**** BEGIN LOGGING AT Thu Aug 13 02:59:58 2009 Aug 13 04:25:42 What are people using for wireless configuration on touchscreen devices? Aug 13 04:26:00 Any good interfaces to wpa_supplicant that also do general network setup? Aug 13 04:28:04 grg: at Openmoko we started to experience with conman (from moblin) Aug 13 04:28:52 grepping for wpa in .bb files shows connman and networkmanager as the only two options Aug 13 04:29:03 connman Aug 13 04:29:03 i'll look into connman :) Aug 13 04:29:20 grg: hrw recently posted a screenshot of a Gtk+ based app to configure it Aug 13 04:31:31 zecke, hmmm i don't see anything Aug 13 04:31:38 got a link? Aug 13 04:33:35 http://marcin.juszkiewicz.com.pl/2009/07/16/hacking-ui-for-small-resolutions/ Aug 13 04:36:55 cool Aug 13 04:37:24 i'm blessed with an 800x600 display, so vga is not a problem Aug 13 06:53:53 morning Aug 13 07:16:49 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r2bc4ef8c97 10openembedded.git/ (5 files in 2 dirs): Add framework for generating board build directories and capability to build multiple boards Aug 13 07:44:03 03Koen Kooi  07org.openembedded.dev * r857baa5b0f 10openembedded.git/conf/distro/include/angstrom-2008-preferred-versions.inc: angstrom: prefer a version of python that actually builds Aug 13 07:45:05 03Steve Sakoman  07org.openembedded.dev * rca8509346c 10openembedded.git/ (conf/checksums.ini recipes/gnome/libgnomecanvas_2.26.0.bb): libgnomecanvas: add 2.26.0 Aug 13 07:53:59 03Koen Kooi  07org.openembedded.dev * r7d61ab869f 10openembedded.git/recipes/u-boot/ (u-boot-git/new-pinmux.patch u-boot_git.bb): u-boot git: update beagleboard pinmux file Aug 13 08:20:19 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r5b95b24f46 10openembedded.git/scripts/ (MAKEALL board_list.sh environment.sh make-local.conf.sh): Build boards under the 'boards' directory Aug 13 08:20:20 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r9c1acf9930 10openembedded.git/recipes/linux/linux-2.6.30.2/at91sam9g45ek/defconfig: Remove power management from sam9g45ek defconfig which breaks kernel build Aug 13 08:24:56 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r5ce9fa6dc2 10openembedded.git/scripts/MAKEALL: Do not comment away the real work\! Aug 13 09:01:44 03Koen Kooi  07org.openembedded.dev * rd193340b35 10openembedded.git/recipes/gnome/gdm_2.26.1.bb: gdm 2.26.1: add missing dep on xinput Aug 13 09:10:56 i see a bug for guile-native 1.8.2 (#5261) Aug 13 09:11:04 someone reverted this? Aug 13 09:15:22 is there any way to speed up OE on amd64, like with psyco on i386? Aug 13 09:18:29 how many BB_THREADS do you have? and how many "make -j 4"? Aug 13 09:20:27 I have 4 and 4, it's maxing my quadcore out Aug 13 09:20:50 the psyco speedup would be nice when I'm playing around with recipes and configs though (since re-generating the cache is the most slow task) Aug 13 09:21:48 and correct me if I'm wrong but I'm pretty sure the caching stuff is singlethreaded Aug 13 09:23:30 th1_: IIRC there are some bitbake tests to get a faster parser Aug 13 09:23:46 th1_: take a look to the bitbake git repo Aug 13 09:24:53 ok, I might later but right now I want a stable build :) Aug 13 09:50:34 th1_: then you've to wait ;) (or bought a 8-core machine) Aug 13 10:09:35 thebohemian: ping? Aug 13 10:09:38 zecke: ? Aug 13 10:09:44 mickey|sports: yes Aug 13 10:09:50 telco? Aug 13 10:10:00 (it's 12 MESZ) Aug 13 10:10:08 mickey|sports: didn't manage to get skye working..., can I follow on irc? Aug 13 10:10:29 hmm, k, then join #oe-meeting Aug 13 10:11:02 what is that meeting? Aug 13 10:11:08 should I join too? Aug 13 10:11:22 is that the meeting discussed on the ml? Aug 13 10:11:23 Gnutoo: if you are a member of the OE e.V. board Aug 13 10:11:26 in order to bootstrap the ev? Aug 13 10:11:28 no I'm not Aug 13 10:11:44 hi mickey|sports, zecke Aug 13 10:11:59 should I become one or is there enough members? Aug 13 10:12:15 hi pb__ Aug 13 10:12:16 Gnutoo: relax and wait :) Aug 13 10:12:21 ok thanks Aug 13 10:12:29 the e.V. exists, we are just not there to easily add new members :} Aug 13 10:12:32 Gnutoo: definitely join, but we have some bearocracy to fix Aug 13 10:12:39 beforehand Aug 13 10:12:46 ok Aug 13 10:13:03 were the status translated into english? Aug 13 10:16:37 pb__: could I talk you into looking at the posted glibc patch to use glibc-ports 2.9 when building glibc-2.9? Aug 13 10:16:57 zecke: yup, I can look at taht Aug 13 10:16:59 that, too Aug 13 10:20:29 could someone try to build guile-native? the 1.8.2 version doesn't compile to me Aug 13 10:30:29 zecke: Do you have an idea how to easily manipulate OE or bitbake to skip any hg/git/svn/cvs SRC_URI in do_fetch? Aug 13 10:31:25 Laibsch: skip? it tries mirrors first anyway? Aug 13 10:31:29 Laibsch: skip == fail? Aug 13 10:31:31 methril|work, what's your setup...that was my commit Aug 13 10:31:50 someone else had it failing Aug 13 10:31:51 zecke: I thought those SRC_URI wouldn't try a mirror? Aug 13 10:31:56 I've cloned his setup Aug 13 10:31:59 and it worked Aug 13 10:32:01 strange.... Aug 13 10:32:28 Laibsch: they at least used to try a mirror Aug 13 10:32:32 zecke: I want to skip, not fail. I'm in the process of setting up an OE source mirror and I did not want to waste the considerable HD space for the SCM sources. Aug 13 10:32:57 zecke: Oh, then we misunderstood each other yesterday. I asked and you said they don't. Aug 13 10:33:23 so, it make sense to mirror the tarballs generated from SCMs? Aug 13 10:33:33 Laibsch: sorry, we will try to get a mirrored checkout/repo to take load off the repo's Aug 13 10:33:45 Laibsch: little, only the ones we point to a given SRCREV Aug 13 10:33:48 ? Aug 13 10:34:46 Laibsch: sometimes, e.g. glibc 2.5 we built from CVS as there was no release. So this rev is pretty stable and it would make sense to mirror Aug 13 10:34:57 Laibsch: on the other hand, source that changes every day... Aug 13 10:35:10 OK, understood Aug 13 11:12:09 Laibsch, have dns changes gone through? Aug 13 11:12:33 let me see Aug 13 11:13:38 Crofton: yes, looks good. Thank you. Aug 13 11:13:56 I think, we will eventually have more than one IP for mirrors. Aug 13 11:18:46 Gnutoo: sorry i'm using oe.dev and shr from shr/import Aug 13 11:19:08 me too Aug 13 11:19:29 in the oe.dev repo with the shr conf? Aug 13 11:19:37 is what i'm doing right now :) Aug 13 11:23:51 brb, food Aug 13 11:27:26 03Koen Kooi  07org.openembedded.dev * r8ded6c1580 10openembedded.git/ (6 files in 4 dirs): Aug 13 11:27:26 linux-omap*: evm and beagle updates Aug 13 