**** BEGIN LOGGING AT Mon Oct 05 02:59:57 2009
Oct 05 06:47:56 <khem`> gregoiregentil: try using gcc 4.4
Oct 05 06:48:09 <khem`> gregoiregentil: what board do you have 
Oct 05 06:48:44 <m4t> hey khem, i have those diffs
Oct 05 06:48:56 <khem> m4t: good work. 
Oct 05 06:49:01 <m4t> its huge, though, do you have a way to split them?
Oct 05 06:49:08 <khem> hmm
Oct 05 06:49:14 <m4t> the commits are *mostly* logical
Oct 05 06:49:33 <m4t> i did it recipe-by-recipe, except for the first few, which took place before i really knew how to use git
Oct 05 06:49:33 <khem> generally you should club them in individual commits
Oct 05 06:49:53 <m4t> yea, 90+% of them are, with comments
Oct 05 06:50:00 <khem> should be ok. Actually you can rebase them 
Oct 05 06:50:05 <khem> on a different branch 
Oct 05 06:50:20 <khem> and rearrange them as you like
Oct 05 06:50:21 <m4t> yea i should do that probably
Oct 05 06:50:29 <m4t> one sec,
Oct 05 06:50:31 <khem> git rebase is very powerful
Oct 05 06:51:53 <m4t> 104177 lines in the diff
Oct 05 06:51:57 <m4t> though, some of that isn't mine
Oct 05 06:52:01 <khem> hmmm
Oct 05 06:52:03 <m4t> my first commit was like aug 28th
Oct 05 06:52:23 <m4t> the majority of that is probably the patches to the walnut kernel
Oct 05 06:52:26 <khem> actually you can publish your branch too for review 
Oct 05 06:52:52 <khem> yeah usually kernel is mostly the one
Oct 05 06:52:57 <khem> modified most
Oct 05 06:54:38 <m4t> http://pastebin.com/d7724a200
Oct 05 06:54:48 <m4t> is the git log, starting right before my first commit
Oct 05 06:56:46 <m4t> i need to --name-only or smth
Oct 05 06:56:50 <m4t> to get the files on there
Oct 05 06:57:22 <khem> your email address is phony in the commits you need to correct that too
Oct 05 06:58:11 <khem> commits are also not per format 
Oct 05 06:58:29 <m4t> heh :/
Oct 05 06:58:32 <m4t> http://pastebin.com/d20eaa615 has the files
Oct 05 06:58:43 <khem> I think you should rebase them and make changes 
Oct 05 07:00:03 <m4t> alright, i'll look into that
Oct 05 07:00:19 <khem> yeah changes look ok
Oct 05 07:00:24 <khem> otherwise
Oct 05 07:00:36 <m4t> yea mostly just version bumps
Oct 05 07:00:41 <m4t> a couple new recipes
Oct 05 07:00:52 <khem> if you want people to look at them then make them managable so people can look into parts they would like to
Oct 05 07:01:05 <khem> yeah
Oct 05 07:02:32 <m4t> i got a couple new boards on the way
Oct 05 07:02:40 <m4t> sigma designs smp86xx based
Oct 05 07:02:46 <khem> cool.
Oct 05 07:03:12 <m4t> yea. the sdk i've found is based on an older version of buildroot
Oct 05 07:03:29 <m4t> gcc 3.x / kernel 2.6.15
Oct 05 07:03:35 <khem> oh
Oct 05 07:03:41 <khem> is it ppc ?
Oct 05 07:03:54 <m4t> 300mhz mips soc
Oct 05 07:04:09 <m4t> with security coprocessor, and hardware video/audio decoding
Oct 05 07:04:19 <khem> cool.
Oct 05 07:04:33 <khem> do you intend to use OE on it ?
Oct 05 07:04:53 <m4t> i might try to
Oct 05 07:05:18 <m4t> i built up a toolchain, userland image, kernel, and kernel modules / playback software tonight
Oct 05 07:41:18 <hyc> gregoiregentil: here's a workaround if gcc 4.4 doesn't work: http://svn.chezphil.org/libpbe/trunk/include/AtomicCounter.hh
Oct 05 07:57:55 <recalcati> g.m.
Oct 05 08:37:13 <hrw> morning
Oct 05 08:39:12 <ant_work> hrw: hi
Oct 05 08:40:23 <florian> hi hrw 
Oct 05 08:40:30 <sgh> Hi!
Oct 05 08:41:04 <florian> hi sgh 
Oct 05 08:41:05 <sgh> Apparently DEFAULT_PREFERENCE does not work for me. I'm setting it to "-1" but oe still want to build by recipe .... 
Oct 05 08:41:21 <florian> if its the only one... 
Oct 05 08:41:52 <XorA> or if PREFERRED_VERSION is set
Oct 05 08:42:27 <sgh> Welll ... it is a kernel, os there are plenty of alternatives ....
Oct 05 08:43:05 <sgh> It  looks like this : http://pastebin.com/d7cdf06f5
Oct 05 08:44:11 <akheron> sgh: is it "linux_2.6.31.bb" ?
Oct 05 08:44:58 <sgh> akheron: welll ... it is my own test-example I did it in the stable 2009 branch becasue there where no 2.6.31 kernel available 
Oct 05 08:45:26 <akheron> but what's the file name?
Oct 05 08:45:43 <sgh> linux_2.6.31.bb
Oct 05 08:46:08 <akheron> ok, so is there DEFAULT_PREFERENCE override for your machine or distro or architecture or something?
Oct 05 08:46:27 <akheron> try bitbake -i and use the command "peek linux DEFAULT_PREFERENCE"
Oct 05 08:46:38 <sgh> akheron: ok .... I'll try that ....
Oct 05 08:47:16 <hrw> guys: do we have working RDP client in OE?
Oct 05 08:47:31 <sgh> akheron: it says -1, but which "virtual/kernel" still says it  is going to build 2.6.31
Oct 05 08:49:01 <sgh> akheron: If I remove my local linux_2-6-31.bb it wants to buiid oe's 2.6.31-kernel ..
Oct 05 08:50:18 <akheron> hrm..
Oct 05 08:50:41 <akheron> is there a PREFERRED_VERSION for the kernel
Oct 05 08:51:40 <sgh> akheron: no .... PREFERED_PROVIDERS only contains some qt- and gcc-stuff 
Oct 05 08:52:22 <akheron> PREFERRED_VERSION is not the same as PREFERRED_PROVIDERS
Oct 05 08:54:11 <akheron> so you should really check the PREFERRED_VERSION, I don't really which is the precedence between it and DEFAULT_PREFERENCE
Oct 05 08:54:39 <sgh> akheron: akheron ok ... will check
Oct 05 08:54:39 <akheron> but I would assume that PREFERRED_VERSION wins over DEFAULT_PREFERENCE
Oct 05 08:55:29 <akheron> sgh: so what you should grep for is DEFAULT_VERSION_virtual/kernel lines in machine and distro configs for your setup
Oct 05 08:55:39 <sgh> akheron: ok
Oct 05 08:57:56 <woglinde> hi
Oct 05 08:58:02 <woglinde> ~seen mrmoku
Oct 05 08:58:03 <ibot> mrmoku is currently on #htc-linux #oe #openmoko-cdevel. Has said a total of 224 messages. Is idling for 1h 54m 33s, last said: 'Heinervdm: morning'.
Oct 05 08:58:28 <mrmoku> woglinde: hi :)
Oct 05 08:58:32 <woglinde> ah
Oct 05 08:58:48 <woglinde> mrmoku whats the softkeyboard you are using on shr?
Oct 05 08:58:59 <mrmoku> the illume keyboard
Oct 05 08:59:10 <woglinde> ah okay as I thought
Oct 05 08:59:16 <woglinde> the one from raster right?
Oct 05 08:59:18 <mrmoku> yep
Oct 05 09:02:44 <woglinde> thanks mrmoku
Oct 05 09:02:50 <mrmoku> np :)
Oct 05 09:30:30 <ant_work> woglinde: I rewrote the code using getline using fgets. Later I'll replace fscanf.
Oct 05 09:37:17 <woglinde> ant *g*
Oct 05 09:43:40 <ant_work> woglinde: I'll test with glibc before...
Oct 05 09:48:57 <sgh> I'm trying to build a kernel. The kernel is fetched from here : git://git.denx.de/linux-mpc512x.git
Oct 05 09:49:44 <sgh> The problem is that it is checked out in the wring way. It only gives med a dir called "1"...checking it  out manually works fine.
Oct 05 10:24:04 <pb__> sgh: does it really matter what the directory is called?  OE doesn't really care that much.
Oct 05 10:24:29 <sgh> pb__: The problem is that the branch is not checked out..
Oct 05 10:24:49 <pb__> oh, I see.  you mean you literally just get a directory called 1 with nothing in it?
Oct 05 10:24:59 <sgh> pb__: yes ;(
Oct 05 10:25:04 <pb__> doh, that is no good
Oct 05 10:25:09 <pb__> what's the full SRC_URI you are using?
