**** BEGIN LOGGING AT Sun May 16 02:59:57 2010
May 16 03:50:41 <IgorK> khem: thank u kde terminal works fine now
May 16 05:39:03 <IgorK> ALL, is it possible to issue "bitbake <packagename>" so that it will use existing source tree?
May 16 06:37:10 <dm8tbr> IgorK: look at bitbake documentation, specifically the -f -c part of it. You can reuse a previously populated work directory that way
May 16 07:02:40 <ericben> hi
May 16 07:48:15 <khem> ericben: hello
May 16 07:51:17 <CIA-2> 03Khem Raj <raj.khem@gmail.com> 07org.openembedded.dev * r4d76ed2903 10openembedded.git/ (3 files in 3 dirs): 
May 16 07:51:17 <CIA-2> linux-2.6.33: Make 2.6.33 as default kernel for qemux86
May 16 07:51:17 <CIA-2> Signed-off-by: Khem Raj <raj.khem@gmail.com>
May 16 07:55:58 <ericben> khem: hi
May 16 07:56:25 <khem> ericben: are you in .de ?
May 16 07:56:38 <ericben> khem: .fr (I'm near Bordeaux)
May 16 07:57:27 <ericben> 
May 16 07:58:16 <ericben> khem: is the following fix correct of can the problem be related to problems other people seems to have with gtk ? http://patchwork.openembedded.org/patch/2048/
May 16 07:59:12 <ericben> it seems the is a dependency problem when building gtk native (or gtk) with several threads
May 16 08:18:12 <khem> ericben: yeah seems like we have libmimedir for gnome
May 16 08:18:16 <khem> so that dep is ok
May 16 08:22:14 <ericben> khem: ok, when updating things like this in recipes, should I always increase the PR ?
May 16 08:25:04 <khem> yes look at the wiki
May 16 08:25:20 <khem> there are several docs which help in writing good patches
May 16 08:27:13 <ericben> ok so there is no case where we may mdoify a recipe without increasing the PR ?
May 16 08:27:23 <ericben> (just to be sure ;-)
May 16 08:27:59 <khem> yes 
May 16 08:28:10 <ericben> ok thanks
May 16 08:28:12 <khem> when there is no changes in the output packages
May 16 08:28:22 <khem> then the PR remains same
May 16 08:28:36 <khem> for example adding a new machine to  linux kernel
May 16 08:28:58 <ericben> ah OK
May 16 08:29:00 <khem> if the resulting code changes then PR bump is needed
May 16 08:29:14 <khem> btw. I dont think this fix should be needed
May 16 08:29:24 <khem> I wonder why gobjects are being pulled in
May 16 08:29:43 <ericben> in libmime ?
May 16 08:29:46 <khem> yeah
May 16 08:30:05 <ericben> there seems to be soomething strange with gobject, GNUtoo also has problems with this
May 16 08:30:09 <khem> are we using some option to enable it
May 16 08:30:34 <khem> hmm
May 16 08:33:11 <ericben> in libmimedir this seems to be required as a default
May 16 08:33:22 <ericben> by mimedir-attachment
May 16 08:33:22 <khem> hmm
May 16 08:33:38 <khem> and is there a way we can disable it
May 16 08:33:41 <ericben> I don't know anything about libmimedir, that just happenned when trying to build a sdk
May 16 08:33:49 <ericben> I'm looking in the sources now
May 16 08:34:58 <ericben> and there don't seem to be an easy way to remove it as it's in libmimedir_la_SOURCES in Makefile.am
May 16 08:35:31 <ericben> and I don''t see any configure option nor define to prevent it from being built
May 16 08:35:41 <ericben> so this may be a compulsory part on the lib 
May 16 08:35:52 <ericben> of the lib sorry
May 16 08:38:47 <khem> ok 
May 16 08:39:10 <khem> but then it should be called somethin like gnome-libmime or somesuch
May 16 08:40:06 <ericben> I think glib is not only used for gnome no ?
May 16 08:42:53 <ericben> in fact libmimedir-dev is present in task-sdk-base and there don't seems to be any gnome thing in this recipe. Users of libmimedir are gpe tools and vtype-utils
May 16 08:44:20 <CIA-2> 03Eric BENARD <eric@eukrea.com> 07org.openembedded.dev * r31402cd6f9 10openembedded.git/recipes/psplash/ (12 files in 6 dirs): (log message trimmed)
May 16 08:44:20 <CIA-2> psplash : update to latest SVN
May 16 08:44:20 <CIA-2> - update to latest SVN
May 16 08:44:20 <CIA-2> - update file names : psplash-hand-img.h -> psplash-poky-img.h
May 16 08:44:20 <CIA-2> - fix comments
May 16 08:44:20 <CIA-2> - fix psplash-18bpp.patch (missing break in the switch)
May 16 08:44:20 <CIA-2> Signed-off-by: Eric Benard <eric@eukrea.com>
May 16 08:44:32 <CIA-2> 03Eric BENARD <eric@eukrea.com> 07org.openembedded.dev * r7c196e6061 10openembedded.git/recipes/evtest/evtest_1.25.bb: 
May 16 08:44:33 <CIA-2> evtest: add libxml2 in DEPENDS
May 16 08:44:33 <CIA-2> else I get the following log as libxml2 is not yet built :
May 16 08:44:33 <CIA-2> | evtest-capture.c:33:29: error: libxml/encoding.h: No such file or directory
May 16 08:44:33 <CIA-2> | evtest-capture.c:34:30: error: libxml/xmlwriter.h: No such file or directory
May 16 08:44:33 <CIA-2> Signed-off-by: Eric Benard <eric@eukrea.com>
