**** BEGIN LOGGING AT Thu Oct 29 02:59:57 2009 Oct 29 07:18:07 03Jeremy Lainé  07org.openembedded.dev * r8452d20bc5 10openembedded.git/recipes/linux/linux-2.6.31/boc01/ (boc01.dts boc01.dts.v1): linux-2.6.31: declare boc01 LED in devicetree Oct 29 07:18:09 03Jeremy Lainé  07org.openembedded.dev * r31c440b89c 10openembedded.git/: Merge branch 'org.openembedded.dev' of ssh://git@git.openembedded.net/openembedded into org.openembedded.dev Oct 29 08:21:55 good morning Oct 29 08:25:17 03Sebastian Spaeth  07shr/import * r267b8f6a49 10openembedded.git/ (3 files in 3 dirs): tzdata: 2009n->2009p in shr/import as upstream src disappeared. Not for upstreaming. Oct 29 08:28:41 good morning Oct 29 08:33:11 good morning Oct 29 08:34:58 morning Oct 29 08:35:21 I'm doing a "bitbake -c clean u-boot" because of a compile error. How can I also tell it to clean everything that u-boot depens on? Oct 29 08:36:07 mickeyl: good morning Oct 29 08:36:35 morning pb_ Oct 29 08:47:56 A version that builds fine on my colleagues computer fails building u-boot on mine. How can I clean everything u-boot related? Oct 29 08:49:44 Hello ! Oct 29 09:02:55 tasslehoff: there is a local.conf setting to tell bitbake to clean all the deps out also but use with care (and I can't recall what it is off the top of my head) Oct 29 09:09:03 mickeyl: did you manage to capture any more information about your crashy python? Oct 29 09:09:27 good morning Oct 29 09:09:30 pb_: unfortunately not yet, but that's on my schedule for later note Oct 29 09:09:32 florian: good morning Oct 29 09:09:33 err Oct 29 09:09:34 mickeyl: righto Oct 29 09:09:36 s/note/today/ Oct 29 09:09:57 * mickeyl captured in some nasty SMS PDU issues atm. Oct 29 09:10:05 ~lart 3gpp specs Oct 29 09:10:05 * ibot steals 3gpp specs's mojo Oct 29 09:10:54 heh... sms is evil Oct 29 09:23:31 I got hit by checksums.ini again.... Oct 29 09:25:31 morning Oct 29 09:26:27 I created a new recipe and it builds successfully. How do I verify that everything became as I wanted? Ie, can I extract it without actually install it? or just print out what It would do if I installed it? Is this done with opkg? Oct 29 09:26:55 Morning XorA Oct 29 09:30:27 Is it ok to send xorg-7.5 patch now to ML? updated preferred versions and all missing recipes in one patch? Oct 29 09:30:58 jovox: an ipk is just an ar archive with two tar files in it Oct 29 09:31:08 you can always ar x your.ipk Oct 29 09:31:16 or use dpkg-deb Oct 29 09:31:16 then look at the tar.gz file Oct 29 09:31:32 JaMa: checksums present? Oct 29 09:31:37 but before sumbitting I would strongly encourage to test it on a target ! Oct 29 09:31:46 or use dpkg-deb :-) Oct 29 09:31:59 JaMa: why not use short live branch? Oct 29 09:32:10 (oldtimers use ar and tar and so on :-) ) Oct 29 09:32:59 hrw: I don't have commit access so short live branch would be needed to maintain by someone else in OE repo.. Oct 29 09:33:08 eFfeM: so I am not oldtimer as I use dpkg Oct 29 09:33:16 hrw: we're using it in shr distribution on om-gta02.. Oct 29 09:33:24 hrw: so its already in shr/merge branch.. Oct 29 09:33:30 JaMa: I think that you should apply for r/w Oct 29 09:34:43 hrw: OK, I'll ask someone for asking r/w for me :) Oct 29 09:35:07 JaMa: write about it by yourself Oct 29 09:35:19 JaMa: I will vote for you Oct 29 09:36:25 hrw: ok thanks Oct 29 09:36:31 hrw, guess your not an old-timer then Oct 29 09:36:46 * eFfeM is from the unix V7 era Oct 29 09:36:59 everything after that is newspeak :-) Oct 29 09:37:14 eFfeM: I started using unix in 1996 Oct 29 09:38:06 hrw, i beat you by a mile, first user experience in 1983 or so, first real programming work late 1984 Oct 29 09:38:28 eFfeM: I am 1976 - thats why. Oct 29 09:38:47 eFfeM: and eastern europe so it was worse when it comes to computers Oct 29 09:39:17 (port nroff from php11 with 16 bit addresses to mc680x0 with 32 bit addresses, all kind of sizeof(int) != sizeof(char *) issues Oct 29 09:39:29 * eFfeM is actually eFfeM58 Oct 29 09:40:33 hrw, actually you still beat me as you started with linux at 20, I was 26 when I did start with unix Oct 29 09:40:45 but *I* still had an email address with a bang path Oct 29 09:40:45 eFfeM: I said unix, not linux Oct 29 09:40:52 hrw, oops Oct 29 09:40:58 it was sunos4 Oct 29 09:41:23 for most people that started after 95 unix == linux Oct 29 09:41:35 HP2623A terminals were ugly beasts but they were students proof Oct 29 09:41:52 sunos 4, ah, used that as well (actually in '83 I did something with sunos2 iirc) Oct 29 09:43:15 eFfeM: worse thing is Unix disnt improve since 83, even in linux variety :-D Oct 29 09:44:21 writing python scripts is hard sometimes Oct 29 09:44:51 XorA: I tend to agree, seems the latest thing that was really an improvement was sockets and iirc that was bsd 4.0 or so (don't think v7 had them natively) Oct 29 09:45:17 * eFfeM thinks writing python scripts is always hard (and the same goes for perl) Oct 29 09:45:58 eFfeM: I have to learn python one day Oct 29 09:48:12 now I need multidimension table Oct 29 09:50:32 hrw: guess I have to learn it too one day :-( Oct 29 09:50:37 already have a book Oct 29 09:50:52 fortunately real programmers can write C in any language :-) Oct 29 09:52:07 I did a dpkg-deb -c to list the contents of my new ipk but it appears to be empty. My recipe is really simple, more or less just a install -m 0644 ${WORKDIR}/index.php ${D}qmdc/. There were no errors or when it built so how come it's empty? Oct 29 09:53:54 jovox: i think php files are not added to the recipe by default Oct 29 09:53:57 try something like Oct 29 09:54:23 FILES_${PN} += "qmdc/index.php" or so Oct 29 09:54:45 (peek at another recipe for exact syntax and details) Oct 29 09:55:26 and/or read http://docs.openembedded.org/usermanual/usermanual.html Oct 29 09:55:26 hmm oki.. I thought "install" was more or less a copy.. Oct 29 09:55:50 eFfeM: thats why bitbake looks like C Oct 29 09:55:51 install is a copy to the install dir, but does not add it to the package Oct 29 09:56:10 only some files are automatically added Oct 29 09:56:29 ah ok I see.. Thanks Oct 29 09:56:31 hrw, is that a :-) or a :-( ? Oct 29 09:57:39 ;D Oct 29 10:00:56 the ipk is still empty.. oki.. I will look at other recipes and see if I can figure out whats wrong Oct 29 10:01:43 jovox: guess it is the part after the += Oct 29 10:03:24 can I do something like: Oct 29 10:03:25 EXTRA_MYTHTVCONF = "--cpu=${arch}" Oct 29 10:03:25 EXTRA_MYTHTVCONF_armv7a = "--cpu=cortex-a8" Oct 29 10:03:42 and will the 2nd line then override for armv7a ? Oct 29 10:03:52 my SRC_URI="file://index.php" (the index.php is located in my_package/files) and FILES_${PN} += "index.php" and finally install -m 0644 ${WORKDIR}/index.php ${D}qmdc/ Oct 29 10:05:00 jovox: then in the install dir it will end up in qmdc/index.php so you need to add that one Oct 29 10:05:18 also I would expect you'd have ${D}/qmdc (so with a slash) Oct 29 10:06:06 what do you mean with "have to add that one"? Oct 29 10:06:33 FILES_${PN} += "qmdc/index.php" Oct 29 10:06:36 so add the dir Oct 29 10:07:10 btw iit might be that you have gotten some error messages or warnings (i'm not too versed in this area) Oct 29 10:07:16 cd .. Oct 29 10:11:30 http://hrw.pastebin.ca/1647621 - new checksums.ini format Oct 29 10:12:42 the idea is to remove duplicate entries which differ only by url and to be able to check checksums even if your url is different Oct 29 10:13:03 eFfeM: now I get "NOTE: the following files were installed but not shipped in any package: /qmdc/index.php" Oct 29 10:13:57 jovox: exactly that is because your FILES_${PN} is not correct, that is the line that specifies that you need to add this file Oct 29 10:14:10 JaMa: voted Oct 29 10:14:13 jovox, i am also worried by the / in /qmdc.... Oct 29 10:14:36 hrw, format looks ok to me Oct 29 10:14:53 hrw but what I really would like to have is an easy mechanism to add new checksums. Oct 29 10:15:02 you mean the missing /? Oct 29 10:15:10 jovox the starting / Oct 29 10:15:21 seems it is in the root dir Oct 29 10:15:31 echeck your install dir Oct 29 10:15:37 eFfeM: more info? Oct 29 10:15:57 ${D}/qmdc/ <- thats how my install looks now Oct 29 10:16:20 hrw, now i often end up having to add it twice (since someone added somethign