**** BEGIN LOGGING AT Wed Nov 21 02:59:56 2007 Nov 21 04:04:59 Crofton: pong Nov 21 04:29:37 03coredump2 07org.oe.dev * r119306d1... 10/ (1 packages/cryptsetup/cryptsetup_1.0.5.bb): cryptsetup: Update RECOMMENDS and DEPENDS Nov 21 04:29:43 03coredump2 07org.oe.dev * rdbb50c12... 10/ (3 files in 3 dirs): linux: Update ix4xx defconfig to enable cryptsetup on NSLU2 Nov 21 06:18:37 koen: good morning :) Nov 21 07:06:13 hello, is Nov 21 07:06:56 it possible to enable packet injection with zaurus cxx with kernel 2.4.x? Nov 21 07:17:26 morning! Nov 21 07:17:44 morning Nov 21 07:18:58 hey lumag, can you give some help about packet injection with zaurus? Nov 21 07:19:50 d3d3: no. Didn't tried it :) Nov 21 07:20:18 it's ok,thanks a lot. =) Nov 21 07:23:54 !seen Spyro1 Nov 21 07:23:56 lumag, I don't remember seeing spyro1. Nov 21 07:24:50 !seen Spyr1 Nov 21 07:24:53 Spyr1 (n=ian@i-195-137-94-162.freedom2surf.net) was last seen parting #oe 8 days, 7 hours, 31 minutes ago stating "{}". Nov 21 07:53:41 ~seen RP Nov 21 07:53:44 rp is currently on #maemo (1d 19h 14m 49s) #handhelds (1d 19h 14m 49s) #oe (1d 19h 14m 49s) #openmoko (1d 19h 14m 49s) #openzaurus (1d 19h 14m 49s). Has said a total of 12 messages. Is idling for 21h 18m 4s, last said: 'koen: Thats broadly along the lines I was thinking too'. Nov 21 08:09:34 koen: I tried but can't figure out how the CROSS directory was populated/installed for gcc Nov 21 08:43:58 good morning all Nov 21 08:46:30 morning Nov 21 08:49:20 good morning steliosk Nov 21 08:49:53 esben cyberdeck : hi ! Nov 21 08:51:03 heated discussion on the makefile topic on the list :-) Nov 21 08:53:29 looks like it. Both "sides" have valid arguments. IMHO It all comes down on how you use OE (maintain a distro vs write/develop apps) Nov 21 09:31:04 * * OE Bug 3344 has been created by autobuild(AT)openembedded.org Nov 21 09:31:06 * * openmoko-feedreader2-0.0.1+svnr62"3080"-r1-do_fetch Nov 21 09:31:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3344 Nov 21 10:45:19 03xora 07org.oe.dev * r66fc0044... 10/ (4 files in 4 dirs): Nov 21 10:45:19 openmoko-session2.bb, sane-srcrevs.inc : version boost for all openmoko Nov 21 10:45:19 stuff. Also it is now using phonekit. Nov 21 10:59:09 how do you debug a kernel ? Nov 21 10:59:23 printk and jtag Nov 21 10:59:25 * nud needs a flash course on "how to debug memory issues on ARM for a stock driver kernel" Nov 21 11:00:47 morning Nov 21 11:02:10 morning hrw Nov 21 11:02:45 lumag: I saw that you cleaned tosa-pm a bit Nov 21 11:03:06 hrw: Just few ifdefs. Nov 21 11:04:03 I'll upload a patch in a minute Nov 21 11:04:25 http://imgs.xkcd.com/comics/brick_archway.png Nov 21 11:05:31 hrw: http://pastebin.ca/792404 Nov 21 11:06:27 it's a bit dirty, but it allow building 480216 bytes kernel Nov 21 11:07:44 brb Nov 21 11:09:11 lumag: prepare mtn diff with it? Nov 21 11:09:55 lumag: blkdev.h part as separate and send it to lkml? Nov 21 11:10:53 lumag: are you aware there are official fixes for wm97xx touch devices for 2.6.23 in Wolfson git? Nov 21 11:11:13 lumag: broonie is in charge of those fixes :-) Nov 21 11:18:17 I'm having problems with libX11 building since it's looking for its dependencies on /usr/lib/*.la instead of staging area. Could someone give me any hint how to debug it? Nov 21 11:26:16 someone know how to check does compiled binary is armv4/5/6/7 etc? Nov 21 11:28:33 hrw: look for illegal instructions Nov 21 11:29:42 hrw: AFAIK you have to objdump and grep for instruction armv5 only, armv6 only. armv7 only Nov 21 11:30:17 is there a point keeping tree 1.5.1 in org.oe.dev since it's not available anymore ? Nov 21 11:30:46 is there a way to set OE up to build all deb packages instead of ipk? I realize ipk is just a light weight version of deb and can even be renamed to deb and work, I'm interested in using OE to generate deb packages to install on an existing embedded OS. Nov 21 11:31:31 svolpe_gerrath: yes, something like ANGSTROM_PACKAGE_FMT Nov 21 11:31:43 svolpe_gerrath: cant remember the right name off top of my head Nov 21 11:32:32 svolpe_gerrath: INHERIT += "package_deb" is other way Nov 21 11:33:37 XorA: I hoped that there is other way Nov 21 11:33:47 hrw, I would just put that in my distro file? Nov 21 11:34:03 hrw: Im sure I went through this before, but linux executables dont have a flag for arm version Nov 21 11:34:14 hrw: but I would be glad to be wrong in this case Nov 21 11:34:56 hrw, or conf file correct? Nov 21 11:35:45 svolpe_gerrath: distro or any config Nov 21 11:36:21 XorA: I have few builds here with different flags and searc for a way to check binaries Nov 21 11:36:42 armv4/5/6, soft/hard-float, thumb/no-thumb etc Nov 21 11:36:53 hrw: can you tell me quickly what N770 functionnalities I'd loose if I installed poky or angstrom on it ? Nov 21 11:37:03 I think you were thinking about that yesterday Nov 21 11:39:43 Thanks XorA and hrw Nov 21 11:43:04 nud: need to hack to get wifi module working, do not remember does all keys work Nov 21 11:48:02 bb in 30m Nov 21 11:50:09 hrw: not now. It should be split/merged w/ other patches Nov 21 11:50:21 XorA: now. Where is that tree? Nov 21 12:03:53 ok I have the deb packages building but when I run bitbake package-index it only does the repository indexes for the ipk packages, is there a way to build the indexes for the deb packges as well or do I need to do that manually/ Nov 21 12:20:51 r Nov 21 12:20:58 e Nov 21 12:25:24 hi rob_w Nov 21 12:25:37 hi magumbade Nov 21 12:25:43 dorian ? Nov 21 12:25:49 yep Nov 21 12:26:07 ok ,, check #rtl8186 ;-) Nov 21 12:27:34 03xora 07org.oe.dev * ra02782dc... 10/ (5 files in 4 dirs): xserver-kdrive_1.3.0.0.bb : add Xglamo server for Neo1973 Nov 21 13:00:12 03tmbinc 07org.oe.dreambox * r335eff19... 