**** BEGIN LOGGING AT Fri Apr 19 02:59:58 2013 Apr 19 07:12:11 Good morning Apr 19 07:33:52 morning all, hi apelete Apr 19 07:41:25 Hi silvio__ Apr 19 08:37:04 Finally I build a workind toolchain SDK !!! Thanks all for help Apr 19 08:38:26 congrats Apr 19 08:39:41 morning all Apr 19 08:40:10 mornin Apr 19 08:56:13 Hi bluelightning, waynr Apr 19 09:24:36 morning all Apr 19 09:26:24 Hi silviof Apr 19 09:28:09 hi bluelightning , waynr , silviof Apr 19 09:29:22 I want warn who need use xmlrpc-c, pay attention, I had trouble in the linking due to xmlrpc-c-config Apr 19 09:38:05 hi Apr 19 10:07:26 hi hrw Apr 19 10:34:10 hi all Apr 19 10:34:41 hi pb_ Apr 19 10:34:48 morning bluelightning Apr 19 11:00:47 good morning Apr 19 11:10:30 hi pb_ Apr 19 11:29:48 * pb_ stabs the kernel Apr 19 11:29:49 arm-oe-linux-gnueabi-ld: internal error in segment_precedes, at ../../gold/layout.cc:3100 Apr 19 11:32:59 pb_: using the gold linker for the kernel is supported now? here, it failed always with low level stuff with custom linker scripts :( Apr 19 11:34:27 yeah, I was rather hoping it worked nowadays but apparently not. Apr 19 11:34:35 never mind, I guess I can patch it to use ld.bfd easily enough Apr 19 11:40:27 llvm nanananaaaa Apr 19 11:51:52 pb_: you seem to stab something different of the toolchain each day ;) Apr 19 12:08:53 the bad thing is, that switching between bfd and gold linker is difficulty. In kernel it is easy because LD is called directly. But when ld is called by gcc (e.g. in grub2), much more work is required to call the working ld Apr 19 12:20:47 Do you have any idea why the download of tcl8.4.19-src.tar.gz results in an HTML file instead of the gz? Thank you. Apr 19 12:26:11 volto: what recipe are you using? Apr 19 12:28:40 waynr, I fixed it by copying the file by hand,but I think it was: tcl_8.4.19.bb Apr 19 12:28:55 I am using openembedded with angstrom Apr 19 12:29:50 I am quite a beginner yet, but I am interested in contributing if I can, for example for telling about missing files in repositories Apr 19 12:36:26 ensc|w: yeah, indeed. to fix this particular case, it looks as though patching KERNEL_LD in kernel-arch.bbclass is all that's needed. Apr 19 12:36:47 I'm slightly surprised that this hasn't already been fixed in oe-core. I guess not all that many people are using gold still. Apr 19 12:37:07 BlindMan: indeed, every day I find myself unpleasantly surprised by some or other piece of brokenness Apr 19 12:41:56 pb_: for grub I use http://pastebin.com/raw.php?i=KR7NbHX0 hack Apr 19 12:43:51 volto: when I try downloading it using wget there are no problems Apr 19 12:45:12 Yes, it is strange. Well, I copied it by hand and the build goes further. I will check whether I have a proxy set in wget Apr 19 12:45:18 what might be confusing the server Apr 19 12:48:16 ensc|w: hah, right. you might be able to use something like CC += "-B ${S}/.bin/gcc" which might be fractionally less hacky, perhaps. Apr 19 12:48:40 Ouch, I think I found the problem, in my ~/.wgetrc I once set (years ago): header = User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:2.0b10) Gecko/20100101 Firefox/4.0b10 Apr 19 12:48:44 but I dunno, I can't remember whether -B goes on the start or the end of the search path. Apr 19 12:50:51 doh, alsa-utils depends on bash? Apr 19 12:50:53 * pb_ stabs oe-core Apr 19 12:52:09 volto: maybe wget in OE should use an explicitly-defined wgetrc Apr 19 12:52:27 pb_: thx; will look into it Apr 19 12:52:58 