else inbetween) and I like to do a new pull before I commit and push Oct 29 10:16:39 also i need to manually sort myself (ofc I use a script for that) Oct 29 10:17:21 but guess I'd like something like newchecksums.ini or so that will be merged automatically when the commit is there Oct 29 10:17:29 eFfeM: I check some other recipes.. Oct 29 10:17:33 jovox, have you looked in your work dir ? Oct 29 10:17:43 there is an install dir in there Oct 29 10:17:49 is your file in that dir ? Oct 29 10:18:15 eFfeM: there was a patch on oe ml for multiple sources of checksums.ini - can you look at it? Oct 29 10:18:45 hrw, guess I missed it, will try to hunt it down from the archive, how long ago was it? Oct 29 10:20:00 btw and if i am sometimes slow it is because i am triple tasking Oct 29 10:20:11 [oe] [PATCH] base.bbclass: Combine all checksum.ini files found in BBPATH Oct 29 10:20:17 eFfeM: but looks like it got merged Oct 29 10:20:25 eFfeM: it works now, added the /. Thanks you very much! Oct 29 10:20:27 hrw will look Oct 29 10:20:46 jovox, np, we all started to learn things the hard way Oct 29 10:30:09 03Sebastian Spaeth  07shr/merge * r777c741b82 10openembedded.git/recipes/eject/eject_2.1.5.bb: eject: temporarily use angstrom as SRC_URI as geocities.com whent down. Not for upstreaming. Oct 29 10:30:15 hrw, looked at the patch, it is indeed merged already, not fully sureo on how to actually use it Oct 29 10:30:48 03Marcin Juszkiewicz  07shr/merge * r2278afa81b 10openembedded.git/ (conf/checksums.ini recipes/tzdata/tzdata_2009p.bb): tzdata: added 2009p Oct 29 10:31:11 eFfeM: create conf/checksums.ini in builddir? Oct 29 10:35:45 hrw with builddir do you mean tmp (there is one autogerenated there) or do you mean another dir? Oct 29 10:35:56 also not sure what to do to tell it to merge Oct 29 10:38:30 eFfeM: place you keep conf/local.conf Oct 29 10:38:56 hrw, ok Oct 29 10:39:06 and then it is automatically merged? Oct 29 10:39:12 looks like Oct 29 10:39:35 jovox had a quick look at it apparently the / between ${D} and qmdc is not needed Oct 29 10:41:50 hrw will give it a try next time Oct 29 10:48:25 JaMa: welcome to the team :-D Oct 29 10:50:52 XorA: :) Oct 29 10:51:30 only a week left to sneak in under the old rules :-D Oct 29 10:55:51 hello Oct 29 10:56:32 I am trying to finish building a task that was begun yesterday and interrupted. Oct 29 10:57:33 XorA: new rules for r/w access after oedem? Oct 29 10:58:11 hrw: well I assume the newly voted TSC will either ratify existing rules or make new ones Oct 29 10:58:34 hrw: either way the core team no longer exists so the "rules" have to change in some form Oct 29 10:58:35 It's a svn recipe, even after removing all the associated work/arm-angstrom ..../package-date dirs and doing bitbake -c clean package && bitbake -c rebuild packages, it starts at do_configure, but fails because the source code is not present. Oct 29 10:59:09 XorA: I seriously wan't sure whether you were being sarcastic with you avahi mail the other day :-) Oct 29 10:59:27 spaetz: I think I misread your email Oct 29 10:59:36 spaetz: and was having a bad day, sorry Oct 29 10:59:44 XorA: ok, np ;) Oct 29 10:59:55 spaetz: I read is as avahi stops my sound working Oct 29 11:00:01 can happen (does happen) to everyone Oct 29 11:00:05 spaetz: which I suspect is now what it said Oct 29 11:00:12 not Oct 29 11:00:15 bloody keyboard Oct 29 11:00:25 how do I make bitbake rebuild this package from the VERY beginnning (so that it really does unpack the source) ? Oct 29 11:00:33 hehe, nope. That would have been a surprising effect indeed (avahi stopping sound working) Oct 29 11:01:02 hrw: What did you change in your blog? Oct 29 11:01:33 florian: nothing on last days, theme before Oct 29 11:02:37 hrw: The error was gone yesterday at 12:00 Oct 29 11:03:33 no idea then Oct 29 11:03:56 Longfield: bitbake -c rebuild Oct 29 11:04:59 XorA: with bitbake -c rebuild it directly goes to do_configure with an empty src directory (and then of course fails) Oct 29 11:05:14 Longfield: try -c clean then Oct 29 11:05:25 Longfield: if that works moan at whoever broke -c rebuild Oct 29 11:05:54 bitbake -c clean && bitbake -c rebuild does not do the trick (it's a svn based recipe) Oct 29 11:06:52 your sure your SRC_URI and SRCREV are correct? Oct 29 11:07:01 rev 1 tends to be empty in svn Oct 29 11:07:19 XorA: that's why I'm asking, -c clean && -c rebuild should remove everything Oct 29 11:12:04 XorA: SRC_URI is ok, it usually works very well. but I don't define SRCREV (I have PV = "svn${SRCDATE}" that I have copied from another svn recipe) Oct 29 11:13:43 XorA: I am going to try with PV = "svnr${SRCREV}" Oct 29 11:16:05 Longfield: if you dont do SRCREV= though, it will default to rev 1 Oct 29 11:17:07 Hi all Oct 29 11:17:23 I have problems building dbus in oe.dev Oct 29 11:17:54 The error is that the linking phase of do_compile searches in wrong place for libICE. Oct 29 11:18:18 http://tinderbox.openembedded.net/packages/214613/ Oct 29 11:18:53 The report is for 1.2.1, but the same occurs now for 1.2.3 Oct 29 11:20:30 What I figured out is that I have set TMPDIR = "${TOPDIR}/tmp${DISTRO_PR}", but log suggests the linker looks for in just tmp, without the suffix Oct 29 11:21:41 Creating a soft link tmp -> tmp${DISTRO_PR} helps Oct 29 11:22:05 but where the problem could be, in dbus or libice rather? Oct 29 11:31:02 XorA: there was a problem in the svn directory, it was empty. Removing the svn directory in downloads made the trick Oct 29 11:31:10 XorA: but thanks for the help Oct 29 11:39:04 ok, new checksums.ini format sent to ML Oct 29 11:39:46 * XorA wonders who at OEDEM he can fob off some hardware on Oct 29 11:40:25 XorA: pb_ if nobody else, just leave stuff in the venue :) Oct 29 11:41:30 doh Oct 29 11:42:28 zecke: good afternoon Oct 29 11:42:35 XorA: zauruses again? Oct 29 11:43:18 pb__: can you read my recent mail about new checksums.ini? I am interested in your opinion Oct 29 11:43:35 hrw: yeah, they re just rotting away here Oct 29 11:44:10 XorA: ok, have to remember to avoid you as much as possible - otherwise I can end with pockets full of zauruses ;D Oct 29 11:44:42 hrw: some of the stuff already came from you :-D Oct 29 11:45:03 XorA: small amount of them - just 2 wifi cards Oct 29 11:45:34 * XorA wonders why thesing didnt push his 2.6.31 kernel yet Oct 29 11:46:21 hi rkirti|afk Oct 29 11:49:27 hrw: I dont see where url?= is ever used Oct 29 11:51:14 if I have a tar.gz in my SRC_URI, do I need to have a "install" line for each file in that tar.gz or can I use some kind of wildcard? Oct 29 11:57:19 I figured it out.. Oct 29 11:58:11 XorA: nowhere in OE, can be used by scripts for fetching sources for mirroring Oct 29 12:01:06 hrw: extend it by a new SRC_URI chk://archive.tag.gz; Oct 29 12:01:28 hrw: then we can finally have multi sourced verified fetches Oct 29 12:16:39 To narrow down my problems with dbus and wrong "tmp" dir (http://tinderbox.openembedded.net/packages/214613/), I tracked it down to the exact line that causes the problem; it has something to do with libSM.la file Oct 29 12:17:22 it contains wrong dependency_libs line Oct 29 12:17:40 also, the file date is the same as my older build from August! Oct 29 12:19:58 Am I suspecting right, that DEPLOY_DIR_PSTAGE is used as a single cache for different TMPDIR-s, probably causing OE to misconfigure things? Oct 29 12:22:47 jest: by default D_DIR_PSTAGE is inside of TMPDIR... Oct 29 12:23:40 well, not in KaeilOS defaults... So I assume I should never mix them for e.g. different OE branches and distros? Oct 29 12:25:33 I do not follow KaeilOS too much Oct 29 12:26:51 eFfeM: you updated linux-kirkwood to 2.6.32-rc4 right? Oct 29 12:37:39 hrw, yes Oct 29 12:37:49 is there an issue with it ? Oct 29 12:37:55 jest: never mix anything for different distros Oct 29 12:39:35 and DEPLOY_DIR_PSTAGE is very tied to the tmp/ is came from Oct 29 12:44:13 XorA: this are not different distros Oct 29 12:44:18 eFfeM: works stable on sheeva? Oct 29 12:44:42 e.g. different OE branches and distros? Oct 29 12:44:47 the problem is more elaborate: packages in pstage have embedded tmp paths they were build from Oct 29 12:44:58 jest: yes that is correct Oct 29 12:45:02 jest: dont mix and match Oct 29 