11:27:26 * the 'zippy' daughtercard now has working mmc, rtc and ethernet Aug 13 11:31:18 03Klaus Kurzmann  07shr/import * raec1d462f3 10openembedded.git/recipes/freesmartphone/libgsm0710_git.bb: Aug 13 11:31:18 libgsm0710: switch to SRCPV Aug 13 11:31:18 Signed-off-by: Klaus Kurzmann Aug 13 11:31:20 03Klaus Kurzmann  07shr/import * r2d549a6af6 10openembedded.git/recipes/freesmartphone/fso-gpsd_git.bb: Aug 13 11:31:20 fso-gpsd: switch to SRCPV Aug 13 11:31:21 Signed-off-by: Klaus Kurzmann Aug 13 11:31:25 03Klaus Kurzmann  07shr/import * rf875d62bbb 10openembedded.git/recipes/pyphonelog/pyphonelog_git.bb: Aug 13 11:31:28 pyphonelog: switch to SRCPV and adjust PV Aug 13 11:31:30 Signed-off-by: Klaus Kurzmann Aug 13 11:31:32 03Klaus Kurzmann  07shr/import * rd4710d39f7 10openembedded.git/recipes/freesmartphone/libgsm0710mux_git.bb: Aug 13 11:31:35 libgsm0710mux: switch to SRCPV Aug 13 11:31:37 mrmoku: spammer! :-) Aug 13 11:31:37 Signed-off-by: Klaus Kurzmann Aug 13 11:31:39 03Klaus Kurzmann  07shr/import * r0e9deaf1f7 10openembedded.git/recipes/freesmartphone/dbus-hlid_git.bb: Aug 13 11:31:42 dbus-hlid: switch to SRCPV and adjust PV Aug 13 11:31:44 Signed-off-by: Klaus Kurzmann Aug 13 11:31:46 03Klaus Kurzmann  07shr/import * r5dd097e304 10openembedded.git/recipes/freesmartphone/cornucopia.inc: Aug 13 11:31:49 conrnucopia.inc: switch to SRCPV Aug 13 11:31:51 Signed-off-by: Klaus Kurzmann Aug 13 11:31:57 03Klaus Kurzmann  07shr/import * r11e5b32aec 10openembedded.git/recipes/freesmartphone/mickeydbus_git.bb: Aug 13 11:32:00 mickeydbus: switch to SRCPV Aug 13 11:32:02 Signed-off-by: Klaus Kurzmann Aug 13 11:32:04 03Klaus Kurzmann  07shr/import * r9e11a63bfc 10openembedded.git/recipes/freesmartphone/mickeyterm_git.bb: Aug 13 11:32:07 mickeyterm: switch to SRCPV Aug 13 11:32:09 Signed-off-by: Klaus Kurzmann Aug 13 11:32:43 Ainulindale, yup Aug 13 11:32:55 and answer on core Aug 13 11:33:13 yup Aug 13 11:57:59 thebohemian: hey Aug 13 11:58:08 thebohemian: have you any idea about MIPS? cacao and mips? Aug 13 12:16:38 re Aug 13 12:31:44 i have linux and rs232 serial port. pins 2,3 are closed by jumper(RX, TX). I have a simple test program like while(1) {write(a); read(b); compare(a,b);}.... it is working until I remove jumper... then i'm getting read timeout errors. If I return jumper back I'm still getting read timeouts untlil test app isn't restarted. How can I do this without restarting app ? Aug 13 12:33:42 CruX|: run without flow control? fix your serial driver? Aug 13 12:34:19 zecke: this problem is on all serial ports Aug 13 12:34:32 i don't know how to fix it in userspace Aug 13 12:34:53 only close and open works here but that's ugly Aug 13 12:37:04 03Roger Monk  07org.openembedded.dev * r43100bfa8e 10openembedded.git/recipes/gnome/gnome-games_2.24.0.bb: gnome-games: fix paths to host SDL, bump PR Aug 13 12:38:05 CruX|: all means what? Aug 13 12:39:01 zecke: ttyS0 and ttyUSB0 on my laptop and ttyS0 on at91SAAM... atmel evaluation board Aug 13 12:39:43 CruX|: leaves one question unanswered... Aug 13 12:41:11 zecke: http://openpaste.org/en/16129/ here is my serial port setup not sure right now if it's ok Aug 13 12:42:40 CruX|: okay, no idea then Aug 13 12:43:09 hmm strange Aug 13 13:44:46 Hi. I am trying to build the minimal distro (task: bootstrap-image) on stable/2009 and I am having problems with the distro-feed-configs recipe. Can anyone help me? Aug 13 13:45:59 This is what I get: http://openpaste.org/16133/ Aug 13 13:48:07 JavaJim: PACKAGE_EXTRA_ARCHS not being set? what kind of x86 machine do you have? Aug 13 13:48:49 zecke: MACHINE = "x86" Aug 13 13:49:30 I have tried to set PACKAGE_EXTRA_ARCHS = "x86" but then I got a sanity check error. Aug 13 13:52:15 JavaJim: can you post bitbake -e -b your/distro-feed-configs.bb? Aug 13 13:53:22 JavaJim: for me on .dev it is "CONFFILES_distro-feed-configs="/etc/opkg/all-feed.conf /etc/opkg/x86-feed.conf" Aug 13 13:54:38 Hi, I'm trying to use OE to build an Angstrom image. Bitbake base-image fails with: ERROR: function do_compile failed Aug 13 13:54:38 ERROR: see log in /media/sdb/oe//angstrom-dev/work/armv4t-angstrom-linux-gnueabi/curl-7.19.0-r2/temp/log.do_compile.10458 Aug 13 13:54:53 hunternet93: yes, what does it tell? Aug 13 13:55:13 just a min... Aug 13 13:55:15 zecke: I am afraid I can't follow. Do you mean the output of "bitbake -e -b openembedded/recipes/meta/distro-feed-configs.bb" ? Aug 13 13:55:36 JavaJim: yes, and grep for ^CONFFILES Aug 13 13:55:52 hi all ! Aug 13 13:56:10 hi Aug 13 13:56:15 Ok, here's the error: arm-angstrom-linux-gnueabi-libtool: compile: ccache arm-angstrom-linux-gnueabi$ Aug 13 13:56:15 ../arm-angstrom-linux-gnueabi-libtool: line 976: ccache: command not found Aug 13 13:56:27 I guess I'll install ccache and try again. Aug 13 13:56:40 zecke: No match Aug 13 13:57:34 question: I want to compile a generic image (x11-image) for a x86_64 architecture (AMD phenom in this case). What MACHINE I need to define? Aug 13 13:58:27 zecke: "# expansion of CONFFILES_distro-feed-configs threw " Aug 13 13:58:49 zecke: Maybe that's why there is no match? Aug 13 13:59:14 JavaJim: yes, could you paste me the CONFFILES_${PN} line of the recipe? Aug 13 13:59:24 JavaJim: I have never used stable Aug 13 13:59:48 zecke: CONFFILES_${PN} += '${@ " ".join( [ ( "${sysconfdir}/opkg/%s-feed.conf" % feed ) for feed in "all ${PACKAGE_EXTRA_ARCHS} ${MACHINE_ARCH}".split() ] ) }' Aug 13 14:00:15 zecke: git diff HEAD -- ... showed not difference Aug 13 14:00:24 s/not/no/ Aug 13 14:01:17 JavaJim: do you want to diff stable to dev? this command will not do that Aug 13 14:02:24 zecke: erm, yes, that was my intention. But I am not familiar with git... Aug 13 14:02:38 JavaJim: I have asked you something more easy Aug 13 14:03:23 hunternet93: weird, OE should detect the presence of ccache and only use it then... in the back of my head I remember ccache problems on fedora/RHEL... Aug 13 14:04:56 zecke: What do you mean? I already pasted the CONFFILES_${PN} line... Aug 13 14:05:32 ccash wasn't in my $PATH, Aug 13 14:05:47 I added it and now it has another error.... Aug 13 14:05:59 JavaJim: right, sorry. the diff command you wanted was git diff branchA..branchB -- file :) Aug 13 14:06:04 but it doesn't show a diff Aug 13 14:06:30 I don't suppose there are any utilities/guides for taking files generated by "apt-get source " on debian and making bitbake recipes out of them? Aug 13 14:06:47 Here the error in the log: WARNING: "s3c24xx_udc_set_platdata" [arch/arm/mach-s3c2440/rx3000/rx3000_udc.ko] undefined! Aug 