**** BEGIN LOGGING AT Sat Feb 17 02:59:57 2007 Feb 17 04:24:51 In case anyone has some insights into this, I've been stuck on this all afternoon: http://www.pastebin.ca/359980 Feb 17 04:35:32 hillct: Install libpopt-dev? Feb 17 04:36:01 hillct: If it's not picking it up from staging properly, it might just grab the host system's headers Feb 17 04:36:03 uhh... Feb 17 04:36:59 I know it's a hack, but it's worked for me before :) Feb 17 04:37:01 hadn't considered the possibility that popt-dev was a separate package. popt seems ot include the headers, which is what's reported as missing in the error Feb 17 04:37:24 will try it though Feb 17 04:40:16 yeah. popt-dev doesn't exist. all the libraries and headers are provided in the popt package Feb 17 04:40:54 hillct: Hmm, I just had a prod around the .bb files, it's certainly the right package, popt.. a short-term fix may be to install popt-dev on the host, failing that, check that popt's built in tmp/stamps? Feb 17 04:41:10 And also check that tmp/staging/armeb-linux/include has popt.h? Feb 17 04:42:22 oh, it's there Feb 17 04:42:41 that was of course the first thing I checked Feb 17 04:43:24 is there something wrong with CFLAGS_prepend such that the value isn't being inserted into the cc args lines? Feb 17 04:43:25 The only thing that worries me is that the gcc line doesn't have the -I/..../ Feb 17 04:43:39 yeah Feb 17 04:43:52 that's what I've been beating my head against all day Feb 17 04:44:13 docs say CFLAGS_prepend should do the trick Feb 17 04:44:15 Nuke the _prepend part Feb 17 04:44:18 Try this instead Feb 17 04:44:28 CFLAGS += "-I{STAGING_INCDIR} " Feb 17 04:44:53 I believe I tried that in one of these incarnations. will try again though Feb 17 04:46:44 same Feb 17 04:46:53 Hmm Feb 17 04:46:55 it just doesn't make it into the compiler args Feb 17 04:47:47 I tried variants of overriding do_compile() as well, with no luck Feb 17 04:49:37 hillct: I get a vastly different result on my machine Feb 17 04:49:52 http://www.pastebin.ca/359992 Feb 17 04:54:11 lets see. did the one I posted have EXTRA_OECONF uncommented? Feb 17 04:54:21 I tihnk that's what generated that particular brokenness Feb 17 04:54:50 I've gone through many iterations of this Feb 17 04:55:59 er Feb 17 04:56:43 actually that error relates to EXTA_OEMAKE which in teory is not needed and was just done to see if any of these vars were being picked up in the command args Feb 17 04:56:51 ah Feb 17 04:56:55 I'll have a play with it Feb 17 04:57:09 so if you comment out EXTRA_OEMAKE, you should get the error I was seeing Feb 17 04:58:04 and since this is a cross-compile, I presume it should never pull headers from the host system as you mentioned before Feb 17 04:58:50 I also found that install-native is not required as a dependency Feb 17 04:59:14 Yep, I'm getting the error now Feb 17 04:59:22 I'll have a poke at it, see what fixes it up Feb 17 04:59:32 thanks Feb 17 05:00:04 I've been reading OE docs for the past 12 hours so I'm about to call it a night (midnight here) Feb 17 05:00:58 I may be able to keep my eyes open another hour or so but I'm just so frustrated at this point I'm not sure how useful I can be Feb 17 05:04:56 How can I avoid building a kernel in OE? Feb 17 05:05:21 I don't need a kernel, only the rootfs. Feb 17 05:06:20 I used : ASSUME_PROVIDED += "virtual/kernel" Feb 17 05:06:24 But it didn't work. Feb 17 05:07:13 hillct: I just had some luck by omitting -I${STAGING_INCDIR} entirely Feb 17 05:07:42 just using libdir? Feb 17 05:08:31 I had just assumed that not being able to find headers, meant it needed them specifically Feb 17 05:08:42 testing now Feb 17 05:08:42 Er, let me have a poke at it again Feb 17 05:08:46 Doesn't look reproducible Feb 17 05:10:33 might it have pulled the headers from your host enviroment? Feb 17 05:11:03 I don't think so; what' Feb 17 05:11:35 I don't think so; what's happening is this - first time I compile, it complains about missing popt.h. Second build, it ignores that error entirely (making it look like compilation succeeded), and instead complains about -lpopt Feb 17 05:12:04 so it needs both incdir and libdir Feb 17 05:12:19 or, perhaps just the parent directory STAGING_DIR Feb 17 05:12:26 which i just tried with no luck Feb 17 05:12:30 Yeah, looking at other packages that depend on popt though, they don't appear to specify it.. for example, gphoto2_2.1.4.bb Feb 17 05:12:47 ? Feb 17 05:13:14 popt is a trivial library. it shouldn't have any dependencies of it's own Feb 17 05:13:21 hmm Feb 17 05:13:57 checking the redhat popt spec file since they originated the library Feb 17 05:17:26 oh, you mean others packages that make use of popt Feb 17 05:17:28 yeah Feb 17 05:17:37 hillct: It looks like the -L should go in LDFLAGS += "-L${...