12:45:16 so any renames/ moves/ shares make things... go awry :) Oct 29 12:45:26 sure, now I know Oct 29 12:45:36 Am I missing something, why would you ever mix and match package stage stuff. Oct 29 12:45:43 03Koen Kooi  07org.openembedded.dev * reac32cefd9 10openembedded.git/recipes/asio/asio_1.1.1.bb: asio: add boost to DEPENDS Oct 29 12:45:46 I'll mail Koan to correct instructions in the web Oct 29 12:45:54 03Koen Kooi  07org.openembedded.dev * r1314ecdc7f 10openembedded.git/recipes/gnome/goffice_0.7.14.bb: goffice: fix packaging Oct 29 12:46:01 hrw, it is not doing much at the moment, but I have no problems either: Oct 29 12:46:03 oot@sheevaplug:~# uptime Oct 29 12:46:03 12:45:17 up 1 day, 6:44, 2 users, load average: 0.00, 0.01, 0.00 Oct 29 12:46:03 root@sheevaplug:~# cat /proc/version Oct 29 12:46:03 Linux version 2.6.32-rc4 (frans@linux-suse) (gcc version 4.3.3 (GCC) ) #1 PREEMPT Sat Oct 24 21:38:26 CEST 2009 Oct 29 12:46:10 jest: tmp location is a fixed point, either symlink it or restart clean if you need to move it. Oct 29 12:46:11 * XorA would recomand Angstrom over Kaelios :-D Oct 29 12:46:45 I had a lot of compiling problems with Angstrom+OE.dev; Kaelios just went Oct 29 12:46:59 jest: really? Oct 29 12:47:03 hrw if you want to discuss more about sheeva & kirkwood, page me after 7 pm or tomorrow, need to rush to work now Oct 29 12:47:28 or leave a msg or im here Oct 29 12:48:00 eFfeM-away: ok Oct 29 12:48:28 DJWillis: the problem was that Kaelios by default proposes to set pstage as a fixed dir. As I started with it without any experience in OE, I wasn't aware of potential problems until they happend Oct 29 12:49:42 As I have to clean everything now, I'll try with Angstrom+OE.dev once again Oct 29 12:51:25 I just built images for Angstrom for 3 platforms :-D Oct 29 12:51:32 so it should work Oct 29 12:53:59 with OE dev branch? Oct 29 12:54:50 XorA: just 3? Oct 29 12:55:00 hrw: 4 is now on its way Oct 29 12:55:08 hrw: zaurus stuff wont build Oct 29 12:55:42 * XorA only actually has 3 devices on/under his desk these days Oct 29 12:55:59 10 here Oct 29 12:56:47 10 devices, 7 types of them, 3 archs, 6 subarchs Oct 29 12:58:08 thats *on* desk Oct 29 12:58:53 *as* desk at that rate Oct 29 12:59:34 jest: what's the matter with KaeilOS ? Oct 29 13:00:20 broonie: I have big desk Oct 29 13:08:21 mckoan: I followed instructions on the webpage, including setting DEPLOY_DIR_PSTAGE to a fixed location. The problem occurred to me, when I tried to compile KaeilOS with a different OE branch. The branch was ABI-incompatible, so I had to set up another TMPDIR. Then compilation problems occurred, as pstage is "shared" between two different TMPDIR's. Oct 29 13:08:28 hi florian! Oct 29 13:08:45 good morning #oe Oct 29 13:10:21 mckoan: Because files in pstage have absolute paths to TMPDIR's files from which they were built, I consider a pstage to be TMPDIR-dependent. Hence, IMO it would be much safer to set DEPLOY_DIR_PSTAGE with TMPDIR reference Oct 29 13:10:40 e.g. DEPLOY_DIR_PSTAGE="${TMPDIR}/pstage" Oct 29 13:12:04 jest: thanks for your hint, I'll check it :-) Oct 29 13:12:54 03Koen Kooi  07org.openembedded.dev * rd6f6d5222f 10openembedded.git/ (3 files in 2 dirs): abiword: add 2.8.0 Oct 29 13:13:00 mckoan: just a thought; I'm a beginner, but the problem I encountered was very misleading and hard to locate Oct 29 13:13:44 jest: a different point of view is a richness in each thing, thank you Oct 29 13:14:20 03Koen Kooi  07org.openembedded.dev * r9fb659d894 10openembedded.git/recipes/angstrom/angstrom-uboot-scripts/ (dsplink.cmd touchbook-vrfb.cmd touchbook.cmd): angstrom-uboot-scripts: add more scripts Oct 29 13:14:20 03Koen Kooi  07org.openembedded.dev * rc5e3b9033e 10openembedded.git/recipes/tasks/task-beagleboard-demo.bb: task-beagleboard-demo: switch to nm-applet and abiword-meta Oct 29 13:14:21 03Koen Kooi  07org.openembedded.dev * r3c816a2d62 10openembedded.git/recipes/angstrom/angstrom-gpe-task-settings.bb: angstrom-gpe-task-settings: remove comman-applet Oct 29 13:15:11 rkirti: how goes your maemo-oe work? Oct 29 13:18:29 jest: BTW when you chanche branch are you creating a new tmp too ? Oct 29 13:18:38 s/chanche/change Oct 29 13:19:41 mckoan: yes, bitbake requires this, as ABI changes from 2 (OE stable/2009) to 3 (OE dev) Oct 29 13:20:11 JaMa: shr/merge is mergable with .dev? I want to test newest X.org on one device Oct 29 13:20:20 hrw: pretty much stalled :( I have been busy with my univ. course work here. Not much change since September. Oct 29 13:20:28 hrw: I'll send you patch for xorg.. Oct 29 13:22:21 jest: fine, that's the correct behaviour. Anyway I would change pstage in the meantime although I understand this may not be intuitive. Oct 29 13:22:49 hrw: sent in 1 patch.. Oct 29 13:24:04 JaMa: thx Oct 29 13:24:07 mckoan: Maybe a note in KaeilOS' local.conf like "If you change , also DO change " Oct 29 13:24:55 hrw: tested on freerunner armv4t and spitz armv5 Oct 29 13:25:46 hrw: there is maybe still problem while redrawing bellow cursor.. (we had segfaults while typing on illume keyboard fast), workaround is to disable cursor.. (used in shr/merge Xserver) Oct 29 13:25:51 hi Oct 29 13:26:36 jest: sure. BTW I'm evaluating to 'embed' pstage into tmp Oct 29 13:27:39 JaMa: I want to test on bug Oct 29 13:28:08 jo raster Oct 29 13:28:38 hrw do you already know when you will arrive on friday in berlin? Oct 29 13:29:15 hrw: If you have keylaunch in image it won't probably compile and you need another patch for xf86-input-tslib if you're not using evdev Oct 29 13:30:10 JaMa: there is tslib 0.0.6 now Oct 29 13:30:22 woglinde: ~14 on sxf I think Oct 29 13:30:32 hm directly? Oct 29 13:30:35 yes Oct 29 13:30:37 woglinde: boooyah! ho ho Oct 29 13:30:41 not hauptbahnhof? Oct 29 13:30:53 maybee rober and I could you hook up Oct 29 13:30:55 robert Oct 29 13:31:33 woglinde: I arrive by airport bus Oct 29 13:32:27 I get an error message when I build my recipe. "install: omitting directory `/home/jovox/oe/tmp/work/i486-angstrom-linux/device-controller-web-1.0.0-r0/www/scripts'" I pasted my recipe here: http://pastebin.com/d573f9e3 The tar.gz has some file in the root and two sub directories, images and scripts with both contains files which I want to be installed. Is there anything I need to do to make it install then recursively? Oct 29 13:32:34 hrw hm oh strange Oct 29 13:32:48 woglinde: no, thats my standard way to go to berlin Oct 29 13:32:50 hrw: yes.. there is also patch for tslib_0.0.6 in shr patchwork http://patchwork.dev.bearstech.com/patch/347/ Oct 29 13:33:17 hrw didnt know there is a bus diretcly from poland Oct 29 13:33:42 I think I will be 2 p.m. there too Oct 29 13:33:47 woglinde: there are few companies which transport people from Szczecin to Berlin and back Oct 29 13:34:28 I though train is best option Oct 29 13:34:32 thought Oct 29 13:34:54 woglinde: next time I will try train Oct 29 13:35:07 woglinde: bus takes me from home Oct 29 13:37:15 does someone here tried X.org with more then one fbdev? Oct 29 13:44:35 svn: Unknown hostname 'svn.projects.openmoko.org' while parsing /home/hrw/devel/oe/openembedded/recipes/openmoko-3rdparty/om-neon_svn.bb Oct 29 13:44:43 JaMa: thats shr/merge branch Oct 29 13:45:57 hrw: hmm hostname 'svn.projects.openmoko.org' is known to my dns.. strange Oct 29 13:47:09 JaMa: NOTE: multiple providers are available for virtual/libgl (mesa, mesa-dri, mesa-xlib); Oct 29 13:47:22 JaMa: just notifing Oct 29 13:47:31 hrw hm Oct 29 13:47:43 hm libgl should be mesa Oct 29 13:48:06 not mesa-dri Oct 29 13:48:14 mesa should pull in the right dri lib Oct 29 13:48:29 woglinde: why not? we're using mesa-dri as virtual/libgl for om-gta02 now.. Oct 29 13:48:48 whats the difference btw. mesa and mesa-dri? Oct 29 13:48:54 and mesa-xlib Oct 29 13:50:01 mesa_7.4 just requires mesa-dri_7.4 Oct 29 13:50:34 *g* Oct 29 13:50:46 whats with mesa 7.6= Oct 29 13:50:47 ? Oct 29 13:51:09 the same for mesa_7.2 but mesa_7.6 is only in mesa-dri_git.bb Oct 29 13:52:04 hm for me mesa is one building block which should split up packages into libGL libGLw dri and so on Oct 29 13:53:31 JaMa: xtrans 1.2.5: | configure: WARNING: unrecognized options: --enable-malloc0returnsnull Oct 29 13:55:54 hrw: hmm its appended to