**** BEGIN LOGGING AT Tue Apr 07 02:59:58 2009 Apr 07 03:38:14 * flyback is starting to despise his country Apr 07 03:56:39 http://www.youtube.com/watch?v=gSwha_Ug7sk&feature=related# <--- freaking awesome :P Apr 07 06:51:06 @last Apr 07 06:51:06 Sleep-Walker: [03:56:39] http://www.youtube.com/watch?v=gSwha_Ug7sk&feature=related# <--- freaking awesome :P Apr 07 06:51:37 argh Apr 07 11:00:09 morning Apr 07 12:53:21 hey ppl Apr 07 14:39:38 emfoxzhou: hi Apr 07 14:39:57 emfoxzhou: do you use Xfbdev or common X.org? Apr 07 14:40:19 I'm fighting with that debian more than I'd like :/ Apr 07 14:54:56 X.org Apr 07 14:55:22 so bigger fatter and slower brother of Xfbdev :/ Apr 07 14:55:50 I'm fighting with running X as service Apr 07 14:56:04 it seems that it is not covered by init.d scripts Apr 07 14:56:24 nothing else I did, just apt-get Apr 07 14:57:24 what's your problem? Apr 07 14:57:56 it took me a while till I found that I need to have in sources.list stable, unstable and testing at the same time Apr 07 14:58:14 my problem is that Debian use different Display Manager for GPE Apr 07 14:58:19 gpe-dm is missing Apr 07 14:58:26 emfoxzhou, do you use jabber ? can you give me your jabber id ? Apr 07 14:58:36 so the way of running is different Apr 07 14:59:25 marex: I use gtalk Apr 07 14:59:40 what's your address then ? Apr 07 15:01:05 marex: wanna forward your collection of pen1s enlargement commercials again? Apr 07 15:01:41 Sleep_Walker: :) Apr 07 15:02:21 Sleep_Walker: you build gpe yourself or use binary elsewhere? Apr 07 15:02:49 emfoxzhou: I used GPE in Angstrom before Apr 07 15:04:57 Sleep_Walker: you can try my may: first run 'Xorg &', then run the desired one, xfce-session or such thing Apr 07 15:05:12 emfoxzhou: |miska| build most of images Apr 07 15:06:02 emfoxzhou: I've got `Xfbdev -screen 320x320 -mouse evdev -keybd evdev &' and matchbox-session Apr 07 15:06:22 Xfbdev is really faster :) Apr 07 15:06:27 but not working in Debian Apr 07 15:06:32 is that work? Apr 07 15:06:42 or beyond my configuration and googling skills Apr 07 15:07:00 I tried X.org and it worked Apr 07 15:07:12 maybe I should go back :/ Apr 07 15:07:28 hmmm, beyound my skill too, I never used Xfb Apr 07 15:07:59 I think there is missing tslib as input driver, but I can be wrong Apr 07 15:08:08 X.org is not so slow... you should consider the feeling of 650 users :) Apr 07 15:08:39 if you are using sid, there's a known bug Apr 07 15:08:57 emfoxzhou: I'm not debianist, Apr 07 15:09:09 emfoxzhou: I'm using _something_ :) Apr 07 15:10:50 you used nearly everything :) Apr 07 15:11:25 s/TslibDevice/Device/ Apr 07 15:11:35 I think it works Apr 07 15:12:05 emfoxzhou: you're talking about X.org config? Apr 07 15:12:17 the newest source indicated that somebody what a unified naming Apr 07 15:12:19 yes Apr 07 15:12:21 emfoxzhou: I started X.org succesfully Apr 07 15:12:26 no problem with that Apr 07 15:12:41 and touch screen works? Apr 07 15:12:42 but I missed testing packages, so I didn't have installed matchbox Apr 07 15:12:48 yeah Apr 07 15:12:55 with tslib driver Apr 07 15:13:18 strange... tslib in unstable has been broken Apr 07 15:13:27 emfoxzhou: btw. what about mapping Esc to option+home? Apr 07 15:13:35 I'm missing this key... Apr 07 15:14:29 not bad Apr 07 15:14:44 * Sleep_Walker finally found '=' :) Apr 07 15:14:56 matchbox is not in unstable? Apr 07 15:15:07 I don't know Apr 07 15:15:13 that apt is so messy Apr 07 15:15:42 have you done 'apt-get update' Apr 07 15:15:49 it shows me how much files I have on filesystem, but it doesn't tell me how many packages is left to be installed :b Apr 07 15:15:50 yes Apr 07 15:15:55 also dist-upgrade Apr 07 15:17:51 I use unstable only and things go all well... Apr 07 15:17:58 emfoxzhou: that's why I asked you for some wiki page :( Apr 07 15:19:00 Sleep_Walker: it's a little strange that, for debian user, it's no need to write such a page Apr 07 15:19:10 I know Apr 07 15:19:21 but it's like