} " Feb 17 05:17:50 oh Feb 17 05:18:03 for libpath Feb 17 05:18:07 yeah Feb 17 05:18:16 I believe I tried that at one point as well Feb 17 05:21:18 in your testing, is CFLAGS showing up in the compile args, because that's still not working for me, independant of any other error Feb 17 05:24:45 Nah, it's not Feb 17 05:26:51 Oh, I think I may have it Feb 17 05:26:56 Just let me test this.. Feb 17 05:29:11 ..or maybe not Feb 17 05:31:07 the docs make reference to CCFLAGS as distinct from CFLAGS Feb 17 05:31:10 hmm Feb 17 05:32:37 Yeah, CC for C++, CF for C Feb 17 05:33:16 I'm getting somewhere; the MAKE env var isn't what you want Feb 17 05:34:02 oh, right Feb 17 05:34:15 * hillct is not much of a makefile hacker Feb 17 05:35:38 hillct: From what I can see, 'inherit autotools' shouldn't be there, unless the package uses autoconf, which logrotate doesn't seem to Feb 17 05:35:51 Now it's just a matter of adjusting the .bb to specify what should be packaged Feb 17 05:36:02 oh Feb 17 05:36:06 ok Feb 17 05:36:23 this is my first attempt, so I'm learning Feb 17 05:36:23 Just working out what's up with that ATM Feb 17 05:36:31 Yeah, I've only really packaged minor things before Feb 17 05:36:53 I started on this because I figured it was going to be minor Feb 17 05:39:24 Almost there, I'm just having a slight issue with the makefile Feb 17 05:57:12 hillct: I think I've got it.. oe_runmake sets CFLAGS (from environment) to be higher priority than that in the makefile Feb 17 05:57:55 meaning... you have to override oe_runmake? Feb 17 05:57:58 hillct: I've got it going apart from a -D that's set in the makefile, which is overridden by the OE environment Feb 17 05:58:05 Well, get it to stop setting -e at least Feb 17 06:06:21 hillct: Compiling works, now just tweaking the linking Feb 17 06:06:40 wow Feb 17 06:06:58 and I thought this was going to be a good first build project Feb 17 06:07:09 thanks for all your help Feb 17 06:08:57 No probs, it's been educational for myself as well :) Feb 17 06:12:16 Muahaha! I think it's working Feb 17 06:12:23 !!! Feb 17 06:13:32 http://zaurus.tucuxi.org/logrotate/ Feb 17 06:13:34 Give that a burl Feb 17 06:14:13 What I did, pretty much, was this - remove the autoconf bit, add a patch to fix the Makefile that was clobbering CFLAGS/LDFLAGS, and fiddle with CFLAGS/LDFLAGS Feb 17 06:17:21 cool Feb 17 06:17:25 testing now Feb 17 06:19:51 well, I'll get the install and package going tomorrow Feb 17 06:19:56 thanks so much Feb 17 06:20:04 I probably would never have worked this out Feb 17 06:20:08 hillct: It should be installing ok from that one.. the binary and .conf at least Feb 17 06:20:15 Again, no probs :) Feb 17 06:20:26 Look in tmp/deploy/glibc/ipk/logrotate for the package Feb 17 06:20:33 binary seems to be staging someplace other than sbindir Feb 17 06:21:08 Hmm, shouldn't be too hard to fix, just fiddle with the do_install () Feb 17 06:21:13 yeah Feb 17 06:21:17 should be trivial Feb 17 06:21:20 install: cannot create regular file `/home/slug/slugosbe/tmp/work/armv5teb-linux Feb 17 06:21:20 /logrotate-3.7.1-r0/image/usr/sbin/logrotate': No such file or directory Feb 17 06:21:45 weird.. do_install worked fine here Feb 17 06:47:00 NAbyss_: install needed -D Feb 17 06:47:06 Ah :) Feb 17 06:47:12 when built on an FC6 host anyway Feb 17 06:47:44 ${SBINDIR} seems to be /usr/sbin Feb 17 06:48:07 is there a var that is /sbin by default, or should I just reset it in the .bb file? Feb 17 06:48:43 No idea TBH.. try running 'set' during do_install and look at the variables available Feb 17 06:49:08 mm Feb 17 06:49:21 oh, yeah. in the run script Feb 17 06:49:29 that's pretty cool stuff Feb 17 06:49:32 checking Feb 17 06:58:22 ah. base_sbindir Feb 17 09:08:12 hi all Feb 17 09:19:56 hello Feb 17 09:34:22 ~ping koen Feb 17 09:34:35 pong Feb 17 09:34:58 pong koen Feb 17 