Oct 05 10:25:17 <sgh> SRC_URI = "git://git.denx.de/linux-mpc512x.git;protocol=git"
Oct 05 10:25:29 <sgh> git clon on that works perfectly
Oct 05 10:25:36 <sgh> clone
Oct 05 10:26:05 <pb__> ok.  and what does the fetcher print when it starts?
Oct 05 10:26:06 <XorA> sgh: did you set a SRCREC?
Oct 05 10:26:09 <XorA> SRCREV
Oct 05 10:26:13 <sgh> no
Oct 05 10:26:57 <XorA> try SRCREV_pn-bbname = "${AUTOREV}" in local.conf. where bbname is the name of your recipe
Oct 05 10:27:24 <XorA> 1 is the default SRCREV
Oct 05 10:28:20 <sgh> XorA: ah ok SRCREV = "${AUTOREV}" in the bb-file would also do it i guess?
Oct 05 10:28:29 <XorA> sgh: yes
Oct 05 10:28:39 <sgh> XorA: will try that ...
Oct 05 10:28:43 <XorA> sgh: but we dont like those when you commit
Oct 05 10:29:06 <sgh> XorA: Of cause ...... :D
Oct 05 10:44:24 <hrw> wow.. my current image has 988 packages installed
Oct 05 10:44:31 <hrw> 204MB unpacked
Oct 05 10:44:37 <hrw> ~hail 2GB microsd
Oct 05 10:44:38 * ibot bows down to 2GB microsd and chants, "I'M NOT WORTHY!!"
Oct 05 10:45:19 <XorA> hehe
Oct 05 10:46:08 <hrw> time to split card into parts... I need few rootfs images
Oct 05 10:54:04 <tasslehoff> I want to create a custom image based on Angstrom. I want to start as light as possible, and add stuff when I find out I need it. Is a copy of recipes/images/console-image.bb a sensible starting point?
Oct 05 10:54:40 <sgh> XorA: it worked jsut excellent. Thanks a lot.
Oct 05 10:55:05 <hrw> tasslehoff: depends how you define 'as light as possible'
Oct 05 10:55:13 <pb__> tasslehoff: none of the angstrom images are really what I would describe as "as light as possible", but console-base-image might be a reasonable place to start
Oct 05 10:55:28 <tasslehoff> hrw: true :)
Oct 05 10:58:18 <tasslehoff> pb__: thanks. 
Oct 05 10:59:52 <tasslehoff> The first goal is something like: ssh in, and use mplayer to play a video into a framebuffer (on a beagleboard)
Oct 05 11:03:19 <zecke> pb__: sorry, one more GNU question. Imagine I have either a) BSD licensed code or b) copyright assignment with GNU, it will be possible to get another memcpy function into glibc?
Oct 05 11:03:57 <pb__> zecke: I don't think BSD-licensed code is good enough, but if you have a copyright assignment (and you are the author) then it should certainly be possible.
Oct 05 11:05:07 <zecke> pb__: ah okay, so my own code then and being able to give up copyright on it
Oct 05 11:05:32 <pb__> right, that is the ideal situation.  the gnu project likes to own all its own code so that it can enforce the license if that were to become necessary.
Oct 05 11:05:54 <pb__> there is some bsd code in glibc (e.g. the sunrpc bits) but I believe that adding new bsd code is not considered desirable.
Oct 05 11:06:45 <pb__> is this your neon thing?
Oct 05 11:07:18 <pb__> obviously you are welcome to add it as a local patch in oe if you want, that would be fine.  you might also find that eglibc has a slightly lower threshold of acceptance than the real glibc does.
Oct 05 11:08:37 <tasslehoff> I'm just getting started with OE, so I don't understand everything (anything?) in the recipes yet. Am I correct that console-image just adds wifi and bluetooth support to a console-base-image?
Oct 05 11:09:46 <zecke> pb__: lower, but not in terms of licensing
Oct 05 11:09:54 <zecke> pb__: add a question mark please :)
Oct 05 11:10:11 <pb__> zecke: I'm not sure what the eglibc licensing requirements are exactly.
Oct 05 11:10:47 <zecke> pb__: from what I think I have read: They want to be able to merge everything back to glibc
Oct 05 11:12:00 <zecke> pb__: yeah, I want a faster memcpy, and I want it in the GNU libc. For the next months I'm paid to make QtWebKit faster on ARM, and scrolling the surface is currently using memcpy/memmove and it is #1 in (o)profiles. Being me I want the fix to be in glibc :)
Oct 05 11:12:56 <zecke> pb__: licensing: http://www.eglibc.org/mission
Oct 05 11:22:12 <woglinde> re
Oct 05 11:22:52 <broonie> If it's for ARM real glibc isn't an option anyway :)
Oct 05 11:24:10 <woglinde> pb did you look at the kexec-dietlibc bin?
Oct 05 11:28:12 <pb__> zecke: ah, right.  "copyright for the library to be held by the fsf"
Oct 05 11:28:16 <pb__> so, yeah, you need an assignment
Oct 05 11:29:07 <pb__> zecke: of course, for your own purposes you could use an LD_PRELOAD intercept library, pending integration of the code into glibc proper.
Oct 05 11:29:17 <pb__> woglinde: not yet.  on the other hand, I have now installed eclipse 3.5.
Oct 05 11:29:34 <zecke> pb__: yeah, or just call it fast_memcpy and change the WebKit code, I will try to get code into glibc
Oct 05 11:29:50 <pb__> at least, I think it's 3.5.  it says "Build id: 20090920-1017"
Oct 05 11:29:53 <pb__> zecke: right, or that
Oct 05 11:30:16 * pb__ -> meeting, bbiab
Oct 05 11:33:58 <hrw> someone know does u-boot/ext2load has a problems with ext3?
Oct 05 11:38:38 <tasslehoff> "bitbake omapfbplay", has until now taken > 20 minutes, and that seems long for such a small project. I have a local.conf setup for Angstrom. Could that be the reason it does so much preprocessing?
Oct 05 11:39:06 <akheron> tasslehoff: prepare to wait for two hours if you build from scratch
Oct 05 11:39:35 <akheron> it's got nothing to do with ångström
Oct 05 11:40:28 <tasslehoff> akheron: even if the first thing I do after setting up my oetree is to compile just that one small package?
Oct 05 11:40:52 <akheron> yes, because it compiles the whole cross toolchain for you
Oct 05 11:41:05 <akheron> after that, things are faster
Oct 05 11:41:08 <tasslehoff> akheron: aaaaah. of course. forgot about that :) thanks
Oct 05 11:42:12 <woglinde> pb yes thats the latest 3.5.1
Oct 05 11:47:23 <hrw> ifup eth0
Oct 05 11:47:24 <hrw> SIOCSIFFLAGS: Cannot assign requested address
Oct 05 11:47:31 <hrw> argdh
Oct 05 11:50:27 <recalcati> good morning
Oct 05 11:54:36 <recalcati> I have a funny question about OABI system call and gcc4. Is there, it seems, a project angstrom legacy for this kind of implementation. Aby help will be appreciated. 
Oct 05 12:05:10 <eFfeM> pb_ looking for flights to oedem, how is connectivity from london city airport ?
Oct 05 12:06:21 <hrw> eFfeM: doable
Oct 05 12:06:46 <hrw> eFfeM: city airport -metro-> train station -train-> cambridge
Oct 05 12:07:30 <tasslehoff> I have started with the console-base-image like suggested, and for starters I want to add mplayer, but where. In my custom image-recipe, or should I create a custom task-recipe to go with it? All the OE-variables are a bit overwhelming to me :)
Oct 05 12:08:13 <pb__> eFfeM: less good than stansted, but not too bad.
Oct 05 12:09:26 <hrw> tasslehoff: http://pokylinux.org/doc/poky-handbook.html#usingpoky-extend-customimage-custombb basically
Oct 05 12:09:43 <hrw> tasslehoff: the second way (strace example)
Oct 05 12:12:19 <tasslehoff> hrw: thanks. that recipe is not at all overwhelming :)
Oct 05 12:14:11 <hrw> fucking crap
Oct 05 12:14:44 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * r0ecc03866f 10openembedded.git/packages/enigma2/enigma2.bb: enigma2.bb: push srcdate
Oct 05 12:14:44 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * r761a1358f2 10openembedded.git/: Merge branch 'org.openembedded.dreambox.release_25' of git://git.openembedded.net/openembedded into org.openembedded.dreambox.release_25
Oct 05 12:14:45 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * r1080bb56d0 10openembedded.git/packages/dreambox/dreambox-secondstage.bb: dreambox-secondstage.bb: update 7025 2nd stage loader to 74 because of support for new oled display
Oct 05 12:14:47 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * rc245f9156e 10openembedded.git/packages/gstreamer/gst-plugins-bad_0.10.10.bb: gst-plugins-bad_0.10.10.bb: fix patch include
Oct 05 12:14:50 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * r73b660d24a 10openembedded.git/packages/enigma2/enigma2.bb: enigma2.bb: another srcdate push (for 4506 support)
Oct 05 12:14:53 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * r6dc3bf31f0 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules.bb: update dm8000 drivers (add support for broadcom 4505/4506 tuner)
Oct 05 12:15:03 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * r7fb0d1d728 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: (log message trimmed)
Oct 05 12:15:06 <CIA-1> dreambox-dvb-modules.bb: update dm800 and dm8000 drivers
Oct 05 12:15:08 <CIA-1> fix 1080p24/30 synchronisation problems via AV Receiver
Oct 05 12:15:10 <CIA-1> fix 720p24 output via dvi
Oct 05 12:15:12 <CIA-1> fix loss of dvi(hdmi) audio on framerate change
Oct 05 12:15:14 <hrw> root@bug:~# ifconfig -a
Oct 05 12:15:14 <CIA-1> another small deinterlacer auto handling fix
Oct 05 12:15:14 <hrw> eth0      Link encap:Ethernet  HWaddr FF:FF:FF:FF:FF:FF
Oct 05 12:15:15 <XorA> pb__: BTW is there parking at your office for visitors?