writing ebuild for you Apr 07 15:19:37 but for others, I feel strange people will start to learn debian by treo.... Apr 07 15:20:02 I don't think there is needed more than some examples of config, packages to install etc Apr 07 15:20:05 there's tons of debian manual or starting pages Apr 07 15:20:42 emfoxzhou: well Apr 07 15:20:55 1] if you want users to read much, they won't test Apr 07 15:21:19 2] If I'll have to struggle with Debian, I can't fight with kernel Apr 07 15:21:21 ok. I'm persuaded. I'll write some later Apr 07 15:21:53 emfoxzhou: if you can, create just some tarball I can put on ext2 and use Apr 07 15:22:32 but, you're on GPRS, so it's not way Apr 07 15:22:38 no, not recently... Apr 07 15:23:00 right, the problem is gprs Apr 07 15:23:43 I'm fighting with gsm mux all day, but little progress Apr 07 15:24:20 hacking that package or configuring? Apr 07 15:24:59 porting a gsm0710muxd to treo Apr 07 15:25:27 the one working on neo freerunner Apr 07 15:25:28 ok Apr 07 15:27:11 I have to learn much more new things, glib, dbus,vala... Apr 07 15:27:52 btw, how could I kick out the damn 'emfox' from the server? Apr 07 15:29:12 emfoxzhou: if you have registered nickname, you can "ghost" him Apr 07 15:29:34 like /ns ghost emfox Apr 07 15:31:07 then what happen? Apr 07 15:31:44 nick server told me successfully ghosted Apr 07 15:31:54 now you can change nick Apr 07 15:33:23 I'm back & Apr 07 15:33:49 :) Apr 07 15:34:40 I just got your msg from our bot Apr 07 15:34:49 ah, right Apr 07 15:35:10 if you want start X directly, I think 'nodm' is an option Apr 07 15:35:56 have you tried GPE? Apr 07 15:37:03 no... no suitable binarys Apr 07 15:37:35 you know, I have no build-env Apr 07 15:38:21 wait, you mean GPE in Debian? Apr 07 15:38:45 yes Apr 07 15:40:10 not yet. the last time I install nodm right after X, and it gave me a xterm directly after reboot Apr 07 15:42:31 any workaround of the hwclock epoch? I refer the time start from 1939 Apr 07 15:43:42 emfox: it is known that PalmOS clock begins in 1901, it can probably be done in machine_init section Apr 07 15:43:55 but I think that ntp would be preffered :b Apr 07 15:47:42 if you can describe well why you need it, it can be added Apr 07 15:47:56 but it means that you need to set clock in PalmOS Apr 07 15:50:59 Sleep_Walker: for a treo, it's not normal for a startup ntp Apr 07 15:51:38 emfox: you can also get time from GSM network on startup, can't you? Apr 07 15:52:45 no, GSM networks varys... some don't support time fetching Apr 07 15:53:08 e.g. ChinaUnicom which I'm using Apr 07 15:53:11 if you can set up time in PalmOS, you can do it in Linux, right? Apr 07 15:53:33 if you set up clock in Linux, what time is in PalmOS/ Apr 07 15:53:35 ? Apr 07 15:54:18 the same, nothing changed in PalmOS Apr 07 15:54:49 seems linux does not affect PalmOS Apr 07 15:55:18 emfox: how can that be possible? where is that time stored? Apr 07 15:55:54 maybe PalmOS use another place ... Apr 07 15:57:44 I think the most direct way is to patch hwclock to support --setepoch under arm too Apr 07 15:58:26 currently it's only available on Alpha Apr 07 15:59:59 we already have that in old 2.6.21 kernel tree Apr 07 16:00:20 I'm not sure if this belongs to arm-devel or vanilla Apr 07 16:00:34 you can try Apr 07 16:01:36 I'd better to do more important works :) such as go asleep right now Apr 07 16:01:57 good night Apr 07 16:02:09 I'll give you some info about battery soon (I hope) Apr 07 16:02:15 see you Apr 07 18:02:56 hm ... Im really curious about the result of rebuilding Angstrom with gcc 4.3.3, binutils 2.19.1 and glibc 2.7 :) Apr 07 18:03:34 for now it'd be just good if the cross toolchain built OK Apr 07 18:03:52 I hear ya, having a hell of a time myself getting oe to work heh Apr 07 18:04:11 fails to build anything and gives no error log to boot :/ Apr 07 18:47:02 NOTE: Running task 268 of 487 (ID: 60, /mnt/data/OpenEmbedded/openembedded/recipes/file/file-native_4.21.bb, do_configure) Apr 