**** BEGIN LOGGING AT Mon Sep 15 02:59:57 2008 Sep 15 03:02:45 <_cjc_> anyone home? Sep 15 06:50:03 morning all Sep 15 08:47:53 florian: good morning Sep 15 08:49:22 good morning! Sep 15 09:33:40 03  07org.openembedded.dev * rb032860a6e 10OE.dev/packages/linux/linux-omap/omap3evm/defconfig: linux omap git: sync beagle and omap3evm defconfig Sep 15 09:33:47 03  07master * rb032860a6e 10OE.dev/packages/linux/linux-omap/omap3evm/defconfig: linux omap git: sync beagle and omap3evm defconfig Sep 15 09:39:09 morning Sep 15 09:40:04 03  07org.openembedded.dev * r24417023d3 10OE.dev/conf/machine/omap3evm.conf: omap3evm: switch to linux-omap kernel, it works better Sep 15 09:40:12 03  07master * r24417023d3 10OE.dev/conf/machine/omap3evm.conf: omap3evm: switch to linux-omap kernel, it works better Sep 15 10:33:32 hail zecke Sep 15 11:16:18 mornin Sep 15 11:17:04 hi ant_work Sep 15 11:20:14 pb_: I tried to add 'gdb' to my image Sep 15 11:20:27 but it does not appear in /usr/bin .... Sep 15 11:20:31 ant_work: IMAGE_INSTALL += "gdb" Sep 15 11:25:37 hm..I tried IMAGE_INSTALL += and ANGSTROM_EXTRA_INSTALL... Sep 15 11:25:46 cleaning and rebuilding... Sep 15 11:26:24 then copied the gdb binary by hand :-/ Sep 15 12:01:25 hi woglinde Sep 15 12:06:41 hi ant Sep 15 13:22:01 hi woglinde Sep 15 13:23:13 he pb Sep 15 13:32:19 lack of reset key in laptops is big mistake Sep 15 13:33:07 hrw: It's usually there if you press the power button for an extended period. Sep 15 13:33:38 broonie: you are talking about ATX Power Off funtionality Sep 15 13:34:00 Yup, it does the job well enough I find. Sep 15 13:34:07 has someone a pointer for rm -rf in cpp Sep 15 13:35:46 writing 3.6M iso on cdrw x4 is pain Sep 15 13:36:45 hrw hm the reset is hit the powerkey for 5 seconds Sep 15 13:37:07 hm press Sep 15 13:37:11 maybee better Sep 15 13:37:19 hi kgilmer Sep 15 13:37:34 hey there woglinde Sep 15 13:37:52 woglinde: how about: #define D(x,y) R(x) -R(y); #define R(x) r##x; D(m,f) Sep 15 13:38:13 pb *g* Sep 15 13:38:21 hm seems Sep 15 13:38:28 there are now high api functions Sep 15 13:38:30 *sigh* Sep 15 13:39:10 okay Sep 15 13:39:16 I will steal from coreutils Sep 15 13:44:06 fsck Sep 15 13:44:49 bah Sep 15 13:44:54 that cant be true Sep 15 13:45:12 why is that such a pain Sep 15 13:45:17 deleting dirs Sep 15 13:45:28 making dirs is so easy Sep 15 13:53:47 woglinde: what exactly are you trying to do? Sep 15 13:56:22 * mwester looks around to see if he still has anything that's 4x CD-R -- that's old... Sep 15 13:56:41 er, CD-RW even Sep 15 13:57:34 mwester: I do not remember where I have 8cm dvdrw ;D Sep 15 14:01:08 pb the app I have makes a directory for a session Sep 15 14:01:15 after the session the dir is useless Sep 15 14:01:24 and has to be deleted Sep 15 14:01:34 so no cruft is rotten on the disk Sep 15 14:05:44 hm boost has some stuff Sep 15 14:06:33 woglinde: one recursive function will remove dir with whole tree Sep 15 14:09:29 woglinde: are you too proud to invoke /bin/rm to do this job? Sep 15 14:09:49 pb std::system Sep 15 14:09:56 is bad behavior Sep 15 14:46:48 morning all! Sep 15 15:18:48 have a nice evening/day Sep 15 15:20:31 gn Sep 15 18:10:06 re Sep 15 18:10:15 re Sep 15 18:15:48 could anyone help me with build problems in OE? Sep 15 18:16:04 that depends on what problems you are having Sep 15 18:16:11 i get this: http://pastebin.com/d178f7b10 