13 14:06:47 make[1]: *** [__modpost] Error 1 Aug 13 14:06:47 make[1]: Leaving directory `/media/sdb/oe/angstrom-dev/work/rx3000-angstrom-linux-gnueabi/linux-handhelds-2.6-2.6.21-hh20-r25/kernel26' Aug 13 14:08:12 zecke: I removed ${PACKAGE_EXTRA_ARCHS} from the CONFFILES_${PN} line and now it works. Not sure, if this is the right solution Aug 13 14:08:20 JavaJim: really, interesting... Aug 13 14:09:11 JavaJim: if you search for PACKAGE_EXTRA_ARCHS in the output of bitbake -e -b what do you see? Aug 13 14:09:49 hunternet93: it means that people having built rx3000 machine recently, and that you qualified to make it compile again :) Aug 13 14:10:10 ? Aug 13 14:10:17 What does that mean? Aug 13 14:11:44 zecke: http://openpaste.org/16136/ Aug 13 14:12:41 hunternet93: you have hardware that not many people own, this means things break when the kernel moves forward... Aug 13 14:12:57 hunternet93: but having the hardware qualifies you to fix it Aug 13 14:13:44 zecke: How do I fix it? I have no C experience... Can I write a kernel in Python ;) Aug 13 14:14:53 hunternet93: It is the perfect oppurtunity to learn. E.g. start grepping the kernel tree if this symbol still exists Aug 13 14:15:02 hunternet93: and afterward google on how it is called now Aug 13 14:15:18 JavaJim: really weird, I would blame bitbake or python... no idea Aug 13 14:16:14 zecke: OK, thanks anyway. Shall I report this somewhere or just ignore it? Aug 13 14:16:42 JavaJim: feel free to post to the mailinglist and ask... Aug 13 14:17:28 zecke: OK, will do. Thanks for your help. Aug 13 14:18:30 hunternet93: e.g. you can start by doing bitbake -c devshell virtual/kernel. this should open a gnome-terminal for you Aug 13 14:18:45 hunternet93: and then grep for the above symbol Aug 13 14:18:51 ok... Aug 13 14:19:57 03Klaus Kurzmann  07shr/import * re72e9fcae5 10openembedded.git/conf/checksums.ini: Aug 13 14:19:57 checksums.ini: add PackageKit-0.5.1 Aug 13 14:19:57 Signed-off-by: Klaus Kurzmann Aug 13 14:24:56 I'm at the terminal, but do I search the files for the symbol or do I search for a file with the symbol in it's name? Aug 13 14:25:38 hunternet93: grep searches content inside files... if you should search for a filename I would ask you to use "find" :) Aug 13 14:26:37 zecke: So I do "grep s3c24xx_udc_set_platdata *"? Aug 13 14:27:57 hunternet93: -rn for recursiv and to give you the line number too Aug 13 14:28:05 zecke: Ok thanks. Aug 13 14:28:38 It gives a bunch of files. What am I looking for? Aug 13 14:29:28 hunternet93: the definition... of the function... not the places were it is called Aug 13 14:29:36 hunternet93: it should be in arch/arm or such Aug 13 14:32:30 03Klaus Kurzmann  07shr/import * rb8585eeab0 10openembedded.git/recipes/python/python-pyrtc_git.bb: Aug 13 14:32:30 python-pyrtc: use SRCPV Aug 13 14:32:30 Signed-off-by: Klaus Kurzmann Aug 13 14:32:33 03Klaus Kurzmann  07shr/import * r6eb257c61b 10openembedded.git/recipes/freesmartphone/libframeworkd-glib_git.bb: Aug 13 14:32:33 libframeworkd-glib: use SRCPV Aug 13 14:32:33 Signed-off-by: Klaus Kurzmann Aug 13 14:32:39 zecke: I found the function... Now what? Aug 13 14:34:51 03Klaus Kurzmann  07shr/import * r467ec404b0 10openembedded.git/recipes/freesmartphone/fso-gsm0710muxd_git.bb: Aug 13 14:34:51 fso-gsm0710muxd: use SRCPV Aug 13 14:34:51 Signed-off-by: Klaus Kurzmann Aug 13 14:36:21 I found the function s3c24xx_udc_set_platdata(struct s3c2410_udc_mach_info *pd). What do I do now? Aug 13 14:40:32 zecke? Aug 13 14:43:16 guess I'll google it sometime... Thanks for pointing me in the right direction. Bye! Aug 13 14:44:07 oh Aug 13 15:00:44 * CoreDump just learned about the total awesomeness of VPNs Aug 13 15:01:32 CoreDump: watched CNN? Aug 13 15:02:14 better, I set one up between work and home Aug 13 15:02:24 03Sebastian Krzyszkowiak  07shr/import * re6cf32c744 10openembedded.git/recipes/packagekit/ (2 files in 2 dirs): packagekit: fix segfaults when hald is not running Aug 13 15:12:49 03Koen Kooi  07org.openembedded.dev * rdd641b8a0b 10openembedded.git/recipes/clutter/ (clutter-0.9_git.bb clutter-gst-0.9_git.bb): clutter-0.9: move SRCREV a bit forward so that GLES builds again Aug 13 15:14:53 03Khem Raj  07org.openembedded.dev * r94a9ba7988 10openembedded.git/: Merge branch 'org.openembedded.dev' of git@git.openembedded.org:openembedded into org.openembedded.dev Aug 13 15:14:54 03Michael Smith  07org.openembedded.dev * r4c5f1096fa 10openembedded.git/recipes/pciutils/pciutils_3.1.2.bb: (log message trimmed) Aug 13 15:14:54 pciutils: fix generation of pci.ids.gz Aug 13 15:14:54 Set ZLIB=yes to prevent auto-detection by the lib/configure script, Aug 13 15:14:56 which only looks in /usr/include. Aug 13 15:14:58 Ship pci.ids in pci-ids-uncompressed for users of old hal versions. Aug 13 15:15:00 Pass PREFIX to Make and use the install-lib target to save work. Aug 13 15:15:02 Signed-off-by: Michael Smith Aug 13 15:21:49 arg Aug 13 15:22:04 kergoth: argh? Aug 13 15:22:46 i hate it when i miss something obvious and have to go back and reimplement a good chunk of what i was working on Aug 13 15:25:11 kergoth: yeah life sucks :) Aug 13 15:25:17 hehe Aug 13 15:25:42 kergoth: I'm still sitting on my SCCP code... had to rewrite size handling of buffers... Aug 13 15:25:55 :( Aug 13 15:32:40 kergoth: is there a known issue with the current git tree? tcp_wrappers_7.6.bb is failing Aug 13 15:33:11 if you expect people to know, off the top of their head, about every of the 6,000+ recipes that might be failing to compile, you're mistaken Aug 13 15:33:26 kergoth: no, but I figure people building git current might have some idea, that's all Aug 13 15:33:57 tcpwrappers' libwrap can't find innetgr and yp_get_default_domain, I was going to disable the tcpwrappers recipe but things like ssh seem to want it too Aug 13 15:35:21 tzanger: which ver of ssh are you building? Aug 13 15:35:36 msmith_: um, let me check. it's whatever is default in the current git tree Aug 13 15:36:07 openssh though, right? Aug 13 15:36:12 yes, correct Aug 13 15:37:00 i've built openssh and it's not doing anything with libwrap. Aug 13 15:37:15 I wonder if maybe the configure script is finding something on your host system that it shouldn't, and then trying to link to it. Aug 13 15:38:09 hmm Aug 13 15:39:06 the openssh-3.8p1 recipe seems to want it Aug 13 15:39:12 tzanger: using uclibc? Aug 13 15:39:26 lsh, nfs-utils and tftp really Aug 13 15:39:28 zecke: I believe so yes Aug 13 15:39:45 tzanger: well, uclibc has no yp support... you need to