10/ (3 files in 3 dirs): enigma: backport subtitle memleak fix Nov 21 13:09:49 bye! Nov 21 13:17:28 lumag|gone: in hopes you read backlock, http://opensource.wolfsonmicro.com Nov 21 13:30:29 koen, friggin OSK loads u-boot into RAM such that I can't dl the X11 images into RAM to prepare for flashing ... Nov 21 13:31:19 hrw: grep for 'bx' (armv4t) and 'clz' (armv5te) Nov 21 13:31:35 * koen returns for a wild goose chase Nov 21 13:31:50 "no, you're expected here tomorrow, not today" Nov 21 13:32:59 XorA: how's Xglamo doing these days? Nov 21 13:33:04 Crofton|home: that sucks Nov 21 13:33:19 yeah Nov 21 13:33:39 I guess I need to see if I can flash from CF card ... Nov 21 13:33:47 or see about relocating u-boot Nov 21 13:34:09 koen: can you tell armv6 and armv6-vfp mnemonics too? Nov 21 13:35:06 good morning Nov 21 13:35:21 hmm..... no support for EXEEXT (.exe on cygwin) in bitbake for do_package (movefile) Nov 21 13:35:43 koen: try it and see :-D Nov 21 13:38:57 is something like ser_cont = (len & 0x07) | 0x10; likely to break on different endianness ? Nov 21 13:40:01 nud: no Nov 21 13:40:08 hrw: not without googling :) Nov 21 13:40:20 XorA: send me a gta02 :) Nov 21 13:41:31 koen: gah, you never solve problem yourself :-D Nov 21 13:41:34 XorA: kernel updates in your recent commit? Nov 21 13:41:49 mwester: gta02 only Nov 21 13:41:56 http://rafb.net/p/aI4VJH14.html <-- how hard would it be to fix such a failure ? Nov 21 13:42:13 koen: ping Nov 21 13:42:15 Ok. That makes it easier for me to sync up today, then. :) Nov 21 13:42:24 mickey|berlin: pong Nov 21 13:43:49 koen: how much bandwidth does the mtn server consume per month? Nov 21 13:44:11 mickey|berlin: a neglible amount Nov 21 13:44:26 I would be surprised if it was >2GB/month Nov 21 13:44:29 Anyone have any comments on using the latest binutils with an older glibc? Is this an ok think to do usually, or rather like trying to ride two horses at the same time? Nov 21 13:44:57 would it make sense to define TARGET_EXEEXT, HOST_EXEEXT and BUILD_EXEEXT, all of-course defaulting to empty string and then define them to .exe for cygwin? Nov 21 13:48:18 koen: i need an exact number. do you have stats? Nov 21 13:49:41 * Crofton|home is thinking the only way to get stats is with some kind of IP flow monitoring .... Nov 21 13:50:49 someone know how to tell quilt that when I refresh patch I want to get function names in resulting patch? Nov 21 13:50:58 like 'diff -p' one does Nov 21 13:51:10 mickey|berlin: don't have exact stats, the box run tinderbox and various other things Nov 21 13:51:45 mtn pull gives a number of bytes xfer'ed, you could estimate from that ... Nov 21 13:53:07 hmm Nov 21 13:53:15 (74K + 46K) x ndevs for daily traffic Nov 21 13:53:20 my last pull was yesterday Nov 21 13:53:22 koen: ok, i need an upper boundary i can commit to Nov 21 13:53:29 hrw: QUILT_DIFF_ARGS Nov 21 13:53:33 less than 10G/m? Nov 21 13:53:38 broonie: thx Nov 21 13:53:45 (IIRC, anyway) Nov 21 13:53:47 mickey|berlin: certainly Nov 21 13:53:54 good. Nov 21 13:53:55 broonie: it is confusing as there is also QUILT_DIFF_ARGS Nov 21 13:54:04 what is ndevs? Nov 21 13:54:22 ndevs is tough to estimate Nov 21 13:54:31 we have 60+ keys in the db Nov 21 13:54:32 guess :) Nov 21 13:54:40 but we also have a lot of users Nov 21 13:54:40 and 619 devs Nov 21 13:54:42 so lets use 40 devs Nov 21 13:54:54 and 200K of revs per day Nov 21 13:55:04 oops Nov 21 13:55:10 forgot about ro Nov 21 13:55:18 200 users Nov 21 13:55:33 mickey|berlin: we could just use the scheme we had in the good old bk days Nov 21 13:55:43 one server for devels, one for users Nov 21 13:55:48 200K x 200 devs x 30 days = 9,012,000 K Nov 21 13:55:58 the devel server can only be accessed by developers Nov 21 13:55:59 where would the right place be to check TARGET_OS and BUILD_OS for *cygwin* and conditionally define _EXEEXT, in bitbake.conf ? Nov 21 13:56:05 rough estimate Nov 21 13:56:34 koen: i'm going to get us a fat ass server Nov 21 13:56:42 koen: i just need the bandwidth boundary Nov 21 13:56:45 heh Nov 21 13:57:01 Don't forget that the nslu2-linux.org server syncs regularly (multiple times per hour) Nov 21 13:57:02 mickey|berlin, what are the bw choices? I would guess they are discrete numbers Nov 21 13:57:10 mickey|berlin: fat ass + one-thread-monotone... Nov 21 13:57:22 mwester, but the number of bytes is low for each sync .... Nov 21 13:57:24 mwester: does it still run own automerger? Nov 21 13:57:29 mwester: we're going to change to autosynchronization soon Nov 21 13:57:29 Crofton|home: if it's < 100G/m, i'm ok Nov 21 13:57:45 hrw: it should not be automerging; that should be happening only on the oe server. Nov 21 13:57:52 my estimate was 10M/month :) Nov 21 13:58:20 so even if I wrong by an order of magnitude, you should be ok Nov 21 13:58:46 hrw, at least the one thread is only occupied for a shorter period of time .... Nov 21 13:58:50 Crofton: 9,000,000K = 9000M Nov 21 13:59:17 ok, so I ma wrong by an order of magnitude .... Nov 21 13:59:25 several in fact Nov 21 13:59:41 3 Nov 21 13:59:47 right Nov 21 13:59:54 * Crofton|home goes back to dealing with email Nov 21 14:00:34 03koen 07org.oe.dev * r890f2107... 