Via C3 600 MHz board for that May 30 20:32:01 yeah, that would be easy May 30 20:32:02 The hard drives consume quite a bit of course May 30 20:32:20 and the php stuff can be sluggish May 30 20:32:35 It would also get the size down considerably May 30 20:32:36 you would need to use usb hard drive May 30 20:32:45 or flash May 30 20:32:48 Well, doesn't it have 2G storage? May 30 20:32:57 2G would be plenty for me May 30 20:33:18 Anything bigger goes to the NAS May 30 20:35:40 I don't think it is 2G of flash May 30 20:35:56 that is what the manual said ;-) May 30 20:36:02 1G RAM May 30 20:36:06 2G flash May 30 20:36:12 hmmm May 30 20:36:13 at least that is what I understood May 30 20:36:16 I would re-read May 30 20:36:18 hang on May 30 20:36:33 let us pray to FSM that I hooked up power properly May 30 20:38:33 and a bit lart to the hardware design that would let you hook it up wrong May 30 20:39:16 anyone using madwifi on little endian xscale platforms? May 30 20:41:02 Crofton: http://www.beagleboard.org/uploads/Beagle_HW_Reference_Manual_A_5.pdf page 14 May 30 20:41:28 Fique: SlugOS/LE does... don't know that anyone has tried any recent builds with that, though. May 30 20:42:06 thanks, I haven't had any success on my own and was looking to talk to someone... i'll take a look there May 30 20:42:11 http://rafb.net/p/xNJeVJ73.html May 30 20:42:17 spake u-boot May 30 20:46:02 I'm going to check the manual May 30 20:46:14 need to stop typing my friggin arm is cramping May 30 20:49:33 hi Laibsch (and others) May 30 20:51:35 any reason why the configure step (using autotools) would spit out "cross compiler not found"? May 30 20:55:45 Laibsch, the number on page 14 are giga bits ... May 30 20:56:22 Ah May 30 20:56:34 Well, that reduces things quite a bit, doesn't it ;-) May 30 20:56:59 :) May 30 21:00:51 * XorA|gone wonders which arm Crofton moans about now May 30 21:01:47 my arm May 30 21:02:06 Crofton: but you have a desk of arms :-) May 30 21:02:09 I fell off my bike and scraped it May 30 21:02:15 an it is sore May 30 21:02:46 ah, make the others arms fall off things, only painful to wallet then May 30 21:03:03 heh May 30 21:03:10 I've avoided seeing the doctor May 30 21:08:31 clear May 30 21:33:34 This look familiar to anyone? May 30 21:33:34 ERROR: Required build target 'console-image' has no buildable providers. May 30 21:33:34 Missing or unbuildable dependency chain was: ['console-image', 'task-base-extended', 'kernel-module-zd1211rw', '${INITRAMFS_IMAGE}'] May 30 21:33:34 the kernel module comes from task-base May 30 21:33:34 urg did it again May 30 21:33:37 grr May 30 21:33:48 I am having focus issues need to go get beer May 30 21:34:15 Crofton: that was the old initramsfs breakage May 30 21:34:22 Crofton|work: I think thesing fixed it May 30 21:34:25 it just happened to me May 30 21:34:34 ehm... May 30 21:34:40 I just pulled ..... May 30 21:34:46 maybe he didnt May 30 21:34:52 :-( May 30 21:35:03 crap, I am going drinking May 30 21:35:08 sensible idea May 30 21:35:15 I wanted to run the build over night May 30 21:35:28 Crofton|work: happy beer-30 :) May 30 21:35:34 l8r May 30 21:35:41 kick thesing for me :) May 30 21:35:47 Crofton|work: commet out -do_builtin_initramfs in kernel.bbclass May 30 21:35:55 Crofton|work: I got that change locally May 30 21:36:06 I'll wait May 30 21:36:11 arm is cramping again May 30 21:36:22 just set INITRAMFS_IMAGE = "" May 30 21:36:29 ? May 30 21:40:18 urgh.../oe/build/tmp/angstrom/deploy/glibc/images/c7x0/initramfs-bootmenu-image-c7x0.cpio.gz 2.6MB May 30 21:43:48 Laibsch: please look at this grep-output http://www.pastebin.org/39869 May 30 21:46:33 can someone explain to me how I just did a fresh pull of the oe.dev tree and now my kernel build requires openzaurus-sa? May 30 21:46:48 when i'm not doing anything with zaurus May 30 21:54:29 xjqian: hi May 30 21:54:43 hi May 30 21:55:02 great work with navit, compiled flawlessy May 30 21:56:05 now I have to test it .. the udev regression did let me without touchscreen May 30 21:56:13 :-/ May 30 22:00:27 how about an easier question, could someone recommend how I search the dependency tree so I can find where openzaurus-sa is being flagged for build? May 30 22:07:30 great... I cant even make dependency graphs because of the unbuildable dependency May 30 22:08:08 Fique: I build for Zaurus but don't see the issue (yet ?) May 30 22:09:12 i'm sorry I'm not being clear May 30 22:09:17 I have machine set to cm-x270 May 30 22:09:31 but somewhere along the dependencies May 30 22:09:37 it looks for openzaurus-sa May 30 22:09:41 and says its unbuildable May 30 22:10:21 did a checkout of oe.dev, bitbake updated, local conf with only machine, distro set May 30 22:10:30 and virtual/kernel build target fials May 30 22:11:15 Fique: I see some builds http://tinderbox.openembedded.net/search/ May 30 22:11:31 hrw did it for that machine May 30 22:11:48 interesting May 30 22:11:55 ahh, it's Poky May 30 22:12:01 I can't understand how my setup is wrong then, will have to look deeper May 30 22:12:20 it was working just fine up till I did a fresh checkotu of oe.dev May 30 22:12:48 I don't see nothing dangerous at first sight May 30 22:12:51 http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=shortlog May 30 22:13:39 there has to be something simple that i'm missing May 30 22:13:44 thanks for the answers though May 30 22:13:49 ehm when was that last checkout? eons ago ? May 30 22:14:01 about 8 months May 30 22:14:07 then just been doing updates since then May 30 22:14:25 but since i'm trying to debug madwifi, I wanted a simple buildable image May 30 22:14:29 but now I cant even do that heh May 30 22:14:56 you could revert the update May 30 22:16:03 I don't think its the tree, I really think I have something messed up in my environ May 30 22:16:17 now I see a collie kernel unbuildable target May 30 22:16:33 just switching to DISTRO_TYPE=release May 30 22:17:47 well thanks again :) May 30 22:37:36 Is anyone alive? May 30 23:41:44 Hello, is anyone alive here? May 30 23:50:18 hi, all! is it normal having about 90 BogoMIPS on arm926ej-s based at91sam9260 at 180Mhz or something is misconfigured there? May 31 01:48:24 03Laibsch 07org.oe.dev * rc2448ed3... 10/ (3 files in 3 dirs): xkeyboard-config: initial commit with version 1.3. Based on work provided by Otavio. Partly closes 4340. May 31 01:48:29 03otavio 07org.oe.dev * r32cf7349... 10/ (1 packages/xfce/xfdesktop.inc): May 31 01:48:29 xfdesktop: improve packaging and include desktop files required for plugins to work May 31 01:48:29 * commit by Laibsch May 31 01:48:34 03otavio 07org.oe.dev * r07f17fd7... 10/ (1 packages/xfce/xfce4-panel.inc): May 31 01:48:34 xfce4-panel: improve packaging and include desktop files required for plugins to work May 31 01:48:34 * commit by Laibsch **** ENDING LOGGING AT Sat May 31 02:59:57 2008