when i try to compile base-image Sep 15 18:16:35 the command is malformed (see "| |" near the end) Sep 15 18:16:56 i have no idea where this came from Sep 15 18:16:57 which shell you have as /bin/sh symlink? Sep 15 18:17:37 $ /bin/sh --version Sep 15 18:17:37 GNU bash, version 3.2.39(1)-release (i486-pc-linux-gnu) Sep 15 18:17:37 Copyright (C) 2007 Free Software Foundation, Inc. Sep 15 18:19:48 yes, that does look malformed. I guess this must be some libtool malfunction. Sep 15 18:20:30 you will probably have to look at the script and see what it was intending to put at that point in the pipeline. probably it's meant to be "... | ${something} | ..." and that variable is accidentally expanding to nothing. Sep 15 18:21:12 what script should i look at? Sep 15 18:21:17 libtool Sep 15 18:21:46 hmm... ok, and it must be libtool made by OE, right? Sep 15 18:21:54 right Sep 15 18:22:12 ./arm-angstrom-linux-gnueabi-libtool Sep 15 18:22:13 that one Sep 15 18:22:55 where is installed? Sep 15 18:23:05 in your build directory Sep 15 18:23:29 see: Sep 15 18:23:31 # Sep 15 18:23:32 | make[4]: Entering directory `/opt/OE/oe_tmp/work/xscale-angstrom-linux-gnueabi/glib-2.0-2.16.4-r0/glib-2.16.4/glib' Sep 15 18:23:32 # Sep 15 18:23:32 | /bin/sh ../arm-angstrom-linux-gnueabi-libtool --tag=CC ... Sep 15 18:23:52 oh Sep 15 18:24:00 i'm blind... Sep 15 18:27:32 here is it Sep 15 18:27:36 export_symbols_cmds="\$NM \$libobjs \$convenience | \$global_symbol_pipe | \$SED 's/.* //' | sort | uniq > \$export_symbols" Sep 15 18:27:53 seems that global_symbol_pipe is the cause Sep 15 18:28:49 it's forced to be a null string Sep 15 18:30:17 and in my distribution libtool sets this variable to Sep 15 18:30:18 global_symbol_pipe="sed -n -e 's/^.*[ ]\\([ABCDGIRSTW][ABCDGIRSTW]*\\)[ ][ ]*\\([_A-Za-z][_A-Za-z0-9]*\\)\$/\\1 \\2 \\2/p'" Sep 15 18:30:44 do you have any ideas how to fix libtool generation? Sep 15 18:38:51 hmm, in angstrom stable, /etc/modules.conf is generated by update-modules, and there is a text inside that its autogenerated and that one should look at the files in /etc/modutils Sep 15 18:38:58 but there is no /etc/modutils Sep 15 18:39:14 I'm trying to figure out the correct way to autoload particular kernel modules Sep 15 18:39:36 or would the usual modprobe.conf do the trick? Sep 15 18:40:13 Jin^eLD, you can specify modules to autoload in your kernel recipe Sep 15 18:40:30 an example: module_autoload_g_ether = "g_ether" Sep 15 18:40:46 kgilmer_: well, but how would I do this once the kernel is already built? Sep 15 18:40:49 for exampel during development Sep 15 18:40:56 Jin^eLD: I guess that would be a question for #angstrom Sep 15 18:41:03 indeed.. Sep 15 18:41:11 thx for the hint, let me ask there :> Sep 15 18:41:12 yeah not sure there Sep 15 18:48:24 seems I have to create a file in /etc/modules/textfile which contains the list of the modules and then call update-modules, seems to do the trick Sep 15 18:50:22 aem, /etc/modutils I mean Sep 15 18:59:53 Jin^eLD: 'man update-modules' on debian system? Sep 15 19:00:19 hrw: did not have a debian system at hand Sep 15 19:00:21 :P Sep 15 19:01:49 hrw: found an ubuntu machine.. its man page says: "update-modules is an obsolete command which does nothing. calling it should be fixed." Sep 15 19:01:50 :) Sep 15 19:03:41 ;D Sep 15 19:17:26 bye Sep 15 19:17:34 o mickeyl Sep 15 19:17:39 mickeyl: alive? Sep 15 