May 16 08:44:36 <CIA-2> Signed-off-by: Khem Raj <raj.khem@gmail.com>
May 16 08:44:37 <CIA-2> 03Eric BENARD <eric@eukrea.com> 07org.openembedded.dev * r89c9d4781e 10openembedded.git/recipes/svgalib/ (4 files in 2 dirs): (log message trimmed)
May 16 08:44:37 <CIA-2> svgalib : add svgalib-1.9.25
May 16 08:44:37 <CIA-2> - original recipe comes from :
May 16 08:44:37 <CIA-2> http://lists.linuxtogo.org/pipermail/openembedded-devel/2010-January/016607.html
May 16 08:44:37 <CIA-2> - added fix for compilation against recent kernel
May 16 08:44:38 <CIA-2> - updated and tested on x86
May 16 08:44:38 <CIA-2> - disable helper module
May 16 09:05:15 <CIA-2> 03Martin Jansa <martin.jansa@gmail.com> 07org.openembedded.dev * rf77e87c45f 10openembedded.git/recipes/eglibc/ (eglibc_2.10.bb eglibc_2.11.bb eglibc_2.9.bb): 
May 16 09:05:15 <CIA-2> eglibc: append SRCPV to PR
May 16 09:05:15 <CIA-2> * Automatically pick SRCREV changes without bumping PR manually
May 16 09:05:15 <CIA-2> Signed-off-by: Martin Jansa <martin.jansa@gmail.com>
May 16 09:05:15 <CIA-2> Signed-off-by: Khem Raj <raj.khem@gmail.com>
May 16 10:53:40 <GNUtoo> Tartarus, hi, I bitbaken pango-native,cairo-native and atk-native and no changes at all
May 16 11:04:06 <ericben> hi GNUtoo
May 16 11:12:23 <lgopher> Hi folks, I'm building dev, distro angstorm. I have the PARALLEL_MAKE = "-j4" and BB_NUMBER_THREADS = "4". I can see parallel threads running, but make is running with 1 thread. When I use devshell then I see all my 4 cores busy, but running the same recepie with bitbake makes only 1 core busy. BTW bitbake 1.8.18. Does anynone has any idea ?
May 16 11:21:59 <GNUtoo> ericben, hi
May 16 11:22:04 <ericben> hi lgopher, here with -j8 the 8 cpu* are > 90%
May 16 11:22:22 <GNUtoo> ericben, now that you have commit access you will be able to push things like svgalib
May 16 11:22:43 <ericben> GNUtoo: Khem did this morning
May 16 11:22:49 <GNUtoo> ok
May 16 11:22:57 <lgopher> ericben: hm, this is what I see when I build my package from devshell
May 16 11:23:49 <lgopher> ericben: and in devshell I see -j4 set
May 16 11:25:58 <CIA-2> 03Michael 'Mickey' Lauer <mickey@vanille-media.de> 07org.openembedded.dev * r8b933151ce 10openembedded.git/recipes/nmon/nmon_13g.bb: nmon: new recipe; flexible ncurses-based performance monitor
May 16 11:25:59 <CIA-2> 03Michael 'Mickey' Lauer <mickey@vanille-media.de> 07org.openembedded.dev * r6c0c0f7dac 10openembedded.git/recipes/tasks/task-cli-tools.bb: task-cli-tools: add nmon
May 16 11:47:47 <CIA-2> 03Koen Kooi <koen@openembedded.org> 07org.openembedded.dev * r1773014a1a 10openembedded.git/recipes/libdiscid/libdiscid.inc: libdiscid: convert to new style staging
May 16 11:58:35 <ericben> it seems that my psplash patch that got applied this morning brokes psplash for angstrom, (and other distro/machines I think) I'm preparing a fix
May 16 13:03:12 <lgopher> Folks, I need a help with my new shiny recepie http://gist.github.com/402865#file_icee.inc It depends on mcpp-native and I see the mcpp-native compiled and installed in the sysroot.  The problem I'm facing that linking says there is no such library -lmcpp. I assume that Makefile doesn't respect oe flags for native packages. Is there some standard way to convince to look in the sysroot ?
May 16 13:56:21 <dcordes> what does PR = "n" in a git kernel recipe refer to?
May 16 14:11:33 <playya__> lgopher, i think you have to add the necessary -L and -I option to the CFLAGS
May 16 14:12:20 <lgopher> playya__: yep, I figured that out, the Makefile overwrites LDFLAGS
May 16 14:12:38 <playya__> eh. yes LDFLAGS
May 16 14:13:19 <lgopher> playya__: now I have another problem "new rpath '$ORIGIN/../../../../staging/x86_64-linux/usr/lib' too large; maximum length 46" not sure, what's this about.
May 16 14:18:32 <dcordes> GNUtoo, ping
May 16 14:22:01 <GNUtoo> dcordes, hi
May 16 14:22:45 <dcordes> GNUtoo, hello. can you help me with my question? I need to find out the correct way to determine the correct PR in kernel (git) recipes.
May 16 14:23:01 <GNUtoo> dcordes, ah ok
May 16 14:23:19 <GNUtoo> dcordes, the kenrel is autorev or something like that?
May 16 14:23:41 <dcordes> GNUtoo, no it's fixed revision by hash
May 16 14:23:56 <GNUtoo> ah ok
May 16 14:23:58 <dcordes> GNUtoo, recipes/linux/linux-qsd8_git.bb
May 16 14:24:04 <GNUtoo> ok I'll look
May 16 14:24:49 <GNUtoo> PR = "r13"
May 16 14:25:04 <GNUtoo> I think it work as usual
May 16 14:27:59 <dcordes> GNUtoo, I switched to branch htc-msm-2.6.32 and updated SRCREV to latest in that branch and put correct PV
May 16 14:29:36 <dcordes> GNUtoo, and changed PR to 9 since Makefile: EXTRAVERSION = .9
May 16 14:29:54 <GNUtoo> Makefile: EXTRAVERSION = .9 ???
May 16 14:29:56 <GNUtoo> what's that?