patch it out Aug 13 15:39:52 aha Aug 13 15:40:08 thank you, that's the kind of insight I was hoping for :-) Aug 13 15:41:24 tzanger: keeping this building with uclibc is kind of a struggle... upstream doesn't test with it Aug 13 15:41:40 * tzanger nods and makes a note :-) Aug 13 15:42:08 tzanger: if you could do a patch it would rock, then I could still the approach and use it on libpam :) Aug 13 15:42:33 :-) I'll see what I can do Aug 13 15:51:14 * kergoth sighs Aug 13 15:55:56 * kergoth realizes he was stupid with the prototype staging reimplementation. since he used tarballs, there's no tracking of indirect rdeps... automake-native needs autoconf-native, but autoconf-native rdeps on m4, and my staging population bit didn't end up pulling in m4-native Aug 13 15:56:11 * kergoth shakes head and starts from scratch Aug 13 15:56:59 i was trying to keep the packaging level bits out of it, so the user could package/repackage however they like going from the build output archive, itd be indepednent of those chagnes, so they woudlnt need to rebuild the recipe to re-package.. Aug 13 15:57:06 * kergoth grumbles Aug 13 15:58:38 kergoth: now? Aug 13 16:01:53 * kergoth has no idea what to do now.. time to get more caffeine Aug 13 16:05:07 night Aug 13 16:05:45 zecke: gn Aug 13 16:22:17 03Sebastian Krzyszkowiak  07shr/import * r40f2805c2f 10openembedded.git/recipes/tasks/task-shr-feed.bb: task-shr-feed: add xcompmgr Aug 13 16:35:19 Any idea why FILES_${PN} += "some_files" is ignored and FILES_${PN}-doc += "some_files" works as expected? :/ Aug 13 16:37:12 JaMa: I'd suggest running bitbake -e to check if things are set the way you think they're sent. bitbake -e | grep FILES_ | less or something Aug 13 16:37:16 s/sent/set/ Aug 13 16:40:25 kergoth: you're right .. FILES_pisi is default FILES_pisi="/usr/bin/* /usr/lib/* /usr/lib/python2.6/*" i guess its because I inherited setuptools and it replaced FILES_pisi? Aug 13 16:40:59 most likely, yeah. maybe you just need to do your += after the inherit rather than before? :) Aug 13 16:41:16 (ordering is a factor, inherit puts the inherited bits *right there* in your file) Aug 13 16:41:21 I tried to move inherit up, but its the same Aug 13 16:41:56 wow now it looks good Aug 13 16:42:57 kergoth: thanks Aug 13 16:43:01 np Aug 13 16:43:11 bitbake -e is your friend, i use it *all* the time to check sanity Aug 13 16:45:10 i was insane enough to try also specify FILES_pisi instead FILES_${PN} and interestingly the first didn't work even after reordering inherit up Aug 13 16:45:39 oh yeah, thatll bite you Aug 13 16:45:53 its because variable expansion in variable names isn't immediate, its lazy Aug 13 16:46:11 it happens later on, and when that happens, it can overwrite the existing variable that was already expanded, if you see what i mean Aug 13 16:46:20 so the FILES_${PN} will end up overwriting FILES_pisi Aug 13 16:46:40 yes now i see :) Aug 13 16:46:43 if you do INHERIT += "recipe_sanity" and then bitbake -c recipe_sanity pisi, it will warn you if one of those overwrites is happening Aug 13 16:46:56 (among other checks) Aug 13 16:49:33 i wonder what i did with that prototype class that i was using to generate a git repo from a recipe Aug 13 16:49:35 hrmph Aug 13 16:49:38 too many trees Aug 13 16:55:44 * kergoth sighs Aug 13 16:58:06 i wonder if its worth it to try to keep the staging aspects independent of runtime packaging.. i thought it was, but .. hrmph Aug 13 17:07:23 anyone building meta-toolchain for a ppc603e based board? I haven't seen it build for efika/lite5200/n1200/storcenter machines - seems to fail on libtool-sdk. Doesn't seem like there are any maintainers of boards with the ppc603 cpu Aug 13 17:20:53 03Klaus Kurzmann  07shr/import * rcb0cbdc131 10openembedded.git/recipes/tasks/task-shr-feed.bb: Aug 13 17:20:53 task-shr-feed: add man and man-pages Aug 13 17:20:53 Signed-off-by: Klaus Kurzmann Aug 13 17:20:55 03Klaus Kurzmann  07shr/import * r642ceedec0 10openembedded.git/recipes/freesmartphone/libfsobasics_git.bb: Aug 13 17:20:55 libfsobasics: use SRCPV and update PV Aug 13 17:20:56 Signed-off-by: Klaus Kurzmann Aug 13 17:21:00 03Klaus Kurzmann  07shr/import * r5a9b852111 10openembedded.git/recipes/freesmartphone/libfsoframework_git.bb: Aug 13 17:21:03 libfsoframework: use SRCPV Aug 13 17:21:05 Signed-off-by: Klaus Kurzmann Aug 13 17:21:07 03Klaus Kurzmann  07shr/import * re4ec9ae889 10openembedded.git/recipes/freesmartphone/libfsotransport_git.bb: Aug 13 17:21:10 libfsotransport: use SRCPV Aug 13 17:21:12 Signed-off-by: Klaus Kurzmann Aug 13 17:21:47 tharvey: whats the error Aug 13 17:27:27 khem, Applying patch cross_compile.patch Aug 13 17:27:27 can't find file to patch at input line 11 Aug 13 17:29:23 I've got an earlier snapshot of OE that I can build meta-toolchain for on a custom machine config for gcc-4.3.1 but run into issues linking with libstdc++ with undefined refs to __gcc_qadd@GCC_4.1.0, __gcc_qsub@GCC_4.1.0 and __gcc_qmul@GCC_4.1.0 - no idea where the GCC_4.1.0 comes from Aug 13 17:29:45 tharvey: thats just symbol versioning Aug 13 17:29:48 dont worry Aug 13 17:31:48 don't worry? linker fails to link because of the unresolved symbols Aug 13 17:32:03 this means you have soft float but proabably its looking for 128 bit long double Aug 13 17:32:05 support Aug 13 17:33:30 is there a compiler switch to disable that? Aug 13 17:35:03 yes Aug 13 17:35:11 tharvey: libtool one is a bug Aug 13 17:36:05 ya I noticed libtool recipe includes a common cross_compile.patch which probably isn't appropriate for all Aug 13 17:37:34 FILESPATHPKG=. "libtool-${PV}:" Aug 13 17:37:41 add this in the libtool_2.2.6a.bb Aug 13 17:39:21 for gcc I think you can use --disable-long-double-128 or somesuch Aug 13 17:43:03 jo khem Aug 13 17:43:12 hey Aug 13 17:47:21 khem, FILESPATHPKG didn't help - it does find the cross_compile.patch but the patch doesn't apply (can't find file to patch at input line 11) Aug 13 17:47:55 tharvey: hmm actually add the same in libtool.inc Aug 13 17:48:01 having trouble finding the arcane possible gcc cmdline opts - sounds like your saying my toolchain may be just fine... just need to direct gcc better Aug 13 17:48:24 and remove the FILESDIR line Aug 13 17:48:39 tharvey: yeah Aug 13 17:48:50 you are targetting ppc603 right Aug 13 17:48:56 which is a freescale one IIRC Aug 13 17:49:16 not really knowledgable about gcc and toolchain myself - ya ppc603 (freescale) board is almost identical to lite5200 ref design Aug 13 17:49:28 ok Aug 13 17:49:40 which gcc glibc binutils combo are