waynr, that would be a good idea: to preset the environment as much as possible. Apr 19 12:53:15 ah, added by some wind river dude back in january. Apr 19 12:53:17 * pb_ takes it out again Apr 19 12:55:46 ensc|w: according to the manual -B does go on the front of the path, so it should do what you need. "For each subprogram to be run, the compiler driver first tries the -B prefix, if any." Apr 19 12:58:17 argh GCCMULTILIB Apr 19 12:59:32 waynr: but you can set proxy etc in own wgetrc to not set it again Apr 19 12:59:52 meta/recipes-devtools/gcc/gcc-configure-target.inc:GCCMULTILIB = "--enable-multilib" Apr 19 13:00:09 is there any way (other than bbappend) to overwrite it? Apr 19 13:01:05 well, you can do it in your distro.conf if you want. Apr 19 13:01:13 (or any other .conf file obviously) Apr 19 13:01:32 pb_: how? Apr 19 13:01:38 GCCMULTILIB_mydistro = "" Apr 19 13:02:04 pb_: problem is that I do not have distro and need this only when external toolchain is used Apr 19 13:02:06 or any other override that you know will be applied. if this is local.conf or something then you can use _forcevariable Apr 19 13:02:34 well then, you will have to invent some override that gets applied when the external toolchain is in use. Apr 19 13:03:13 GCCMULTILIB_forcevariable = "--disable-multilib" Apr 19 13:03:15 works Apr 19 13:03:19 pb_: thanks a lot! Apr 19 13:03:42 or use GCCMULTILIB_forcevariable = "${@compute_multilib()"} and write a python function to return --enable/--disable according to whatever criteria you like. Apr 19 13:04:01 er, with the } and " the right way round, of course Apr 19 13:04:05 pb_: I added that into external-linaro-toolchain.inc Apr 19 13:04:17 right, very good Apr 19 13:04:35 if you have a dedicated .inc file for the external toolchain then that's ideal. Apr 19 13:04:38 one more workitem to mark as done Apr 19 13:05:07 now libxml-parser-perl ftfbs and images which want locale files which do not exist ;D Apr 19 14:54:56 Hello. It happens when OE doesn't recompile my package even after bitbake -c clean package ? Apr 19 15:11:50 vadmeste: it'll be restoring it from shared state Apr 19 15:12:24 vadmeste: use -c cleansstate to remove the sstate package as well Apr 19 15:12:56 usually this is not needed since the system will detect whenever you change something in the recipe the next time you ask for it to be built and rebuild it automatically Apr 19 15:14:56 bluelightning: okay thanks Apr 19 16:11:17 moin Apr 19 18:31:11 ah i see why qt.conf wasn't working, it has to be colocated with the qmake2 binary Apr 19 19:32:46 hrw: there is a broonie right here :-) Apr 19 19:33:11 hrw: I think you meant bronie :-D Apr 19 21:29:17 ant_home, hi Apr 19 21:30:22 what's the size of kexecboot compiled with klibc? Apr 19 21:30:31 hi GNUtoo-m4a785t-m Apr 19 21:30:32 and how does it compare to uclibc Apr 19 21:32:06 the klibc static bin is 84kb (armv4) Apr 19 21:32:15 ok Apr 19 21:32:21 it would be for x86 Apr 19 21:32:44 only dietlibc gave similar sizes Apr 19 21:33:04 ok Apr 19 21:33:10 with eglibc I've something huge: Apr 19 21:33:17 -rw-r--r-- 1 gnutoo gnutoo 1.8M Apr 17 23:51 initramfs-kexecboot-image-x60-20130417214636.rootfs.cpio.gz Apr 19 21:34:03 the issue is that the klibc version didn't compile out of the box Apr 19 21:34:09 here 97kb lzma compressed Apr 19 21:34:16 ok Apr 19 21:34:32 hm.. what is failing? Apr 19 21:34:42 if I reduce kernel size I've a full image of 3M and my target is less than 2M Apr 19 21:35:36 