19:18:32 think so Sep 15 19:18:44 mickeyl: whats up? Sep 15 19:19:10 preparing for .tw trip Sep 15 19:23:04 mickeyl: again? Sep 15 19:24:32 last time is 12 months ago Sep 15 19:24:37 so it's time to come there agian Sep 15 19:25:03 every now and then, it is good to get people in the same place :) Sep 15 19:26:25 Whee Sep 15 19:26:41 I might, maybe, have package.bbclass modified to not have foo-dev depend on foo, if foo is an empty package Sep 15 19:26:48 * Tartarus rebuilds a package that does have a main package now Sep 15 19:27:20 Crofton|work: indeed Sep 15 19:27:38 do someone has build failues on glibc 2.6.1 package? Sep 15 19:45:23 Tartarus: very good **** ENDING LOGGING AT Mon Sep 15 19:57:33 2008 **** BEGIN LOGGING AT Mon Sep 15 20:22:16 2008 Sep 15 20:26:08 Well, I thought I had it working, bah Sep 15 20:36:38 _hc: there should be a command line tool named "bl" Sep 15 20:37:03 iirc, the kernel interface is an ioctl rather than /proc, but I could be wrong about that. Sep 15 20:37:18 <_hc> right, just checked the minilite code... Sep 15 20:37:38 <_hc> bl is it! thanks! Sep 15 20:38:56 <_hc> do you know of an eequivalent to control the sound levels/ Sep 15 20:38:56 <_hc> ? Sep 15 20:39:17 setmixer? Sep 15 20:39:36 <_hc> hmm, that's not on this familiar 0.8.4 image Sep 15 20:40:34 ah. in that case there probably isn't one as standard. Sep 15 20:45:47 Trying to build a image for a samsung smdk440 board, but the git repository for this kernel does not exist. Builds with the git://opensource.wolfsonmicro.com/linux-2.6-asoc in place of git://opensource.wolfsonmicro.com/linux-2.6-asoc-ggdev. Is this a correct workaround? Sep 15 20:47:44 ift it works why not Sep 15 20:49:18 I don't flashed the board yet Sep 15 21:06:13 <_hc> pb__: I got a little bit closer, esd-utils allows me to turn on and off the sound, but not change the levels... Sep 15 21:13:52 re florian Sep 15 21:15:15 re Sep 15 21:40:35 how does one convince Xfbdev to work with a USB keyboard? Sep 15 21:49:21 <_hc> pb__: do you happen to know which package includes setmixer? Sep 15 21:53:04 pb_, pb__, around? Sep 15 22:24:22 <_hc> angstrom Sep 15 22:24:29 <_hc> oops, sorry, wrong window :) Sep 15 22:42:00 anyone know how I specify kernel module params when using udev? Sep 15 23:06:39 tharvey, it's a modprobe thing, iirc /etc/modprobe.conf or conf.d/ Sep 15 23:08:29 Tartarus, oh, so I'll add the appropriate config file and have the recipe install that as well - looks like /etc/modprobe.d/ just like on a desktop - thankx Sep 15 23:32:36 Tartarus, do you know what kicks off the loading of modules on a system? ie is there something in init that runs modprobe? Sep 15 23:35:46 there is a modules script in /etc/rcS.d Sep 15 23:36:24 ah... of course... ok thanks Sep 16 01:21:31 what's the bitbake target to clean out an OE tree? Sep 16 01:37:36 aum: there is none. use rm -rf build/tmp **** ENDING LOGGING AT Tue Sep 16 02:05:38 2008 **** BEGIN LOGGING AT Tue Sep 16 07:02:31 2008 Sep 16 08:53:55 morning Sep 16 08:56:41 hello, any knows if theres an x.org driver for the smi501 grafic chicp wired via localbus and not pci? Sep 16 09:04:04 nik0n: that would be a question for #freedesktop Sep 16 09:09:59 sm501 as silicon motion? Sep 16 09:31:49 hrw: yes Sep 16 09:32:16 I only have device which has it on pci bus Sep 16 09:32:37 nik0n: check Simtec patchsets - they have s3c based device with