10/ (1 packages/cryptsetup/cryptsetup_1.0.5.bb): cryptsetup: run through oe-stylize and remove postinst hack Nov 21 14:01:03 Crofton|home: its not your fault, you just cant stand all those orders of magnitude dropping off the dollar :-) Nov 21 14:01:09 grrr Nov 21 14:02:19 mickey|berlin: and we use a compressed ssh tunnel for the mtn traffic anyway :) Nov 21 14:02:48 koen: there is a ~ switch for ssh to give the stats isnt there? Nov 21 14:03:04 koen: then you just divide by how long tunnel has run Nov 21 14:03:06 XorA: dunno Nov 21 14:03:44 XorA, is jealous because i am an attractive consultant due to the currenncy issues Nov 21 14:04:21 koen: actually not, I though ~# also gave amount of traffic per connect Nov 21 14:05:14 Crofton|home: you work for pennies instead of pounds? Nov 21 14:05:50 koen: ya, but this is a crude hack. i want to get away from that Nov 21 14:05:51 Crofton|home: so are you outsourcing india? Nov 21 14:05:51 right Nov 21 14:05:59 soon! Nov 21 14:06:00 hi Nov 21 14:06:10 mickey|berlin: a tunnel is faster than switching dns ;) Nov 21 14:06:16 * XorA probably should stup this before Crofton|home fits him for orange boiler suit Nov 21 14:06:52 I heard that even topmodels and US music 'stars' us euro instead of dollars now Nov 21 14:11:14 anyone with half a brain that has business outside the US Nov 21 14:12:15 The sun newver sets on the American Empire! Nov 21 14:12:24 er wait, wrong empire .... Nov 21 14:13:29 mickey|berlin: http://www.dilbert.com/comics/dilbert/archive/images/dilbert2002222371121.gif Nov 21 14:13:57 http://rafb.net/p/aI4VJH14.html <-- do someone have an idea how such a bug could be fixed ? (usbvision on ARM pxa270) Nov 21 14:14:47 nud: try disabling preempt Nov 21 14:15:54 hrw: *nod* Nov 21 14:17:10 koen; do you know how to do that ? Nov 21 14:17:25 mmh must be in .config or so Nov 21 14:18:07 hrw: Dilbert is a documentary Nov 21 14:20:23 * nud wonders what disabling PREEMPT will imply Nov 21 14:21:24 nud: in worst situation your buildhost will explode Nov 21 14:22:05 in a great burst of sparks and a shower of PC components Nov 21 14:22:50 letting out the magic blue smoke that makes it work Nov 21 14:35:21 all: gta02 available when? Nov 21 14:35:46 Crofton: belated pong Nov 21 14:37:10 morning Nov 21 14:37:16 morning Nov 21 14:38:18 sakoman, gm Nov 21 14:38:28 I build a glibc gusmtix last night Nov 21 14:38:42 hopefully with ext2 Nov 21 14:39:01 Cool! Did it boot? :-) Nov 21 14:39:14 dunno, need to go over to school to try that Nov 21 14:39:31 I see you checked in some wifistix stuff Nov 21 14:39:36 yeah Nov 21 14:39:39 that was easy Nov 21 14:40:00 Hopefully we can drive the diffs down Nov 21 14:40:59 the basic setup is pretty good Nov 21 14:41:16 The other thing you ought to consider is cron -- should have INITSCRIPT_PARAMS = "defaults 65" to work right Nov 21 14:41:23 especially having a place for users to put stuff Nov 21 14:41:38 Yeah, need to write the "how to's" on that Nov 21 14:41:40 * Crofton|home is not a cron guy :) Nov 21 14:41:50 I actually have a use for it Nov 21 14:43:26 I had to add ENABLE_BINARY_LOCALE_GENERATION = "0" to my local.conf since I can not (easily) build qemu on my build machines Nov 21 14:44:22 That sucks! What's the issue? Nov 21 14:44:43 qemu needs gcc-3.4 to build Nov 21 14:45:13 there are ways of dealing with this, but since I do not care about LOALE stuff, I just skip the problem :) Nov 21 14:45:40 Works for me :-) Nov 21 14:45:44 yeah Nov 21 14:46:05 Aloha! Nov 21 14:46:06 for everyone in the "One language to rule them, in the land of Mordor ..." crowd Nov 21 14:46:19 oddly enough, some peopl elike to use their native languages Nov 21 14:46:29 Crofton: Feel free to send more suggestions if you see things that can be improved Nov 21 14:46:38 will do Nov 21 14:46:47 Crofton|home: :) Nov 21 14:47:05 * * OE Bug 3345 has been created by olvaffe(AT)gmail.com Nov 21 14:47:07 * * ecore-native stages unusable libraries Nov 21 14:47:09 * * http://bugs.openembedded.org/show_bug.cgi?id=3345 Nov 21 14:47:29 Crofton: A tough problem to make the newbies and experienced OE folks both happy Nov 21 14:47:39 yeah Nov 21 14:48:14 I like having your changes in an overlay, makes it much easier to find stuff that can move into .dev Nov 21 14:48:22 A few test subjects seem to have gotten up & running with minimal pain Nov 21 14:48:36 without a makefile even :) Nov 21 14:48:39 Yeah, I love overlays! Nov 