all xorg-lib recipes.. can I set EXTRA_OECONF in that recipe? Oct 29 13:56:26 hrw: if someone will add some new options to EXTRA_OECONF in xorg-lib-common.inc then it would be ingored in xtrans.. Oct 29 13:56:41 where can I read about the bitbake install cmd? Oct 29 13:56:51 XorA, http://www.phdcomics.com/comics.php?n=1244 Oct 29 13:56:55 bitbake doesn't have an install cmd Oct 29 13:56:58 will I have this problem? Oct 29 13:57:27 * kergoth yawns Oct 29 13:57:34 kergoth, function? Oct 29 13:57:40 ? Oct 29 13:58:15 are you talking about the do_install *task*, or the "install" that's often run in install tasks, i.e. install foo ${D}${bindir}/ Oct 29 13:58:15 hi kergoth Oct 29 13:58:25 I'm talking about "install -m 0644..." Oct 29 13:58:25 if the former, it's OE, not bitbake. if the latter, its coreutils, not bitbake. Oct 29 13:58:30 ok Oct 29 13:58:31 so coreutils Oct 29 13:58:38 is it really that hard to "man install"? Oct 29 13:58:39 :P Oct 29 13:59:08 *g* Oct 29 13:59:16 not all knows dpkg -S Oct 29 13:59:34 dpkg -S is awesome. so is apt-file Oct 29 13:59:48 or autoapt Oct 29 13:59:51 *g* Oct 29 14:00:23 indeed. auto-apt's configure wrapping is cute too, ever used that? it monitors what files are accessed by the configure script and prompts to install them Oct 29 14:00:25 hehe Oct 29 14:02:03 hrw: Hi ! Oct 29 14:02:37 hrw: I added 2.6.30 to OE and I am surprised to see that both 2.6.28 and 2.6.30 are in rootfs Oct 29 14:03:07 noglitch: 2.6.28 is marked as default Oct 29 14:03:20 noglitch run bitbake -g your_target and look where it sneaks in Oct 29 14:03:22 ah Oct 29 14:03:23 noglitch: some modules were in 2.6.28 and not in 2.6.30 and got fetched Oct 29 14:03:37 in my 2.6.30.bb there is DEFAULT_PREFERENCE_at91sam9263ek = "30" Oct 29 14:03:51 hrw: ah ? ok... Oct 29 14:04:12 JaMa: bump util-macros to 1.3.0 in next ver of patch Oct 29 14:04:26 JaMa: and add checksums;D Oct 29 14:05:32 I don't get this.. I've tried all kind of stuff now but still doesnt manage to make my recipe to install the entire content of my tar.gz Oct 29 14:05:46 hrw: the solution to remove them is to modify the distribution conf files ? Oct 29 14:06:15 noglitch: pastebin usr/lib/opkg/status please Oct 29 14:06:43 jovox hm? Oct 29 14:06:58 jovox what building system your source code is using? Oct 29 14:09:41 Crofton|work: :-D Oct 29 14:09:54 noglitch: linux_2.6.30.bb:DEFAULT_PREFERENCE_at91sam9263ek = "-1" Oct 29 14:10:02 noglitch: linux_2.6.28.bb:DEFAULT_PREFERENCE_at91sam9263ek = "28" Oct 29 14:10:15 my opkg/status http://paste.debian.net/50266/ Oct 29 14:10:18 noglitch: which means: I do not want 2.6.30 by default but rather 2.6.28 Oct 29 14:10:38 hrw: hmm its there.. just missing checksums maybe.. Oct 29 14:12:15 hrw: ? but I want 2.6.30 and I guess as "30" is greater than "28" the 2.6.30 is chosen... Oct 29 14:12:47 hrw: It seems that 2.6.28 is used to give bluetouth & ext3 modules... not realy useful... Oct 29 14:13:02 noglitch: do you have 2.6.30 version of kernel-module-gadgetfs for example? Oct 29 14:13:16 noglitch: I use BT dongle in at91sam usb port Oct 29 14:13:38 hrw: ahh maybe more checksums are missing.. I just found why sometimes my script for checksums adding doesn't work :) Oct 29 14:13:48 noglitch: and some of my usb dongles are ext3 Oct 29 14:14:33 JaMa: easier is: disable checksums, bitbake x11-image -cfetchall, cat tmp/checksums.ini >>oe/conf/checksums.ini, sort, commit Oct 29 14:16:11 noglitch: your tmp/deploy contains modules for 2.6.28 and 2.6.30 kernels. OE task-base/boot recommends some kernel modules and is fine when they are not available. in your situation you have some modules for 2.6.28 but not for 2.6.30 so thats why 2.6.28 kernel got into rootfs Oct 29 14:18:49 JaMa: also bump libx11 in angstrom (as separate commit) Oct 29 14:19:03 woglinde: I got a tar.gz which contains some files in the "root" and two directories (also containing files). All I want todo is to copy the files and directories into my package. I'm not compiling anything so it's just about copying files. This is my log and bb file http://pastebin.com/m2fe84372 Oct 29 14:19:30 hrw: ok, I understand that. Thanks for your explanation. So, if I do not want those modules, I can just remove the tmp/deploy 2.6.28 stuff. If I want them, I have to modify my defconfig... I understand that you vote for an update of my defconfig ;-) Oct 29 14:20:25 noglitch: first remove all 2.6.28 packages Oct 29 14:22:01 kergoth: heh, you're up early Oct 29 14:22:26 oh, wait, you're still on DST over there, aren't you Oct 29 14:23:16 7:20am here, same as pacific at this time of year due to the lack of daylight savings Oct 29 14:23:29 * kergoth is usually at work by 7:45 nowadays, but is working from home today Oct 29 14:24:00 JaMa: NOTE: Task failed: Error: /home/hrw/devel/oe/openembedded/recipes/xorg-lib/libx11/./x11_disable_makekeys.1.6.3.patch not found. Oct 29 14:24:52 ah, yeah, no DST in arizona. I always forget that. Oct 29 14:25:41 this week is unusual because daylight savings is over in europe but still active in (most of) the usa, so the time difference is an hour less than normal Oct 29 14:25:57 hrw: what about removing preferred versions of xorg stuff from angstrom-2008-preferred-versions.inc? as they are set from that xorg include? I switch them to weak assignments here.. to make it compiled.. but then they are not usefull at all Oct 29 14:26:37 JaMa: angstrom do not use any of preferred-x11 files Oct 29 14:27:17 03Koen Kooi  07org.openembedded.dev * r966dfe4909 10openembedded.git/ (11 files in 9 dirs): Oct 29 14:27:17 omap3517-evm: omap3517 was the pre-production name, AM3517 is the production name, rename machine to avoid confusion Oct 29 14:27:17 Signed-off-by: Koen Kooi Oct 29 14:27:34 hrw: yes it does Oct 29 14:27:50 pb__: ah, interesting, i didn't realize they switched at different times Oct 29 14:27:55 orly? Oct 29 14:28:15 hrw: I added them when I was paid to by openmoko Oct 29 14:28:43 XorA: and later angstrom stopped using them Oct 29 14:29:07 hrw: sent by mail Oct 29 14:29:09 15:31 hrw@home:distro$ grep preferred-xorg * -r|grep angstrom Oct 29 14:29:09 include/angstrom-2007-for-openmoko.inc:require conf/distro/include/preferred-xorg-versions.inc Oct 29 14:30:08 hrw: well that was dumb of us Oct 29 14:30:42 jovox when I remeber correctly install dont works well with wildcard Oct 29 14:30:55 JaMa: "for recipe in oe/recipes/xorg*/*.bb;do bitbake -cfetch -b $recipe;done" is other way to generate checksums Oct 29 14:32:18 hrw: thats what I do.. but problem was, tak I had already fetched few packages in workdir.. Oct 29 14:33:09 hrw: usually i use recipe list from git add and then fetch all recipes added by last git add Oct 29 14:33:36 many methods exists ;D Oct 29 14:33:42 hrw: so I need -c clean before or -f :) Oct 29 14:34:32 JaMa: -f should be fine Oct 29 14:46:16 hrw: the standard way of giving a linux defconfig is to put it in a machine directory ? What about using the kernel defconfig file ? Oct 29 14:46:29 noglitch: we do not use kernel configs Oct 29 14:46:55 noglitch: users can easily adapt defconfigs when they are in metadata Oct 29 14:47:15 ok, I see Oct 29 14:47:41 or even provide own ones etc Oct 29 15:09:41 JaMa: I will push all new checksums into .dev Oct 29 15:10:43 hrw: ok Oct 29 15:10:52 hrw: you missed the most important reason, 99% of the in kernel defconfigs are broken Oct 29 15:11:58 XorA: I ignore that Oct 29 15:13:49 1 files changed, 328 insertions(+), 0 deletions(-) Oct 29 15:13:54 lot of x11 stuff Oct 29 15:16:01 03Marcin Juszkiewicz  07org.openembedded.dev * r16ecf3fa60 10openembedded.git/conf/checksums.ini: checksums.ini: added X11 r7.5 entries Oct 29 15:16:25 JaMa: I did not merged your checksums patch Oct 29 15:19:01 how can I prevent udev loading pccardd? Oct 29 15:22:39 woglinde, I thought so too.. Did you see my paste? Oct 29 15:27:33 hi Oct 29 15:29:42 pb_, i've came up with this solution to honour DEFAULT_PREFERENCEs when choosing between multiple providers. What do you think? http://uclibc.org/~aldot/bitbake/0001-sort-eligible-providers-by-DEFAULT_PREFERENCE.patch Oct 29 15:33:34 pb_, with this i can have coreutils-native at preference 0 and set DEFAULT_PREFERENCE=-1 for busybox-native and anybody who doesn't explicitly