May 16 14:30:21 <GNUtoo> basically each time you modify the recipe you should bump PR by one
May 16 14:30:31 <GNUtoo> so it get rebuild
May 16 14:30:51 <GNUtoo> not shure how it works with git but I bet it's the same thing
May 16 14:30:56 <GNUtoo> also git is not sortable
May 16 14:31:00 <dcordes> GNUtoo, line 4 from the kernel Makefile !!
May 16 14:31:04 <GNUtoo> ok
May 16 14:31:18 <GNUtoo> dcordes, the one who shows in uname -a?
May 16 14:31:20 <dcordes> ah ok. now I get the meaning of PR
May 16 14:31:58 <GNUtoo> ok
May 16 14:31:58 <mickeyl> PR is what you want. It's a revision of a recipe, nothing more, nothing less. If something changes the output of said recipe, it's good practice to increment PR.
May 16 14:32:29 <GNUtoo> dcordes, sorry the question should have been: "what is PR", rather than "hello. can you help me with my question? I need to find out the correct way to determine the correct PR in kernel (git) recipes."
May 16 14:32:47 <GNUtoo> this confused me,sorry
May 16 14:32:52 <mickeyl> i.e. while PR _can_ resemble a part of the original software version, it not necessarily has to.
May 16 14:33:15 <dcordes> GNUtoo, I'm no robot
May 16 14:33:40 <dcordes> mickeyl, alright thanks
May 16 14:33:43 <GNUtoo> s/first sorry/sorry but
May 16 14:33:47 <GNUtoo> dcordes, np
May 16 14:34:02 <GNUtoo> dcordes, what's the status of leo+nexusone?
May 16 14:34:43 <GNUtoo> do someone have nexusone btw? because if so you could copy the bootloader part of htcdream 
May 16 14:34:54 <dcordes> GNUtoo, don't have nexusone and I'm working on leo status. need to finish this console image :)
May 16 14:35:01 <GNUtoo> ok
May 16 14:35:02 <mickeyl> kensan in #htc-linux has a nexus IIRC
May 16 14:35:06 <GNUtoo> ok
May 16 14:35:15 <dcordes> the kernel failed with gcc 4.4.4 I'm hoping for it to work with 2.6.32. that's why I updated the recipe
May 16 14:35:33 <dcordes> it's important to use recent gcc with that platform
May 16 14:35:48 <GNUtoo> ok I eared that
May 16 14:35:53 <GNUtoo> I bet it
May 16 14:35:57 <GNUtoo> 's an armv7
May 16 14:36:07 <GNUtoo> that's why you need a recent gcc
May 16 14:36:16 <GNUtoo> QSD is armv7 no?
May 16 14:38:38 <dcordes> that's right
May 16 14:40:04 <dcordes> mickeyl, I'm looking forward to his success with fso2 on that phone :)
May 16 14:41:30 <mickeyl> or french colleagues from hackable devices already did phone calls
May 16 14:41:44 <mickeyl> looks like it works perfectly with the htc_qualcomm_msm abstraction
May 16 14:42:22 <GNUtoo> phone calls on leo or on nexus?
May 16 14:42:29 <mickeyl> nexus
May 16 14:42:34 <GNUtoo> wow nice
May 16 14:42:53 <GNUtoo> I saw a page about n900 on the wiki of FSO
May 16 14:43:08 <dcordes> mickeyl, do you have a link?
May 16 14:43:31 <mickeyl> dcordes: just a mailing list post on smartphones-userland yesterday on friday
May 16 14:43:58 <mickeyl> GNUtoo: yes, electranox in #openmoko-de is working on it, it's a major task though, so i'm not sure whether it's going to go anywhere
May 16 14:44:11 <GNUtoo> ah ok
May 16 14:50:59 <lgopher> could somebody help me with following error "new rpath '$ORIGIN/../../../../../../staging/x86_64-linux/usr/lib' too large; maximum length 46"
May 16 14:59:59 <CIA-2> 03Michael 'Mickey' Lauer <mickey@vanille-media.de> 07org.openembedded.dev * r2e51f95364 10openembedded.git/recipes/rfkill/rfkill_0.4.bb: rfkill: new recipe; rfkill configuration utility
May 16 15:04:02 <eFfeM1> mickeyl: can you remove the eFfeM/sanity branch from git for me ? decided not to venture further on this and after my vacation it is probably grossly outdated too
May 16 15:05:39 <mickeyl> eFfeM: ok
May 16 15:06:46 <mickeyl> done
May 16 15:07:04 <eFfeM> ty
May 16 15:31:19 <CIA-2> 03Koen Kooi <koen@openembedded.org> 07org.openembedded.dev * r1c32fe23f5 10openembedded.git/recipes/psplash/ (files/angstrom/psplash-poky-img.h psplash.inc): psplash: fix angstrom logo
May 16 15:47:33 <jesusmad3> is there any problem with webcams or with v4l2 in Angstrong? I can only get 10-15 frames per second.. while tryng the exact thing in my PC I get 30
May 16 15:55:15 <CIA-2> 03Stiffmaster <luke-g@laptop.(none)> 07org.openembedded.dev * r6b4f7c1fbe 10openembedded.git/recipes/linux/ (linux-qsd8/htcleo/defconfig linux-qsd8_git.bb): linux-qsd8: move to new branch
May 16 16:10:43 <mickeyl> whoever "stiffmaster" is, please fix your git config
May 16 16:14:10 <ericben> hi, may anyone with commit access push this patch which unbreaks psplash ? http://thread.gmane.org/gmane.comp.handhelds.openembedded/32716 
May 16 16:14:39 <CIA-2> 03Khem Raj <raj.khem@gmail.com> 07org.openembedded.dev * r08983f3421 10openembedded.git/recipes/psplash/ (5 files in 5 dirs): 
May 16 16:14:39 <CIA-2> psplash: Rename HAND_* defines to POKY_* in all remaining psplash-poky-img.h
May 16 16:14:39 <CIA-2> Signed-off-by: Khem Raj <raj.khem@gmail.com>
May 16 16:14:46 <woglinde> didnt khem pushed your psplash fix?