Oct 05 12:15:16 <CIA-1> improved locking times for alps-bsbe1-c/d01a tuner
Oct 05 12:15:18 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * ra980303bca 10openembedded.git/packages/gstreamer/gst-plugin-dvbmediasink.bb: gst-plugin-dvbmediasink.bb: push srcdate
Oct 05 12:15:20 <hrw> now I understand..
Oct 05 12:15:21 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox * r4c0fbdb8d2 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: dreambox-dvb-modules.bb: update dm8000 drivers (add support for broadcom 4505/4506 tuner)
Oct 05 12:15:29 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox * r28ad388170 10openembedded.git/packages/dreambox/dreambox-secondstage.bb: dreambox-secondstage.bb: update 7025 2nd stage loader to 74 because of support for new oled display
Oct 05 12:15:33 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox * r0e0cef1fdb 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: (log message trimmed)
Oct 05 12:15:38 <CIA-1> dreambox-dvb-modules.bb: update dm800/dm8000 drivers
Oct 05 12:15:40 <CIA-1> changes both:
Oct 05 12:15:42 <CIA-1> - implement FE_TIMEDOUT flags in all tuner drivers (used in e2 to speedup the service searching a little bit)
Oct 05 12:15:44 <CIA-1> - add new option to change scaler sharpness (/proc/stb/vmpeg/0/pep_scaler_sharpness)
Oct 05 12:15:46 <CIA-1> - fix kernel freeze caused by demux problem when try to use more the 64 section filter (i.e. much recordings on the same transponder are running)
Oct 05 12:15:49 <CIA-1> - deinterlacer optimizations
Oct 05 12:15:53 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * rd646a6b599 10openembedded.git/packages/enigma2/enigma2-plugins.bb: enigma2-plugins.bb: push srcdate
Oct 05 12:15:56 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox.release_25 * re6433dd4a5 10openembedded.git/packages/gstreamer/ (2 files in 2 dirs): packages/gstreamer: backport mpegpsdemux speedup
Oct 05 12:16:03 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox * reb2094bc9f 10openembedded.git/packages/dreambox/dreambox-dvb-modules.bb: (log message trimmed)
Oct 05 12:16:06 <CIA-1> dreambox-dvb-modules.bb: dm8000 driver update
Oct 05 12:16:08 <CIA-1> changes:
Oct 05 12:16:10 <CIA-1> --------
Oct 05 12:16:12 <CIA-1> video output core update
Oct 05 12:16:14 <CIA-1> add some picture enhancement options to /proc/stb/vmpeg/0 (disabled by default)
Oct 05 12:16:16 <CIA-1> pep_sharpness, pep_contrast, pep_saturation, pep_hue, pep_brightness
Oct 05 12:16:20 <CIA-1> 03Felix Domke <felix.domke@multimedia-labs.de> 07org.openembedded.dreambox * r233b01e164 10openembedded.git/: Merge branch 'org.openembedded.dreambox' of git://git.opendreambox.org/git/ghost/opendreambox-1.5 into org.openembedded.dreambox
Oct 05 12:16:24 <CIA-1> 03ghost <andreas.monzner@multimedia-labs.de> 07org.openembedded.dreambox * rd7e72cbc37 10openembedded.git/packages/dreambox/dreambox-secondstage.bb: 
Oct 05 12:16:27 <CIA-1> dreambox-secondstage.bb: update dm800 and dm8000 secondstage loader
Oct 05 12:16:33 <CIA-1> small webif bugfix
Oct 05 12:24:19 <recalcati> I have a funny question about OABI system call and gcc4. Is there, it seems, a project angstrom legacy for this kind of implementation. Any help will be appreciated. 
Oct 05 12:25:10 <pb__> I don't quite understand your question.  What exactly are you trying to do?
Oct 05 12:25:27 <recalcati> I need to link old libraries
Oct 05 12:25:31 <pb__> "OABI system call" and "gcc4" seem to be two unrelated things.  GCC doesn't do any system calls.
Oct 05 12:25:43 <hrw> recalcati: you want to do angstrom build for oabi?
Oct 05 12:25:46 <pb__> well, other than the trivial cacheflushing thing in libgcc, which you could easily enough override.
Oct 05 12:25:55 <recalcati> hrw:  yes
Oct 05 12:26:13 <hrw> recalcati: MACHINE=collie bitbake something
Oct 05 12:26:36 <hrw> recalcati: easiest way
Oct 05 12:26:56 <pb__> XorA: yes
Oct 05 12:27:09 <recalcati> collie is gcc4 but OABI? really. I check. Then I need pxa270 and iwmmx
Oct 05 12:27:14 <XorA> pb__: awesome, I may drive down then
Oct 05 12:28:14 <hrw> recalcati: iwmmxt and oabi works at all?
Oct 05 12:28:51 <zecke> recalcati: it will be a lot more easy to recompile the library
Oct 05 12:29:22 <zecke> recalcati: or in case of closed source (you should have known...) write some assembly to call the library and hope it will work
Oct 05 12:29:47 <recalcati> hrw: I don't understand. I had an old gcc3, but it hasn't iwmmx capability. So I move to gcc4, I add CONFIG_AEABI to k2.6. mplayer works, but I have an old libray, without sources, that I need to link
Oct 05 12:29:48 <eFfeM> hrw, pb__, thanks for the info, might have to fly with KLM in order to keep my award points and they only fly to lhr and lcy
Oct 05 12:30:19 <pb__> eFfeM: righto.  lhr is simpler to get to but takes a bit longer.
Oct 05 12:30:23 <recalcati> zecke: "assembly to call"   ....  if possible it will be the solution
Oct 05 12:30:45 <zecke> recalcati: well, you ask about closed source in a opensource forum, it is really not appropriate :)
Oct 05 12:30:58 <zecke> and gcc3.4 supports iwmmxt
Oct 05 12:31:04 <pb__> lhr -> cambridge involves a 50-minute subway ride, then one change, then another 50-minute train journey.  lcy->cambridge requires at least one extra subway change, possibly two.
Oct 05 12:31:10 <recalcati> ops, you are right
Oct 05 12:31:47 <hrw> recalcati: and lib is oldabi?
Oct 05 12:31:49 <recalcati> I have to check gcc3.4, so.
Oct 05 12:32:11 <recalcati> hrw: intel ipp
Oct 05 12:32:53 <recalcati> I'm sorry about my question.
Oct 05 12:35:17 <pb__> ah, just one extra change.  take dlr to bank, then change to northern line for kings cross.  total 35 minutes according to the tfl website.
Oct 05 12:35:33 <Jay7> well..
Oct 05 12:35:43 * Jay7 is going to make tosa usable
Oct 05 12:35:52 <XorA> DLR bloomin clockwork railway
Oct 05 12:35:56 <pb__> piccadilly line from lhr takes about 50 mins to kings cross, so dlr is 15 mins faster in theory.  whether that actually makes a difference obviously depends on the timings of the trains from kings x.
Oct 05 12:36:41 <pb__> kx -> cbg runs every 30 mins so the extra 15 minutes might or might not make the difference between getting on an earlier train or not,
Oct 05 12:37:36 <pb__> so, all in all, probably not much to choose between them.  whichever airport gives you the best flight time is probably the one to go for.
Oct 05 13:01:09 <jovox> Hi. I'm still working on my initramfs. What I'm trying todo now is to see if there is a usb-stick available. If it is, I will copy some files from it before switching into my "main" rootfs. In the /init script of the initramfs I see that it creates the nodes in /dev. All partitions on my harddrive is created but not the one on the usb memory stick. But the strange thing is that if I run the script (the one that creates the /dev nodes) manually after the init scri
Oct 05 13:03:09 <jovox> ah, thats why I need to pass a rootdelay as an agrument to the kernel!