set PREFERRED_PROVIDER_virtual/coreutils-native = "busybox-native" will end up choosing coreutils-native Oct 29 15:33:55 khem: several of us are having a discussion about mono on beagle -- its no longer running on arm platforms. Does not look like any of the recent changes could have broke anything, so might be a gcc or lib incompatibility. Oct 29 15:34:09 (on #beagle) Oct 29 15:36:36 blindvt: I am a bit surprised that this patch is necessary. You might check with überhackers zecke or rp, they understand that bitbake code better than I do. Oct 29 15:41:45 x11 question: how to list input devices from x11 session? Oct 29 15:42:14 pb_, it doesn't take the _provider_'s DP into account (AFAICS), just the DP of each version of the possible providers of a package Oct 29 15:42:17 as I would like to know how many keyboards/touchscreens does x11 see Oct 29 15:42:44 pb_, now it does correctly (the NOTE is just to see if that triggers anywhere else): http://paste.debian.net/50272/ Oct 29 15:43:18 denix: I have some pb on dm365-evm with arago. U-Boot 1.3.4 (Mar 2 2009 - 11:50:39), uImage-dm365-evm-r26.bin (md5 bb962a3be08448c4d3ba495f414bed0c uImage-dm365-evm-r26.bin), arago-base-image from your site. echo "COMPONENT" > /sys/class/davinci_display/ch0/output and echo "720P-60" > /sys/class/davinci_display/ch0/mode, cat /dev/urandom > /dev/fb[012] .. nothing appears Oct 29 15:44:58 RP, zecke, any comment on the above? Oct 29 15:48:05 JaMa: libxaw 1.0.7 does not builds Oct 29 15:48:44 hrw: because of docs? Oct 29 15:48:53 yes Oct 29 15:48:54 hrw: something about troff/groff? Oct 29 15:48:57 yes Oct 29 15:49:07 | troff: fatal error: can't find macro file s Oct 29 15:49:18 hrw: there is missing groff-native package.. it builds ok if groff is installed on buildhost.. Oct 29 15:49:35 or some kind of --disable-docs? Oct 29 15:49:37 hrw: working on it just now.. (I had to remove groff from my buildhost to test it..) Oct 29 15:50:02 hrw: the same problem will be later in libx11 so I'll create groff-native Oct 29 15:50:24 hrw: not sure if there is --disable-docs for libxaw Oct 29 15:51:16 JaMa: --enable-docs Enable building of Xaw documentation Oct 29 15:57:15 JaMa: EXTRA_OECONF += " --disable-docs " Oct 29 15:57:19 JaMa: that helped Oct 29 15:58:25 hrw: groff-native_1.20.1 seems working too Oct 29 15:59:40 uf.. newer tslib driver works better Oct 29 16:01:44 blindvt`: default preference is the weakest form of preference so it seems strange to apply that filter later like that? Oct 29 16:02:21 03Marcin Juszkiewicz  07org.openembedded.dev * ra1c155c79f 10openembedded.git/recipes/xorg-driver/xf86-input-tslib_0.0.6.bb: xf86-input-tslib: added 0.0.6 Oct 29 16:03:39 blindvt`: I understand the problem - its that bitbake does give preference to a provider with the same name. Oct 29 16:03:58 Its a question of whether default preference sorts higher or lower than "same name" Oct 29 16:09:42 RP, not sure if i understand what you say.. Did i make my problem clear? Oct 29 16:10:56 blindvt`: yes, you did Oct 29 16:12:07 blindvt`: The trouble is that the default provider for "busybox-native" should be "busybox-native" since they have the same name. Should the DEFAULT_PREFERENCE have a higher or lower influence than the same name Oct 29 16:12:09 ? Oct 29 16:13:01 DEFAULT_PREFERENCE is defined as the weakest preference possible Oct 29 16:13:33 So the current behaviour is "same name", then sort by preferences Oct 29 16:13:40 03Thomas Zimmermann  07shr/merge * r807f703c2e 10openembedded.git/recipes/vagalume/ (files/index.theme vagalume_0.7.1.bb): Oct 29 16:13:40 shr/merge: vagalume: add index.theme again Oct 29 16:13:40 Signed-off-by: Klaus Kurzmann Oct 29 16:13:41 03Thomas Zimmermann  07shr/merge * r31d6debdb5 10openembedded.git/recipes/connman/ (connman_git.bb files/shr/connman): Oct 29 16:13:45 shr/merge: connman: do not disable usb0 on startup Oct 29 16:13:47 Signed-off-by: Klaus Kurzmann Oct 29 16:13:49 03Thomas Zimmermann  07shr/merge * r19aadbbae9 10openembedded.git/ (conf/distro/shr.conf recipes/tasks/task-shr-minimal.bb): Oct 29 16:13:52 shr/merge: make LIBC configureable Oct 29 16:13:54 Signed-off-by: Klaus Kurzmann Oct 29 16:13:56 03Martin Jansa  07shr/merge * r5b38a7a2e0 10openembedded.git/ (3 files in 2 dirs): Oct 29 16:13:59 shr/merge: Merge xf86-video-glamo-kms to xf86-video-glamo as kms branch is now merged to master. Oct 29 16:14:01 Signed-off-by: Klaus Kurzmann Oct 29 16:14:05 03Klaus Kurzmann  07shr/merge * rb90b9e2cfe 10openembedded.git/: Merge branch 'shr/merge' of ssh+git://git@git.openembedded.net/openembedded into shr/merge Oct 29 16:22:39 RP, there's imho one missing bit in the chain: if a virtual package has two providers, then the preference is applied to each of the providers to pick the suitable version Oct 29 16:23:15 RP, this is fine, since we see that virtual/coreutils is provided by coreutils-1.2.3 and busybox-3.2.1 (let's say). Oct 29 16:24:40 RP, but then there is no attempt made to see if any of that two providers have a higher preference, it just picks the first one in the list and suggests to set a PREFERRED_PROVIDER. Oct 29 16:25:30 RP, and not attempting to look at each preference is the missing bit, AFAIU Oct 29 16:26:21 hi blinder Oct 29 16:27:28 RP, a preferred provider would of course override that, but if there is no preferred provider set then it should pick according to the preference and not according to alphabetical sorting of the names (or some such) Oct 29 16:27:32 woglinde, heya Oct 29 16:28:44 RP, i think that the actual name is of no concern there, just the default preferences of eligible would pick the correct thing, agree? Oct 29 16:29:28 blindvt`: As far as I can work out your patch would override any set PREFERRED_PROVIDER? Oct 29 16:31:23 bye Oct 29 16:32:50 RP, well, i'm not entirely happy with that patch although it works as advertised. Not sure if that's ok, but if a PREFERRED_PROVIDER is set (to busybox which has DP -1) then it first NOTE: eligible changed: to coreutils, busybox (since that's what's hinted by DP) but instantaneously "DEBUG: adding busybox-native to satisfy virtual/coreutils-native" which has the desired effect in the end Oct 29 16:33:29 blindvt`: It fixes your use case, I think it will break other things as PREFERRED_PROVIDER_X = "y" will no longer do what you expect Oct 29 16:33:53 That whole set of functions needs ripping up and strting again really :/ Oct 29 16:34:22 Along with someone writing down clearly what the different scenarios are and what the expected behaviour is Oct 29 16:34:38 03Jeremy Lainé  07org.openembedded.dev * r715e7333f3 10openembedded.git/recipes/linux/linux-2.6.31/boc01/ (011-091028-gpio.patch boc01.dts boc01.dts.v1): linux-2.6.31: initialise GPIO states on boc01 Oct 29 16:35:41 RP, so the provided PREFERRED_PROVIDER wins in any case, but if nobody specifies a PREFERRED_PROVIDER then it choosed based on the DEFAULT_PREFERENCE. That's essentially what i'd need and which would not change the previously existing logic Oct 29 16:35:42 RP hehe topic to discuss at oedem? among with RDEPENDS virtual/ stuff? Oct 29 16:36:28 woglinde: Yes, someone should note these on the agenda I guess Oct 29 16:39:41 RP, ultimately i think choosing due to default preference is cleaner than to clutter all other distros with preferred providers or doing ugly stuff in bitbake.conf, but i'm completely unfamiliar with both bitbake and oe ATM, so i can't really judge :) Oct 29 16:40:06 blindvt`: Imagine we have packages A (DP 0) and B (DP1) where b provides A. Now imagine what "bitbake A" does before and after your patch. Oct 29 16:40:55 Currently there is a slight preference given to A because its PN matches the name of the target we want. You are disabling that preference and hence changing the behaviour Oct 29 16:41:57 couldn't we just set the DP of A if it's PN is A and if it doesn't have an explicit DP set, rather than doing extra magic on top of DP? Oct 29 16:41:57 blindvt`: I'm not saying there is not a problem, just that I think there are some subtle effeccts of that path that are not intended Oct 29 16:42:13 then explicitly setting it would override the behavior Oct 29 16:42:15 hm. It would bake B unless you said you'd prefer A to supply B, i fear Oct 29 16:42:32 