21 14:48:54 Hey -- I've been asked for one :-) Nov 21 14:49:03 makefile that is! Nov 21 14:49:15 what I do not like you having a copy of .dev in svn Nov 21 14:49:25 but, I do not have a better alternative atm Nov 21 14:49:29 nutters Nov 21 14:49:58 Me either -- gave it lots of thought, but it seems the only rational approach Nov 21 14:50:23 I am hopefully we could have a stable Angstrom branch and make tarballs availabel Nov 21 14:50:23 Got to have it locked down or support will be hell! Nov 21 14:50:28 yeah Nov 21 14:50:36 koen says no plans for that Nov 21 14:50:50 so perhaps mine will be a default "stable" Nov 21 14:51:01 we need a sane way to update your copy of .dev Nov 21 14:51:14 it's called 'monotone' Nov 21 14:51:15 without gumstix and oe guys going crazy :) Nov 21 14:51:32 heh Nov 21 14:51:49 indeed Nov 21 14:52:01 well, got to run. have a flat tire to fix Nov 21 14:52:02 * koen wonders what actual gumstix users think of continuously being called "stupid aol users" Nov 21 14:52:15 l8r Nov 21 14:52:24 For the record I have *never* said anything like that Nov 21 14:52:33 koen, I suspect the ones that fit the description do not know we say it Nov 21 14:52:39 I don't even think it is remotely true Nov 21 14:52:43 and the others understand where it comes from :) Nov 21 14:52:56 craig kept saying that to justify his OE decisions Nov 21 14:53:24 He made different decisions than I have Nov 21 14:53:53 Hopefully I can keep all ends of the spectrum relatively happy Nov 21 14:54:00 good lcuk :) Nov 21 14:54:10 you chose hard problems Nov 21 14:54:31 Or at least keep a relatively flat level of unhappiness across the spectrum :-) Nov 21 14:54:49 fix the tire :) Nov 21 14:54:55 OK, l8r Nov 21 14:57:05 03xora 07org.oe.dev * r9752a217... 10/ (2 files in 2 dirs): linux-openmoko : commit missing patch Nov 21 14:57:10 03xora 07org.oe.dev * r704814d5... 10/ (3 files in 2 dirs): xserver-kdrive : commit missing patches Nov 21 15:03:44 hrw, I set FEED_ARCH_nokia800 = "armel" in the distro file but when I build a package it still has "Architecture: arm Nov 21 15:03:44 " in the deb control file, however it does now build the file under "depoy/glibc/deb/armel/" how do I change what gets writen into DEBIAN/control for the arcitecture? Nov 21 15:09:50 PACKAGE_ARCH I told.. Nov 21 15:10:54 hrw, I'm sorry, I just saw that :-/ Nov 21 15:14:32 hrw, do you know why the FEED_ARCH_nokia800 is set to armv6 and not armel by default? (Angstrom.inc) Nov 21 15:18:21 because nokia n800 is armv6 Nov 21 15:19:24 hrw, that is what I thought but a lot of the packages for the N800 are listed as armel and I have been able to find very little about armel (google). Nov 21 15:20:21 * XorA goes to sort out some banking Nov 21 15:20:40 ~armeabi Nov 21 15:20:48 ~eabi Nov 21 15:20:49 i guess eabi is the new "Embedded" ABI by ARM ltd which allow mixing soft and hardfloat code. Not yet supported for armv4l cpu's like StrongARM. Embedded distro with EABI: Angstrom. More info: http://wiki.debian.org/ArmEabiPort and http://www.angstrom-distribution.org Nov 21 15:20:56 svolpe: http://wiki.debian.org/ArmEabiPort Nov 21 15:21:59 koen: disabling PREEMPT doesn't change anything in the error message: http://rafb.net/p/aI4VJH14.html :-/ Nov 21 15:22:19 nud: line 10 Nov 21 15:22:33 koen: yes I saw Nov 21 15:22:39 but I've disabled CONFIG_PREEMPT Nov 21 15:22:53 the error says otherwise Nov 21 15:23:14 how could you check that on a running kernel ? Nov 21 15:24:11 hrw, so armel is just arm EABI. Nov 21 15:24:50 root@terra:~$ zcat /proc/config.gz | grep PREEMPT Nov 21 15:24:51 # CONFIG_PREEMPT is not set Nov 21 15:26:05 koen: looks like it is disabled... Nov 21 15:27:09 koen: I think I pasted the wrong link ;-) http://rafb.net/p/453cfw53.html Nov 21 15:27:37 svolpe: in short: yes Nov 21 15:27:50 nud: ok, so it's not preempt Nov 21 15:28:09 Ciao to everyone! :-) I'm starting a job on a handheld rfid reader with windows mobile 6 on board...OE could be used for developing application for it? Nov 21 15:28:12 hrw, from what I read it seems like you would want armel since it is more efficient. Nov 21 15:28:29 nud: tried decoding it with ksymoops yet? Nov 21 15:29:20 sbaturzio: OE is for linux devices Nov 21 15:29:43 hrw: ah...I see Nov 21 15:30:07 koen: ksymoops are 2.4 iirc Nov 21 15:30:09 and there aren't free alternatives for developing for windows mobile? Nov 21 15:30:31 sbaturzio: I never owned wm powered device so no idea Nov 21 15:30:44 koen: I didn't Nov 21 15:31:01 hrw: afaik it works on 2.6 as well, but you need to run it on the target, since it's using /proc/ksyms Nov 21 15:31:08 hrw: ok, thank you anyway ;-) Nov 21 15:32:09 sbaturzio: ms visual studio express does not support wm? Nov 21 15:32:15 nud: ipkg install ksymoops ; ksymoops -k /proc/kallsyms Nov 21 15:33:08 ipkg install ksymoops ; ksymoops -k /proc/kallsyms file-with-oops.txt Nov 21 15:33:11 that is Nov 21 15:33:27 hrw: don't know, but I think my handheld SDK need evc++, not the express. But I'll give a look to vs-express anyway Nov 21 15:34:57 koen: has it to be directly on the device ? Nov 21 15:36:18 * chouimat|work is