Oct 05 13:03:15 * jovox checks
Oct 05 13:09:06 <eFfeM> jovox: also take care that all functionality is in the kernel, it could be that if things are modules they are not loaded yet
Oct 05 13:09:57 <Jay7> hm..
Oct 05 13:10:17 <Gnutoo> hi, I've an oops...can ksymoops produce something like gdb backtrace(that is to say with all functions names and code lines)
Oct 05 13:10:26 <Jay7> tosa x11-image builds xorg-server and hal
Oct 05 13:10:58 <Jay7> I can't find from where it takes this
Oct 05 13:11:15 <Gnutoo> I've tried increasing the debug level with -ddddd but still got assembly
Oct 05 13:11:44 <Gnutoo> things like [ 2034.526245] [<c00b5fe8>] (sys_ioctl+0x0/0x64) from [<c0026da0>] (ret_fast_syscall+0x0/0x2c)
Oct 05 13:11:50 <jovox> eFfeM, how do I know that the modules are loaded? The default script just sleeps for a number of seconds but that doesnt feel like a very safe way..
Oct 05 13:12:16 <Gnutoo> and I've no jtag and no serial
Oct 05 13:12:27 <Gnutoo> and the kernel is 2.6.27
Oct 05 13:12:33 <Gnutoo> s/27/29
Oct 05 13:13:11 <Gnutoo> maybe I should use kgdb over udp
Oct 05 13:13:26 <Gnutoo> I hope there's kgdb for this kenrel
Oct 05 13:13:51 <XorA> Jay7: if there is no XSERVER in machine.conf angstrom defaults to xorg now I think
Oct 05 13:14:18 <Jay7> XorA: XSERVER ?= "xserver-kdrive-fbdev"
Oct 05 13:14:22 <Jay7> from x11-image.bb
Oct 05 13:14:35 <Jay7> i.e. it should be defined elsewhere
Oct 05 13:14:42 <Jay7> ah.. may be from angstrom configs
Oct 05 13:14:47 <Jay7> well..
Oct 05 13:14:56 <XorA> Jay7: yeah its in anstrom-2008.1.conf
Oct 05 13:15:53 <Jay7> XorA: can you suggest what is better - stick to kdrive 1.3 or fix xorg-server + hal? :)
Oct 05 13:16:09 <XorA> Jay7: fix? its working for other machines
Oct 05 13:16:23 <ant_work> XorA: ah, and that udev script ?
Oct 05 13:17:00 <ant_work> I've seen thesing ended up with rw in fstab
Oct 05 13:17:07 <XorA> although I guess tosa will go wrong with Xorg due to that word only writes problem
Oct 05 13:17:18 <Gnutoo> mmm the one in kernel doesn't support ethernet and anyway I've no ethernet
Oct 05 13:17:23 <XorA> udev caching code is broken by design
Oct 05 13:17:30 <Gnutoo> so I'll enable frame pointers
Oct 05 13:17:36 <Gnutoo> and hope it'll produce better traces
Oct 05 13:17:36 <XorA> requires rw rootfs before rootfs is mounted rw
Oct 05 13:17:53 <ant_work> nasty, isn't?
Oct 05 13:17:56 <Gnutoo> indeed and it takes too much space as they said on the mailing list
Oct 05 13:18:45 <tasslehoff> hrw: trying the example from the link you gave me earlier. should I just add "require console-base-image.bb" in my recipe? I get "ERROR: Could not include required file console-base-image.bb while parsing foo.bb".
Oct 05 13:19:31 <Gnutoo> wow there's also -g
Oct 05 13:22:13 <jovox> If I put in a delay on 60 seconds the device is created in /dev..  I need to find a faster way to determine if there is a usb stick attached
Oct 05 13:23:31 <ant_work> jovox: use rootwait
Oct 05 13:23:41 <eFfeM> jovox as first cmd in the initscript you could do an lsmod > somefile
Oct 05 13:24:09 <jovox> rootwait sounds interesting.. will google it
Oct 05 13:25:48 <jovox> eFfeM, how would that help?
Oct 05 13:30:01 <jovox> rootwait doesnt seem to have any affect
Oct 05 13:37:18 <Jay7> XorA: there is no keyboard definition for tosa
Oct 05 13:37:26 <hrw> shit. uboot do not list mmc partitions
Oct 05 13:37:30 <Jay7> I can't use Fn+keys
Oct 05 13:37:39 <Jay7> only generic keys
Oct 05 13:37:47 <Jay7> touchscreen is not working too
Oct 05 13:38:06 <XorA> touchscreen should be using tslib like normal
Oct 05 13:39:59 <hrw> shit again
Oct 05 13:40:02 <hrw> ~mondays
Oct 05 13:40:03 <ibot> Sounds like _somebody_'s got a case of the Muhhndays!
Oct 05 13:40:38 <Jay7> XorA: I know that is _should_ :)
Oct 05 13:42:35 <jovox> I googled for roootwait and it looks like what I'm looking for.. But as I said, it doesnt wait for my devices to become ready. Does the kernel have to be prepared for supporting rootwait or something?
Oct 05 13:47:33 <ant_work> jovox: the kernel must have some stuff compiled in and not as module in order to boot from SD/MMC/USB
Oct 05 13:47:47 <ant_work> with or without rootwait
Oct 05 13:49:02 <jovox> ant_work, I'm not booting from SD/MMC/USB. I'm using the USB for upgrading the system. My initramfs will look for a upgrade file on the usb disk and the copy it into the main rootfs
Oct 05 13:51:53 <pb__> jovox: rootwait won't make any difference if your rootfs is not actually on the usb stick.
Oct 05 13:52:01 <ant_work> well, the initramfs init script could probably need some sleep() here and there
Oct 05 13:52:13 <pb__> it only waits until the root device becomes available, and if you are booting from initramfs it will be instantly available, no need to wait
Oct 05 13:53:02 <pb__> if you're trying to mount the thumbdrive from your own linuxrc script then you will have to account for the device settling time yourself using some kind of loop or delay.
Oct 05 13:53:08 <eFfeM> but if you have usb as modules they will not be loaded when you start your initramfs script
Oct 05 13:53:20 <eFfeM> sleep(1)
Oct 05 13:53:48 <pb__> well, sure, it's up to you to load any modules that are necessary.  obviously the thumbdrive will never appear if the driver isn't loaded.
Oct 05 13:54:14 <pb__> in this kind of situation it would make more sense to compile usb into the kernel statically though
Oct 05 13:55:32 <XorA> if your using mdev inside the initramfs you will get an event when the driver appears and you can run a script
Oct 05 14:06:21 <dth> der ist zwar teurer, hat aber den besseren temperaturbereich
Oct 05 14:06:23 <dth> http://focus.ti.com/docs/prod/folders/print/tvp5150am1.html
Oct 05 14:06:49 <dth> sorry, wrong window
Oct 05 14:07:22 <tasslehoff> I'm trying to make my own image based on console-base-image. In my foo.bb I put "require console-base-image.bb", and get "ERROR: Could not include required file console-base-image.bb while parsing foo.bb". Any ideas?
Oct 05 14:10:24 <eFfeM>  tasslehoff that is because there is no console-base-image.bb file
Oct 05 14:10:36 <eFfeM> oops, there is, in images
Oct 05 14:11:34 <eFfeM> what about using DEPENDS ?
Oct 05 14:12:15 <pb__> that won't help for images.  all that DEPENDS will achieve is building console-base-image before your own image, which is unlikely to be useful.
Oct 05 14:12:49 <pb__> if "require" isn't working then it might indicate that console-base-image.bb is simply not in the path where bitbake is looking for it.  it's hard to say anything very intelligent about that without knowing where the .bb file that is doing the requiring is located.
Oct 05 14:13:12 <pb__> on the other hand, console-base-image.bb is simple enough that I would be inclined just to copy and modify it rather than requiring it.
Oct 05 14:16:11 <tasslehoff> pb__: that's what I did as well, since I couldn't make it work :)
Oct 05 14:17:59 <pb__> heh
Oct 05 14:20:09 <Gnutoo> hi
Oct 05 14:20:17 <Gnutoo> how do I debug an oops without:
Oct 05 14:20:21 <Gnutoo> *jtag
Oct 05 14:20:23 <Gnutoo> *serial
Oct 05 14:20:32 <Gnutoo> so no jtag probe and no kgdb
Oct 05 14:20:36 <Gnutoo> *ethernet
Oct 05 14:20:40 <pb__> do you have any console output at all?
Oct 05 14:20:50 <woglinde> re
Oct 05 14:20:56 <tsjsieb> tasslehoff: it works fine here though, with foo.bb in the same dir as console-base-image.bb
Oct 05 14:20:59 <Gnutoo> hi woglinde 
Oct 05 14:21:44 <pb__> and is it just an oops, or is it a full kernel crash?
Oct 05 14:22:10 <pb__> if it is a full crash and you don't have any way of getting at the diagnostic output, you are basically stuffed.