you using Aug 13 17:49:53 * khem can try to build it at lunch hour Aug 13 17:50:01 keep throwing toolchain over the fence to the team developing apps for it and they had issues with gcc_4.1.x and gcc_4.2.x and want a gcc_4.3.x but I end up with something that won't link with libstdc++ Aug 13 17:50:34 libtool.inc - same issue - are you thinking its choosing the wrong cross_compile.patch? Aug 13 17:50:53 tharvey: for freescape you should use eglibc Aug 13 17:51:11 it has complete soft float support for it Aug 13 17:51:20 and gcc 4.3 to be safe Aug 13 17:51:22 why do you say that? apps team won't accept anything but glibc Aug 13 17:51:31 but the patches are also put in gcc 4.2 Aug 13 17:51:43 cant remember which gcc 4.2 Aug 13 17:52:16 tharvey: eglibc is nothing but glibc for architectures that glibc maintainers dont care Aug 13 17:52:20 so whats the difference Aug 13 17:52:45 IMO openembedded should start thinking of making eglibc as primary libc Aug 13 17:52:55 like debian or ubuntu Aug 13 17:53:00 are doing as of now Aug 13 17:53:03 dunno diffs myself... just know that its a religious argument here - resistant to change :) Aug 13 17:53:21 I know but you can convince them and you should Aug 13 17:53:23 didn't know that deb/ubuntu are moving Aug 13 17:53:34 they are and for very same reason Aug 13 17:53:46 because they want to address world beyong desktop Aug 13 17:54:39 likewise uses eglibc on many of his powerpc platforms Aug 13 17:55:17 would be a hard sell, but I'll kick off a build using eglibc and see how it goes Aug 13 18:00:47 OK Aug 13 18:00:50 I think it will work Aug 13 18:01:14 I remember getting ppc build for 603 going some months ago Aug 13 18:05:43 khem, you going to be able to look at the libtool thing? Aug 13 18:06:14 tharvey: yes I just posted a patch Aug 13 18:06:16 try that out Aug 13 18:06:23 and let me know if it works for you Aug 13 18:07:11 How do I enable a new kernel module? Aug 13 18:07:52 I have changed the defconfig Aug 13 18:08:01 and put =m on the module Aug 13 18:08:08 but i dont get an ipkg Aug 13 18:08:25 khem, posted a patch to maillist? haven't seen it yet but when I do I'll try it out Aug 13 18:09:08 khem, back on the glibc issue - so you were saying that apparently glibc doesn't have full soft-float support for ppc603. I should be able to direct it to use hard-float then right? Aug 13 18:14:41 tharvey: I mean e500 support too Aug 13 18:14:52 for single precision and double precision Aug 13 18:15:12 IMO if you use gcc 4.3+ eglibc 2.9+ binutils 2.19+ you should be all set Aug 13 18:16:29 tharvey: I posted the patch like 10 min back so it might be some delay before it appears on ml Aug 13 18:17:56 I'm just thinking if I have a difficult sell with eglibc - I'm wondering why the toolchain is trying to use hard float by default. Not clear why glibc lack of full soft-float for ppc603e should be an issue as it has an FPU Aug 13 18:18:47 Updated OE this morning, but bitbake beagleboard-demo-image build fails on gnome-games. Here is the error log: http://pastebin.com/m2d37f41b Aug 13 18:20:37 tharvey: glibc lacks it because maintainer dont care and submitter is happy that its there in eglibc Aug 13 18:21:46 tharvey: actually eglibc will ease your and your devs pains Aug 13 18:22:00 you can also take the eglibc patches and apply then to glibc Aug 13 18:22:18 you have to chose what way you want to do it Aug 13 18:23:04 and eglibc always syncs with upstream glibc every week mostly Aug 13 18:23:23 so eglibc = glibc + bits that glibc doesnt care but embedded guys doo Aug 13 18:23:32 is that hard to sell :) Aug 13 18:23:42 eglibc is glibc + some patches thats it Aug 13 18:24:09 and yeah it has nice patches that enables cross testing Aug 13 18:24:24 and you can also include/drop features like uclibc Aug 13 18:24:49 locales can be generated using cross localedef Aug 13 18:25:07 so many pluses for embedded devs :) Aug 13 18:27:04 khem, ok so I understand better diffs between eglibc and glibc but am still not understanding choice of hard vs soft float in gcc - what would default be (depends on how gcc or g/elibc was built?) Aug 13 18:27:19 have to run out for a bit, back in a while Aug 13 18:33:25 tharvey: for your system FPU is E500 right ? Aug 13 18:33:38 you dont have some IEEE compliant FPU or do you Aug 13 18:34:18 if you want to use E500 that will be enabled Aug 13 18:34:32 re Aug 13 18:35:50 wb Aug 13 18:44:00 john3909: is it on .dev ? Aug 13 18:44:28 khem: yes Aug 13 18:45:09 strange where does it get SDL from Aug 13 18:45:28 do you have your do_run* scripts somewhere Aug 13 18:45:52 khem: In the same temp folder? Aug 13 18:45:57 yes Aug 13 18:46:24 khem: let me paste in to pastbin Aug 13 18:46:48 ok Aug 13 18:47:16 khem: http://pastebin.com/m46f774cd Aug 13 18:50:16 this one gives no idea Aug 13 18:50:30 run.do_compile ? Aug 13 18:56:03 khem: this was run.do_compile Aug 13 19:03:40 khem: does this not have something to do with the earlier post from Roger Monk? (05:37:05 AM) CIA-1: Roger Monk org.openembedded.dev * r43100bfa8e openembedded.git/recipes/gnome/gnome-games_2.24.0.bb: gnome-games: fix paths to host SDL, bump PR Aug 13 19:09:05 03Sebastian Krzyszkowiak  07shr/import * r153911f09f 10openembedded.git/ (3 files in 3 dirs): shr-splash-theme-niebiee: new recipe Aug 13 19:29:15 khem, not sure what FPU mpc5200 has - its double precision Aug 13 19:30:49 its actually an mpc5200b - http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPC5200B Aug 13 19:31:48 datasheet says mpc603e series e300 core Aug 13 20:12:34 how can i change image size from config files? Aug 13 20:25:06 tharvey: it seems it has real FPU Aug 13 20:25:11 with DFP Aug 13 20:25:18 DFP is supported in eglibc Aug 13 20:26:54 you've sold me on looking at eglibc Aug 13 21:04:58 jo laibsch Aug 13 21:05:32 * Laibsch just returned from a glisssssssssssssendo performance Aug 13 21:05:38 AMAZING! Aug 13 21:05:51 what? Aug 13 21:06:23 look around youtube for "ulik & le snob" Aug 13 21:06:33 absolutely incredible Aug 13 21:10:49 re Aug 13 21:10:53 mickey|dinner: ping Aug 13 21:10:56 jo florian Aug 13 21:15:43 kergoth: does COLLECTIONS take more than 2 overlays? Aug 13 21:16:07 i do builds with 6+ dirs in my COLLECTIONS pretty regularly. it doesn't care Aug 13 21:16:20 space separated list, highest priority to lowest, left to right Aug 13 21:16:43 kergoth: I tried 3 and bitbake chokes, saying blah is not a BitBake file while parsing blah Aug 13 21:16:57 where blah is a first collection in the list Aug 13 21:17:03 it's a directory, not a file Aug 