I think it's kexec-tools-klibc that fails Apr 19 21:36:00 http://paste.debian.net/hidden/a0f939cf/ Apr 19 21:36:00 ah, interesting, I'm working on last version Apr 19 21:36:06 ok Apr 19 21:37:05 should I upgrade the recipe somehow? Apr 19 21:37:12 like change it to git? Apr 19 21:38:16 I'm working on 2.04 Apr 19 21:38:53 let me look Apr 19 21:38:58 note the patchset need to be rebased Apr 19 21:39:08 and some fixes went upstream Apr 19 21:39:26 ok Apr 19 21:40:04 ah you work on it without oe? Apr 19 21:40:11 or do you have the recipes? Apr 19 21:40:42 I have some recipes needing refresh Apr 19 21:40:50 2) also there is no guide for configuring the kernel for it Apr 19 21:40:52 ok Apr 19 21:40:56 I'm concentration on kernel those days... Apr 19 21:41:45 ah, I found a 2.0.3 Apr 19 21:41:53 if you want I can send it Apr 19 21:42:15 can't remember the status, though... Apr 19 21:42:33 yes Apr 19 21:42:35 it would be nice Apr 19 21:42:45 btw the patches seem to lack the header for oe Apr 19 21:42:49 don't be scared... Apr 19 21:42:52 like UpstreamStatus: Apr 19 21:43:25 yes, there have been no changes after switch to oe-core Apr 19 21:43:33 next version will have it Apr 19 21:44:05 ok nice Apr 19 21:45:07 also is there some plans for 2) or can I help? because since there is no shell....debugging is harder Apr 19 21:45:17 you can't just inspect the filesystem Apr 19 21:45:26 although I've logs on the serial port Apr 19 21:46:48 sent to @no-log.org Apr 19 21:47:37 about kernel, well, the idea would be to create some specific fragments and modularize linux-yocto-tiny Apr 19 21:47:58 I'm still working on the main recipe, linux-yocto_3.8 Apr 19 21:48:38 the actual l-y-t-k_3.2 is abusing of yocto kernel tools :) Apr 19 21:49:35 ok Apr 19 21:49:42 thanks Apr 19 21:52:04 iirc when I was working on kexec-tools I had issues with the headers changing under my nose Apr 19 21:52:45 hmm bits32 Apr 19 21:53:21 iirc I fixed this for x64 too but can't find the tree..maybe on another disk... Apr 19 21:55:02 GNUtoo-m4a785t-m, argh..what I sent you is incomplete... Apr 19 21:55:14 pls fix PACKAGES and FILES Apr 19 21:55:39 (I took oe-core recipe 'cause I was too lazy to edit the checksums ;) Apr 19 22:19:29 ok Apr 19 22:53:05 kexec/ifdown.c:40:2: warning: implicit declaration of function 'if_nameindex' [-Wimplicit-function-declaration] Apr 19 22:53:10 the include is there Apr 19 22:53:22 and then it fails like that: error: dereferencing pointer to incomplete type Apr 19 22:53:24 hmmm Apr 19 22:53:35 it's probably klibc specific Apr 19 22:53:45 I mean the failure Apr 19 22:59:41 yes klibc seem to lack that struct Apr 19 22:59:55 *klibc includes Apr 19 23:15:53 how do i override the value of a variable that gets set in a class file? Apr 19 23:16:13 seems like just doing it from recipe context doesn't work Apr 19 23:16:42 the variable in question that i want to override is OE_QMAKE_QMAKE Apr 19 23:17:06 essentially i want to symlink Apr 19 23:17:42 that executable into $WORKDIR and use the symlink to run it since otherwise the qt.conf file that gets created in $WORKDIR isn't very useful Apr 19 23:18:17 so instead of running qmak2 from $STAGING_BINDIR or whatever it's called, run it from $WORKDIR Apr 19 23:21:17 $STAGING_BINDIR_NATIVE* Apr 19 23:40:35 Do I need to do something special on a dependency-only package? Apr 19 23:40:46 I'm getting an "Unable to resolve package " error when I build. **** ENDING LOGGING AT Sat Apr 20 02:59:57 2013