May 16 16:14:50 <woglinde> hihi
May 16 16:15:01 <woglinde> ericben no rw access yet?
May 16 16:15:33 <ericben> well it seems we were 3 to update this
May 16 16:16:27 <ericben> woglinde: I think it's not yet enabled
May 16 16:16:46 <woglinde> hm
May 16 16:16:51 <woglinde> mickeyl?
May 16 16:17:01 <ericben> khem: I'm going to rebase my patch as it fix other things especialy for pandora
May 16 16:23:42 <ericben> woglinde: I've just sent a v4 to the list, as I'm not sure if I can push my you please do it ?
May 16 16:24:15 <woglinde> ericben test it
May 16 16:24:29 <ericben> ok soo I must read the doc ;-)
May 16 16:24:45 <ericben> to avoid to push unwanted things
May 16 16:24:53 <woglinde> right
May 16 16:25:04 <woglinde> try
May 16 16:25:06 <woglinde> git push origin org.openembedded.dev:org.openembedded.dev
May 16 16:25:21 <woglinde> or a pull
May 16 16:25:27 <woglinde> with ssh credential
May 16 16:25:33 <woglinde> which really harmless
May 16 16:25:39 <ericben> ok I'll try
May 16 16:28:32 <Tartarus> There's always git push --dry-run, too :)
May 16 16:28:41 <ericben> woglinde: ok the clone works, I was not aware my key was already active, I was waiting for the email I read on the wiki ;-)
May 16 16:28:53 <woglinde> *g*
May 16 16:29:04 <woglinde> seems mickeyl did it in silience
May 16 16:29:36 <woglinde> ericben now you can push as you want
May 16 16:29:42 <woglinde> besides core stuff
May 16 16:29:49 <woglinde> which should be discussed on ml
May 16 16:30:38 <ericben> woglinde: ok thanks, I'll try in a few minutes, I'm reading gitphrasebook
May 16 16:30:53 <woglinde> okay
May 16 16:35:07 <CIA-2> 03Eric Benard <eric@eukrea.com> 07org.openembedded.dev * r2abd81d7a8 10openembedded.git/recipes/psplash/ (6 files in 3 dirs): 
May 16 16:35:07 <CIA-2> psplash: unbreak recipe
May 16 16:35:07 <CIA-2> psplash was broken by my previous patch for several machines / distro
May 16 16:35:07 <CIA-2> http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=31402cd6f9d1e7887926bfde465bb25b07f6dbea
May 16 16:35:07 <CIA-2> this patch should fix this and is compile tested for kaeilos and angstrom distro
May 16 16:35:08 <CIA-2> and for boc01 and omap3pandora machines
May 16 16:35:09 <CIA-2> Signed-off-by: Eric Benard <eric@eukrea.com>
May 16 16:35:28 <ericben> ok that works :-)
May 16 16:35:31 <woglinde> ah ericben
May 16 16:35:34 <woglinde> commit msg
May 16 16:35:51 <woglinde> try packaganame/file: shortdesc
May 16 16:35:57 <woglinde> than blank line
May 16 16:36:10 <ericben> ok the blank line will prevent to have so many things here ?
May 16 16:36:11 <woglinde> and than starting with "* "
May 16 16:36:17 <woglinde> right
May 16 16:36:26 <woglinde> and is better for searching
May 16 16:36:51 <ericben> ok I take note (I already read this somewhere ... but didn't apply)
May 16 16:36:51 <woglinde> and than starting with "* " the long description if neccessary
May 16 16:36:55 <woglinde> re mickeyl
May 16 16:56:42 <GNUtoo> ericben, you're involved in pandora?
May 16 17:02:01 <IgorK> Hi ALL
May 16 17:02:19 <IgorK> I compiled wxwidgets using the dev shell
May 16 17:02:39 <IgorK> Now I need to popul;ate them and put them into the x11-gpe-image
May 16 17:02:54 <IgorK> is it possible to do with just one command?
May 16 17:09:03 <ericben> GNUtoo: no
May 16 17:09:11 <GNUtoo> ok
May 16 17:11:27 <ericben> one detail : is a é permitted in commit log or commiter's name ?
May 16 17:15:11 <woglinde> ericben what?
May 16 17:18:20 <ericben> a 'é' (e with accent)
May 16 17:18:43 <woglinde> ericben hm
May 16 17:18:48 <woglinde> good question
May 16 17:18:52 <woglinde> ask it on the ml
May 16 17:18:57 <IgorK> something like "bitbake -f -c use-wxwidgets-precompiled x11-gpe-image"? ;-)
May 16 17:19:00 <ericben> the e of my name is a e with accent usually ;-)
May 16 17:19:00 <woglinde> I am clearly for utf8
May 16 17:19:12 <woglinde> but dont know what the othet thinks
May 16 17:19:19 <woglinde> or if git has problems with it
May 16 17:21:26 <ericben> ok I've checked and it's already the case on OE so yes it seems ok (there is not git problem as it works fine on kernel.org)
May 16 17:24:38 <woglinde> okay
May 16 17:28:11 <GNUtoo> any news on g_realloc?
May 16 17:28:17 <GNUtoo> I bet none
May 16 17:30:25 <flo_lap> re
May 16 17:30:42 <florian> GNUtoo: gstreamer?