09:36:17 koen : glibc 2.5 fails when compiled for a ppc without fp. (like the 405). Have a look at this -> http://sourceware.org/bugzilla/show_bug.cgi?id=2444 Feb 17 09:38:26 stupid glibc Feb 17 09:40:03 i get alot of these -> warning: -msoft-float and -mlong-double-128 not supported Feb 17 09:40:08 and the build fails Feb 17 09:40:59 the only option i see at the moment is to enable fp emulation in the kernel for ppc405 and compile with hardfloat Feb 17 09:41:19 sed out -mlong-double-128 ? Feb 17 09:41:24 but its very slow Feb 17 09:42:16 you have added the -mlong-double-128 for efika right ? Feb 17 09:43:21 yes since it failed to build without Feb 17 09:45:23 the same guy has submited this patch -> http://gcc.gnu.org/ml/gcc-patches/2006-02/msg01589.html Feb 17 09:45:35 maybe is the way out of the problem Feb 17 10:01:57 Koen: seen my message on the mailing list ? Feb 17 10:02:10 about the donations? Feb 17 10:02:12 yes, I read it Feb 17 10:02:34 Yeah and about the pressrelease towards linuxdevices I want to do ? Feb 17 10:03:23 I think that's a good idea Feb 17 10:03:37 Okay :) Feb 17 10:03:48 Will write one and send it to them Feb 17 10:06:09 gerwinin: you are the organisational genius Feb 17 10:07:45 gerwinin: I will be available for booth manning but due to illness havent organised my schedule yet, Ill fill in Wiki tomorrow probably Feb 17 10:15:37 drat Feb 17 10:15:46 all the interesting talks are at the same time Feb 17 10:16:03 tinymail/gegl/exmap/etc Feb 17 10:16:28 koen: cloning time Feb 17 10:21:29 hrm Feb 17 10:21:42 the insane.bbclass update forces a rebuild Feb 17 10:23:01 thats insane :-) Feb 17 10:24:22 someone, make patch for bitbake to ignore any "dependencies" Feb 17 10:37:15 koen : more stuff -> http://sourceware.org/bugzilla/show_bug.cgi?id=2749 Feb 17 10:38:01 koen : i have send an email to steven munroe for help on the subject Feb 17 10:39:34 Xora thanks Feb 17 10:39:44 It would be cool if you are there ! Feb 17 10:42:21 * gerwinin is happy his scoop, spectrum analyser and function generator arrived Feb 17 12:40:49 in the chatlogs of this channel, I read about a wifi Hermes II driver for Linux 2.6, does anyone know more about this? Feb 17 13:01:49 hi mickeyl Feb 17 13:01:57 cheers pb_ Feb 17 13:03:40 morning Feb 17 13:17:33 koen: ping Feb 17 13:26:30 please close bugs 1321 and 1172. they are both with familiar as distribution and old. Feb 17 13:26:59 CoreDump|home: key added Feb 17 13:27:05 mr_nice: you can do that yourself as well Feb 17 13:27:10 koen: thanks! Feb 17 13:27:27 koen: how? Feb 17 13:27:31 CoreDump|home: I suspect that answers your ping? Feb 17 13:27:38 indeed Feb 17 13:27:44 mr_nice: mark them as resolved or invalid Feb 17 13:28:20 koen: ok - thx Feb 17 13:42:08 03koen 07org.oe.dev * re16f199d... 10/ (1 packages/cups/cups_1.2.7.bb): cups: fix packaging Feb 17 13:47:29 03koen 07org.oe.dev * r80f4fea0... 10/ (1 classes/insane.bbclass): insane.bbclass: don't do bb.fatal is a .so is present in ${PN}, lots of application ship plugins with that extension (e.g. perl) Feb 17 13:48:48 heh Feb 17 13:49:08 right after I change insane.bbclass, zecke enters the channel Feb 17 13:49:28 hehe Feb 17 13:49:40 changed bb.fatal to bb.error? Feb 17 13:49:55 [15:20] CIA-20: koen org.oe.dev * r80f4fea0... / (1 classes/insane.bbclass): insane.bbclass: don't do bb.fatal is a .so is present in ${PN}, lots of application ship plugins with that extension (e.g. perl) Feb 17 13:49:58 only one Feb 17 13:50:03 I'm fixing the rest Feb 17 13:50:20 been in bed, caught a cold/flu whatever Feb 17 13:50:41 zecke: get well soon! Feb 17 13:51:05 glibc-dbg failed for me yesterday, but when I wanted to look into it I had to rebuild glibc :} Feb 17 14:00:23 koen: the insanity tester does not work on the second run, for some weird reason... Feb 17 14:00:59 question, which files i have to check & modify to change the installed packages in the angstrom-gpe-image ? Feb 17 14:01:44 ERROR: QA Issue: non debug package contains .debug directory: libgsmd-tools Feb 17 14:05:38 hi Feb 17 14:06:24 can sb help me with the following prob? http://rafb.net/p/soNXKX27.html Feb 17 14:07:18 this is a code-paste-page Feb 17 14:07:48 i really get tired.. :( Feb 17 14:08:14 jrn_: you're missing DISTRO="angstrom-2007.1" Feb 17 14:08:51 03koen 07org.oe.dev * r86d18b1b... 