blindvt`: right Oct 29 16:42:38 kergoth: please no ;-) Oct 29 16:43:06 why not? it would give you the slight preference while giving back the user control over it Oct 29 16:43:32 kergoth: Set the DP of A to what? Oct 29 16:44:26 i'd suggest 0, but the key is just picking a sane value and then document it.. this behavior should really be documented already, but i doubt it is Oct 29 16:44:41 kergoth: 0 is the default for everything Oct 29 16:45:01 kergoth: DP is always set effectively Oct 29 16:45:24 so make it 1. or come up with a better method. Oct 29 16:46:06 Its all just about sorting order and we just need to define which mechanisms have which priority on the sort Oct 29 16:46:27 currently its mostly consistent now but not well documented Oct 29 16:46:50 currently there's no way to override your "slight preference" at all. not exactly flexible Oct 29 16:47:20 kergoth: PREFERRED_PROVIDER? Oct 29 16:47:39 PREFERRED_SORT_ORDER :) Oct 29 16:48:01 that's exactly what blindvt`is trying to avoid. the whole point of DEFAULT_PREFERENCE is to set the defaults for the provider preferences Oct 29 16:48:12 and right now they aren't obeyed, i'd call that a bug Oct 29 16:48:25 * Crofton|work is pulling your leg :) Oct 29 16:49:08 kergoth: The question is whether the preference numbers make sense amongst a set of versions of a given PN or whether they make sense amongst a group of providers Oct 29 16:49:19 I have always assumed the former, not the latter Oct 29 16:50:17 seems pointless to limit it, particularly when the name of the variable doesn't imply that behavior Oct 29 16:50:24 * kergoth shrugs Oct 29 16:51:07 kergoth: I don't disagree as such, I just don't like the fact we keep changing the rules each time we find a corner case - that code is a problem area Oct 29 16:51:30 The "slight preference code" was extremely useful in solving a number of problems we had with it Oct 29 16:52:38 blindvt`: Can you send a summary of the problem and the proposed patch to the OE+bitbake lists please Oct 29 16:53:01 blindvt`: It needs some thought and I'll put that on my todo list Oct 29 16:53:05 hmm Oct 29 16:54:10 At least the order of thefiles on disk no longer influences the provider bitbake chooi Oct 29 16:54:14 schooses Oct 29 16:54:35 I loved that reproducibility problem :/ Oct 29 16:55:11 hehe Oct 29 16:55:22 hello all Oct 29 16:55:29 The mess of our do_stage functions is also causing me major pain at the moment Oct 29 16:55:46 RP, let me try to retain your "slight preference code" by fiddling eligible _before_ it. Oct 29 16:56:18 blindvt`: Since you're using a virtual that will fix your test case but not the general problem Oct 29 16:56:26 blindvt`: :) Oct 29 16:57:06 kergoth: How would you feel about using the results of do_install for staging and dropping do_stage entirely? Oct 29 16:58:01 i like the idea, i was experimenting with that a while back. just need to make sure we do our mangling / have a mechanism to add mangling, i'd think Oct 29 17:09:14 morning all Oct 29 17:09:30 thesing: why you not pushed 2.6.31 for collie? Oct 29 17:10:06 XorA: because my patches are based on .32-pre* Oct 29 17:10:15 I know that they work there. Oct 29 17:10:16 thesing: push that then :-D Oct 29 17:10:54 is it really worth it? .32 should be released soon. Oct 29 17:11:07 thesing: I would like to test sometime :-D Oct 29 17:11:20 isnt .32 still months away> Oct 29 17:11:21 ? Oct 29 17:11:23 yo thesing Oct 29 17:11:33 it seems to be 6 months between release at the moment Oct 29 17:12:01 not sure. Oct 29 17:12:14 But if you find time to test I will push asap. Oct 29 17:12:31 thesing: next week I want to spend a couple of hours on collie Oct 29 17:13:09 I will rebase my patches to .31 and put them in oe. Oct 29 17:13:14 probably this evening. Oct 29 17:13:26 *g* Oct 29 17:13:53 * XorA threatens to send thesing a pile of zaurus stuff Oct 29 17:14:20 what stuff do you have in mind? Oct 29 17:14:40 a collie and a tosa + gps + random BT and wifi cards Oct 29 17:15:03 its all rotting in my flat Oct 29 17:15:06 never used Oct 29 17:15:08 gathering dust Oct 29 17:15:28 that would mean that you completely drop out of zaurus stuff. Oct 29 17:15:42 that would leave me still with my own personal collie Oct 29 17:16:14 and there is a second OE tosa in Edin as well Oct 29 17:16:33 what kind of bt card is it? the DTL-1 one? Oct 29 17:16:48 socket Oct 29 17:17:17 I got that one from hrw too ;) Oct 29 17:17:30 this came from USA somewhere Oct 29 17:17:41 But I'm interested in the gps card. Oct 29 17:17:52 are you at OEDEM? Oct 29 17:18:08 unfortunately not. Oct 29 17:18:15 hm Oct 29 17:18:19 I could take it Oct 29 17:18:29 and give it thesing Oct 29 17:18:38 you live in berlin, don't you? Oct 29 17:18:43 :) Oct 29 17:18:48 woglinde: that would be cool, Ill even post it to phil or get lrg to carry it, I dont think Ill make it Oct 29 17:19:33 XorA: didn't Jay7 contact you for the collie? Oct 29 17:19:55 ah! that is one I forget to do Oct 29 17:19:56 thesing: oh yeah Oct 29 17:20:06 Jay7: sorry man, losing my memory Oct 29 17:20:44 so only one tosa left and some add ons left. Oct 29 17:20:46 RP, could you please sed -i -e s/dependecies/dependencies/ lib/bb/taskdata.py ;# TIA Oct 29 17:20:47 XorA: seems I'm too :) Oct 29 17:21:05 * woglinde gevies xora paper and pen Oct 29 17:21:12 XorA: may be grab your tosa to russia too? for lumag Oct 29 17:21:45 that will be easier to do delivery from one point Oct 29 17:21:55 XorA: that leaves the wifi cards. how much do you have (and which) Oct 29 17:22:44 thesing: erm cant remember offhand, the list is on the wiki somewhere Oct 29 17:23:12 Jay7: fine by me as long as I have some way to get it to you Oct 29 17:23:41 XorA: I'm working on this slowly :) Oct 29 17:23:46 btw Oct 29 17:24:09 Jay7: take a holiday in Scotland Oct 29 17:24:35 http://wiki.openembedded.org/index.php/Project_devices Oct 29 17:24:41 here is list of OE hw Oct 29 17:25:01 XorA: can I ask CE-RH2 too? :) Oct 29 17:25:24 Jay7: got a spitz? Oct 29 17:25:44 XorA: does it working with spitz only? :) Oct 29 17:25:54 afaik it should work with any clamshell Oct 29 17:26:01 XorA: hold on it. It should work on collie too. Oct 29 17:26:06 ;) Oct 29 17:26:11 Jay7: not with c7x0 devices Oct 29 17:26:20 hm.. have no driver? Oct 29 17:26:26 the wiring is different between the two Oct 29 17:26:29 ah.. Oct 29 17:26:33 different rings on the connector Oct 29 17:26:38 well. I have akita anyway :) Oct 29 17:26:51 basically send me emails Oct 29 17:27:01 s/me/the list/ Oct 29 17:27:18 no point in good hardware rotting in Edinburgh Oct 29 17:27:42 anyway time for me to go home Oct 29 17:28:06 bye xora Oct 29 17:28:39 well.. then we will left one tosa for Samuel Ortiz Oct 29 17:42:43 bbl Oct 29 18:30:06 re Oct 29 18:39:02 JaMa: xserver 1.7.1 do not builds with mesa 7.4 Oct 29 18:43:13 03Graeme Gregory  07org.openembedded.dev * r5c1b686ba6 10openembedded.git/site/avr32-common: site/avr32-common : add values for ORbit2 taken from debian avr32 logs Oct 29 18:43:24 03Graeme Gregory  07org.openembedded.dev * rb51c85efe8 10openembedded.git/recipes/proxy-libintl/ (2 files in 2 dirs): Oct 29 18:43:24 proxy-libintl_20080418.bb : add an SONAME to library Oct 29 18:43:24 Found in Ångström build. libintl.so built from this recipe was not getting Oct 29 18:43:24 an soname so the automatic shlibs code was failing making packages that Oct 29 18:43:25 should depend on this not. Ultimately leading to failing images. Oct 29 18:46:27 did i read somewhere that virtual packages aren't working properly in DEPENDS? Oct 29 18:50:02 hrw|gone: have you tried if it works with mesa-dri_git.bb? Oct 29 18:51:29 hrw|gone: and maybe you need to drop lbxutil from DEPENDS in recipes/xorg-xserver/xorg-xserver-common.inc Oct 29 18:58:25 re Oct 29 18:58:31 JaMa: target lacks dri Oct 29 19:00:14 mickey|office: hi Oct 29 19:01:04 JaMa: anyway started mesa-dri build Oct 29 19:55:44 bye Oct 29 20:19:23 Hi all Oct 29 20:20:22 Can I ask for clarification on how to submit patches for OE? I have some corrections to existing recipe, what to do with them? Oct 29 20:20:56 I read different instructions on wiki, bugs and mailing list Oct 29 20:21:44 jest: send your patches to the mailing list Oct 29 20:22:13 but how to generate