impressed by how gnu source code can be crap Nov 21 15:36:56 chouimat|work: there're crap code everywhere ;-) Nov 21 15:37:25 otavio: true Nov 21 15:38:25 nud: sadly, yes Nov 21 15:38:50 * nud builds ksymoops Nov 21 15:39:24 could someone help with a libdir issue? I'm having trouble to understand why libx11 is using a wrong libdir value when running do_configure (local.conf: http://paste.debian.net/43083) Nov 21 15:40:58 otavio: DISTRO=angstrom-2007.1 Nov 21 15:41:02 generic is broken Nov 21 15:41:18 koen: generic is broken? Hell! Nov 21 15:41:52 koen: Ok .. I'll take a look at it then and see if I can spot out what's wrong and provide a patch Nov 21 15:42:31 koen: why angstrom-2007.1? Any special reason? anywhere to look for description of available distros? Nov 21 15:43:24 angstrom 2007.1 is the best maintained distro in OE Nov 21 15:47:11 looks like we still didn't reach our goal of sane defaults yet Nov 21 15:48:23 stupid question but is OE capable of generating a complete sane linux 2.4 image? Nov 21 15:48:34 with uclibc linked agains 2.4 headers and so on? Nov 21 15:48:45 koen: it would be nice if it was documented on the 'Getting started manual' :( Nov 21 15:48:55 koen: thank you a lot :-) Nov 21 15:51:18 ~curse qemu 'configure' script Nov 21 15:51:19 May the fleas of a thousand camels infest your most sensitive regions, qemu 'configure' script ! Nov 21 15:51:51 ln -sf angstrom.conf generic.conf :) Nov 21 15:51:56 Warning (read_ksyms): no kernel symbols in ksyms, is /proc/kallsyms a valid ksyms file? Nov 21 15:51:57 mmmh Nov 21 15:52:11 mickey|berlin, someone just needs to give generic a looking over Nov 21 15:52:26 it would be nice to have a basic distro, without feeds etc Nov 21 15:54:10 chouimat|work: I'm looking what is different and to see if I can fix it Nov 21 15:54:12 op Nov 21 15:54:19 Crofton|home: ^^ Nov 21 15:54:20 cool Nov 21 15:54:27 from overview on the website , man can read "Currently it can run on any Linux distribution, and plans are in place to allow it to work under Windows." Nov 21 15:54:35 who works on this problem ? Nov 21 15:54:42 * Genesis need to do that too Nov 21 15:54:46 otavio, we just need some to do the work :) Nov 21 15:54:47 * Genesis can help Nov 21 15:55:53 I'm just learning how oe works :-) Nov 21 15:56:14 It looks amazing how easy is to do things with it (even I took a days to discover generic.conf was broken ;P) Nov 21 16:00:07 koen : do you know who works on windows OE port ? Nov 21 16:01:49 03koen 07org.oe.dev * r67dfddbb... 10/ (1 conf/local.conf.sample): local.conf.sample: update DISTRO to something that works for 99.9% of the users Nov 21 16:02:10 great idea :) Nov 21 16:06:36 hrw, PACKAGE_ARCH_nokia800 = "armel" Nov 21 16:06:36 causes the ipk's to be built as armel packages but the deb packages are still being built as arm even though OE is building them in deploy/glibc/deb/armel Nov 21 16:08:53 hrw, maybe I need to use DPKG_ARCH. I'm looking at the package_deb.bb right now. Nov 21 16:13:46 it looks like package_deb.bbclass has DPKG_ARCH ?= "${TARGET_ARCH}", shouldn't it be DPKG_ARCH ?= "${PACKAGE_ARCH}" Nov 21 16:14:18 that way if you set PACKAGE_ARCH it will hold true for the deb packages as well? Nov 21 16:15:00 it is TARGET_ARCH because dpkg does not handle multiple archs Nov 21 16:15:31 so all packages for arm variants are done as 'arm' but put in arm/armv4t/armv5te/armv6/armv7 feeds Nov 21 16:18:12 hrw, thanks, that makes sense. Nov 21 16:21:01 hrw, I guess it would be pretty easy to set up a distro in OE that is targeted at building armel.deb packages for a specific n800 os (OS2007, IT2008), all you really need to do is set up all the important lib versions so there are no dependancy issues ie: (requires libc6 >2.6.1 but OS2007 uses libc2.3.5). I'll think about it some more and maybe give it a crack :-) Nov 21 16:22:06 it2008 still isn't eabi? Nov 21 16:22:08 sjeesh Nov 21 16:22:12 slackers Nov 21 16:23:33 koen, not sure about it2008. Nov 21 16:23:43 glibc 2.3.5 isn't eabi Nov 21 16:23:50 it's oabi + small eabi helper Nov 21 16:23:54 koen, that is what OS2007 uses. Nov 21 16:24:05 ah, right, I misread, sorry Nov 21 16:24:16 koen, it2008 is not released yet. Nov 21 16:24:17 its glibc-csl etc Nov 21 16:24:28 svolpe: it2008 was released already Nov 21 16:24:30 I have it2008 on my n800 :) Nov 21 16:24:41 hrw, officially? Nov 21 16:24:42 svolpe: 'n810 firmware leaked' == release ; Nov 21 16:24:59 hrw, yes I knew about the leak but its not official yet, correct? Nov 21 16:25:12 svolpe: n810 is in US stores Nov 21 16:26:17 hrw, yes and sometime in Dec is the official it2008 release for the N800s, although it seems dumb since the N810s are out with it and it has leaked to the N800s as well. Nov 21 16:27:31 the us stores have "out of stock" now Nov 21 16:27:46 svolpe: to tell the truth - I do not care does os2008 was released or not. Nov 21 16:28:17 svolpe: I have nokia770 which is not supported anymore and I do not have n800 or plans to buy n810 Nov 21 16:29:26 svolpe: according to nokia "tablets group" I live in that worst part of europe and I am not supposed to even know that such