10/ (1 packages/gsm/libgsmd_svn.bb): libgsmd: fix packaging Feb 17 14:08:56 hmm and then it will work on a terrier? Feb 17 14:09:29 yes Feb 17 14:10:05 thank you so much!! Feb 17 14:10:59 yeeehaw! looks really nice not to see an error Feb 17 14:10:59 zecke: can insane output the actual error? Feb 17 14:12:36 Hi Feb 17 14:13:09 I need some little help :) Very little :) Feb 17 14:13:32 don't ask to ask, just ask Feb 17 14:13:51 koen: yes, I added that, and then left the office (wanted to look into the glibc-dbg issue) Feb 17 14:14:26 http://bugs.openembedded.org/show_bug.cgi?id=1881 Feb 17 14:15:00 koen: add the config.log :) Feb 17 14:16:16 Thanks. I need to avoid the kernel compilation. I tried with ASSUME_PROVIDED += " virtual/kernel", and also with ASSUME_PROVIDED += " linux-ecbat91-2.6.17-r1", and with ASSUME_PROVIDED += " linux-ecbat91" ... but the kernel is being compiled... Feb 17 14:17:06 okay later :) Feb 17 14:17:06 time to rest Feb 17 14:17:07 zecke: that's still uploading Feb 17 14:17:16 oh, okay will wait for it Feb 17 14:17:22 ah, done Feb 17 14:17:31 koen: mpd and mp3blaster fail this check as well Feb 17 14:18:58 * Arhuaco wonders how to avoid the kernel compilation :) Feb 17 14:19:20 re Feb 17 14:19:30 Arhuaco: why do you want to avoid it? Feb 17 14:19:38 koen: http://svn.gnome.org/viewcvs/gnome-vfs/trunk/configure.in?rev=5284&view=markup Feb 17 14:19:39 Arhuaco: lots of packages need kernel headers Feb 17 14:20:02 AC_ARG_WITH(samba-includes, [ --with-samba-includes=PREFIX Location of samba includes.], Feb 17 14:20:05 with_samba_includes="$withval", with_samba_includes="/usr/include") Feb 17 14:21:13 * koen tries Feb 17 14:21:17 Ah, I see. koen, I wanted to avoid the kernel compilation because the kernel that I'm building is not ready yet. I still need to build u-boot inside OE to tag the kernel (using mkimage). And I just wanted to save time with my tests. Feb 17 14:21:28 koen: same construct with OpenSSL Feb 17 14:22:11 okay, time to sleep :) Feb 17 14:26:16 Arhuaco: you only need uboot-utils for mkimage Feb 17 14:28:39 I see. Well, I'd neel to patch u-boot (http://svn.arhuaco.org/svn/src/emqbit/ECB_AT91_V2/u-boot/u-boot-1.1.4.patch). I will read an example. Jin^eLD, I thought I could avoid the kernel compilation to make the rootfs, but it seems I cannot. Feb 17 14:29:08 Arhuaco: well, if your board is already flashed with uboot, you do not really need to build uboot yourself in OE Feb 17 14:29:16 (unless you really want to :) Feb 17 14:29:16 Aha. Feb 17 14:29:24 Yes! I don't want to. Feb 17 14:29:26 :) Feb 17 14:29:33 It's flashed with u-boot. Feb 17 14:29:50 03koen 07org.oe.dev * r63008622... 10/ (1 packages/gnome/gnome-vfs_2.16.3.bb): gnome-vfs: fix #1881 Feb 17 14:29:55 And I can do it using the serial port. So I don't really need to build it inside OE. Same with the kernel :) Feb 17 14:30:17 you can skip the uboot thing, but kernel does not hurt I think Feb 17 14:30:33 I upload the kernel using u-boot (with tftp), or using a loader (with the serial port). Feb 17 14:30:34 especially since you get various kernel modules as ipk packages and can choose which ones you install Feb 17 14:30:42 yes, I have a same kind of board Feb 17 14:31:07 I also upload the uImage of the kernel via tftp, but then I configured u-boot to mount my rootfs via nfs Feb 17 14:31:09 I see. Well, I really wanted to skip the kernel compilation :) But, it's ok if I canot do it. Feb 17 14:31:14 I see. Feb 17 14:31:34 I think you can hmm, let me see my configs Feb 17 14:31:48 I'm making tests and I just learnt how not to build the toolchain again. And it saves a LOT of time :) Feb 17 14:31:54 Arhuaco: you can skip kernel compilation, but we advice not to Feb 17 14:31:54 :) Feb 17 14:32:37 Ok, koen. I would like to do it in my tests. But later I'll compile it. I just need to debug the rootfs building.... Feb 17 14:32:56 In the final version I'll build it. Feb 17 14:33:01 I think ASSUME_PROVIDED += " virtual/kernel " will do the trick Feb 17 14:34:56 Jin^eLD: I tried ASSUME_PROVIDED += " virtual/kernel" (without the space before the "), and it did not. Feb 17 14:35:16 mhm Feb 17 14:35:32 (I added it to local.conf) Feb 17 14:35:37 is some cases (ppc for example) you need uboot utils to process the kernel image Feb 17 14:35:55 yes me too.. I did