May 16 17:31:51 <GNUtoo> no gtk+
May 16 17:43:47 <woglinde> re
May 16 17:44:01 * florian has seen people complaining about gstreamer in this context
May 16 17:44:32 <florian> I guess its related to the version of glib that gets selected
May 16 17:46:14 <woglinde> florin hm right could be
May 16 17:46:20 <woglinde> didnt tought about it
May 16 17:50:16 <lgopher> now that, a lot more active people here, may someone could help me to identify the problem: during do_stage I get like 100th of  "new rpath '$ORIGIN/../../../../../../staging/x86_64-linux/usr/lib' too large; maximum length 46", but after that everything seems to be fine. Any idea ?
May 16 17:51:30 <lgopher> I even don't know where to start. google didn't return anything helpful.
May 16 17:53:15 <lgopher> I get this message for every file in sysroot/bin and sysroot/lib. 
May 16 18:32:06 <khem> lgopher: chrpath problem
May 16 18:32:15 <khem> lgopher: whats your host distro
May 16 18:32:35 <lgopher> khem: debian lenny x86_64
May 16 18:34:13 <lgopher> khem: funny, that all packages are compiling  fine, the new one, which I try to add is causing this problem.
May 16 18:36:25 <khem> lgopher: I get these too on ubuntu but 
May 16 18:37:20 <khem> GNUtoo: whats issue with g_realloc
May 16 18:38:09 <lgopher> khem: could I just ignore it, or there is some fix to apply ?
May 16 18:38:38 <khem> yes so far I have been ignoring too
May 16 18:43:24 <lgopher> khem: the thing that makes me nervous, is that during stage of my package, I get this error for every file in the sysroot/bin sysroot/lib
May 16 18:45:05 <lgopher> khem: another thing is, oe tells me some like: using old style for staging or something like that. May this cause the problem ?
May 16 18:46:29 <khem> lgopher: generally not 
May 16 18:46:40 <khem> but such recipes should be convered to new staging
May 16 18:47:57 <lgopher> khem: hmmm, the next question, obviously, is how do I do this ? 
May 16 18:48:49 <khem> lgopher: go through ml or git log there are plenty of examples
May 16 18:48:55 <khem> thats can get you started
May 16 18:50:00 <lgopher> khem: ok, looking already.
May 16 18:51:40 <dcordes> khem, everything builds fine with the gcc 4.4.4 and binutils 2.20.1
May 16 18:51:48 <dcordes> for htcleo machine
May 16 18:53:48 <khem> dcordes: congrats :)
May 16 18:55:16 <dcordes> khem, unfortunately the binaries won't run
May 16 18:55:31 <khem> I am not surprised
May 16 18:55:37 <khem> what errors do u get
May 16 18:55:42 <dcordes> none
May 16 18:56:47 <dcordes> I think -mfpu=neon might be a problem
May 16 18:58:14 <khem> do you have neon enabled in kernel 
May 16 18:58:18 <khem> if not then yes
May 16 19:05:03 <lgopher>  khem: concerning the new style. The patches I see so far, seem to remove do_stage and replace them with do_install. I've already doing the do_install staff, but it seems like I get "make install" at the do_stage stage anyway. It's kind of very confusing. Could you take a glance on http://gist.github.com/402865#file_icee.inc ?
May 16 19:06:01 <GNUtoo> khem, I bitbtke glib-2.0-native,atk-native,pango-native,cairo-native then I clean gtk+-native then I re-bitbake it and I've that :
May 16 19:07:46 <GNUtoo> Cannot load module TMPDIR/work/i686-linux/gtk+-native-2.20.0-r8.1/gtk+-2.20.0/modules/input/im-xim.la: TMPDIR/work/i686-linux/gtk+-native-2.20.0-r8.1/gtk+-2.20.0/modules/input/.libs/im-xim.so: undefined symbol: g_realloc_n
May 16 19:07:52 <GNUtoo> with TMPDIR beeing my temp dir
May 16 19:18:13 <khem> lgopher: look ok
May 16 19:18:28 <khem> you prolly dont need those configure bits
May 16 19:19:02 <khem> GNUtoo: hmmm thats a libglib function
May 16 19:19:25 <khem> GNUtoo: do you have libglib-native of some sort installed
May 16 19:19:36 <GNUtoo> I'll look thanks a lot
May 16 19:25:16 <dcordes> khem, that was a valuable hint
May 16 19:25:30 <dcordes> khem, it is disabled. somehow kernel doesn't compile with it
May 16 19:25:55 <GNUtoo> khem, I've the glib-2.0-native not libglib-native
May 16 19:29:06 <GNUtoo> khem, by installed you mean staged or isntalled on the host distribution?
May 16 19:48:01 <dcordes> khem, the problem is not in the kernel but in the OE toolchain.
May 16 20:00:53 <dcordes> khem, sorry got disconnected
May 16 20:00:58 <dcordes> khem, did I miss anything?
May 16 20:01:44 <dcordes> khem, with a 4.4.0 toolchain the kernel builds with neon enabled but the OE binaries can still not be executed
May 16 20:09:25 <dcordes> khem, I don't understand why the gcc-4.4.4 provided by OE can't build leo kernel with neon and a random 4.4.0 one can.
May 16 20:10:54 <GNUtoo> no
May 16 20:11:03 <GNUtoo> you didn't miss anything
May 16 20:19:09 <khem> dcordes: what error do u get
May 16 20:20:09 <dcordes> khem, well I see no errors. init just dies with no output. if I put init=<randombinary> I will see no output either
May 16 20:21:49 <khem> dcordes: I mean when compiling kernet with 4.4.4 toolchain
May 16 20:22:04 <dcordes> just a sec
May 16 20:22:45 <dcordes> khem, http://privatepaste.com/f653586f92
May 16 20:24:38 <khem> dcordes: can you add V=1 and paste the full compiler invocation
May 16 20:24:51 <dcordes> ok
May 16 20:27:47 <dcordes> khem, http://privatepaste.com/db2e37b200
May 16 20:28:03 <dcordes> http://groups.google.com/a/chromium.org/group/chromium-os-dev/browse_frm/thread/d1e37e4758bbb0cf
May 16 20:33:06 <khem> dcordes: hmmm its passing soft-float to kernel
May 16 20:33:41 <dcordes> khem, should I try rebuilding with bitbake?