hardware exists Nov 21 16:29:26 hrw, I completely understand. The n800 was not my first choice with all the closed stuff, but it WAS the only one in the price range I wanted to stay :-/ Nov 21 16:30:34 hrw, Neo1973 would have been my first choice :-) Nov 21 16:31:11 hrw, wow thats harsh. Nov 21 16:32:26 hrw, what the "tablets group" say, that is. Nov 21 16:32:52 svolpe: officially nokia tablets do not exists in Poland - you will not find them on nokia website or in nokia shops Nov 21 16:36:19 So, the N770 you have is contraband? Nov 21 16:36:21 ;) Nov 21 16:36:46 * mwester imagines the Nokia Secret Police breaking down hrw's front door... Nov 21 16:37:04 mwester: ;) Nov 21 16:37:23 mwester: I do not want to think what will happen when it will get broke. Nov 21 16:37:51 no official service in Poland... but who cares - I do not even have warranty card Nov 21 16:40:26 hrw, I guess I'm not very up on the politics, why would Nokia take that stance? Nov 21 16:46:44 no idea Nov 21 17:06:18 evening! Nov 21 17:07:30 hi lumag Nov 21 17:07:53 XorA|gone, yes, that's pretty interesting. Nov 21 17:09:27 lumag: we need to push more code upstream then Nov 21 17:10:59 currently we are discussing the TMIO patchset with Ian. Nov 21 17:11:26 tosa-power is a mess. Nov 21 17:12:10 ko Nov 21 17:13:27 tosa-keyboard needs a cleanup to use generic gpio-keys Nov 21 17:14:45 tosa-bluetooth is ready IMO. Nov 21 17:16:55 how do you autoload a module in OE ? Nov 21 17:17:04 back in october's OE Nov 21 17:17:52 lumag: what's the plan for merging the tosa tmio version with the one from ian? Nov 21 17:17:53 autoload_modulename = 'modulename' Nov 21 17:18:29 hrw: in a bb file ? Nov 21 17:18:44 in the one which builds module Nov 21 17:19:26 tosa-tmio-lcd should be probably rewritten to use backlight class Nov 21 17:19:36 mh this means in the kernel bb file Nov 21 17:20:24 arm-dma-coherent and usb-ohci-hooks should go via the tmio patchset Nov 21 17:21:08 re Nov 21 17:21:43 koen: I followed your suggestion and tried midori on gpe, but it doesn't even load, gives a segfault Nov 21 17:22:22 tosa-pxaac97 can go Nov 21 17:23:11 lumag: backlight is easy Nov 21 17:23:30 koen: He plans to resubmit core changes. After that I'll try to get most our changes ported to new tmio core Nov 21 17:23:53 has anyone played with OpenEmbedded on a Sigma Designs device? Nov 21 17:24:26 wm97xx patch should probably go via wolfsonmicro? Nov 21 17:25:14 jamie___: as 'build on' or 'build for'? Nov 21 17:25:15 wm9712-* patches look like hacks and probably should not be merged Nov 21 17:25:56 who works on windows port ? Nov 21 17:26:04 i could help him Nov 21 17:28:36 hrw: as in 'build for'. you don't run GCC on these things, they're a bit wimpy for that :-) Nov 21 17:29:12 jamie___: you checked OE/conf/machine/ directory already? Nov 21 17:29:34 jamie___: and I built angstrom under angstrom one day Nov 21 17:29:53 but it was in chroot on my x86 desktop Nov 21 17:30:44 heh. my (non-oe) tree takes ~40 minutes to build on a 2GHz Core Duo. the device is a 166MHz ARM w/ 16MB RAM and no MMU. I'm not wanting to try it. Nov 21 17:31:18 16MB ram and no mmu? Nov 21 17:31:24 jamie___: arm7 cpu? Nov 21 17:31:43 yes, arm7 glued onto an unclear number of specialised coprocessors Nov 21 17:31:58 and still stuck in the land of 2.4.26-uc0 Nov 21 17:33:14 heads up toolchain ppl: http://gcc.gnu.org/viewcvs/trunk/gcc/fold-const.c?r1=130238&r2=130258&pathrev=130258 Nov 21 17:34:45 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34130 Nov 21 17:38:13 http://upload.wikimedia.org/wikipedia/commons/7/73/Ana.b747.pokemon.arp.750pix.jpg Nov 21 17:38:45 yuck Nov 21 17:38:55 I flew in a plane to the Linux realtime conference in Austria, and the plane was labeled "OE-LTD" :-) Nov 21 17:39:18 hrw: what do you mean by 'backlight is easy' ? :) Nov 21 17:39:22 http://www.airliners.net/open.file?id=1278924&WxsIERv=Qr%20Univyynaq%20Pnanqn%20QUP-8-314%20Qnfu%208&Wm=1&WdsYXMg=Nhfgevna%20Neebjf%20%28Glebyrna%20Nvejnlf%29&QtODMg=Senaxsheg%20nz%20Znva%20%28Eurva-Znva%20NO%29%20%28SEN%20%2F%20SES%20%2F%20RQQS%29&ERDLTkt=Treznal&ktODMp=Ncevy%2015%2C%202007&BP=1&WNEb25u=Gubenys%20Qbruevat&xsIERvdWdsY=BR-YGQ&MgTUQtODMgKE=&YXMgTUQtODMgKERD=198&NEb25uZWxs=2007-10-11 Nov 21 17:39:23 %2021%3A25%3A32&ODJ9dvCE=&O89Dcjdg=400&static=yes&width=1024&height=721&sok=JURER%20%20%28ert%20%3D%20%27BR-YGQ%27%29%20%20BEQRE%20OL%20cubgb_vq%20QRFP&photo_nr=1&prev_id=&next_id=1241076 Nov 21 17:39:27 wooo Nov 21 17:39:30 sry Nov 21 17:40:03 likewise: it seems you volunteered to add the patch :) Nov 21 17:40:23 * koen foresees the autobuilder keeping the lab a bit warmer the next few days Nov 21 17:40:50 :-) Nov 21 17:43:51 lumag: writing backlight drivers Nov 21 17:44:05 ahh :) Nov 21 17:44:35 lumag: my first code contribution to kernel was backlight driver for progear tablet Nov 21 17:44:41 :) Nov 21 17:44:53 lumag: and backlight-corgi is really simple Nov 21 17:45:24 Good evening Nov 21 17:45:39 Any bitbake gurus online? Nov 21 17:46:19 tosa-backlight is a bit 'messy' but only a bit :) Nov 21 17:46:56 anttix: no guru, but maybe you want to ask a FAQ? Nov 