13 21:17:17 I changed back to BBFILES/BBFILE_* and it works Aug 13 21:17:45 do I need any updates to bitbake by any chance? mine is couple months old Aug 13 21:17:53 all collections does is sets BBPATH, BBFILES, and BBFILE_* Aug 13 21:18:01 i'll check and see if i have anything i need to push for it Aug 13 21:18:10 haven't had time to push much lately, busy working toward the mvl6 release Aug 13 21:18:41 same here - I'm sitting on some changes I need to push back to OE as well :) Aug 13 21:19:17 * kergoth 's losing track of what has and hasn't been pushed. git isn't very good at tracking changes with a lot of cherry picking in long lived branches Aug 13 21:20:13 speaking of BBPATH - I used to setting it outside of bitbake or local.conf, usually in a shell before calling bitbake. is it still correct? Aug 13 21:21:28 thats how you should be setting it. in order to make setting BBPATH from collections work correctly, i had to re-exec bitbake with the new bbpath, to force it to re-parse the configuration metadata using the new search path Aug 13 21:21:38 ugly, but works :P Aug 13 21:22:16 wow. is it in OE or in your private branch? Aug 13 21:22:55 i think the OpenEmbedded version has it, but i doubt anyone's making use of that aspect of it. it only does the bbpath re-set/re-exec if it changes due to the collections processing, so if BBPATH was already set correctly, it won't mess with it Aug 13 21:23:20 ah, interesting. thanks! Aug 13 21:24:10 the mvl content server delivers the collections as tarballs, including the collection that has the bitbake.conf and all, so i had to work out a way to break the recursion there ;) now i have the conf/bitbake.conf in bitbake load a version of collections.inc from the bitbake install dir to bootstrap the collection processing & unpacking, then it re-execs to pull the config from the unpacked collections Aug 13 21:24:22 pretty twisted, but neat Aug 13 21:25:36 :) Aug 13 21:25:52 let me know if you find anything missing in COLLECTIONS processing Aug 13 21:28:08 for now I switched back to using BBFILES, which strangely works fine Aug 13 21:28:43 bbl Aug 13 21:29:07 * kergoth nods, will look into it this evening most likely and get back to you Aug 13 21:34:35 florian: ... Aug 13 21:36:33 mickey|dinner: success... looks.... ugly :) Aug 13 21:36:42 well... FAX :) Aug 13 21:36:52 i can scan it and email it, if you think it's better Aug 13 21:36:57 true Aug 13 21:37:22 no... will hand it in and if they complain we waste more time with it. Aug 13 21:37:35 meinst du mikey|dinner? Aug 13 21:37:37 righto Aug 13 22:21:32 denix0: you're still building on Gentoo? Aug 13 22:22:45 I discovered now ccache perversely affects my rebuilds from packaged staging... Aug 13 22:22:55 I thought G Aug 13 22:23:07 entoo was immune from the ccache issues Aug 13 22:33:15 grrr Aug 13 22:33:26 I'm so close to booting this :-| Aug 13 22:38:10 ant__: ccache from gentoo you mean Aug 13 22:38:23 yes, the implementation Aug 13 22:38:48 it seems OE fails horribly on some distros, just compiling Aug 13 22:38:56 hmm Aug 13 22:39:01 on Gentoo, at least, I have issues only from pstage :) Aug 13 22:39:13 may be we should provide ccache-native Aug 13 22:40:44 oh, a new issue from packaged-staging... Aug 13 22:40:47 | touch: cannot touch `/oe/build/tmp/deploy/glibc/ipk/Packages': No such file or directory Aug 13 22:41:12 of course..I'm rebuilding from bare /deploy/glibc/pstage ... Aug 13 22:43:15 I don't know...kergoth says he can use packaged staging Aug 13 22:43:39 here I see touch /oe/build/tmp/deploy/glibc/ipk/Packages failing... Aug 13 22:43:51 perhaps touch /oe/build/tmp/deploy/glibc/ipk before Aug 13 22:44:03 eh.. mkdir Aug 13 22:45:50 ok, is log.do_package_update_index_ipk Aug 13 22:46:10 or mkdir -p Aug 13 22:48:23 package_ipk.bbclass Aug 13 22:49:44 I'm pretty sure packaged staging is intended to work just keeping th e'pstage' dir and deleting all the rest in /tmp (even /ipk) Aug 13 22:49:50 RP confirmed it lately Aug 13 22:51:08 let see where does ${DEPLOY_DIR_IPK} come from.. Aug 13 22:53:24 well, is in ./bitbake.conf:DEPLOY_DIR_IPK = "${DEPLOY_DIR}/ipk" Aug 13 23:07:23 ant__: how will it populate without ipks Aug 13 23:08:29 from the ones in pstage? Aug 13 23:11:33 hm..ah.. is created in the wrong place... in ../tmp/deploy/glibc/ipk.lock Aug 13 23:11:43 what are you working on? Aug 13 23:11:53 ./classes/package_ipk.bbclass:do_package_update_index_ipk[lockfiles] = "${DEPLOY_DIR_IPK}.lock" Aug 13 23:12:19 I built from scratch yestaerday and was trying a rebuild from packaged staging, without ccache Aug 13 23:13:25 http://tinderbox.openembedded.org/public/logs/task/2395111.txt Aug 13 23:13:25 our default deploy dir does not differentiate between library types (uclibc eglibc/glibc) Aug 13 23:14:03 it would be nice that we put the uclibc specific stuff under /something/uclibc/ Aug 13 23:14:32 oh it does it seems hmm Aug 13 23:15:05 kergoth: ./bitbake.conf:DEPLOY_DIR ?= "${TMPDIR}/deploy" Aug 13 23:15:35 but I forgot to say I'm using Angstrom: ./distro/include/angstrom.inc:DEPLOY_DIR = "${TMPDIR}/deploy/${LIBC}" Aug 13 23:16:27 btw the guys of 'micro' should be warned: ./distro/micro.conf:DEPLOY_DIR ?= "${TMPDIR}/deploy/${LIBC}" Aug 13 23:16:39 ?= Aug 13 23:17:24 usually people don't mess with these in local.conf (I don't) Aug 13 23:20:56 ant__: they do Aug 13 23:21:01 should just use =, the user can still override in local.conf using OVERRIDES Aug 13 23:21:15 DEPLOY_DIR_local = "/go/where/I/want/you/to/go/you/bastard" Aug 13 23:21:16 ant__: if someone wants to put DEPLOY_DIR out of buildtree Aug 13 23:21:38 I'm still hit by the package_stagefile_Shell in that case :( Aug 13 23:22:14 some hackish recipes (zaurus-updater) Aug 13 23:22:19 kergoth: hmm so this can be used in local.conf ? Aug 13 23:23:11 of course. Aug 13 23:23:28 you can use override specific variable assignments or appends/prepends anywhere. bitbake doesn't care Aug 13 23:23:50 bitbake -e | grep \^OVERRIDES= Aug 13 23:24:31 the last three win over all (local.conf, distro.conf, machine.conf) Aug 13 23:25:41 it helps if you remember that bitbake/oe was designed to be layered, so that more specific things can override less specific things, and that this layering is implemented in two ways: config file load & ordering, and OVERRIDES Aug 13 23:25:56 so _local is the most specific assignment using OVERRIDES, generally Aug 13 23:26:13 hmm cool Aug 13 23:27:06 kergoth: someone should copy & paste the explanations in the OE Manual Aug 13 23:27:32 very well said Aug 13 23:28:45 