them? Just "git format-patch"? Oct 29 20:22:20 if you can send them in a format that git-am will accept then that would be ideal, but it isn't vital. Oct 29 20:22:27 yes, git format-patch would be fine. Oct 29 20:25:19 jest: git-format-patch HEAD^n where n is the number of commits you have, then git-send-email 00*.txt Oct 29 20:29:28 you don't need to do HEAD^n Oct 29 20:29:36 git format-patch -n will do Oct 29 20:35:57 does anyone know which package i need to bake to get kernel module device-mapper? Oct 29 20:39:38 cheese, virtual/kernel will give you a kernel and modules but you might need to edit the defconfig file to actually get the module Oct 29 20:40:14 thanks, patch is on the way :) Oct 29 20:42:00 kergoth: good to know :) Oct 29 20:48:04 Hmm... I wonder what I missed here: http://labs.trolltech.com/blogs/2009/09/10/cross-compiling-qtx11/ Oct 29 20:55:12 hi, i'm a little bit puzzled, I need a file to be put in staging in a recipe, but it is not; is adding a do_stage with a autotools_stage_all line the solution ? Oct 29 21:01:23 :q Oct 29 21:01:27 florian: ping Oct 29 21:04:11 Jay7: pong Oct 29 21:04:35 florian: are you admin of LTG? Oct 29 21:04:47 Jay7: yes Oct 29 21:04:57 florian: I see no wiki there.. Oct 29 21:04:59 is it ok? Oct 29 21:05:44 we have wiki for kexecboot before Oct 29 21:06:46 and I can't find place in interface where I can re-enable it Oct 29 21:07:06 Jay7: that would be a bug... you used the GForge wiki? Oct 29 21:07:59 ehm.. I've used wiki for kexecboot project on LTG before.. I don't know was it GForge or something else :) Oct 29 21:08:13 but anyway I'll prefer mediawiki engine Oct 29 21:08:32 I see "New wiki plugins (using MediaWiki or phpWiki)" on main page ;) Oct 29 21:09:06 btw, homepage link is broken too.. Oct 29 21:09:24 http://kexecboot.home.linuxtogo.org/ is not found Oct 29 21:11:03 oh well... sounds like I have to work on this again Oct 29 21:12:56 florian: this is not a high priority task :) Oct 29 21:14:16 Jay7: well... I would prefer to have things running again. Oct 29 21:14:32 :) Oct 29 21:14:47 Time is very limited unluckily Oct 29 21:14:59 weekend is coming :) Oct 29 21:15:36 Jay7: not really Oct 29 21:18:07 hum... that's a dns issue. Oct 29 21:18:08 its thursday Oct 29 21:18:17 thats like the weekend already Oct 29 21:27:40 anyone know which recipe provides the non-busybox dd ? Oct 29 21:27:51 i could grep but i imagine dd would come up a lot Oct 29 21:28:14 util-linux I think Oct 29 21:28:21 but thats from memory Oct 29 21:28:31 thx XorA|gone i'll look Oct 29 21:28:53 XorA|gone: I have the recipe for .31 ready. Oct 29 21:29:10 thesing: awesome Oct 29 21:29:20 but I didn't test yet. Oct 29 21:29:21 coreutils: /bin/dd Oct 29 21:29:49 I'm waiting for my arm eabi for collie build to finish. Oct 29 21:30:31 thesing: ah, bumped your gcc and stuff to 4.4.X Oct 29 21:30:43 its not that easy. Oct 29 21:30:49 woglinde: got it, thanks Oct 29 21:30:53 you need to change a couple of other places. Oct 29 21:31:29 but I think I found all for micro (well woglinde did) and angstrom. Oct 29 21:32:16 well lets see it it runs on the machine.. Oct 29 21:32:20 hehe Oct 29 21:32:24 hi, the mythtv recipe puts a lot of files in install and image dirs (in usr/include/mythtv); what is the best way to get these in staging too Oct 29 21:32:33 thesing I had some aligments error Oct 29 21:32:39 but console stuff worked Oct 29 21:32:43 eFfeM: add it to do_stage Oct 29 21:33:09 woglinde: wie gehts Oct 29 21:33:24 khem, i know but how ? Oct 29 21:33:30 khem sleepy Oct 29 21:33:49 don't know the env var to reference and not sure if I may do cp -r in do_stage Oct 29 21:33:54 eFfeM: uh, look at the *thousands* of other recipes that use do_stage? Oct 29 21:34:57 woglinde: hmmm es ist 11:30 uhr Oct 29 21:34:58 kergoth: they seem to copy from $(S) which I am not too sure is the proper way Oct 29 21:35:12 or they do not mention a path at all Oct 29 21:35:15 of course it is. where else are you going to get it, /tmp? Oct 29 21:35:26 don't want to do plenty of trial and error runs Oct 29 21:35:38 kergoth: expected them to come from image or install dirs Oct 29 21:35:40 so you'd rather make your laziness our problem Oct 29 21:35:46 03Graeme Gregory  07org.openembedded.dev * rf410685d5f 10openembedded.git/site/avr32-common: site/avr32-common : fix search replace damage in last commit Oct 29 21:36:14 well I guess there is an env var for install or image, but I can't find it Oct 29 21:36:20 and that is what I want to know Oct 29 21:36:38 eFfeM: autotools_stage_all Oct 29 21:36:39 you should be installing from the build area, not image/install. Oct 29 21:37:05 files are moved from image to install, and install is broken up by subpackage. you'd have to hardcode references to your packaging layout in the task, which would be idiotic Oct 29 21:37:05 khem, i know about autotools_stage_all, but mythtv does not use autotools Oct 29 21:37:48 then you have to do some typing Oct 29 21:37:52 ok that explains why install is not a good idea Oct 29 21:37:59 but the whole dir is in image Oct 29 21:38:16 XorA|gone: btw. what are you doing these days? Oct 29 21:39:37 eFfeM: eventually do_stage will go away entirely in favor of automatic staging population based on do_install. that day is not today Oct 29 21:39:47 thesing: IT and working for lrg Oct 29 21:39:57 thesing: PXA drivers at the moment :-( Oct 29 21:40:00 kergoth: I would love to have that Oct 29 21:40:24 XorA|gone: are drivers a bad thing? Oct 29 21:40:35 khem: richard brought it up earlier, i experimented with it when i wrote my prototype private staging implementation Oct 29 21:40:39 would be nice indeed Oct 29 21:40:40 thesing: I keep forgetting how much pxa mainstream sucks Oct 29 21:40:43 kergoth: ok, I understand; issue is that there are many .h files exported by mythtv and they come from different places from $(S) so doing this manually is quite labourous (and if I may say so, I'd rather spent my time on something more useful) Oct 29 21:41:02 I though that eric does a pretty good job. Oct 29 21:41:04 then read bitbake.conf for 10 seconds and see what variable image is Oct 29 21:42:05 woglinde: how do I verify that my system is really eabi? Oct 29 21:42:22 thesing: look in sanity.bbclass Oct 29 21:42:32 thesing: there is a check in there Oct 29 21:42:43 thesing build the kernel with eabi flag set remove oabi Oct 29 21:42:47 compat Oct 29 21:42:49 and start Oct 29 21:42:51 *g* Oct 29 21:43:00 kergoth: thanks for the pointer, i don't know my way too well in the dirs Oct 29 21:43:07 and ofc it is just $(D) Oct 29 21:43:12 thesing: you can pick a binary and dump it Oct 29 21:43:15 bitbake.conf is the main configuration file, and the only one that bitbake parses directly. Oct 29 21:43:25 it's what includes the distro/machine config files, local.conf, etc Oct 29 21:43:30 ah ok, didn't know that Oct 29 21:43:40 it's *the* master config file, it defines the defaults for nearly every variable Oct 29 21:44:10 khem: is 'file' enough? Oct 29 21:44:36 thesing: no I think -readelf -e Oct 29 21:44:44 will show you the flags Oct 29 21:44:57 aahh found this in the taglib recipe: Oct 29 21:44:59 install -m 0644 ${D}/${includedir}/${PN}/*.h ${STAGING_INCDIR}/${PN} Oct 29 21:45:09 that'll get me started Oct 29 21:45:40 ugh, the packaging code is so ugly... really need to refactor this mess Oct 29 21:48:04 khem: "ABI Version: 0" says its oabi? Oct 29 21:48:12 thesing: yes Oct 29 21:48:32 eabi would be 1? Oct 29 21:48:57 kergoth go go Oct 29 21:49:38 thesing: it would be 4 or 5 Oct 29 21:50:12 actually 0x4000000 or 0x5000000 Oct 29 21:51:32 oabi is 0x200 IIRC Oct 29 21:55:52 well this is pretty bad. Oct 29 21:56:05 thesing hm? Oct 29 21:56:10 calling it a day, cya Oct 29 21:56:28 nite effem Oct 29 21:57:44 woglinde: well my supposed eabi system is oabi. Oct 29 21:58:01 hm Oct 29 21:58:14 didnt look into the binarys Oct 29 21:58:24 thesing angstroem? Oct 29 21:58:42 but the kernel builded eabi with my toolchain Oct 29 21:58:52 okay tv now Oct 29 21:58:56 Yet Li Oct 29 21:58:59 http://www.pastebin.ca/1648450 is my patch Oct 29 22:14:39 hi, are there still some gcc4.3 or 4.4 related issues in OE ? Oct 29 22:14:59 I think I spoke with someone some time ago and I was told to stick with 