sm501 Sep 16 09:32:41 that would be to easy .... tq modules wired it via localbus :) Sep 16 09:32:53 ok, gonna give it a try Sep 16 10:26:54 mickey|tv: right, flashed my gta01. it seems to boot at least so I guess this means my u-boot is ok :-) Sep 16 12:20:42 hi Sep 16 13:09:50 good morning Sep 16 13:18:43 morning Sep 16 14:05:29 I am building procps agaiinst glibc on an i686 arhc and at runtime ps throws an Illegal Instruction error. This is often due to linking against a different glibc but I * think* Sep 16 14:05:40 that my glibc's match. any thoughts? Sep 16 14:06:15 waite: for which target is this? Sep 16 14:06:53 cbrake: the 1.4 xserver in the dev tree fixed the fixed font issue Sep 16 14:07:02 jfyi Sep 16 14:07:31 nik0n: excellent :-) Sep 16 14:09:30 it is my own MACHINE and distro. The machine conf is at: http://pastebin.com/m7f54622f Sep 16 14:17:02 thebohemian, any thoughts on what might be going wrong? My distro call out preferred glibc = 2.6.1 and binutils 2.18. Sep 16 14:18:08 I saw that ther is a vmware machine configuration. is a vmdk image generated automatically when using this machine? Sep 16 14:29:38 waite: sorry, no idea what went wrong then Sep 16 14:48:58 waite: what actually is the illegal instruction? Sep 16 14:49:06 and what cpu are you using? Sep 16 14:53:26 pb_ I can't tell. ps just return "Illegal Instruction" Sep 16 14:54:02 pb_ I am using a Celeron M Sep 16 14:55:20 waite: run it under gdb and see what instruction it stops on Sep 16 14:56:14 pb_, my bad. ps gives me a SIGILL (signal 4) Sep 16 14:57:45 has anyone an idea why the prompt of my busybox looks like this: "\u@\h:\w " how could i fix it? Sep 16 15:04:09 horizon: what do you want it to look like? Sep 16 15:08:18 Anyone know how to set "Metadata branch" and "Metadata revision" on http://tinderbox.openembedded.net/builders/testing@freesmartphone.org/ ? Sep 16 15:08:49 waite: that's the same thing. as before, use gdb. Sep 16 15:09:40 pb_: it is just so that the tab completion is gone and the prompt seems not to be expanded. i dont have special wishes how the prompt should look like, i only need it working again Sep 16 15:10:05 horizon: try: export PS1='$' Sep 16 15:10:19 maybe better with a space: export PS1='$ ' Sep 16 15:11:12 mickeyl: good morning Sep 16 15:11:25 yes, already treid that, but the shell lets me still go back vith backspace and "deletes" the prompt Sep 16 15:11:50 the history also doesn't work Sep 16 15:12:26 the "Expand prompt string" option of busybox is set Sep 16 15:12:47 what shell are you using? Sep 16 15:12:55 but for some strange reason it doesn't work anymore :/ Sep 16 15:12:58 ash Sep 16 15:13:19 mm, that is odd. are you sure you haven't accidentally started using a different shell? Sep 16 15:13:32 I don't think ash should behave like that even if expansion were off Sep 16 15:14:07 BusyBox v1.11.2 (2008-09-16 16:38:29 CEST) built-in shell (ash) Sep 16 15:14:19 pb_, gdb shows me nothing else. I really think it is at the link stage. I also don't have ps built with debugging so it is harder to be sure. Sep 16 15:15:03 waite: I don't understand what you mean by "at the link stage" Sep 16 15:15:22 horizon: heh, right, evidently that is indeed ash Sep 16 15:15:28 :) Sep 16 15:15:59 sounds like you might have encountered a bug in busybox somehow. you might want to take it up with the h4x0rs in #elinux Sep 16 15:16:08 I think