21 17:47:03 I was wondering about the dependency mechanism of bitbake. I have a recepie that has to copy some stuff from kernel tree before patching. However the problem is that do_unpack and do_patch get called before kernel compile and thus KERNEL_SOURCE is undefined. Nov 21 17:47:25 if I compile kernel manually everything works. Nov 21 17:47:46 anttix: which kernel bitbake recipe? Nov 21 17:47:58 likewise, how was the conference? Nov 21 17:47:59 likewise: in-house Nov 21 17:48:04 It looks like this: Nov 21 17:48:41 lumag: its part of tmiofb? Nov 21 17:49:03 ... DEPENDS = "virtual/kernel" Nov 21 17:49:03 inherit module Nov 21 17:49:03 addtask copy_sources after do_unpack before do_patch Nov 21 17:49:03 do_copy_sources() { Nov 21 17:49:03 mkdir -p ${S}/drivers Nov 21 17:49:04 mkdir -p ${S}/fs Nov 21 17:49:06 cp -r ${KERNEL_SOURCE}/drivers/mtd ${S}/drivers Nov 21 17:49:18 Crofton|home: many talks, 1/3rd extremely interesting, 1/3rd nice to know, 1/3 crap Nov 21 17:49:32 sounds normal :) Nov 21 17:49:39 no.. arch/arm/mach-pxa/tosa_lcd.c Nov 21 17:49:58 Crofton|home: I'm keeping an eye on the conference papers/presentations. I can point you to some really good ones, if they appear. Nov 21 17:50:15 thanks Nov 21 17:50:37 Unfortunately, 4 tracks in parallel... so I missed a few. Nov 21 17:51:26 that can also help with the avoiding crap aspect though :) Nov 21 17:51:39 Crofton|home: For the RTLW (which was also with ELCE this year) you can d/load the conference CD from their site: http://www.realtimelinuxfoundation.org/events/rtlws-2007/rtlws-2007.iso Nov 21 17:52:03 Now when kernel has not been compiled yet, KERNEL_SOURCE is undefined, but bitbake will nevertheless execute tasks do_unpack and do_patch before even looking at the DEPENDS line. Any way to make it reconsider? Nov 21 17:52:37 anttix: and where is KERNEL_SOURCE defined? Nov 21 17:52:48 lumag: indeed - backlight driver will be quite simple Nov 21 17:53:28 likewise: module-base.bbclass Nov 21 17:53:40 KERNEL_SOURCE = "${@base_read_file('${STAGING_KERNEL_DIR}/kernel-source')}" Nov 21 17:53:53 which is absolutely guaranteed to fail if the file doesn't exist yet ;) Nov 21 17:54:03 Actually it's just empty. Nov 21 17:55:46 bye all Nov 21 17:55:58 cya hrw|gone Nov 21 17:56:09 likewise: I could copy sources in do_configure_prepend , but then I have to apply patches manually which pretty much defeats the purpose of automated tasks :( Nov 21 17:57:44 anttix: So you depend on the location of the kernel source code, for building your module? Nov 21 17:57:45 likewise: and I can't patch the kernel directly since the whole idea of the package is to provide a patched module for a stupid variation of our hardware that needs the hack. Nov 21 17:58:18 anttix: I see. How does madwifi handle this, for example? Nov 21 17:58:41 likewise: Hmm, good idea, I'll check that. Nov 21 17:59:37 likewise: They depend on kernel sources on configure and build stages. Nov 21 18:00:16 likewise: my problem is that I need kernel tree to copy sources from before I can execute do_patch Nov 21 18:00:29 likewise: and bitbake apparently doesn't think that do_patch can depend on anything :P Nov 21 18:01:31 anttix: no idea. Maybe leave a question on the list about your problem, and maybe someone will come up with a really neat solution. Nov 21 18:03:03 likewise: Well I need a solution today :P I might just try some ugly hacks ;) Nov 21 18:03:56 likewise: I think bitbake is not meant to work like this so I'm pretty much on my own trying to force it to do something unnatural. Nov 21 18:12:44 anttix: with bitbake 1.8 you can state in the recipe that your do_unpack depends on the do_ stage of the kernel compile Nov 21 18:13:21 koen|away: like this: do_copy_sources[depends] = "virtual/kernel:do_populate_staging" ? Nov 21 18:13:47 do_copy_sources[recrdeptask] was what I was thinking Nov 21 18:13:57 but I'm not really into that area of bitbake :) Nov 21 18:14:14 koen|away: I was just reading base.bbclass and was wondering about similar lines before do_unpack there. Nov 21 18:14:27 koen|away: what's the difference? between depends and recrdeptask ? Nov 21 18:14:37 * koen has no idea Nov 21 18:15:55 koen|away: recrdeptask is used in base.bbclass like this: do_fetchall[recrdeptask] = "do_fetch" Nov 21 18:16:19 My guess is that it's meant to be recursive. Whatever that means of course ;) Nov 21 18:20:37 Well at first glance it seems to work, at least my build didn't die right away :P I'll see the results when it's finished. Thanx a lot! Nov 21 18:22:32 BTW: Running this test with BB_NUM_THREADS=8 and PARALLEL_MAKE="-j4" on a dual quad core machine with 6Gb of RAM and 15K RPM SAS hard disks. Too bad I only have this baby for a couple of more days before it goes to crunch some stupid database :) Nov 21 18:24:02 I know it's nothing compared to a proper build farm, but hey! We have a budget ;) Nov 21 18:25:23 03mwester 07org.oe.dev * r96fb7b25... 10/ (3 files in 3 dirs): Nov 21 18:25:23 portmap: SlugOS only - do not build with "-pie -fpie" due to binutils problems. Nov 21 18:25:23 (This change should be reverted when we get SlugOS working with binutils 1.18) Nov 21 18:25:27 03mwester 07org.oe.dev * r368c5337... 