May 16 20:34:15 <khem> how did you build it ?
May 16 20:35:15 <dcordes> manually specifying the OE toolchain with CROSS_COMPILE
May 16 20:35:23 <khem> hmm ok
May 16 20:35:50 <khem> CONFIG_NEON=y is set in .config too ?
May 16 20:36:00 <dcordes> yes. without it, it compiles
May 16 20:36:34 <dcordes> I could set CONFIG_NEON in the defconfing in OE metadata and rebuild
May 16 20:36:47 <dcordes> problem about that is OE makes zImage but I need Image
May 16 20:37:07 <khem> hmmm ok
May 16 20:38:22 <khem> you need to pass -mfpu=neon -mfloat-abi=softfp flags
May 16 20:38:28 <khem> to kernel build 
May 16 20:38:40 <khem> these are not default
May 16 20:38:59 <dcordes> how is that done?
May 16 20:42:20 <khem> dcordes: pass KBUILD_CFLAGS=" -mfpu=neon -mfloat-abi=softfp" on commandline
May 16 20:42:46 <CIA-2> 03Martin Jansa <Martin.Jansa@gmail.com> 07org.openembedded.dev * rb0adb60c4c 10openembedded.git/conf/distro/shr.conf: 
May 16 20:42:46 <CIA-2> shr: drop PREFERRED_VERSIONs for toolchain, now in sync with sane-toolchain.inc again
May 16 20:42:46 <CIA-2> Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
May 16 20:45:53 <dcordes> khem, http://privatepaste.com/e185e2e6fb
May 16 20:46:14 <dcordes> how to make OE build Image rather than zImage ?
May 16 20:46:30 <dcordes> brb
May 16 20:53:54 <khem> dcordes: you can set it in your machine.conf
May 16 20:54:10 <dcordes> khem, what's the var?
May 16 20:54:36 <khem> KERNEL_IMAGETYPE = "zImage"
May 16 20:54:42 <khem> or whatever you want
May 16 20:54:44 <dcordes> awesome
May 16 20:57:22 <GNUtoo> khem, should I install a newer glib-2.0 on my host,I'll look if it's in the requirements
May 16 20:59:09 <dcordes> khem, http://privatepaste.com/831db9c10a let's see what happens
May 16 20:59:27 <GNUtoo> not there
May 16 21:09:21 <khem> GNUtoo: well I think glib-2.0-native should have provided it
May 16 21:11:43 <GNUtoo> indeed
May 16 21:11:46 <GNUtoo> it's very strange
May 16 21:11:52 <GNUtoo> I don't understand the reason
May 16 21:13:32 <GNUtoo> maybe packaged staging is playing me tricks?
May 16 21:14:58 <khem> could be 
May 16 21:16:17 <GNUtoo|oeee> ok
May 16 21:16:30 <GNUtoo|oeee> so what could I do? bump pr locally?
May 16 21:16:38 <dcordes> khem, http://tinderbox.openembedded.net/packages/573463/
May 16 21:17:13 <khem> dcordes: same error hmm
May 16 21:17:36 <khem> dcordes: try doing manual build inside that tree
May 16 21:17:42 <khem> with the options I mentioned
May 16 21:17:48 <dcordes> khem, I did that. see previous paste.
May 16 21:18:02 <dcordes> http://privatepaste.com/e185e2e6fb
May 16 21:18:40 <khem> dcordes: ok
May 16 21:18:49 <khem> lets hack it in differently
May 16 21:19:11 <dcordes> I don't understand why kernel must be compiled with neon flag
May 16 21:19:54 <khem> CC="/home/lgorris/openembedded/tmp/cross/armv7a/bin/arm-angstrom-linux-gnueabi-gcc -mfpu=neon -mfloat-abi=softfp"
May 16 21:20:21 <khem> dcordes: actually I think the problem is that the asm file should have .fpu neon in there
May 16 21:20:24 <khem> then it will work
May 16 21:24:01 <khem> dcordes: btw which kernel version are you using
May 16 21:24:28 <dcordes> 2.6.32-9
May 16 21:25:53 <GNUtoo|oeee> I'll sleep bye
May 16 21:26:44 <dcordes> khem, http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=Makefile;h=ec932b2155b54be9e2ae502b8467a60ccd6f4ea5;hb=refs/heads/htc-msm-2.6.32
May 16 21:26:55 <khem> dcordes: hmmm
May 16 21:27:06 <dcordes> maybe you're right and they missed out that .fpu neon thingy
May 16 21:29:09 <dcordes> I will compare arch/arm/mach-msm/idle-v7.S with one from different tree for same device
May 16 21:29:30 <khem> #ifdef CONFIG_VFP VFPFSTMIA r0, r1              /* Save VFP working registers */
May 16 21:29:41 <khem> insert it between these two line
May 16 21:30:01 <dcordes> just line with ".fpu neon" ?
May 16 21:30:06 <khem> .fpu neon 
May 16 21:30:08 <khem> yeah to try
May 16 21:30:14 <dcordes> ok let's see
May 16 21:30:51 <dcordes> if this will work it will not answer the question why it compiles with 4.4.0 toolchain
May 16 21:30:56 <khem> crappy code
May 16 21:31:25 <khem> it compiles with 4.4.0 toolchain ?
May 16 21:31:29 <khem> from where ?