Oct 05 14:22:28 <pb__> if it is just an oops, you can recover the error message from dmesg or /proc/kmsg
Oct 05 14:22:40 <tasslehoff> tsjsieb: putting it in the same directory worked here as well
Oct 05 14:22:42 <Gnutoo> pb__, for now just an oops
Oct 05 14:22:56 <pb__> in that case there should be no problem, just check dmesg for the details.
Oct 05 14:23:05 <Gnutoo> pb__, but dmesg only print function + assembly
Oct 05 14:23:17 <Gnutoo> I want something like in gdb bt
Oct 05 14:23:20 <Gnutoo> if possible
Oct 05 14:23:32 <pb__> if your kernel was built with backtraces then you should get them from dmesg.
Oct 05 14:23:52 <Gnutoo> [   92.138519] [<c00b7d38>] (sys_ioctl+0x0/0x64) from [<c0027da0>] (ret_fast_syscall+0x0/0x2c)
Oct 05 14:23:58 <Gnutoo> I'll rebuild it then
Oct 05 14:24:00 <pb__> ah, well, there you are.  that's the backtrace
Oct 05 14:24:16 <Gnutoo> yes but +0x64 is not the line in the C code?
Oct 05 14:24:35 <pb__> no, it's the address of the instruction.  use addr2line to convert it to a c line number.
Oct 05 14:24:45 <Gnutoo> ah yes I remember...thanks a lot!!!
Oct 05 14:24:58 <Gnutoo> I saw addr2line in the embedded linux primer book
Oct 05 14:25:13 <chouimat> morning
Oct 05 14:25:19 <Gnutoo> pb_, thanks a lot!!!
Oct 05 14:25:21 <pb__> or you can run cross-gdb on your host and inspect the kernel (statically) that way
Oct 05 14:25:26 <Gnutoo> ok
Oct 05 14:27:46 <pb__> woglinde: eclipse 3.5 does indeed seem to be a bit smaller than the old one.  only 1.5G at the moment.
Oct 05 14:28:56 <woglinde> :)
Oct 05 14:29:11 <woglinde> ~lart poulsbo
Oct 05 14:29:11 * ibot puts on some milking gloves.  "All right, now, poulsbo, this won't hurt a bit...."
Oct 05 14:32:21 <pb__> hm.  there seems to be something a bit wrong with junit though.
Oct 05 14:34:00 <woglinde> pb hm
Oct 05 14:34:06 <pb__> for some reason it doesn't seem to compile the tests before running them anymore
Oct 05 14:34:09 <woglinde> thats what I not tested
Oct 05 14:34:11 <pb__> which is a bit sad, obviously
Oct 05 14:34:26 <woglinde> maybee a new option?
Oct 05 14:34:50 <pb__> possibly, though it would be a bit of a shame if there is some cryptic "unbreak my tests" option that I need to check.
Oct 05 14:34:55 <woglinde> ~seen rp
Oct 05 14:34:58 <ibot> rp is currently on #oe (2d 11h 48m 26s), last said: 'ulitu: Obviously I go so far with implementing it and got distracted'.
Oct 05 14:35:17 <pb__> for the time being I guess I can work around it by running "mvn test" manually from the shell
Oct 05 14:35:36 <woglinde> hm so eclipse maven plugin is broken?
Oct 05 14:35:41 <woglinde> hi kergoth
Oct 05 14:35:47 <kergoth> hey woglinde 
Oct 05 14:36:15 <pb__> possibly.  I'm not entirely sure how responsibility is divided between eclipse itself and the mvn plugin.
Oct 05 14:37:15 <woglinde> so debugging drm_init now
Oct 05 14:37:24 <pb__> hi kergoth
Oct 05 14:38:09 <pb__> actually, I think I'm going to go back to my old eclipse, this is too tedious
Oct 05 14:38:15 <pb__> my new ram should be here tomorrow anyway
Oct 05 14:38:23 <woglinde> *g*
Oct 05 14:38:26 <kergoth> hey
Oct 05 14:42:31 <ant_work> pb__: it's fun when you have to debug the IDE
Oct 05 15:06:01 <pb__> yeah, that does make java development even more fun than it would be otherwise.
Oct 05 15:06:30 <pb__> ah joy, now I have unresolved compilation errors.
Oct 05 15:06:48 * pb__ stabs scala
Oct 05 15:09:07 <ant_work> pb__: if you want a break I can send you a couple of monstruous mysql queries to improve...(EDI-generators)
Oct 05 15:09:12 <ant_work> :=)
Oct 05 15:10:28 <pb__> heh, no thanks
Oct 05 15:11:01 <pb__> I think I will just go and stick hot needles into my eyeballs for a little while
Oct 05 15:11:06 <gregoiregentil> hyc: Thanks ;-)
Oct 05 15:11:12 <ant_work> every time I look at our database I reckon I should pay someone to rename all the fields...
Oct 05 15:11:54 <Jay7> ant_work: this is because programmers are not sales managers :)
Oct 05 15:12:17 <ant_work> no, its' because sale managers are not good programmers ;)
Oct 05 15:12:32 <pb__> yeah, we have that problem with our database too.
Oct 05 15:12:42 <pb__> the idea of renaming all the wrongly-labelled columns is too awful to contemplate though
Oct 05 15:13:11 <woglinde> pb you are working with scala?
Oct 05 15:13:22 <pb__> woglinde: perhaps more "against" than "with", but yes
Oct 05 15:13:38 <woglinde> hehe its nice features
Oct 05 15:13:44 <woglinde> arg it has
Oct 05 15:15:24 <ant_work> pb__: the original sin of our db is it was prototyped/deployed with M$Acce$$
Oct 05 15:17:51 <Jay7> omg..
Oct 05 15:18:07 <Jay7> ms access
Oct 05 15:18:41 <ant_work> he..I still have some frontends
Oct 05 15:19:34 <ant_work> imagine it was in Informix/SQL in the mid '80...then here was decided to go M$...
Oct 05 15:21:07 <pb__> woglinde: yeah, it's a great language, it's just that its integration with eclipse seems to leave something to be desired
Oct 05 15:21:14 <pb__> quite a lot to be desired, actually
Oct 05 15:21:59 <woglinde> hm can me someone help with kernel debugging
Oct 05 15:22:50 * hrw back to sd and uboot fight
Oct 05 15:23:11 <woglinde> I am gettin this oops
Oct 05 15:23:17 <woglinde> http://pastebin.ca/1594704
Oct 05 15:23:20 * XorA hands hrw #torpedo
Oct 05 15:23:31 <hrw> ?
Oct 05 15:24:01 <woglinde> objdump output is http://pastebin.ca/1594705
Oct 05 15:24:12 <pb__> hrw: for sinking the u-boot, I suppose
Oct 05 15:24:12 <hrw> woglinde: intel gfx?
Oct 05 15:24:13 <woglinde> I am not sure at which command it's breaking
Oct 05 15:24:19 <hrw> ah
Oct 05 15:24:46 <pb__> woglinde: the eip number is the address of the faulting instruction
Oct 05 15:25:02 <pb__> so, 0x38 bytes into drm_init(), whatever that is.
Oct 05 15:25:12 <XorA> hrw: yeah pb__ is right
Oct 05 15:25:15 <XorA> anyway time to go
Oct 05 15:29:39 <woglinde> hm
Oct 05 15:30:00 <woglinde> pb I posted the objdump
Oct 05 15:30:12 <ant_work> bbl
Oct 05 15:30:16 <pb__> groan, now the scala compiler is crashing with a stack overflow
Oct 05 15:30:26 <pb__> I guess it was a dumb plan to try upgrading it.
Oct 05 15:30:36 <woglinde> pb sorry to hear that
Oct 05 15:30:49 * pb__ reverts to scala 2.7.4
Oct 05 15:31:13 <pb__> it is a bit sad that every time I try to upgrade one of these things it seems to get worse in some way.
Oct 05 15:31:35 <pb__> maven 2.2.x is hopeless as well, I am forced to build on an old host with maven 2.0.x installed.
Oct 05 15:35:44 <hrw> ~seen florian
Oct 05 15:35:46 <ibot> florian <n=fuchs@port-217-146-132-69.static.qsc.de> was last seen on IRC in channel #oe, 6h 54m 25s ago, saying: 'if its the only one... '.
Oct 05 15:39:40 <woglinde> pb so its these instruction because of the 0x130?
Oct 05 15:39:43 <woglinde>  pid = (struct pci_device_id *)&driver->pci_driver.id_table[0];
Oct 05 15:39:54 <woglinde>  528:   8b 9f 30 01 00 00       mov    0x130(%edi),%ebx
Oct 05 15:40:09 <woglinde> or pointer
Oct 05 15:41:17 <pb__> woglinde: that sounds like a likely suspect, but it is impossible to tell from that output.  is 0x528 the instruction address that matches drm_init+0x38?
Oct 05 15:42:46 <woglinde> mov I will put objdump for this too
Oct 05 15:43:18 <woglinde> http://pastebin.ca/1594756
Oct 05 15:43:37 <woglinde> dont know how to count
Oct 05 15:44:11 <woglinde> unfornatly kgdb helps not this much
Oct 05 15:45:33 <woglinde> hm is it simple 510+x038?