that when I tried to work with a prebuilt toolchain, which btw turned out not to be the best choice :) Feb 17 14:36:54 I see. Well, I like the toolchain that OE builds, but I just wanted to avoid compiling it again whenever I change local.conf, machine.conf or distro.conf. Feb 17 14:37:20 (with changes that should not affect the toolchain) Feb 17 14:37:33 uhm, then the toolchain is not compiled again I think? Feb 17 14:37:53 It is! :) Feb 17 14:38:04 if the changes do not effect the toolchain is not re-compiled Feb 17 14:38:24 huh? hm.. weird. well, if I change some stuff in my distro conf and those changes are not related to gcc/glibc then the toolchain is untouched Feb 17 14:38:45 so yes, as Ifaistos just said, the toolchain is not recompiled if your changes have nothing to do with it Feb 17 14:39:11 Ok. Weird then... In my case it was built all the time. Feb 17 14:39:20 Using latest OE.mtn and latest stable bitbake. Feb 17 14:39:25 that's strange indeed Feb 17 14:39:29 I see. Feb 17 14:39:41 probably you mean the parsing of the OE metadata that seems to take for ever imho Feb 17 14:39:47 Ha ha ha. Feb 17 14:40:07 No, it was built! I swear. I had to use ASSUME_PROVIDED += " virtual/${TARGET_PREFIX}gcc-initial:gcc-cross-initial" and so on to avoid the recompilation. (in local.conf) Feb 17 14:40:28 It saves me a lot of time. Feb 17 14:41:35 if you changes the machine in local.cong, then yes its going to recompile Feb 17 14:41:50 I didn't change the machine. Feb 17 14:41:51 :) Feb 17 14:42:00 if you change the distro from libc to uclibc, yes it will recompile Feb 17 14:42:14 I swear didn't make big changes. Feb 17 14:42:35 Just added a package to the rootfs and bitbake recompiled it. Feb 17 14:49:38 (I'm using EABI) Feb 17 14:49:57 DISTRO="angstrom-2007.1" ? Feb 17 14:50:18 Perhaps it has something to do with the recompilation. Feb 17 14:50:24 Ah no, but the config is similar. Feb 17 14:50:30 I'm adding a new one. Feb 17 14:50:40 why? Feb 17 14:50:49 * koen is getting sick of NIH Feb 17 14:52:29 Well, because I'm trying a new machine. Would it be better to use Angstrom? I thought about it. Feb 17 14:53:22 Angstrom is actually maintained and supported in OE Feb 17 14:57:27 Yup. I guess I should ise it. Feb 17 14:57:32 I should use it. Feb 17 14:59:32 It doesn't matter if the machine will not use X11' Feb 17 14:59:35 ? Feb 17 15:00:31 actually I am interested in that too.. Angstrom wants X11, right? Feb 17 15:00:33 no Feb 17 15:00:34 no Feb 17 15:00:40 morning Feb 17 15:00:50 angstrom has nothing to do w/ X... Feb 17 15:01:23 I see. Well, I should use Angstrom for the machine. I really thought about and I think it's the best thing to do. Feb 17 15:01:27 currently the bootstrap and x11 image are the ones supported Feb 17 15:01:42 I see, I'd need the bootstrap image. Feb 17 15:01:46 if people want to support other images, please do Feb 17 15:02:11 koen: I think bootstrap is going to be replaced by task-base? Feb 17 15:02:16 Ok :) I want to port to the angstrom. Feb 17 15:02:16 I thought that is Feb 17 15:02:32 Jin^eLD: task-bootstrap was replace, not (angstrom-)bootstrap-image :) Feb 17 15:02:41 ooh ok, now I get the difference :) Feb 17 15:02:46 I see. Feb 17 15:02:47 :) Feb 17 15:03:26 I need to compile for a arm920t machine. Feb 17 15:03:48 I like the idea of using angstrom. Feb 17 15:06:16 FEED_ARCH_ep93xx = "armv4t", FEED_ARCH_h6300 = "armv4t" in angstrom-2007.1.conf are similar machines. Reading..... Feb 17 15:06:24 (similar CPUs) Feb 17 15:07:38 * Arhuaco reads http://www.angstrom-distribution.org/building-angstrom Feb 17 15:09:52 <_law_> which .bb file builds /dev/*devices* ? Feb 17 15:10:15 <_law_> there are some basic files before udev script start Feb 17 15:10:16 <_law_> s Feb 17 15:11:30 _law_: org.oe.dev/files iirc Feb 17 15:11:35 _law_, why don't you read responses to your questions? Feb 17 15:11:49 <_law_> psokolovsky_: ? Feb 17 15:12:46 _law_, read my responses on #angstrom today. they have full info, if grep -r really doesn't work for you Feb 17 15:13:05 Hey there ... ;) Feb 17 15:14:47 ok, /quit is not the way to go :) Hey there again. Feb 17 15:14:51 <_law_> psokolovsky_: you thinks ist devices_table.txt ? Feb 17 15:15:09 <_law_> the devices init script comes after udev init... Feb 17 15:15:16 _law_, you *know* it is