May 16 21:31:39 <khem> it depends on how the toolchain was configured
May 16 21:32:02 <khem> as I said OE toolchain is not configured to enable neon by default
May 16 21:32:15 <khem> but if you did that when configuring 
May 16 21:32:18 <khem> gcc 
May 16 21:32:21 <khem> then it will work
May 16 21:33:09 <dcordes> well I build with the neon flags in OE all the time
May 16 21:33:28 <dcordes> and building manually I did not specify anything but CROSS_COMPILE and ARCH
May 16 21:35:00 <khem> how did you compile the 4.4.0 toolchain
May 16 21:35:03 <khem> thats the question
May 16 21:35:16 <dcordes> I took it from google
May 16 21:35:19 <dcordes> precompiled
May 16 21:35:32 <khem> and CONFIG_VFP says to activate code to enable VFP
May 16 21:35:40 <khem> and here we are using the same in kernel
May 16 21:35:58 <khem> ok
May 16 21:36:01 <khem> show me gcc -v
May 16 21:36:08 <khem> of google gcc 
May 16 21:36:23 <dcordes> wait let me first try adding that line
May 16 21:36:36 <woglinde> why did you take the toolchain from google?
May 16 21:36:46 <khem> woglinde: hallo
May 16 21:36:55 <khem> woglinde: he is trying a droid
May 16 21:37:03 <khem> porting over to OE I guess
May 16 21:37:07 <dcordes> woglinde, because I suspect they are using it to build for htc nexus one (aka passion aka mahimahi eval. board)
May 16 21:37:18 <khem> there you go
May 16 21:37:19 <dcordes> woglinde, nexus has qsd8250 just like my target
May 16 21:38:08 <dcordes> nexus one is a 'droid'. but my target machine comes with windows ce
May 16 21:38:13 <khem> is code aurora forum a google thingy
May 16 21:38:25 <dcordes> no it is merely a qualcomm thingy
May 16 21:38:40 <khem> so open source face of qualcomm ?
May 16 21:38:48 <dcordes> yes face
May 16 21:39:02 <khem> Oh my buddy daniel walker works there
May 16 21:39:05 <dcordes> http://htc-linux.org/wiki/index.php?title=Leo
May 16 21:39:54 <dcordes> that's my target machine. htcleo
May 16 21:41:06 <dcordes> compiling
May 16 21:42:28 <dcordes> is it generally a problem to build kernel with gcc 4.4.0 and userspace binaries with 4.4.4 ? (despite from any fpu stuff)
May 16 21:43:06 <dcordes> khem, ok your patch fixed it.
May 16 21:43:18 <dcordes> khem, it now builds with oe 4.4.4
May 16 21:44:01 <dcordes> so it might be that google gcc is somehow hacked up to ignore that inconsistency ?
May 16 21:47:46 <dcordes>   Kernel: arch/arm/boot/Image is ready
May 16 21:52:25 <khem> dcordes: as I said its possible to configure gcc when it is built to generate VFP code by default OE doesnt do that
May 16 21:52:45 <khem> but if you ask it then you have to either inform it thru options or directives 
May 16 21:52:52 <dcordes> ah gcc build time.. get it
May 16 21:57:32 <khem> yes
May 16 21:57:58 <dcordes> khem, do you think it's better to keep using the 'normal' OE toolchain and patch the kernel or vice versa?
May 16 21:58:17 <khem> Use OE and patch kernel
May 16 21:58:30 <khem> because kernel is assuming something which is not there
May 16 21:58:38 <khem> kernel should demand instead
May 16 21:58:43 <khem> to be more clear
May 16 22:00:48 <dcordes> khem, do you know what is the name of gcc config that's needed to go without that patch?
May 16 22:01:25 <dcordes> the one that will lead to gcc generating VFP code by default
May 16 22:04:21 <khem> --with-fpu=neon
May 16 22:05:10 <khem> --with-float=softfp
May 16 22:05:32 <khem> if you pasted gcc -v of google gcc 
May 16 22:05:39 <khem> I could tell you what they have in it
May 16 22:09:35 <dcordes> wait sec adding the patch.
May 16 22:26:31 <khem> dcordes: did it boot ?
May 16 22:26:57 <dcordes> khem, yes and things changed
May 16 22:27:04 <dcordes> it's running extremely fast
May 16 22:27:19 <dcordes> and I see errors from the angstrom image now
May 16 22:27:36 <dcordes> init starts, udev is executed and then I get a segfault
May 16 22:28:13 <khem> oh well you are far
May 16 22:28:41 <dcordes> khem, did you click that wiki page link by any chance? we've had a discussion on the whole problem in our mailing list
May 16 22:28:52 <khem> dcordes: htc ?
May 16 22:28:58 <dcordes> khem, some guys built images with basic flags and it worked
May 16 22:29:07 <dcordes> khem, but not oe images, some builtroot stuff
May 16 22:29:10 <dcordes> khem, yes
May 16 22:30:55 <dcordes> I will acquire a dmesg
May 16 22:31:49 <dcordes> khem, http://lists.linuxtogo.org/pipermail/mobile-linux-discuss/2010-May/000087.html
May 16 22:37:49 <khem> dcordes: hmm what image did you build ?
May 16 22:37:57 <dcordes> khem, minimal-image
May 16 22:39:56 <dcordes> khem, checking tinderbox link
May 16 22:40:59 <dcordes> khem, it would be interesting to build an image with the same flags as the guy in the mail usese
May 16 22:41:06 <dcordes> khem, but I don't know how to
May 16 22:42:26 <khem> oh and whats your libc ?
May 16 22:42:28 <khem> version
May 16 22:42:56 <dcordes> khem, is it set by distro?
May 16 22:43:50 <khem> dcordes: yes
May 16 22:45:21 <dcordes> PREFERRED_VERSION_glibc 		?= "2.9"
May 16 22:45:37 <khem> seems ok
May 16 22:45:48 <dcordes> angstrom-2008.1.conf
May 16 22:47:04 <dcordes> khem, would you like to look at one of the do_compile logs from my (nonworking) image ?