i'll see if i can flesh it out a bit more. some of this high level description would be helpful. we had a number of people internally at MV that had great difficulty using the system until they grasped the big picture. some people are like that, and it's good to help them do so Aug 13 23:31:32 Woohooo! gcc-4.4.1 configured --with-mips-plt builds an image that actually functions. Aug 13 23:31:48 grg cool Aug 13 23:32:07 although dbus-daemon is *still* an issue Aug 13 23:33:00 it fails to link with -mplt. and fails to run with -mplt -fPIE Aug 13 23:33:51 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r430bcadad5 10openembedded.git/recipes/linux/linux-2.6.30.2/at91sam9263dfc/defconfig: Modify SAM9263 Linux config to detect sound card Aug 13 23:33:52 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r2692e8f4f8 10openembedded.git/recipes/at91bootstrap/at91bootstrap-2.13-rc1/at91sam9g45ekes/ (AT91SAM9G45_RomCode_Replacement.zip README.txt): Add the G45 ROM replacement binary with explanation Aug 13 23:33:53 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r67b8d63dd3 10openembedded.git/recipes/u-boot/ (2 files in 2 dirs): Allow make install for u-boot Aug 13 23:33:56 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * rfcf78a2d1f 10openembedded.git/recipes/qt4/qt4.inc: Configure QT to allow an BGR display Aug 13 23:33:59 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r3b342633a1 10openembedded.git/recipes/linux/linux-2.6.30.2/arch-at91/ (2 files): Add patches inside tree, since www.at91.com is down Aug 13 23:34:02 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r79118ed51e 10openembedded.git/conf/machine/include/AT91.inc: Add uboot as a MACHINE_FEATURE Aug 13 23:34:05 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * rc9356ff01a 10openembedded.git/recipes/at91bootstrap/at91bootstrap_2.13-rc1.bb: Support loading AT91SAM9G45ES ROM replacement code into deploy dir Aug 13 23:34:12 03Ulf Samuelsson  07ulf/linux-2.6.30.2 * r68f2ccca7c 10openembedded.git/recipes/linux/ (linux-2.6.30.2-at91-kernel-config.inc linux_2.6.30.2.bb): Use the in-tree patches in linux Aug 13 23:36:34 * kergoth grumbles Aug 13 23:42:07 good night Aug 14 00:12:38 any opinion on adding a new bitbake commandline option for 'all' tasks? it's a common pattern, the creation of a wrapper task that calls the one you want and has a recrdeptask flag set to it, to run it against the requested providers + their deps Aug 14 00:12:44 i.e. fetchall, buildall, package_stage_all Aug 14 00:12:52 seems like a -c fetch --all or something would be good Aug 14 00:25:41 Only if you can add -c clean --all :) Aug 14 00:26:49 it just seems silly to keep reimplementing the same pattern in various classes when it seems like a pretty common thing to ask bitbake to do Aug 14 00:27:00 run this task against what i want and its deps, rather than just this Aug 14 00:27:07 yeah Aug 14 00:27:29 I'm not sure how buildall would be different from build, but I can see some of the other tasks Aug 14 00:30:03 well, when you run 'bitbake foo' it doesn't necessarily package the deps of foo, it would stage the deps of foo so foo's packaging bits could run, but thats all. buildall runs the build task on the deps too Aug 14 00:30:25 base.bbclass defines a buildall task today Aug 14 00:31:04 how can I find out what root= option I have to provide my kernel? Aug 14 00:31:44 1) off topic, 2) it depends Aug 14 00:35:26 In the kernel I built, mmc sd card access is built in. But root=/dev/mmcblk0p2 returns "cannot open root device or unknown-block". The terminal is scrolling to fast to see wether the partitions were listed and I don't have a serial cable. Aug 14 00:35:36 well, I guess it really is OT Aug 14 00:36:50 J13R, remember not all mmc/sd cards are created equal. linux just plain doesn't like some models and/or vendors Aug 14 00:37:17 well, certain mmc host drivers don't anyway Aug 14 00:38:25 i suggest booting with root somewhere else (nfs is convenient) to confirm whether or not you can manually mount the mmcblk0p2. Aug 14 00:39:21 hmm ... nfs over usb ... maybe Aug 14 00:39:47 grg: you know a good howto? Aug 14 00:40:41 I just tested with another kernel, and it says mmcblk0 has 2 partitions, p1 and p2, so it should be alright Aug 14 00:40:52 ok Aug 14 00:41:37 you can try passing the hex value of the block device as your root= value Aug 14 00:42:10 i.e. root=MMmm where MM is the hex value for the block device's major number and mm is the hex of the minor. Aug 14 00:43:39 grg: the initrd of the other kernel doesn't have the device file Aug 14 00:43:41 so for /dev/mmcblk0p2, maj 179, min 2, you would pass root=b302 Aug 14 00:44:20 you don't need a device file Aug 14 00:44:36 how do I know the major number then? isn't it hardware-dependent? Aug 14 00:44:53 hmm... maybe... i forget Aug 14 00:45:12 see the linux docs dir. there's a file full of numbers there somewhere Aug 14 00:45:23 I'll just try it out Aug 14 00:45:50 for mmc you might need to pass in rootwait Aug 14 00:46:06 yes, rootwait is a good idea too Aug 14 00:46:43 ok, rootwait. should I try root=/dev/mmcblk0p2 or the hex version first? Aug 14 00:47:06 * grg shrugs Aug 14 00:47:17 how long will it take you to change the bootargs? Aug 14 00:47:28 I do it in word :-) Aug 14 00:48:05 as in, ms word? Aug 14 00:48:05 ok, that didn't work, trying with hex ... Aug 14 00:48:13 grg: yes, on the ce Aug 14 00:48:16 :-) Aug 14 00:48:23 ugh Aug 14 00:48:29 config for the bootloader (haret) Aug 14 00:48:36 double ugh Aug 14 00:49:45 didn't like b302 either Aug 14 00:50:06 what filesystem are you using? Aug 14 00:50:12 ext2 Aug 14 00:50:19 should be fine Aug 14 00:51:48 hmm :-( maybe nfs is better anyways to open a ssh shell there Aug 14 00:53:39 J13R, linux/Documentation/filesystems/nfsroot.txt has what you need Aug 14 00:56:11 grg: thanks. might be complicated to get the network connection done though ... Aug 14 00:57:13 i have done nfsroot over usb ethernet in the past. its not impossible Aug 14 00:57:49 omg Aug 14 00:58:03 I just found the magic that makes everything happen in the other kernel Aug 14 00:58:18 the makeitwork=1 kernel parameter? Aug 14 00:58:36 no a shell script in the initrd Aug 14 01:04:38 looks like it is hda is used for the sd card Aug 14 01:06:21 That is retarded. Aug 14 01:06:30 nah I was wrong Aug 14 01:06:41 CompactFlash = hda Aug 14 01:06:42 i take it all back then :) Aug 14 01:09:10 hmm... I put a bunch of things in EXTRA_IMAGEDEPENDS in my local.conf yet they aren't in the filesystem tarball. Aug 14 01:09:40 