Oct 05 15:46:17 <pb__> yes
Oct 05 15:46:37 <pb__> so, that would indicate that the faulting instruction is at 0x548, which does match the code from your original crash (0x8b ...)
Oct 05 15:47:01 <pb__> though, weirdly, "mov    0xc,%ebx" doesn't seem like a very likely instruction to be trapping, nor does the preceding lea.
Oct 05 15:47:10 <pb__> what was the actual error that you saw in the first place?
Oct 05 15:47:43 <woglinde> hm I am thinking something with pid is wrong
Oct 05 15:47:46 <hrw> stupid question I have.
Oct 05 15:48:11 <woglinde> but I am not sure
Oct 05 15:48:12 <Crofton> question stupid have i
Oct 05 15:48:12 <woglinde> pid = (struct pci_device_id *)&driver->pci_driver.id_table[0];
Oct 05 15:48:14 <hrw> SD cards normally formatted keeps partition table in MBR which is first 256 bytes of card - right?
Oct 05 15:48:32 <woglinde> and if you request something from pid the kernel bumps out
Oct 05 15:49:10 <pb__> hrw: no, the partition table is in the second 256 bytes
Oct 05 15:49:17 <hrw> thx
Oct 05 15:49:19 <woglinde> hm will test it again
Oct 05 15:49:31 <pb__> woglinde: what was the error message?
Oct 05 15:49:39 <hrw> pb__: ah.. right - 1st is x86 bootloader
Oct 05 15:50:09 <woglinde> pb which one?
Oct 05 15:50:23 <woglinde> http://pastebin.ca/1594704
Oct 05 15:50:29 <woglinde> or something above this
Oct 05 15:50:44 <pb__> the stuff above that
Oct 05 15:51:54 <woglinde> mom
Oct 05 15:53:09 <woglinde> pb BUG: unable to handle kernel NULL pointer dereference at 0000000c
Oct 05 15:54:07 <pb__> ah, maybe your objdump is broken
Oct 05 15:54:23 <pb__> I wonder if the instruction at 0x548 is actually "mov 0xc(%esi),%ebx"
Oct 05 15:54:44 <pb__> that'd make more sense and it would be consistent with the error
Oct 05 15:55:07 <pb__> in which case, yes, it would appear that pid is NULL in your case
Oct 05 15:55:59 <woglinde> hm can I test it with if (pid) DRM_DEBUG ?
Oct 05 15:56:19 <pb__> something like that, yeah
Oct 05 15:57:42 <eFfeM> ~seen likewise
Oct 05 15:57:43 <ibot> likewise <n=chatzill@82-171-51-231.ip.telfort.nl> was last seen on IRC in channel #oe, 6d 17h 10m 48s ago, saying: 'khem: hack lightly, and enjoy your evening'.
Oct 05 15:57:50 <pb__> ah joy, the scala compiler seems to be incompatible with java5
Oct 05 15:59:29 <woglinde> pb what?
Oct 05 15:59:39 <woglinde> only java6?
Oct 05 15:59:44 <pb__> yeah
Oct 05 15:59:53 <pb__> with java6 it runs fine, with java5 I get this crazy stack overflow
Oct 05 16:00:41 <woglinde> good to know
Oct 05 16:04:11 <woglinde> pb hm
Oct 05 16:04:19 <woglinde> seems as I suggested
Oct 05 16:05:39 <pb__> oh, hm, no, it isn't java5 making it crash.  that is a bit weird.
Oct 05 16:05:48 <woglinde> pb uh
Oct 05 16:05:53 <pb__> it seems to work fine on one computer and crash on the other although they both apparently have the same jvm
Oct 05 16:05:59 <pb__> bizarre
Oct 05 16:06:34 <pb__> I had forgotten to set JAVA_HOME to select java5 on the first one. :-}
Oct 05 16:06:52 <pb__> default appears to be java6 on all our hosts
Oct 05 16:07:33 <pb__> luckily, when I select java5 on the host where it was previously crashing, it now works fine :-)
Oct 05 16:07:48 <pb__> so, apparently I am now all set. 
Oct 05 16:08:54 <pb__> only an hour lost to crazy java issues, I guess that's not so bad.
Oct 05 16:15:13 <woglinde> yeah
Oct 05 16:15:15 <woglinde> thats okay
Oct 05 16:15:35 <pb__> now I can get back to actually fixing my testsuite failures :-}
Oct 05 16:48:51 <CIA-1> 03Jeremy Lainé <jeremy.laine@m4x.org> 07org.openembedded.dev * r280be5d65a 10openembedded.git/recipes/linux/ (linux-2.6.29/boc01/007-090825-lm73.patch linux_2.6.29.bb): linux-2.6.29: update patch for lm73 on boc01
Oct 05 16:59:05 <thesing> morning all
Oct 05 17:03:43 <woglinde> hi thesing
Oct 05 17:14:20 <CosmicPenguin> Say what you want about autotools, it sure is nice that all you have to do to update to a newer version of the X server is change a few version numbers...
Oct 05 17:18:30 <CIA-1> 03Jeremy Lainé <jeremy.laine@m4x.org> 07org.openembedded.dev * r5d021e5237 10openembedded.git/recipes/linux/ (3 files in 2 dirs): linux-2.6.29: update SPI patch for boc01
Oct 05 17:58:36 <Vink__> Question... 
Oct 05 17:59:25 <Vink__> After oe compilation success where does the root file system gets built in? which folder
Oct 05 18:00:29 <mlip> Vink__: tmp/deploy/<libc-type>/images
Oct 05 18:00:49 <mlip> with tmp being usually at the same level as openembedded
Oct 05 18:24:04 <zoid__> any admin here, that could submit 2 patches (rt3070, lighttpd) and a fix for wpa-supplicant? or should i file a bug for it? (or mail?)
Oct 05 18:27:33 <kergoth> zoid__: emailing the list is usually the best bet
Oct 05 18:34:36 <florian> re
Oct 05 18:34:49 <zoid__> kergoth: openembedded-devel ? the last 2 times i got no response there (but always helping ppl in irc ;p)
Oct 05 18:37:48 <kergoth> zoid__: yes.
Oct 05 18:37:51 <kergoth> you sent your patches?
Oct 05 18:38:07 <zoid__> did send 2 git diffs actually , ye 
Oct 05 18:38:07 <kergoth> hrmph, opkg's codebase is .. not pleasant
Oct 05 18:38:12 <kergoth> oh look, a leak..
Oct 05 18:39:55 <florian> hs
Oct 05 18:40:42 <hrw> florian: hi
Oct 05 18:40:43 <hrw> re
Oct 05 18:45:02 <pb_> kergoth: yah, it's pretty grim
Oct 05 18:46:07 <pb_> if oe e.V. ever gets some funds to spend on development bounties, a less sucky package manager would be near the top of my wishlist
Oct 05 18:46:26 <pb_> florian: good evening
Oct 05 18:46:28 * kergoth is making it use the default value rather than prompting if stdin isn't a tty
Oct 05 18:46:31 <kergoth> that would be very nice
Oct 05 18:47:11 <florian> pb_: ack
Oct 05 18:47:46 <florian> kergoth: that's a very good idea. 
Oct 05 18:48:33 <florian> kergoth: i submited a patch to fix selecting packages if one is available from multiple feeds yesterday. 
Oct 05 18:48:41 <kergoth> nice
Oct 05 18:54:23 <zoid__> kergoth: sent git diff to the mailinglist, maybe someone can look over it this time. (or look at the older ones)
Oct 05 19:16:50 <hrw> florian: did xserver-common 1.32 worked for you?
Oct 05 19:17:42 <florian> hrw: yes
Oct 05 19:18:21 <hrw> florian: cool. I will nominate it for stable/2009 as I need it for bug
Oct 05 19:18:56 <florian> hrw: maybe we should try some more devices but in fact i would like to see this too
Oct 05 19:19:07 * florian needs to build images anyway
Oct 05 19:19:48 <hrw> florian: as DEF_PREF=-1 ;)
Oct 05 19:21:20 <florian> :-)
Oct 05 19:29:43 <khem> morning all
Oct 05 19:30:21 <Crofton> gm
Oct 05 19:30:31 <Crofton> I am in Mountain View this week
Oct 05 19:30:55 <khem> Crofton: cool
Oct 05 19:31:09 <woglinde> hm can start a pxe boot from grub?
Oct 05 19:31:10 <khem> we can grab a lunch together if you like
Oct 05 19:31:24 <woglinde> someone has a howto?