libc.so.6 does not match the libc ps was compiled against. Sep 16 15:16:13 ah, thx Sep 16 15:16:26 waite: doesn't match in what way? Sep 16 15:16:34 so long as the ABIs are the same it shouldn't matter Sep 16 15:17:26 I agree but it has the feel of something very strange like that. Sep 16 15:32:07 well, I think the only way to proceed is for you to find out what the actual instruction is that fails. Sep 16 15:33:49 yeah I am on that track now. I am building with debugging Sep 16 15:35:07 ok. I don't think debugging is necessary but it won't hurt. Sep 16 15:37:48 re Sep 16 15:58:48 bye Sep 16 16:26:36 pb_, pb__, trying your suggestion now Sep 16 16:41:31 Tartarus: ok, very good Sep 16 16:41:40 1st pass didn't work tho :( Sep 16 16:42:31 he pb Sep 16 16:46:31 yeah, using getVar(...,0) gives me: Sep 16 16:46:32 NOTE: main_pkg is ${@bb.parse.BBHandler.vars_from_file(bb.data.getVar('FILE',d), Sep 16 16:46:34 d)[0] or 'defaultpkgname'} Sep 16 17:00:14 i'm having trouble getting a USB keyboard to work with Xfbdev 1.4 Sep 16 17:00:19 works find with 1.3 Sep 16 17:00:45 is there anything special I need to do with the -keybd parameter? Does anyone else have this working? Sep 16 17:02:36 Tartarus: you shouldn't need to do that main_pkg thing at all Sep 16 17:03:06 oh, I guess you are looking at the wrong bb.data.getVar() call. Sep 16 17:03:15 the one I meant was the first one I quoted, i.e. the one that retrieves RDEPENDS Sep 16 17:03:26 Ah, hm Sep 16 17:03:42 if you are trying to retrieve ${PN} without expansion then yes, you will lose, but that shouldn't be required. Sep 16 17:04:32 So just: Sep 16 17:04:32 rdepends = explode_deps(bb.data.getVar('RDEPENDS_' + pkg, d, 0) or bb.data.getVar('RDEPENDS', d, 0) or "") Sep 16 17:04:32 if main_is_empty and main_pkg in rdepends: Sep 16 17:04:32 rdepends.remove(main_pkg) Sep 16 17:04:33 ? Sep 16 17:04:38 right Sep 16 17:05:02 Nope Sep 16 17:05:04 so you should find that rdepends now looks like [ '${PN} (= ${PV})'], not ['xproto (= 1.0)'] Sep 16 17:05:45 ah, so what happens? Sep 16 17:05:53 tossing in a bb.note Sep 16 17:07:43 Hmmm, ok Sep 16 17:07:50 NOTE: rdepends: ['${PN} (= ${DEBPV})'] Sep 16 17:08:02 ah, close Sep 16 17:08:13 But won't we need to re-expand rdepends later? Sep 16 17:08:20 yes, but that happens already Sep 16 17:08:32 at least, it should do. Sep 16 17:08:41 I think the expansion at this point has always been bogus, it's just been harmless until now Sep 16 17:08:50 ok, lemme try... Sep 16 17:09:25 I guess what you're seeing now is debian.bbclass screwing you up. Sep 16 17:10:02 oh, no, it's like that in bitbake.conf, I just remembered it wrong Sep 16 17:10:14 in that case, you should be safe to just use the string exactly as it is there to compare against Sep 16 17:10:25 This does work: Sep 16 17:10:31 rdepends = explode_deps(bb.data.getVar('RDEPENDS_' + pkg, d, 0) or bb.data.getVar('RDEPENDS', d, 0) or "") Sep 16 17:10:31 remstr = "${PN} (= ${DEBPV})" Sep 16 17:10:31 if main_is_empty and remstr in rdepends: Sep 16 17:10:31 rdepends.remove(remstr) Sep 16 17:10:53 ok, sounds good Sep 16 17:10:57 * Tartarus builds zeroconf now Sep 16 17:13:00 Nope, now zeroconf is broken Sep 16 17:13:17 wait, hm Sep 16 17:13:21 might have broken it myself earlier :) Sep 16 17:13:29 heh Sep 16 17:13:35 what's the breakage/ Sep 16 17:14:04 yeah, my fault from changing thing wrong call earlier Sep 16 