07 18:48:09 marex@rin:~$ cat /proc/loadavg Apr 07 18:48:10 7.42 6.05 5.07 5/155 17335 Apr 07 18:48:23 ewww .... BB_NUMBER_THREADS is cool magic :-) Apr 07 18:48:55 it makes dual-core near to unusable if set to 2 and parallel-make set to 4 :-) Apr 07 18:49:55 ouch Apr 07 18:50:11 cores+1 not cores*2 :P Apr 07 18:50:31 unless you have hyperthreading, then I guess you could do cores*2 Apr 07 18:50:42 or even cores*2+1 heh Apr 07 18:51:20 the -j flag should be set to core+1 for optimum performance iirc. yea, but whatever Apr 07 18:51:25 I wish I could even get that far heh Apr 07 18:51:28 btw do you know how Intel HT works ? Apr 07 18:51:51 not at a physical level, but from what I hear it's suppose to allow two threads to run on a single core in "parallel" Apr 07 18:51:53 use this link ;) http://wiki.openembedded.net/index.php/Getting_started Apr 07 18:51:58 I did :/ Apr 07 18:52:04 it fails on 0 of xxx Apr 07 18:52:25 can't even build the very first thing it needs to which is shasum-native iirc Apr 07 18:52:30 and the log is empty Apr 07 18:53:01 yea, the HT allows improved usage of the CPU core ... or rather it allows you to use the core units much more Apr 07 18:53:02 wonder if it has anything to do with me not installing psyco Apr 07 18:53:24 the CPU also tells you it has two cores, but that's not true Apr 07 18:53:36 yeah Apr 07 18:53:43 nah ... it works without psyco Apr 07 18:53:50 pastebin the log ;) Apr 07 18:53:51 That's for the sake of schedulers so that they will schedule two threads on one core Apr 07 18:54:27 nah ... it actually behaves like dual-core ;) Apr 07 18:54:29 k, will pastebin shortly Apr 07 18:55:29 Well exactly, this way the OS treats it like two processors instead of one Apr 07 18:55:43 Which is how the hardware is supposedly implemented :) Apr 07 18:56:32 no fpu on these pxa chips? Apr 07 18:57:20 yup, no FPU Apr 07 18:57:31 that's why it's so slow ... but you have iWMMXt and VFP Apr 07 18:57:39 see eg. here http://wiki.debian.org/ArmEabiPort Apr 07 18:58:14 I see Apr 07 18:59:00 pastebin.com/m19a1024a Apr 07 18:59:27 i'm not concerned about kernel errors since i'm going to build my own kernel and modules anyway Apr 07 19:00:26 note the log file is empty thus why it prints nothing from the log heh Apr 07 19:00:36 what's here /home/mekius/Misc/phone/oe/build/tmp/work/x86_64-linux/shasum-native-1.0-r1/temp/log.do_compile.4345 ? Apr 07 19:00:39 I've tried just building a single package as well Apr 07 19:00:50 dang Apr 07 19:01:15 pastebin local.conf Apr 07 19:01:50 wait, should local.conf be in build/conf or in openembedded/conf? Apr 07 19:02:23 openembedded being the git checkout folder Apr 07 19:03:09 OpenEmbedded/build/conf/local.conf Apr 07 19:03:30 ah, hell, that's probably part of it Apr 07 19:03:40 you dont have that one ? Apr 07 19:04:13 no, I left it where the sample conf was Apr 07 19:04:24 then that's the problem Apr 07 19:04:46 just weird that it seemed to know what distro and machince I was building regardless :/ Apr 07 19:05:06 well i'll give it a shot, hopefully it works heh Apr 07 19:05:57 pastebin.com/m89b6d54 Apr 07 19:06:03 that's my local.conf Apr 07 19:07:02 http://pastebin.com/m1f5d521b Apr 07 19:07:15 this one's the relevant part of mine Apr 07 19:07:53 well it's about to start building Apr 07 19:08:03 I wonder if it's the BUILD_ARCH, hmm Apr 07 19:10:24 | ../ncurses/./tty/hardscroll.c: In function '_nc_scroll_optimize': Apr 07 19:10:24 | ../ncurses/./tty/hardscroll.c:254: internal compiler error: Segmentation fault Apr 07 19:10:24 | Please submit a full bug report, Apr 07 19:10:24 | with preprocessed source if appropriate. Apr 07 19:10:27 WOW :D Apr 07 19:13:49 ouch :P Apr 07 19:33:46 marex: btw, same error after moving the config and after taking out the BUILD_ARCH Apr 07 19:36:20 use my config, that should work for you Apr 07 19:38:05 well i'm thinking the issue lies elsewhere Apr 07 19:38:14 the do_compile for this package is a single line Apr 