Oct 05 19:31:50 <khem> Crofton: are you at google .. or somehwere else
Oct 05 19:32:48 <woglinde> hm okay found it
Oct 05 19:33:20 <kergoth> hmm, i wonder if itd be worth sharing a configure cache file amongst the autoconf based recipes, to avoid re-execution of some of the tests.  wonder if the speed increase would be noticable
Oct 05 19:34:03 <khem> kergoth: it is a good idea
Oct 05 19:34:09 <CosmicPenguin> Only if you had a lot of similar packages that did the same tests
Oct 05 19:34:15 <kergoth> course, then simultanious configure runs wouldn't be viable, you'd have to use a file lock around the config.cache
Oct 05 19:34:20 <woglinde> hm now I have to finish the moblin chroot
Oct 05 19:34:22 <khem> CosmicPenguin: most of the tests are common
Oct 05 19:34:27 <khem> like c library functions
Oct 05 19:34:28 <kergoth> CosmicPenguin: that's often the case, though.  the basic macros like AC_PROG_CC run cached tests
Oct 05 19:34:30 <hrw> woglinde: java for moblin?
Oct 05 19:34:30 <kergoth> yeah
Oct 05 19:34:35 <Crofton> khem, Ettus Galactic HQ
Oct 05 19:34:45 <woglinde> hrw nope
Oct 05 19:34:48 <CosmicPenguin> How expensive are those anyway, though?
Oct 05 19:34:52 <CosmicPenguin> I think it would go for stuff like xorg
Oct 05 19:34:54 <woglinde> crappy poulsbo
Oct 05 19:35:02 <woglinde> need to test if our hw differs
Oct 05 19:35:03 <kergoth> that's what i'm wondering, might have to try it out and profile it
Oct 05 19:35:10 <hrw> woglinde: gma500.. you are fucked
Oct 05 19:35:31 <woglinde> iegd should work
Oct 05 19:35:32 * kergoth made a quick shell script to run native configures against a bunch of recipes to gather up the test results, and noticed the configure runs were a bit speedier
Oct 05 19:36:31 <khem> kergoth: I am worried about feeding cached values to packages which may suddenly behave badly
Oct 05 19:36:43 <khem> because the tests work differently etc.
Oct 05 19:36:46 <kergoth> caching test results is an explicit operation.
Oct 05 19:36:51 <kergoth> they arent cached by default
Oct 05 19:36:58 <kergoth> you have to call a macro to do it
Oct 05 19:37:15 <kergoth> so if someone is doing that, its likely the test result is already prefixed to their own namespace and the like
Oct 05 19:37:17 <kergoth> i.e. sudo_cv_..
Oct 05 19:38:41 <khem> ideally yes that should be 
Oct 05 19:39:13 <kergoth> at any rate, one of those things to throw a prototype together of, and see if it's worth the trouble
Oct 05 19:39:27 <kergoth> would be easy enough to implement
Oct 05 19:39:51 <khem> would you keep the cache uptodate as you parse along and find new tests 
Oct 05 19:40:02 <khem> or rather run autoconf along
Oct 05 19:40:45 <kergoth> not sure what you mean.  i'm not suggesting manually maintaining the thing.  just add a --cache-file=${TMPDIR}/config.cache.${HOST_SYS} to EXTRA_OECONF
Oct 05 19:41:07 <khem> oh I see and then we will populate this file manually .
Oct 05 19:41:13 <kergoth> no
Oct 05 19:41:15 <kergoth> why would you?
Oct 05 19:41:24 <kergoth> each ./configure run will write out test results to it
Oct 05 19:41:28 <kergoth> which will be used by subsequent runs
Oct 05 19:41:43 <kergoth> we already have manually maintained test results, its called the site/ files :P
Oct 05 19:41:53 <khem> right thats what I was wondering
Oct 05 19:42:19 <khem> and my first question was about what you mentioned now
Oct 05 19:42:52 <khem> at build time you will have one ${TMPDIR}/config.cache.${HOST_SYS}
Oct 05 19:44:45 <kergoth> ah, right, yes
Oct 05 19:44:49 <khem> so it will be getting overwritten by every recipe
Oct 05 19:44:56 <khem> confugure run
Oct 05 19:45:30 <kergoth> true, it wouldn't likely build up results, so you'd only get the core stuff
Oct 05 19:45:39 * kergoth shrugs, might not be worthwhile in that case
Oct 05 19:45:47 <khem> configure will still execute/compile the tests it doesnt find in the cache
Oct 05 19:46:03 <CosmicPenguin> Yeah, you would want it to build up results over time
Oct 05 19:46:30 <khem> thats what I was thinking collect the results along the way
Oct 05 19:47:04 <khem> I dont know how it will behave in parallel builds
Oct 05 19:47:30 <kergoth> thats why i said you'd want a file lock around the global config.cache, which would mean no more simultanious configure task runs, which may lose you as much as the caching gains you
Oct 05 19:47:36 <hrw> or let autotools.do_configure keep lock on that file, do configure stuff;cat S/config.cache >>TMP/config.cache.HOST_SYS;sort -u TMP/config.cache.HOST_SYS;release lock
Oct 05 19:47:44 <khem> :)
Oct 05 19:48:14 <hrw> it could be fun to run it once and then store for next builds
Oct 05 19:48:30 <hrw> so ok, first run will be slower but then you can split it to site/ files...
Oct 05 19:48:30 <khem> incremental builds
Oct 05 19:49:18 <kergoth> anyone messed with x86_64 targets?  seems the site file is missing like half the tests that are in ix86-common
Oct 05 19:49:22 <khem> assumption is that the cached results yield same configuration second time
Oct 05 19:49:36 <hrw> kergoth: there is a branch for such ones
Oct 05 19:49:41 <kergoth> ah
Oct 05 19:49:49 <hrw> no idea how old
Oct 05 19:50:43 <kergoth> huh, roman's started messing with multilib stuff on his branch
Oct 05 19:51:14 <khem> in past I have had problems building with cached autoconf results but then here we are not supplying artificial results but results from previous runs
Oct 05 19:51:43 <kergoth> might want to grep out the ac_cv_env entries, they cache CC, CFLAGS, etc
Oct 05 19:51:48 * kergoth is doing so manually here
Oct 05 19:53:03 <hrw> hmm.. now just 947 packages in image... was 988 befoe
Oct 05 19:53:11 <woglinde> hrw *g*
Oct 05 19:53:25 <hrw> woglinde: the fun is that I *added* some
Oct 05 19:55:13 <hrw> found. less modules
Oct 05 19:57:16 <Gnutoo> hi just a question
Oct 05 19:57:28 <Gnutoo> if I use addr2line should I remove kenrel obtimizations
Oct 05 19:57:32 <Gnutoo> ?
Oct 05 19:57:38 <Gnutoo> addr2line on vmlinux
Oct 05 20:01:25 <khem> probably
Oct 05 20:01:40 <Gnutoo> ok thanks a lot
Oct 05 20:01:46 <khem> Gnutoo: it relies on debug info
Oct 05 20:01:53 <Gnutoo> because the book said to do that but not with addr2line
Oct 05 20:02:11 <khem> so it comes down to gcc how good info it can generate at higher opt levels
Oct 05 20:02:25 <CIA-1> 03Thomas Zimmermann <zimmermann@vdm-design.de> 07shr/import * r74f40e3fef 10openembedded.git/ (3 files in 3 dirs): 
Oct 05 20:02:25 <CIA-1> python-pygtk: update to 2.16.0
Oct 05 20:02:25 <CIA-1> Signed-off-by: Klaus Kurzmann <mok@fluxnetz.de>
Oct 05 20:02:26 <CIA-1> 03Thomas Zimmermann <zimmermann@vdm-design.de> 07shr/import * rda7339bf75 10openembedded.git/ (3 files in 3 dirs): 
Oct 05 20:02:28 <CIA-1> xserver-xorg: update to 1.7.0
Oct 05 20:02:30 <CIA-1> Signed-off-by: Klaus Kurzmann <mok@fluxnetz.de>
Oct 05 20:02:32 <CIA-1> 03Thomas Zimmermann <zimmermann@vdm-design.de> 07shr/import * r0bbdab0f75 10openembedded.git/recipes/efl1/emotion_svn.bb: 
Oct 05 20:02:35 <CIA-1> emotion: fix staging QA issue
Oct 05 20:02:37 <CIA-1> Signed-off-by: Klaus Kurzmann <mok@fluxnetz.de>
Oct 05 20:02:39 <CIA-1> 03Thomas Zimmermann <zimmermann@vdm-design.de> 07shr/import * rb1f98e6871 10openembedded.git/ (3 files in 2 dirs): 
Oct 05 20:02:42 <CIA-1> python-pygobject: update to 2.20.0
Oct 05 20:02:44 <CIA-1> Signed-off-by: Klaus Kurzmann <mok@fluxnetz.de>
Oct 05 20:02:55 <khem> yeah optimization and debug info has conflicting requirements you can get one on expense of other
Oct 05 20:11:24 <hrw> question... how much work would take adding new shlibs like code?