4.2 on ARM for some reason Oct 29 22:15:12 would someone mind clearing ? Oct 29 22:29:06 03Thomas Kunze  07org.openembedded.dev * rbbe9ae502b 10openembedded.git/recipes/linux/ (18 files in 2 dirs): Oct 29 22:29:06 add collie patches for 2.6.31 kernel Oct 29 22:29:06 moved the at91 from SRC_URI to SRC_URI_append_machine for the other two Oct 29 22:29:06 machines that use this kernel. Patches conflict otherwise. Oct 29 22:42:36 Hello Oct 29 22:46:11 03Otavio Salvador  07org.openembedded.dev * rc99bb48707 10openembedded.git/recipes/lzma/ (10 files in 3 dirs): (log message trimmed) Oct 29 22:46:11 lzma: update to 4.65 and use INC_PR Oct 29 22:46:11 Few header files has been added to ${STAGING_INCDIR}/lzma to allow Oct 29 22:46:11 squashfs-tools to be compiled against it. Oct 29 22:46:12 The gcc optimization flags has been droped for now since it ought to Oct 29 22:46:16 have been fixed since the last test with the recipe, otherwise it Oct 29 22:46:18 should be easy to be added back. Oct 29 22:46:18 I did push the pending stuff we have been using here at O.S. Systems Oct 29 22:46:20 03Otavio Salvador  07org.openembedded.dev * r0456b5ee38 10openembedded.git/recipes/squashfs-tools/ (4 files in 2 dirs): Oct 29 22:46:23 squashfs-tools: add lzma support in 4.0 version Oct 29 22:46:25 This added lzma by default since squashfs-tools is not common to be Oct 29 22:46:29 installed into a image and it doesn't increase the compilation time a Oct 29 22:46:31 lot. Oct 29 22:46:33 Signed-off-by: Otavio Salvador Oct 29 22:46:35 03Otavio Salvador  07org.openembedded.dev * r6375bcefe0 10openembedded.git/conf/bitbake.conf: Oct 29 22:46:38 conf/bitbake.conf: use mksquashfs -lzma for squashfs-lzma images Oct 29 22:46:40 This change only works with the new squashfs-tools, version 4.0. Oct 29 22:46:42 Signed-off-by: Otavio Salvador Oct 29 22:46:42 Please tell me if it breaks anything (it has been running fine here for some time) Oct 29 22:46:46 03Otavio Salvador  07org.openembedded.dev * r7afc325ab1 10openembedded.git/ (7 files in 2 dirs): (log message trimmed) Oct 29 22:46:49 chicken: revamp recipes Oct 29 22:46:51 To allow easy usage of chicken, in embedded environments, we have done Oct 29 22:46:53 some refactoring in the packages: Oct 29 22:46:55 - split it up to chicken, libchicken and libuchickenl Oct 29 22:46:57 - drop 2.6 version since no distro depends on that and is not Oct 29 22:47:03 commonly used anymore; Oct 29 22:47:05 03Mario Domenech Goulart  07org.openembedded.dev * rdc86104dd9 10openembedded.git/recipes/chicken/chicken-native_4.2.0.bb: Oct 29 22:47:08 chicken-native_4.2.0.bb: added Oct 29 22:47:10 Signed-off-by: Otavio Salvador Oct 29 22:47:12 03Mario Domenech Goulart  07org.openembedded.dev * r63f6025496 10openembedded.git/recipes/chicken/ (chicken-4.2.0/soname.patch chicken.inc chicken_4.2.0.bb): Oct 29 22:47:19 chicken: added SONAME support for chicken 4 Oct 29 22:47:21 Signed-off-by: Otavio Salvador Oct 29 22:47:23 03Mario Domenech Goulart  07org.openembedded.dev * r2fdfaad187 10openembedded.git/recipes/chicken/chicken.inc: Oct 29 22:47:26 chicken.inc: fixed vars passed to the build system (LIBRARIAN, TARGET_C_COMPILER) Oct 29 22:47:28 Signed-off-by: Otavio Salvador Oct 29 22:51:00 marex: gcc 4.4 has been good to me so far Oct 29 22:51:15 khem, I have issues compiling uboot with it Oct 29 22:51:27 marex: sure thats uboot's problem Oct 29 22:51:45 khem, well I've been tracing it for two or three weeks now Oct 29 22:51:55 marex: what is the error ? Oct 29 22:52:00 I have no idea where to seek help (uboot people are not wiling to fix it) Oct 29 22:52:34 marex: depends if its a gcc problem surely they are right Oct 29 22:52:40 khem, it doesn't work (it runs, but hangs when reaches a function) if that one particular function (__div64_32) is used Oct 29 22:53:11 marex: is this uboot function ? Oct 29 22:53:23 khem, more interestingly ... it hangs only if that function is longer than one instruction Oct 29 22:53:24 yes Oct 29 22:53:56 what is the parent function Oct 29 22:54:32 lib_general/vsprintf.c (put_dec) or cpu/pxa/timer.c (somewhere there) Oct 29 22:54:46 it's used once in vsprintf.c and twice in timer.c Oct 29 22:55:13 __div64_32 seems to be like a kernel function may be they took the implementation from linux Oct 29 22:55:22 can you check if thats the case Oct 29 22:55:58 ah true Oct 29 22:56:07 I never thought of checking that, thanks :) Oct 29 22:56:40 you're right Oct 29 22:58:10 marex: are they in _sync_ Oct 29 22:58:19 khem, yea, I noticed that Oct 29 22:58:32 khem, I just didn't think of checking kernel for that, thanks a lot! Oct 29 22:59:36 np Oct 29 23:21:29 khem, seems completely fine compared to linux kernel, just missing __attribute__((weak)) ... Oct 29 23:21:38 adding it doesn't fix it obviously Oct 29 23:22:13 ok Oct 29 23:22:33 are u using gcc 4.4.2 Oct 29 23:22:47 4.4.1, but that shouldn't be the case Oct 29 23:22:57 it doesn't work ever since 4.3 Oct 29 23:23:12 I'm diving through git log of div64.c in linux kernel Oct 29 23:24:42 03Richard Purdie  07org.openembedded.dev * rfcee557e1b 10openembedded.git/classes/package.bbclass: Oct 29 23:24:42 package.bbclass: Cleanup getVar syntax to use True, not 1 as third argument (from Poky) Oct 29 23:24:42 Signed-off-by: Richard Purdie Oct 29 23:31:30 marex: you said that if only 1 instruction long this works Oct 29 23:31:47 marex: what does that mean Oct 29 23:32:07 does that mean you are commenting out the function fully Oct 29 23:33:11 khem, well ... dunno if one instruction, but something like that -- if I stick in one 'puts()' (which I assume would be something like 1-3 instructions or not much) it does the 'puts()', but doesn't return from the function anyway Oct 29 23:33:33 khem, yes ... I put debugging 'puts()' everywhere around -- it works till it hits the function Oct 29 23:33:44 and in the function it works only if I comment the rest altogether Oct 29 23:35:54 btw if I put two of those 'puts()' calls there, it does only the first one iirc Oct 29 23:38:01 lib_generic/vsprintf.c seems to have its own version of do_div Oct 29 23:40:31 that's just a macro ... it selects whether to use 32 or 64bit division Oct 29 23:41:08 which one do you use Oct 29 23:41:59 64 bit one Oct 29 23:42:09 include/div64.h Oct 29 23:42:14 yes Oct 29 23:45:42 kernel ones have likely(), unlikely() in do_div I suppose you might try that out too Oct 29 23:46:20 * khem is trying to avoid debugging __div64_32 Oct 29 23:48:25 in kernel arm uses its own version of do_div for arm may be thats why kernel one works Oct 29 23:49:52 khem, hm ... porting that to uboot might show something, hang on Oct 29 23:51:15 if you can generate assembly output of this function with working version and non working version it might help in further diagnoses Oct 29 23:51:33 khem, true, but they are totally different Oct 29 23:51:46 I mean the C version of the function Oct 29 23:51:49 I compared the output of 4.2 and 4.4 already, no luck Oct 29 23:51:53 compiled with gcc 4.2 and gcc 4.4 Oct 29 23:51:59 oh you did Oct 29 23:52:09 are they same Oct 29 23:52:10 yeah, but the output is more than complicated Oct 29 23:52:24 Well may be Oct 29 23:52:28 post it somewhere Oct 29 23:52:29 no, they are not ... gcc4.4 has to use some other weird optimizations Oct 29 23:52:43 You can also tweak opt levels Oct 29 23:52:52 try with -Os and then with -O1 Oct 29 23:52:55 khem, dunno if uboot doesn't depend on it Oct 29 23:53:12 only for this one file Oct 29 23:53:50 * khem drives off Oct 29 23:56:10 khem, marex.hackndev.com/div64-42.S and 44.S Oct 30 02:10:03 03Koen Kooi  07org.openembedded.dev * r703b3f6dc4 10openembedded.git/conf/distro/angstrom-2008.1.conf: angstrom: bump DISTRO_PR and some cleanups Oct 30 02:10:20 03Koen Kooi  07org.openembedded.dev * rea66c8bbc7 10openembedded.git/conf/checksums.ini: checksums: add xf86-input-tslib 0.0.6 Oct 30 02:10:20 03Koen Kooi  07org.openembedded.dev * rb2010b603e 10openembedded.git/recipes/abiword/ (abiword-2.5.inc abiword_2.8.0.bb): abiword 2.8.0: band-aid packaging before overhauling it Oct 30 02:10:20 khem, so after porting the functions from linux kernel, it seems to work Oct 30 02:10:29 khem, thanks but still ... it's weird **** ENDING LOGGING AT Fri Oct 30 02:59:58 2009