07 19:38:46 $CC $CFLAGS $CXXFLAGS -o oe_sha256sum main.c sha256.c Apr 07 19:39:53 And CFLAGS seems a bit screwed up Apr 07 19:39:59 -isystem/home/mekius/... Apr 07 19:40:15 this system part seems weird and could screw it up I'd think Apr 07 19:42:15 it's also using ccache which I'm not sure where it's finding that heh Apr 07 19:55:46 ok ... my lappy is too slow, let's use quadcore Apr 07 19:56:13 apt-get install ccache ;) Apr 07 19:56:27 it has sanity checker so that should be OK Apr 07 19:59:22 I have ccache installed Apr 07 19:59:37 just wondering how it determined it should use that instead of gcc directly :/ Apr 07 20:03:17 ok, that's bizarre, it actually does build the file :/ Apr 07 20:03:47 it's like the script itself is exiting improperly or something Apr 07 20:10:00 mekius, rm -rf OpenEmbedded/build/tmp and load my local.conf Apr 07 20:10:05 ;) Apr 07 20:10:13 with that you should be OK Apr 07 20:10:25 if it doesnt work with that one, we can try debuging further Apr 07 20:13:11 i'll try it, but I don't think this is a conf issue, my conf doesn't differ that much from yours and if I echo the command from do_compile and run it manually it works fine, it's weird Apr 07 20:13:19 but I'll try it Apr 07 20:14:06 mekius, also, preserve the machine set to palmtt5 Apr 07 20:14:37 since if the issue is with palmt680, we'll be able to pinpoint it this way Apr 07 20:19:15 marex: k Apr 07 20:20:33 do_compile still failed :/ Apr 07 20:23:22 so it outputs the run.do_compile.pid file right? Apr 07 20:23:29 well if I run that it works fine... Apr 07 20:24:22 unless it's actually running something different internally heh Apr 07 20:27:46 and again the log is empty :) Apr 07 20:29:42 sucks Apr 07 20:34:56 indeed Apr 07 20:36:58 I mean it's the simplest thing and it just fails :P Apr 07 20:37:19 it's pretty weird Apr 07 20:37:24 can you give me an ssh login ? Apr 07 20:37:32 I guess no ;-) Apr 07 20:37:44 well this is on my laptop so no heh Apr 07 20:37:48 plus I'm at work :) Apr 07 20:38:02 but probably wouldn't do that anyway :) Apr 07 20:40:31 ;) Apr 07 20:40:53 it's still weird it has issues with the native stuff Apr 07 20:41:01 what distro do you run ? Apr 07 20:42:02 archlinux Apr 07 20:42:32 but really this seems like a very fundamental problem somewhere in their build scripts because when i run the run.do_compile.xxxx script it works so something has to be different Apr 07 20:45:37 maybe some of your environment settings are broken Apr 07 20:45:46 which shouldn't really affect OE though Apr 07 20:49:25 well it puts all the env settings into that run.do_compile.xxx script Apr 07 20:49:45 I'm assuming what's in that script is what OE is using Apr 07 20:49:55 if not then something is broken :) Apr 07 20:50:12 but I'll probably have to talk to one of the oe devs to figure this one out since it doesn't seem like a configuration error best I can tell Apr 07 20:53:01 I'm suppose to run bitbake from inside the build directory right? Apr 07 20:54:08 and the build directory should be in the parent directory of the oe git checkout directory? Apr 07 20:54:24 so build and oe.git would be in the same folder or should build be inside the oe.git dir? Apr 07 20:56:27 mekius, join #oe ;-) Apr 07 20:56:55 well you set PATH and BBPATH so it shouldnt matter Apr 07 20:57:42 true Apr 07 21:01:25 marex: will have to mess with this later, already wasting too much time :) Apr 07 21:13:02 mekius, I wasted one hour using the quadcore buildhost Apr 07 21:13:16 looks like I'll stick with working stuff Apr 07 21:13:20 heh Apr 07 22:13:38 http://www.metacafe.com/watch/1389016/trigun_legato_and_the_rodericks/ <--- wouldn't you just love to do this to all the assholes in the world Apr 07 22:53:30 * flyback goes out to get some dinner, bbl Apr 08 02:20:24 * flyback notes the problem he's been having with mental decline is like symptoms of sleep deprevation even when he's getting 10 hrs of sleep **** ENDING LOGGING AT Wed Apr 08 02:59:57 2009