Oct 05 20:12:09 <hrw> Jar files has own way of providing dependency informations
Oct 05 20:13:06 <kergoth> you just add a function and add that function's name to the list of functions to be run by do_package
Oct 05 20:13:26 <kergoth> so the question is how much work is it to extract that information from the jar files
Oct 05 20:14:38 <hrw> extract one file from jar, parse it
Oct 05 20:16:32 <kergoth> so you'd probably have to emit some extra info into tmp and read it back in, to map the jar files dependency info to the recipes that provide them, and do the jar extraction bit.. pretty straightforward
Oct 05 20:17:13 <hrw> will add to todo
Oct 05 20:18:06 <khem> anybody here using llvm in some sort
Oct 05 20:20:20 <thesing> khem: lvm as in logical volume management?
Oct 05 20:20:45 <hrw> thesing: llvm is not lvm
Oct 05 20:20:48 <woglinde> ~lart moblin for not including nfsroot
Oct 05 20:20:48 * ibot slaps moblin upside the head with a wet fish for not including nfsroot
Oct 05 20:24:17 <Gnutoo> hi, what's the kenrel option to do -O0 ?
Oct 05 20:24:26 <Gnutoo> instead of -O2 or -Os
Oct 05 20:33:13 <Gnutoo> khem, so what options should I add....because my traces seem imprecise I've removed premptive but I also want to remove -O2
Oct 05 20:36:30 <khem> yesh remove -O2 and add -O
Oct 05 20:36:34 <khem> which is same as -O1
Oct 05 20:36:43 <Gnutoo> khem, yes but what's the kernel config for that: CONFIG_
Oct 05 20:36:51 <khem> and will give you better results
Oct 05 20:37:23 <Gnutoo> I don't find it in xconfig because I don't know the name of the config option
Oct 05 20:37:37 <woglinde> hm what I have to do do recompile moblin kernel?
Oct 05 20:37:47 <woglinde> rpms are installed
Oct 05 20:37:57 <woglinde> but config is strange
Oct 05 20:38:42 <Gnutoo> woglinde, mobilin can be built with oe?
Oct 05 20:38:53 <hrw> woglinde: read Fedora docs about compiling kernel
Oct 05 20:39:24 <woglinde> hrw okay
Oct 05 20:39:27 <woglinde> gnutoo no
Oct 05 20:39:40 <Gnutoo> or I'm curious what are you doing exactly? porting mobilin to arm?
Oct 05 20:40:09 <woglinde> ah --target I guess
Oct 05 20:40:33 * hrw recompiled fedora kernel once. no more
Oct 05 20:40:39 <khem> Gnutoo: you have to edit the toplevel makefile
Oct 05 20:40:46 <Gnutoo> ah ok thanks a lot
Oct 05 20:41:17 <khem> Gnutoo: look for CONFIG_CC_OPTIMIZE_FOR_SIZE and edit the else part
Oct 05 20:41:25 <Gnutoo> ok
Oct 05 20:41:54 <khem> thesing: llvm.org
Oct 05 20:43:06 <woglinde> moblin is somehow smarter
Oct 05 20:43:18 <khem> woglinde: in what way
Oct 05 20:44:15 <Gnutoo> thanks a lot!!!
Oct 05 20:49:02 <Gnutoo> mmm
Oct 05 20:49:36 <Gnutoo> http://www.pastebin.ca/1595700
Oct 05 20:49:48 <Gnutoo> /tmp/cc3ImTk4.s:2539: Error: selected processor does not support `ldrexb r0,[r3]
Oct 05 20:49:51 <Gnutoo> and it's arm
Oct 05 20:49:55 <Gnutoo> arch_msm
Oct 05 20:49:56 <Gnutoo> etc...
Oct 05 20:50:55 <khem> exclusive ld/st are armv6 and up
Oct 05 20:51:18 <khem> so if you chose ARM family which is older than armv6 thats what you get
Oct 05 20:51:18 <hrw> khem: vfp or not?
Oct 05 20:51:32 <khem> doesnt matter
Oct 05 20:52:53 <Gnutoo> no vfp
Oct 05 20:52:58 <Gnutoo> but armv6
Oct 05 20:53:01 <Gnutoo> ok
Oct 05 20:53:12 <Gnutoo> I've armv6-novfp
Oct 05 20:53:24 <hrw> ah.. msm right
Oct 05 20:53:34 <Gnutoo> yes
Oct 05 20:53:42 <Gnutoo> the -O0 triggered it
Oct 05 20:53:56 <khem> Gnutoo: kernel may not work well with O0
Oct 05 20:54:32 <khem> use V=1 on your commandline and post the commandline invocation for failing case
Oct 05 20:54:47 <Gnutoo> ok I knew v1
Oct 05 20:55:25 <Gnutoo> let's see with -O1
Oct 05 20:55:28 <Gnutoo> seem to pass the error
Oct 05 20:56:52 <Gnutoo> I hope it'll work this time
Oct 05 20:57:21 <khem> I think with optimization the error just gets smeared away it still is there
Oct 05 21:04:15 <hrw> bye
Oct 05 21:11:01 <Gnutoo> nooo it didn't change a thing...I'll have to debug manually
Oct 05 21:11:12 <Gnutoo> I'll pastebin code
Oct 05 21:12:23 <Gnutoo> http://www.pastebin.ca/1595777
Oct 05 21:12:26 <Gnutoo> it fails at:
Oct 05 21:12:44 <Gnutoo> line 8 of the pastebin
Oct 05 21:13:13 <Gnutoo> if I remove the function call before and replace it by -1 it works
Oct 05 21:13:59 <Gnutoo> ah wait a sec
Oct 05 21:16:23 <Gnutoo> maybe I should read the stacktrace manual
Oct 05 21:16:28 <Gnutoo> s/stacktrace/oops
Oct 05 21:20:51 <Gnutoo> ok got the line
Oct 05 21:24:44 <Gnutoo> the line was on top of the trace I didn't saw it
Oct 05 21:24:52 <Gnutoo> s/line/line number
Oct 05 21:35:25 <ant__> woglinde: buh
Oct 05 21:35:27 <CIA-1> 03Andrea Adami <andrea.adami@gmail.com> 07org.openembedded.dev * rd9ec7f1a8d 10openembedded.git/recipes/kexec/files/kexec-tools-2-klibc.patch: kexec-tools_klibc-static_2.0.1: rewrite some functions for klibc compatibility (getline, fscanf). Now it builds...WIP
Oct 05 21:35:40 <ant__> ..but segfaults :D
Oct 05 21:57:01 <woglinde> moo
Oct 05 21:57:05 <woglinde> going home now
Oct 05 22:20:30 <thesing> night all
Oct 05 23:02:59 <denix> anybody alive?
Oct 05 23:03:14 <Jay7> kinda
Oct 05 23:03:34 <denix> half-alive? :)
Oct 05 23:04:11 <denix> I have a Q about bitbake's version sorting...
Oct 05 23:04:17 <Jay7> half-life :)
Oct 05 23:04:33 <Jay7> not sure that I can answer this, but.. :)
Oct 05 23:05:04 <denix> I'm banging my head against the wall today
Oct 05 23:05:32 <denix> bitbake selects 2.6.28-r3 instead of 2.6.28+2.6.29-rc3-r3 :(
Oct 05 23:05:33 <Jay7> known effect :)
Oct 05 23:06:13 <Jay7> hm..
Oct 05 23:17:17 <ant__> thx Jay7 
Oct 05 23:17:18 <CIA-1> 03Andrea Adami <andrea.adami@gmail.com> 07org.openembedded.dev * rc518a888a6 10openembedded.git/recipes/kexec/ (2 files in 2 dirs): kexec-tools-klibc-static_2.0.1: revised patchset/Tested on arm. Bump PR.
Oct 05 23:28:25 * Jay7 -> sleep
Oct 05 23:42:19 <CIA-1> 03Stanislav Brabec <utx@penguin.cz> 07org.openembedded.dev * r06c06f67c7 10openembedded.git/recipes/xorg-xserver/ (files/report-correct-randr12.patch xserver-kdrive_1.3.0.0.bb): xserver-kdrive-1.3.0.0: Report correct XRandR version. Fixes non-working GTK+.
Oct 06 01:00:01 <Vink__> Hi, I have compiled oe using 'bitbake x11-image'
Oct 06 01:00:21 <Vink__> how do I add firefox to my rootfs now?
Oct 06 01:06:28 <grg> Vink__, you can copy the x11-image.bb to a new recipe file and add extra stuff to the IMAGE_INSTALL variable.
Oct 06 01:06:58 <kergoth> or IMAGE_INSTALL_append_pn-x11-image = " foo", or ...
Oct 06 01:07:01 <kergoth> :)
Oct 06 01:11:36 <Vink__> grg, kergoth   thanks
Oct 06 01:12:56 <grg> the append method is probably better if you only want to add a few things. I copied (a different image file) because I wanted to remove stuff as well.
Oct 06 01:13:55 <Vink__> yea, I too want to add/ remove stuffs to it
Oct 06 01:14:18 <Vink__> do we have a list of all the packages?
**** ENDING LOGGING AT Tue Oct 06 02:59:58 2009