not? ;-) Feb 17 15:15:55 _law_, I gave you link to bbclass which handles it. does it do something else but creating static /dev? Feb 17 15:19:58 <_law_> psokolovsky_: mom i?ll try to find the bbclass Feb 17 15:30:01 <_law_> psokolovsky_: dont find it anymore sorry Feb 17 15:30:25 _law_, how can I help you? Feb 17 15:31:09 <_law_> psokolovsky_: in which file can i set /dev/random static ? Feb 17 15:31:45 _law_, device_table*, see channel archives for exact pointer Feb 17 15:31:53 or use find Feb 17 15:32:43 hi Feb 17 15:32:50 <_law_> psokolovsky_: packages/initscripts/initscripts-1.0/device_table.txt ? Feb 17 15:32:54 <_law_> hi florian Feb 17 15:33:16 _law_, possibly Feb 17 15:33:25 _law_, if it's referenced from somewhere Feb 17 15:35:59 morning Feb 17 15:36:38 <_law_> psokolovsky_: /dev/random c 444 0 0 1 8 - - - Feb 17 15:36:49 <_law_> random and uradom are already in... Feb 17 15:36:51 _law_, where's that? Feb 17 15:37:01 <_law_> device_table.txt Feb 17 15:37:55 _law_, what about device_table-minimal.txt? Feb 17 15:47:40 _law_, any news? Feb 17 15:47:45 <_law_> psokolovsky_: i think thats the right one :-) Feb 17 15:48:06 _law_, that would real surprise ;-)) Feb 17 15:48:07 <_law_> psokolovsky_: now i should know who it is setup.. :-) Feb 17 15:48:11 would be Feb 17 15:50:43 <_law_> psokolovsky_: kernel docs? Feb 17 15:51:51 _law_, dunno. I would expect that look at that file should be enough at least for a try to add them, w/o any docs. ymmv Feb 17 16:00:59 03coredump2 07org.oe.dev * r467588e7... 10/ (1 packages/altboot/altboot_wip.bb): altboot_wip.bb: Test commit Feb 17 16:36:03 fosdem looks like a hectic couple of days Feb 17 16:36:24 it is Feb 17 16:36:43 I've gone over my schedule and signed up for some booth slots Feb 17 16:38:20 Hopefully more time appears Feb 17 16:49:07 is it possible to build openmoko w/ .dev natively (ie: w/o downloading and installing the OM svn tree)? Feb 17 16:51:47 any advice on getting from Brussels airport to Astrid? Feb 17 16:54:04 CoreDump|home: yes Feb 17 16:54:26 Crofton|laptop: taxi is the fastest, but iirc you can go by subway or bus as well Feb 17 16:54:55 I am cheap Feb 17 16:55:07 well, I do not fear public transport Feb 17 16:55:26 Crofton|laptop: ask mickeyl, do13, lrg or Xora Feb 17 16:55:31 they come by plain Feb 17 16:55:38 ok Feb 17 16:55:44 you come by train? Feb 17 16:55:47 * Groucho_ has just missed mickey_away :-{ Feb 17 16:56:23 Crofton|laptop: yes, I'll arrive by train and walk from the station to the hotel Feb 17 16:56:35 ok Feb 17 17:00:38 koen, what station is near the hotel? Feb 17 17:01:21 central station Feb 17 17:01:26 ok Feb 17 17:01:35 that's a ~15 minutes walk Feb 17 17:01:35 found it on the web page even :) Feb 17 17:02:08 koen: thanks. MACHINE fic-gta01 / DISTRO openmoko and openmoko-image.bb gives me an image useable by qemu? Feb 17 17:02:45 no Feb 17 17:03:03 for qemu you want MACHINE=qemuarm Feb 17 17:03:25 ahh, but the above create images flashable to a neo then? Feb 17 17:03:27 I didn't add DISTRO=openmoko yet because that needs a hack to base.bbclass Feb 17 17:03:43 CoreDump|home: why do you want to use qemu? Feb 17 17:03:50 koen: i don't actually Feb 17 17:03:54 this isn't crazy ass shit like qt/e Feb 17 17:03:56 just wondering ;) Feb 17 17:03:58 it's just gtk on X Feb 17 17:04:19 so DISTRO=Angstrom then? Feb 17 17:04:36 angstrom-2007.1 or generic or openzaurus-unstable Feb 17 17:05:42 the kernel (well, uboot actually) has a slight buildproblem with angstrom, but that doesn't matter since no-one has an actual phone yet Feb 17 17:05:55 Seems like koen still loves the TrollTech products Feb 17 17:05:57 * Groucho_ hides Feb 17 17:06:02 CoreDump|home: FWIW: all openmoko apps are already in the angstrom feed Feb 17 17:06:05 feeds* Feb 17 17:06:30 koen: I just want to prep my build env for P0, that's all Feb 17 17:06:54 Groucho_: who doesn't... ;) Feb 17 17:06:57 CoreDump|home: in that case, just wait till you get the actualy phone Feb 17 17:08:03 CoreDump|home: I'm trying to convince mickey to drop the FIC bbcollection since everythign (minus DISTRO=openmoko) is in OE Feb 17 17:08:31 koen: I am not landing at Brussels airport Feb 17 17:08:31 IIRC that is the long-term plan Feb 17 17:08:49 make OM a native OE target Feb 17 17:09:04 CoreDump|home: could be, but right now .dev has more bugfixes to openmoko than openmoko svn proper Feb 17 17:09:26 not "install ungodly amount of special SVN trees, fixed ancient OE databases, and apply a ton of menual fixes" Feb 17 17:09:30 * Groucho_ thought that after all these years he has been away people changed their minds and fall in love with the TT community manager :) Feb 17 17:09:37 =) Feb 17 17:09:40 heh Feb 17 17:11:33 OM will be buildable just like angstrom w/ OE once the OM folks cleanup their SDK I guess Feb 17 17:11:37 At least since I heard that Llorn is now back into TT development I am trying to convice people to quickly sell their TT stocks :D Feb 17 17:11:47 Groucho_: hehe Feb 17 17:17:20 whats that supposed to mean? Feb 17 17:24:11 Hi ljp, guessed you were awake :-) Feb 17 17:25:37 Kone i have tried to build openmoko-messages and openmoko-appmanager. Both builds. Could we add them to OE? Feb 17 17:25:49 s/Kone/koen/ Feb 17 17:26:12 aren't they already in? Feb 17 17:26:16 no Feb 17 17:28:28 ./openmoko-apps/openmoko-appmanager_svn.bb Feb 17 17:29:09 sorry that is correct, but not openmoko-messager Feb 17 17:29:28 ljp: That supposed to mean that TT(n) = TT(n-1) - community manager + developer Feb 17 17:29:43 goxboxlive: aaaah, FIC didn't have a .bb for that :) Feb 17 17:29:59 * Groucho_ is wondering what happened to that role ... Feb 17 17:30:05 well, i am still acting as community liaison Feb 17 17:30:26 i c Feb 17 17:30:28 just wanted less to do with marketing people Feb 17 17:31:05 where CC: measn ignore Feb 17 17:32:07 Groucho: oslo is going to hire a 'Trolltech Community' person. closer to europe :) Feb 17 17:33:44 goxboxlive: OM-messages added Feb 17 17:33:59 great thx Feb 17 17:35:11 03koen 07org.oe.dev * r42ab062f... 10/ (1 packages/openmoko-apps/openmoko-messages_svn.bb): openmoko: add messages application Feb 17 17:36:20 koen: do you have some time for me? Feb 17 17:36:54 mr_nice: yeah Feb 17 17:38:30 koen: if a device is in apm -s state what does normaly bring it back? Feb 17 17:38:56 usually a press on the powerbutton Feb 17 17:39:13 or something else configured as wakeup event in the kernel Feb 17 17:39:49 koen: where is that wakeup event defined? Feb 17 17:40:07 no idea on that, sorry Feb 17 17:40:35 koen: no prob - thx. now I know about what I have to search Feb 17 17:42:47 eno: so the oleg toolchain built now, how do I make a specific package? Feb 17 17:42:50 oops wrong window Feb 17 17:42:55 :) Feb 17 18:34:00 hm what was the bitbake command to see which packages is used for virtuals? Feb 17 18:35:58 morning all Feb 17 18:36:11 good morning RP! Feb 17 18:36:29 woglinde: bitbake -s or bitbake -nv should show somwthing Feb 17 18:38:26 woglinde: hi Feb 17 18:38:51 he mr_nice I am trying an opie-image Feb 17 18:38:57 I spotted one error Feb 17 18:39:03 simpad-kernel is only zImage Feb 17 18:39:06 not bzImage Feb 17 18:39:43 woglinde: yes - hrw sayd that he changed it to zImage Feb 17 18:40:26 woglinde: also you need to download the new swithes driver patch form bugreport 1873 Feb 17 18:40:40 hm that is included already Feb 17 18:40:47 or did you made another one? Feb 17 18:41:19 woglinde: yes but the one which is included killes the irq of the touchscreen driver - so better to use the one attached at the bugreport Feb 17 18:41:54 woglinde: just dl and replace it Feb 17 18:42:35 woglinde: hopefully some one is that nice and includes it in oe :) Feb 17 18:43:48 woglinde: please also have a look over conf/simpad.conf if there is a USE_DEV_FS="1" Feb 17 18:44:55 woglinde: It have to be not there Feb 17 18:45:06 mr_nice I know Feb 17 18:45:19 I changed my simpad kernel conf heaviliy Feb 17 18:45:47 woglinde: you can share the improovements :) Feb 17 18:46:04 mr_nice yes I have to remmeber my password Feb 17 18:46:05 woglinde: I am not good in bitbakestuff Feb 17 18:46:11 that only stops me Feb 17 18:46:35 woglinde: hehe Feb 17 18:47:05 ah now qt-mt compiles instead auf qt-mt-satic Feb 17 18:47:07 static Feb 17 18:50:15 woglinde: I have to implement the power button - where do you think is the right place to put the file arch/arm/mach-sa1100/simpad-power.c ? Feb 17 18:50:51 hm Feb 17 18:52:49 simpad.c or power-simpad.c to be in line with leds Feb 17 18:53:04 woglinde: ok Feb 17 18:53:33 woglinde: power-simpad.c will be a good choice I think Feb 