17:14:12 righto Sep 16 17:14:14 * pb_ go home now Sep 16 17:14:36 zeroconf is right, now, rechecking xproto Sep 16 17:14:44 ok, good luck Sep 16 17:14:52 I have to take my wife to the pub now (!) but I will be back a bit later Sep 16 17:15:00 sounds fun :) Sep 16 17:15:25 woop, works Sep 16 17:15:28 i'll post shortly Sep 16 17:25:48 gm all Sep 16 17:32:34 bbl Sep 16 18:15:29 how can you specify with ipkg which version of a pkg to install in the event you have two packages with the same package name? Simply referring to the package with its version such as 'ipkg install pn_pv' surprisingly does not work Sep 16 18:15:59 download or local file? Sep 16 18:16:07 download Sep 16 18:18:55 in the package index: Package=madwifi-ng-modules Version=openwrt-r12613-r0 and on the target an 'ipkg install madwifi-ng-modules_openwrt-r12613-r0' gives me a 'Cannot find package ...' Sep 16 18:19:41 kind of annoying... can't believe that there is no way to specify version of package in the case you have multiple packages with different versions Sep 16 18:34:50 ipkg -i /path/to/it perhaps? Sep 16 18:35:05 * Tartarus checks and sees his ea key hasn't expired, yet Sep 16 18:35:06 ga Sep 16 18:35:14 bonehead bitbake question, how do i specify a specfic version PV I want to build? Sep 16 18:35:29 bitbake mypackage_1.1 Sep 16 18:35:35 kgilmer rais DEFAULT_PREFRENCES Sep 16 18:35:45 in the .bb Sep 16 18:35:53 thanks woglinde, but how can i do it from the command line? can I? Sep 16 18:37:23 hm you could do bitbake -b pathtobb/bla.bb Sep 16 18:37:53 but bitbake moo-version should work too, but I can happen that the package with lower version number is build too Sep 16 18:38:03 it Sep 16 18:38:05 hms Sep 16 18:39:19 oh :) Sep 16 18:39:27 that's odd to me but it works Sep 16 18:39:41 so the recipe is packagename_version Sep 16 18:39:50 but to specify on the command line it's packagename-version Sep 16 18:39:54 yes Sep 16 18:39:55 _ => - Sep 16 18:40:09 thanks woglinde :) Sep 16 18:40:22 no porb Sep 16 18:40:37 but remind my warning Sep 16 18:40:46 about lower version builds too Sep 16 18:41:01 well this is for an image Sep 16 18:41:13 hm? Sep 16 18:41:15 i want to version each release cycle in the recipe Sep 16 18:41:25 package/images/myimage_r1.0 etc Sep 16 18:41:27 hm hm Sep 16 18:41:31 okay Sep 16 18:42:34 and from my testing, after you showed me how to build a specific version, bitbake bug_r1.1 does not try to build bug_r1.0 Sep 16 18:42:42 so looks like it's doing what i want Sep 16 18:43:44 i mean with a '-' rather Sep 16 18:43:47 bug-r1.1 Sep 16 18:44:43 *g* Sep 16 19:20:57 hello Sep 16 20:03:27 hi blindman Sep 16 20:10:28 has anyone seen this:NOTE: package ncurses-native-5.4-r15: task do_qa_staging: started Sep 16 20:10:28 NOTE: QA checking staging Sep 16 20:10:28 ERROR: QA Issue: libpanel.la failed sanity test (workdir) in path /home/david/work/sw/pocketpc/openembedded/OE/angstrom-tmp//staging/i686-linux/usr/lib Sep 16 20:11:22 aum hm ncurses-native worked right here all the time Sep 16 20:12:11 several ppl in angstrom have found libncurses-native broken Sep 16 20:13:18 hm Sep 16 20:13:37 let me see Sep 16 20:15:00 hi rp!!!!!!!!!!!! Sep 16 20:15:16 Shhh!!! Don't frighten him off! Sep 16 20:15:22 hi woglinde Sep 16 20:15:27 morning all Sep 16 20:15:30 gm Sep 16 20:15:32 Greetings Sep 16 20:42:08 aum I build ncurses-native here with out problems Sep 16 