Is there a recommended approach to get stuff into the tarball? Aug 14 01:23:55 I'll probably be more successful with a initrd Aug 14 01:36:26 moin Aug 14 01:36:58 J13R: hey, out of interest i did built tk for arm yesterday... it is searching tcl in the right directory here... Aug 14 01:37:27 zecke: moin Aug 14 01:37:47 zecke: I resolved it by removing python-tk and python-idle Aug 14 01:38:08 zecke: I didn't have the tcl folder in work Aug 14 01:38:19 J13R: well, you don't need one Aug 14 01:38:31 it does -I it Aug 14 01:38:46 J13R: for you, and it would be nice to understand why that is the case Aug 14 01:40:09 zecke: what can I do? you have my pastebin dump Aug 14 01:40:59 J13R: could you paste me your log.do_configure, run.do_configure, log.do_compile of tk? Aug 14 01:41:16 zecke: will do once I'm in windows again Aug 14 01:43:02 thanks Aug 14 01:44:04 what is the "right" way to configure the kernel? Aug 14 01:44:14 editing some recipe? Aug 14 01:44:27 J13R: provide your own defconfig? Aug 14 01:44:52 zecke: defconfig = .config ? Aug 14 01:46:09 J13R: yes, e.g. see linux_2.6.30.bb Aug 14 01:46:51 J13R: in the SRC_URI it has file://defconfig, and the in linux-2.6.30/ you have different directories for machines, so if you create linux-2.6.30/${MACHINE}/defconfig, this file will be used Aug 14 01:47:08 if there is a kernel module, and I want to build it in, can I give it some parameters (that I would normally pass on with insmod)? Aug 14 01:48:08 J13R: for modprobe YOUR_MODULE... yes, I don't remember the name of the variable though Aug 14 01:48:28 zecke: no if I build it in Aug 14 01:48:42 J13R, linux/Documentation/kernel-parameters.txt Aug 14 01:48:56 J13R: sorry, I don't understand this one... you mean modulename.param=value? Aug 14 01:49:04 yeah :) Aug 14 01:49:28 OT: which timezone are you two located at? Aug 14 01:49:39 +0930 Aug 14 01:49:47 ah Aug 14 01:51:43 grg: 9:30? that is east of korea? Aug 14 01:52:01 (stupid question... everything is east of something :}) Aug 14 01:52:17 +12 Aug 14 01:53:23 it is significantly south of Korea Aug 14 01:53:30 zecke: (nz) you're from germany? Aug 14 01:54:16 J13R: I'm from Germany, I live in Taiwan Aug 14 01:54:31 zecke: I'm from Austria, I live in New Zealand Aug 14 01:54:42 hehe Aug 14 01:58:38 NZ has linuxconf next year. Maybe I should take a holiday there in January. Aug 14 01:58:55 grg: wut? link plz Aug 14 01:59:08 http://lca2009.linux.org.au/ Aug 14 02:01:06 brb rebooting Aug 14 02:10:53 my tk files: http://pastebin.com/f313bc17f http://pastebin.com/f2129fe9a Aug 14 02:11:13 staging_helper and do_configure Aug 14 02:13:22 and the log files: http://pastebin.com/f37244637 http://pastebin.com/f7d36740d Aug 14 02:14:10 tell me if you need anything else to check that tk/tcl problem Aug 14 02:15:39 hmm, anyone tested jlibtool? Aug 14 02:16:33 J13R: log.do_compile please, I don't have the link from your yesterdays post anymore Aug 14 02:16:47 kergoth: found the problem Aug 14 02:16:53 ah, what was it? Aug 14 02:17:02 * kergoth just got home from work, didn't have time to look at it yet Aug 14 02:17:03 kergoth: I had no recipes in that overlay Aug 14 02:17:25 huh Aug 14 02:17:35 thats an odd behavior for that situation Aug 14 02:17:58 oh! i remember what it is Aug 14 02:18:08 so, when it handles the cache, since there were no recipes, it was saying the overlay dir "is not a bitbake file" Aug 14 02:18:46 pretty sure its a bitbake bug i found a while back in the way it handles bbfiles. when it processes a dir in bbfiles, it doesn't seem to always remove it after it traverses the dir Aug 14 02:18:51 * kergoth tracks down the commit Aug 14 02:19:26 zecke: http://pastebin.com/f37244637 Aug 14 02:19:37 thanks! for now I just put there an empty recipe :) Aug 14 02:20:14 denix0: http://cgit.openembedded.org/cgit.cgi/bitbake/commit/?h=1.8&id=8f5df854ac53b5d7e7dd2fe2432b781cb195383f Aug 14 02:20:29 denix0: s/BBMASK/BBFILES/ in the log message, it was a typo/thinko Aug 14 02:20:49 zecke: I take the original .config from linux-rp, do make menuconfig and plug it into the recipe, right? because what menuconfig produces is so different from the original .config Aug 14 02:21:24 thanks! I'll try it. I guess I also need to update my bitbake... Aug 14 02:21:29 thinko :-) what a funny word Aug 14 02:22:03 J13R: well, if it wasn't a mistake in typing, but a mistake in thinking.. :) Aug 14 02:22:21 in this case i think it was a copy/paste bug when i was rebasing / fixing up the commits to go upstream, actually Aug 14 02:22:31 miscommunication bewteen brain and keyboard :) Aug 14 02:22:46 it seems to also touch bbmask handling... Aug 14 02:23:28 think that was just a cleanup Aug 14 02:23:29 so, the description just mixes two fixes... :) Aug 14 02:23:40 J13R: just look at the linux-rp recipe... Aug 14 02:23:44 see, it was iterating based on the length, but it never used the counter for the iteration, so it was pointless Aug 14 02:23:54 heh Aug 14 02:24:01 regardless, that should take care of the issue :) Aug 14 02:24:28 * kergoth looks into making packaged staging non-optional Aug 14 02:24:38 J13R: If you currently use linux-rp... you should see an error with "No default configuration for ${MACHINE}..."? Aug 14 02:27:29 does bitbake support something like useflags? Aug 14 02:27:57 J13R: itd be a function of oe, not bitbake. itd be just another variable. but no, we don't have anything like that yet Aug 14 02:28:14 would be nice: python -tk Aug 14 02:28:37 kergoth: thanks, works perfectly! Aug 14 02:28:43 great Aug 14 02:29:07 J13R: (tk): weird... i have no idea why it picks up this dir :} Aug 14 02:29:14 kergoth: also thanks for compressed collection handling Aug 14 02:31:57 J13R: J13R could you also give me unix/config.log? Aug 14 02:32:24 kergoth: btw, downloading of compressed collections - have you looked at repo for something like this? Aug 14 02:32:25 zecke: where would I find that? Aug 14 02:33:41 J13R: tk8*/unix/config.log Aug 14 02:37:38 zecke: http://pastebin.com/f5b2f531d Aug 14 02:39:59 zecke: I think the problem is in the tclConfig.sh : http://pastebin.com/f7e0ffe62 Aug 14 02:40:22 the variables point to work there Aug 14 02:45:26 zecke pasted "tclConfig.sh" at http://paste.lisp.org/display/85361 Aug 14 02:45:44 J13R: okay, and this is mine... so the problem is not within tk but in tcl :) Aug 14 02:51:30 zecke: all logs from tcl: http://pastebin.com/f45efda86 Aug 14 02:53:14 configure creates the .sh file Aug 14 02:54:49 J13R: I will have a look **** ENDING LOGGING AT Fri Aug 14 02:59:57 2009