May 16 22:48:28 <khem> that would not help
May 16 22:48:40 <khem> dcordes: you have to figure out whats crashing
May 16 22:49:29 <dcordes> udev. but debugging that is far above my level
May 16 22:49:54 <khem> hmm
May 16 22:50:18 <khem> does it stop there 
May 16 22:50:23 <khem> or continues 
May 16 22:50:29 <khem> some further
May 16 22:55:35 <dcordes> khem, booting again..
May 16 22:56:31 <dcordes> khem, it says remounting root file system after the two segfault lines
May 16 22:57:04 <khem> ok
May 16 22:57:12 <khem> whats your kernel commandline ?
May 16 22:58:07 <dcordes> now it got way faster
May 16 22:58:24 <dcordes> caching udev devnodes
May 16 22:58:31 <dcordes> populating dev caches
May 16 22:58:41 <dcordes> root: mount point blah no such file
May 16 22:58:45 <dcordes> segfault
May 16 22:58:54 <dcordes> configuring network interfaces
May 16 22:59:02 <dcordes> ifup...
May 16 22:59:06 <dcordes> done
May 16 22:59:17 <dcordes> setting up ip spoofing protection
May 16 22:59:31 <dcordes> Sun May 16 ...
May 16 23:00:20 <dcordes> and cursor keeps blinking (on each boot). so it's not dieing. 
May 16 23:00:32 <dcordes> you can also see that by spaming printks in the dmesg
May 16 23:00:47 <dcordes> (dmesg is optainable by dumping a location in ram after softreset)
May 16 23:01:10 <khem> ok 
May 16 23:01:30 <khem> do u flash minimal image into flash ?
May 16 23:01:41 <khem> how is your boot setup
May 16 23:01:41 <dcordes> I'm booting from uSD. commandline is "init=/sbin/init root=/dev/mmcblk0p2 rootwait=6"
May 16 23:02:05 <dcordes> s/rootwait/rootdelay/
May 16 23:02:31 <khem> ok can you make it "init=/sbin/init root=/dev/mmcblk0p2 rw rootdelay=6" ?
May 16 23:02:33 <khem> and retry
May 16 23:02:52 <dcordes> ok
May 16 23:03:53 <dcordes> btw if I boot some non angstrom rootfs based on buildroot it works
May 16 23:04:04 <dcordes> but it's very tiny only starting busybox I think
May 16 23:04:44 <khem> yeah I sure init=/bin/sh will work with OE too
May 16 23:05:58 <dcordes> so it looks pretty much like above
May 16 23:06:14 <dcordes> with illegal instruction added after populating dev cache
May 16 23:06:25 <dcordes> and segmentation fault
May 16 23:06:53 <dcordes> I will start ash 
May 16 23:08:46 <dcordes> khem, heh you were right
May 16 23:09:07 <dcordes> i guess that's because it's not very computationally complex to start ash
May 16 23:09:21 <dcordes> but the guy with buildroot said he was also able to connect via usb
May 16 23:09:24 <dcordes> usbnet
May 16 23:16:38 <khem> illegal instr
May 16 23:26:24 <dcordes> khem, sorry. 3g link hates me
May 16 23:26:29 <dcordes> khem, are you still there?
May 16 23:29:17 <khem> yes
May 16 23:29:22 <khem> illegal instr
May 16 23:29:36 <khem> could mean that neon is not enabled
May 16 23:29:49 <khem> add debug user_debug=-1 to kernel commandline
May 16 23:31:34 <dcordes> ok
May 16 23:33:50 <CIA-2> 03Lukas Gorris <lukas@htc-linux.org> 07org.openembedded.dev * rbe3ee7ec69 10openembedded.git/recipes/linux/ (linux-qsd8/htcleo/defconfig linux-qsd8_git.bb): linux-qsd8: updates srcrev, htcleo defconfig
May 16 23:33:51 <CIA-2> 03Lukas Gorris <lukas@htc-linux.org> 07org.openembedded.dev * r1b18816f4d 10openembedded.git/conf/machine/include/htc-qsd8.inc: htcleo: use uncompressed kernel iamge
May 16 23:38:46 <dcordes> khem, see nothing new with that line added
May 16 23:39:16 <khem> I guessed so
May 16 23:39:28 <khem> you need to enable debugging in kernel .config
May 16 23:40:21 <dcordes> khem, what's the CONFIG name ?
May 16 23:43:50 <khem> CONFIG_DEBUG IIRC
May 16 23:44:13 <dcordes> there is DEBUG_USER=n
May 16 23:44:29 <khem> yeah set it to -1
May 16 23:44:36 <dcordes> ok
May 16 23:44:41 <khem> or y
May 16 23:44:46 <khem> whatever it wants
May 16 23:45:12 <dcordes> and that's overwritten by the cmdline ?
May 16 23:49:18 <dcordes> compiling
May 16 23:58:46 <khem> ok
May 17 00:00:13 <dcordes> khem, I'm building with org.oe.dev HEAD +DEBUG_USER=-1 in leo defconfig
May 17 00:02:12 <dcordes> NOTE: Tasks Summary: Attempted 621 tasks of which 600 didn't need to be rerun and 0 failed.
May 17 00:03:29 <khem> ok
May 17 00:49:43 <dcordes> khem, http://privatepaste.com/f146433bf8 dmesg from kernel but with gcc_4.4.4 in oe
May 17 00:49:51 <dcordes> s/but/built/
May 17 00:50:42 <dcordes> khem, debugging enabled and starting same oe minimal-image
May 17 00:54:46 <dcordes> init (1): undefined instruction: pc=0000a654
May 17 01:40:39 <khem> hmmm that address is weird
May 17 01:43:47 <dcordes> khem, hi again. why what's wrong about it ?
**** ENDING LOGGING AT Mon May 17 02:46:33 2010

**** BEGIN LOGGING AT Mon May 17 02:48:14 2010
**** ENDING LOGGING AT Mon May 17 02:59:57 2010