20:42:25 hi woglinde Sep 16 20:43:28 jo ant Sep 16 20:44:05 woglinde: help with gdb...http://rafb.net/p/XAzzMe16.html Sep 16 20:44:17 what are these insane paths? Sep 16 20:44:28 (glibc-2.6.1) Sep 16 20:45:28 I basically forgot how to use gdb (long long ago) and I suppose I'm missing something obvious here... Sep 16 20:46:16 but WORKDIR here is fishy... Sep 16 20:46:23 ant hm Sep 16 20:46:49 what are you trying Sep 16 20:46:54 it's about Opie Sep 16 20:47:01 I installed the debug packages Sep 16 20:47:04 run the qpe target binary on the host device? Sep 16 20:47:57 yes, all on Zaurus (ttyS0) Sep 16 20:48:24 at first I had missing symbols, then I installed libqpe-dbg, libopie2-dbg, opie-taskbar-dbg Sep 16 20:48:32 as suggested by bluelightning Sep 16 20:48:48 yes Sep 16 20:49:14 gdb qpe Sep 16 20:49:14 run Sep 16 20:49:40 it doesn't fail.. Sep 16 20:49:45 should do, maybee you need to setup some stuff like /etc/init.d/qpe Sep 16 20:49:58 that's why I'm looking for putting some breakpoints Sep 16 20:50:20 you could run gbd-server Sep 16 20:50:26 gdb Sep 16 20:50:36 and debug over ethernet Sep 16 20:50:44 on build-host you mean? Sep 16 20:50:48 with source-code on the host Sep 16 20:50:52 ah, ok Sep 16 20:51:20 http://dominion.thruhere.net/koen/cms/using-gdbserver Sep 16 20:52:36 woglinde: I did a strace of /etc/init.d/opie restart http://www.pastebin.ca/1201727 Sep 16 20:52:58 but apart a couple of ENOENT... Sep 16 20:53:15 (my friend /etc/localtime for one) Sep 16 20:53:32 only a SIGCHLD (Child exited) @ 0 (0) Sep 16 20:53:36 ... Sep 16 20:54:05 that sounded suspicious to bluelightning Sep 16 20:54:42 ant you should use strace -o moo.txt -ff moo Sep 16 20:54:52 for better seperating threads Sep 16 20:54:55 and childs Sep 16 20:54:57 ah, ok, gtkterm is ugly... Sep 16 20:57:14 here is 'gdb run': http://rafb.net/p/zVeKOo14.html Sep 16 20:57:50 hm Sep 16 20:57:52 *g* Sep 16 20:58:14 he, you see... Sep 16 20:58:39 I had sometimes the same problems Sep 16 20:58:54 and other day and a new build they vanished Sep 16 20:58:59 btw it fails only on c7x0 Sep 16 20:59:18 but not only reported by me... Sep 16 21:07:26 woglinde: last try with strace: http://rafb.net/p/VyxHU965.html Sep 16 21:08:25 hm Sep 16 21:08:27 never seen Sep 16 21:08:30 such an error Sep 16 21:08:42 I'll rebuild from scratch... Sep 16 21:13:22 woglinde: this was the log of strace...http://rafb.net/p/Ziswrh10.html Sep 16 21:13:34 sorry to having bothered you soo long ;-) Sep 16 21:15:57 ant hm there should be more log files Sep 16 21:16:08 each for every thread and child Sep 16 21:16:18 anyone know where I can find an armv5teb rootfs? Sep 16 21:16:22 it's all: from 1 to 362 Sep 16 21:18:33 woglinde: sorry Sep 16 21:18:44 there is a ton of logfiles :-/ Sep 16 21:23:35 woglinde: if you ever happen to be free, here the logs http://rapidshare.com/files/145863590/opie.tar.gz.html Sep 16 21:24:08 *g* Sep 16 21:24:21 ^_^ Sep 16 21:38:05 Hi ... I'm trying to use oe under freeBSD. Where can I tell oe 'patch' command is in fact 'gpatch' ? is that hardcoded or in configuration file ? thx Sep 16 21:39:53 oe patch command is quilt Sep 16 21:42:23 realy ? I got this error : Sep 16 21:42:25 File "patch_do_patch", line 98, in patch_do_patch Sep 16 21:42:25 File "", line 339, in Resolve Sep 16 21:42:25 __builtin__.CmdError: Command Error: exit status: 1 Output: Sep 16 21:42:25 patch: unrecognized option `--dry-run' Sep 16 21:42:25 Usage: patch [options] [origfile [patchfile]] [+ [options] [origfile]]... Sep 16 21:42:56 * cbrake likes the way package staging removes ipk files during a package rebuild :-) Sep 16 21:43:18 that saves a lot of hassle if you go back a version Sep 16 21:44:08 write_erase maybee quilt is using patch Sep 16 21:44:24 so you have to set a QUILT var Sep 16 21:44:31 this might be an oe error Sep 16 21:44:35 under bsd Sep 16 21:44:54 quilt-native_0.45.bb, do_patch failed :-) Sep 16 21:45:01 funny Sep 16 21:45:03 hehe okay Sep 16 21:45:08 yes Sep 16 21:54:02 write_erase please file a bug in the bugtracker and provide a patch if possible Sep 16 22:21:09 'nite Sep 16 22:21:31 nite Sep 16 22:39:28 Apologies for the trouble today. The bot should stop bouncing in and out of your channel now. Sep 16 23:24:27 is there a shell function that can be used in scripts (like oenote) that produces output to bitbake vs output in a log? Sep 17 01:54:11 how many build threads does bitbake use when running OE recipes? Sep 17 01:58:01 As many as you tell it. Sep 17 01:58:21 i didn't see any thread count options in the bitbake manpage Sep 17 01:58:36 BB_NUMBER_THREADS = "2" Sep 17 01:58:40 In your local.conf file. Sep 17 01:58:48 ahh, ok Sep 17 01:58:51 thx Sep 17 01:58:56 Not to be confused with Sep 17 01:59:04 PARALLEL_MAKE = "-j 3" Sep 17 01:59:11 also in your local.conf file Sep 17 01:59:20 what's the difference? Sep 17 02:00:38 The latter is passed into make inside recipes, and lets make do parallel stuff. The former is handled by bitbake, and will let bitbake do multiple bitbake recipes in parallel. So the combination of the two above will result in as many as 6 simulaneous processes beating on your system. Sep 17 02:01:26 Of course all this presumes that you have the horsepower, and that the makefiles themselves are capable of running with the -j option -- many aren't. Sep 17 02:02:05 (The ones that don't are marked in the .bb file, but I keep stumbling over broken makefiles constantly. Sep 17 02:02:07 ) Sep 17 02:03:47 what's the optimal setup for an antique 32-bit single-core Athlon XP 2GHz 1GB RAM box? Sep 17 02:05:09 in terms of safe, or fast? Sep 17 02:05:31 former more than the latter Sep 17 02:06:36 Avoid BB_NUMBER_THREADS (I've had troubles with the git/svn fetchers stomping on each other), and stick with PARALLEL_MAKE = "-j 2" Sep 17 02:06:44 ok Sep 17 02:07:00 should i comment out BB_NUMBER_THREADS, or just set it to 1? Sep 17 02:07:07 Comment it out. Sep 17 02:07:11 k Sep 17 02:11:27 one little weird thing in the angstrom build howto page cost me a full day's processing time: Sep 17 02:11:39 the howto says: Sep 17 02:11:47 TMPDIR = "/home/david/work/sw/pocketpc/openembedded/OE/${DISTRO}-tmp/" Sep 17 02:12:06 but that causes some package builds to break at QA sanity check stage Sep 17 02:12:10 changing it to: Sep 17 02:12:20 TMPDIR = "/home/david/work/sw/pocketpc/openembedded/OE/build/${DISTRO}/tmp/" Sep 17 02:12:22 seems to work Sep 17 02:18:59 hmmm Sep 17 02:19:08 I wonder if not having 'tmp' in the name upsets something Sep 17 02:19:17 I've got a bunch of QA errors I always ignore Sep 17 02:19:40 and I set TMPDIR to ${my_workdir}/build_$arch_$machine Sep 17 02:19:55 Tartarus: in the dev branch, having 'tmp' in the name breaks ncurses-native, for one **** ENDING LOGGING AT Wed Sep 17 02:59:57 2008