10/ (3 files in 3 dirs): autofs: remove dependency on bash, and add rc.update.d stuff to bb file. Nov 21 18:25:31 03mwester 07org.oe.dev * r7179eab2... 10/ (3 files in 3 dirs): Nov 21 18:25:31 slugos-init: turnup - use volatiles to create /var/log/lastlog when turned-up Nov 21 18:25:31 to external storage (NFS, memstick, or disk). This stops openssh complaining. Nov 21 18:25:36 03mwester 07org.oe.dev * r6114e88c... 10/ (1 packages/meta/slugos-packages.bb): slugos-packages: added portmap and nfs-utils back in. Nov 21 18:33:04 * * OE Bug 3346 has been created by likewise(AT)gmx.net Nov 21 18:33:06 * * Upstream GCC C compiler bug Nov 21 18:33:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3346 Nov 21 18:38:36 mickey: I'm assuming you're on gprs now? Did you see the xmms2 client? Nov 21 19:00:51 03likewise 07org.oe.dev * r4bdc2a4d... 10/ (3 files in 3 dirs): gcc-4.1.2: Follow upstream fix for pr34130. -C*abs(D) became abs(-C*D), not good. Nov 21 19:02:06 bonne soirée Nov 21 19:02:47 03likewise 07org.oe.dev * rc971ec89... 10/ (1 packages/gcc/gcc-cross_4.1.2.bb packages/gcc/gcc_4.1.2.bb): gcc-4.1.2: Update PR for gcc(-cross). Mention to update PR of -cross when .bb main changes. Nov 21 19:19:50 hi all ... i have really trouble to have a x86-qemu image buildable ... neither angstrom (2007. and 2008.1) neither 'generic' finish console-image !!! damn !!! i need it for a presentation in a couple of days !!! Nov 21 19:20:41 sincerelly is time to provide some 'release' or 'revision' know to be good or at least really really close to be good !!! Nov 21 19:21:05 gremlin[it]: well, there are images on the website. Nov 21 19:21:19 I dunno about a qemu image though Nov 21 19:22:34 no i need to show OE working ... not a final image ... Nov 21 19:22:47 it should be a intro course on using OE ... Nov 21 19:23:17 ah and qemu fails to build for you? Nov 21 19:23:45 no OE fail, every distro on a different package !!! Nov 21 19:23:51 gremlin[it]: http://www.angstrom-distribution.org/unstable/autobuild/qemux86/ Nov 21 19:23:59 #@COMMENT: This distribution configuration serves as a starting point for Nov 21 19:23:59 #@COMMENT: the integration of new target platforms, machines, or distributions. Nov 21 19:23:59 #@COMMENT: It bases on conservative settings and should always be buildable. Nov 21 19:24:12 taken from generic.conf ... Nov 21 19:24:21 do you have gcc-3.4 on your host? Nov 21 19:24:51 dcordes no my host gcc is a 4.1.2 ... Nov 21 19:25:34 koen thnaks ... but as i said i need to show OE in action ... so i need it at least build a x11 image of some kind ... Nov 21 19:25:40 gremlin[it]: builds fine over here.... Nov 21 19:27:06 koen angstrom 2007.9 mean i have to set DISTRO = "angstrom-2007.1" right ? Nov 21 19:27:17 yes Nov 21 19:29:10 ok let's try again ... Nov 21 19:32:44 koen which bitbake is need ? Nov 21 19:32:59 I use the 1.8 branch Nov 21 19:33:25 i have the 1.8.8 taken from fedora repos .... Nov 21 19:33:42 and now fail to fetch ! (yestarday it worked) ... Nov 21 19:34:57 mhh maybe i found ... strange (not sense) error is reported if source directory is inacessible ... Nov 21 19:35:04 * * OE Bug 3347 has been created by autobuild(AT)openembedded.org Nov 21 19:35:06 * * glibc-2.5-r8-do_compile Nov 21 19:35:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3347 Nov 21 19:35:16 * * OE Bug 3349 has been created by autobuild(AT)openembedded.org Nov 21 19:35:18 * * alsa-lib-1.0.14-r2-do_install Nov 21 19:35:20 * * http://bugs.openembedded.org/show_bug.cgi?id=3349 Nov 21 19:35:28 * * OE Bug 3348 has been created by autobuild(AT)openembedded.org Nov 21 19:35:30 * * glibc-intermediate-2.5-r8-do_compile Nov 21 19:35:32 * * http://bugs.openembedded.org/show_bug.cgi?id=3348 Nov 21 19:49:51 sakoman, attempting to flash glibc gumstix now Nov 21 20:33:07 sakoman, ping Nov 21 22:18:04 * * OE Bug 3350 has been created by gilles.mocellin(AT)free.fr Nov 21 22:18:06 * * gpe-clock applet does not expand to show entire time Nov 21 22:18:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3350 Nov 21 23:14:05 are there any javascript-enabled browsers that can be built for angstrom/glibc? minimo is failing on do_compile Nov 21 23:19:37 scruggs: mozilla looks to be available .. Nov 21 23:19:47 scruggs: i didn't build it yet Nov 21 23:21:12 yeah... I think minimo is just a scaled-down version of mozilla. I'll try the full-blown mozilla build and see what I get Nov 22 01:22:15 Does someone knows if there's a problem with the GIT mirror of oe? It's not updated since four days ago Nov 22 01:42:19 ~seen RP Nov 22 01:42:46 rp is currently on #maemo (2d 13h 3m 50s) #handhelds (2d 13h 3m 50s) #oe (2d 13h 3m 50s) #openmoko (2d 13h 3m 50s) #openzaurus (2d 13h 3m 50s). Has said a total of 12 messages. Is idling for 1d 15h 7m 5s, last said: 'koen: Thats broadly along the lines I was thinking too'. Nov 22 01:43:58 ~seen zecke Nov 22 01:43:59 zecke was last seen on IRC in channel #oe, 2d 11h 10m 2s ago, saying: 'yeah we have a new Makefile...'. **** ENDING LOGGING AT Thu Nov 22 02:59:56 2007