17 18:53:46 03koen 07org.oe.dev * rab8d3cfb... 10/ (1 packages/gpe-session-scripts/gpe-session-scripts_0.67.bb): gpe-session-scripts: clean up dependencies Feb 17 18:59:25 hms opie-image is in a bad state Feb 17 19:03:07 hm, didn't tested opie-image Feb 17 19:08:29 woglinde, http://www.linuxtogo.org/gowiki/OpieWithAngstrom . otherwise, there're lots to do due to OE changes. patches welcome ;-I Feb 17 19:08:43 psokolvosky oh thanks very match Feb 17 19:08:45 will look Feb 17 19:22:39 Is there or was there a known problem with calibration on the tosa? Feb 17 19:23:34 some people reported problems, but those where mostly PEBKAC iirc Feb 17 19:23:40 I met YamaKen yesterday and he reported calibration issues with his tosa Feb 17 19:23:50 e.g. stuff left in /home Feb 17 19:23:58 He said those even appeared on qemu images Feb 17 19:24:27 koen: How did they manifest? He said clicking the calibration screen had no effect at all. Feb 17 19:24:42 opie or gpe? Feb 17 19:24:47 (or something else?) Feb 17 19:24:59 all of them. Feb 17 19:25:13 I was very surprised to hear that. Feb 17 19:25:31 could be a missing kernel driver, a broken touchscreen or misconfigured software Feb 17 19:25:33 I assumed the tosa support had improved significantly since the devs received a machine. Feb 17 19:26:03 Laibsch: to boot qemu images you need to load a different kernel, so that isn't representive for a tosa Feb 17 19:26:08 I guess a broken touchscreen is out if qemu angstrom images exhibit the same behaviour. Feb 17 19:26:52 all my machines calibrate correctly with angstrom Feb 17 19:27:09 but you don't have a tosa, do you? Feb 17 19:27:15 no Feb 17 19:27:18 I'll ask hrw about it. Feb 17 19:27:24 but as I said, qemu can't emulate a tosa Feb 17 19:27:32 OK Feb 17 19:27:43 koen I sadly cant remeber my password for monotone whats the best way to fix this? Feb 17 19:27:55 woglinde: there isn't Feb 17 19:28:01 koen new user? Feb 17 19:28:04 woglinde: don't choose bad passwords Feb 17 19:29:15 hm maybe I can brute force it Feb 17 19:44:31 i think there is a bug when the rootfs is created Feb 17 19:44:46 i have set the rootfs dir in my distro as following -> IMAGE_ROOTFS = "${TMPDIR}/rootfs/${MACHINE}" Feb 17 19:45:33 and i get this...stelios@Athena:/proj/oplinux-0.2$ ls tmp/libc/rootfs/ Feb 17 19:45:33 etc x86 Feb 17 19:46:10 inside the etc is rcS.d Feb 17 19:48:01 stelios@Athena:/proj/oplinux-0.2$ ls tmp/libc/rootfs/etc/ Feb 17 19:48:01 rcS.d Feb 17 19:55:49 eds-dbus packaging is broken. the eds-dbus ipk I just built only contains category .pngs Feb 17 20:12:22 does anyone know how to scroll back in a qemu window to see the kernel boot log? Feb 17 20:15:22 cbpage: shift+page-up Feb 17 20:15:39 and keep your fingers crossed that the kernel maps your keys correctly Feb 17 20:16:23 koen: thx, that works -- was trying ctrl-.. per the qemu docs ... Feb 17 20:18:34 looks like I can also redirect a serial console to stdout -- that seems useful. Feb 17 20:20:55 cbrake yes Feb 17 20:26:52 woglinde: works quite well! :-) Feb 17 20:32:06 pH5: created with bitbake -i? Feb 17 21:46:40 NOTE: generating locale es_NI (UTF-8) Feb 17 21:46:40 Error: f0005 Feb 17 21:46:40 qemu: uncaught target signal 6 (Aborted) - exiting Feb 17 21:46:59 is that a known problem for angstrom / neo1973? Feb 17 21:47:14 is that an OE-built qemu? Feb 17 21:47:18 * CoreDump|afk is using qemu from debian Feb 17 21:47:25 don't do that Feb 17 21:47:38 never ever try to outsmart OE Feb 17 21:47:41 qemu was complaining about a wrong gcc version Feb 17 21:47:52 apt-get install gcc-3.4 Feb 17 21:47:57 that's soooo ahrd Feb 17 21:48:01 hard* Feb 17 21:48:42 yeah, well, assume_provided qemu-native worked just fine in the past. for me anyway Feb 17 21:49:24 I guess I'm .oz354x spoiled... Feb 17 21:49:48 OZ is so last year.... Feb 17 21:50:12 and that is not a good thing IMO Feb 17 23:38:29 Laibsch: I think its actually fair to close 1297 as a WONTFIX as work in 3.5.5 has started Feb 17 23:42:01 Laibsch: actually I just closed it as FIXED as I just checked my stuff has been synced into oz354x Feb 17 23:50:24 XorA|gone: OK, great Feb 18 02:14:58 03Laibsch 07org.oe.documentation * r65e93af2... 10/ (1 usermanual/chapters/recipes.xml): recipes.xml: improve some awkward wording. **** ENDING LOGGING AT Sun Feb 18 02:59:57 2007