**** BEGIN LOGGING AT Tue Nov 27 02:59:57 2007 Nov 27 07:16:55 dear oe .. i am trying to build a uclibc system .. so i set DISTRO = "angstrom-2007.1" and TARGET_OS = "linux-uclibc" but yet it still build a glibc-2.5 .. do i still miss something ? Nov 27 07:18:44 ANGSTROM_MODE = "uclibc" aahh , that tells the faq Nov 27 07:19:37 :p Nov 27 07:50:17 in which phase binaries are installed in "-cross" packages? Nov 27 08:12:03 morning Nov 27 08:39:48 bonjour Nov 27 08:46:39 hello, is it normal that packages compiled by dependencies dont get packed after compilation? Nov 27 08:47:02 seen this on gpm (requires ncurses for compile and runtime) but ncurses package isnt packed into an ipkg Nov 27 08:47:21 same for much of the xserver-kdrive dependencys Nov 27 09:01:41 morning Nov 27 09:27:06 anyone using xserver-kdrive? Nov 27 09:28:14 do i need a xorg.conf for this? Nov 27 09:28:23 xserver starts up on a linux 2.4 kernel but i cant move mice etc. Nov 27 09:28:36 and i only see the b/w background Nov 27 09:58:05 * * OE Bug 3375 has been created by john_lee(AT)openmoko.com Nov 27 09:58:06 * * meta-toolchain does not fix pkgconfig data files path properly Nov 27 09:58:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3375 Nov 27 10:25:49 low hanging bug fruit: http://bugs.openembedded.org/show_bug.cgi?id=3291 Nov 27 10:25:49 could someone apply the patch for me? Nov 27 10:38:03 any1 knows of a kdrive version that can be compiled with linux 2.4 headers? Nov 27 10:38:18 cause my whole system runs on linux 2.4 and everything is build with these headers Nov 27 10:38:21 nik0n: no one tried it with OE Nov 27 10:38:22 including uclibc Nov 27 10:38:37 so its up to me :) Nov 27 10:38:39 nik0n: we used 2.4 kernels with userspace built with 2.6 kernel headers Nov 27 10:38:53 rschuster: please add 'mtn diff' of change... Nov 27 10:39:05 hrw: tryed that too, but then i wasnt able to do fconfig on the target for example Nov 27 10:39:16 its an ugly board Nov 27 10:39:29 fconfig? Nov 27 10:39:36 ifconfig, sorry Nov 27 10:39:44 it always told me: socket not availible Nov 27 10:39:51 cause uclibc was build with 2.6 headers Nov 27 10:42:40 hrw: thanks for tackling this. patch is attached. Nov 27 10:45:26 pushing then Nov 27 11:01:02 rschuster: pushed Nov 27 11:01:49 hrw: thanks alot Nov 27 11:03:41 03thebohemian 07org.oe.dev * r2ea9f4ae... 10/ (1 packages/maemo4/hildon-libs_0.15.1.bb): hildon-libs: add libmatchbox to dependencies - close #3291 Nov 27 11:03:50 hi Nov 27 11:04:17 quick question , what is this error ? gconvert.c:48:2: error: #error GNU libiconv not in use but included iconv.h is from libiconv Nov 27 11:04:23 it fail when i build glib Nov 27 11:05:12 BabelO: you build for glibc system? Nov 27 11:05:53 hrw: i build for trolltech greenphone Nov 27 11:06:17 hrw: yes glibc system Nov 27 11:07:07 did you built libiconv by hand? Nov 27 11:07:16 hmm no Nov 27 11:07:33 before my OE was setup for HTC Blueangel Nov 27 11:07:41 and now i do bitbake -c clean Nov 27 11:07:51 then bitbake bootstrap-image Nov 27 11:07:58 with greenphone conf file Nov 27 11:08:03 libiconv destroys each glibc build once it got built ;( Nov 27 11:08:21 so what i need to do ? Nov 27 11:08:35 restart from scratch ;( Nov 27 11:08:38 hrw: time for CONFLICTS = "glibc" in libiconv.bb ? Nov 27 11:08:53 XorA: sort of Nov 27 11:09:03 or check from uclibc Nov 27 11:09:06 hrw: how can i tell bitbake to do that ? bitbake -c clean is not enought Nov 27 11:09:53 mv tmp old-broken-tmp Nov 27 11:10:08 ok, i do it Nov 27 11:12:01 i think it take the afternoon to build ;) Nov 27 11:12:05 * * OE Bug 3291 has been RESOLVED (FIXED) by Nov 27 11:12:07 * *  hildon-libs needs libmatchbox in dependency Nov 27 11:12:09 * * http://bugs.openembedded.org/show_bug.cgi?id=3291 Nov 27 12:06:06 03xora 07org.oe.dev * r61e3209c... 10/ (3 files in 3 dirs): Nov 27 12:06:06 uboot-openmoko_svn.bb : add glamo ram initialisation patch so stop Nov 27 12:06:06 glamo deadlocks. Nov 27 12:14:30 Could someone take a look at http://bugs.openembedded.org/show_bug.cgi?id=3373 for a Firefox build fix? Nov 27 12:14:50 It has a patch and works for me. Might be useful to others too Nov 27 12:18:26 does resulting firefox works for you? Nov 27 12:19:47 hrw: yes, it does Nov 27 12:20:39 hrw: #3372 and #3371 would be nice to have in too Nov 27 12:23:33 * hrw -> lunch Nov 27 12:47:52 cbrake: ping Nov 27 12:48:48 nud: I'm just heading out -- will be back in 45 min ... Nov 27 12:50:05 cbrake: fine Nov 27 12:50:18 cbrake: I was just wondering if you had ever used PCI on the sb-x270 Nov 27 12:50:30 cbrake: a yes or a no would make it happy for now ;-) Nov 27 13:14:08 otavio: There's no patch attached to #3372. Trying to build ff for openmoko now btw Nov 27 13:18:02 I try to build wrt54oe distro fo an at91sam9263ek. But I get "Unable to determine version for kernel headers" during package uclibc-0.9.28-r9 Nov 27 13:18:16 any hints? Nov 27 13:18:40 the helloworld compiles and run Nov 27 13:22:32 urmel: use angstrom in uclibc mode Nov 27 13:22:52 I have tried this too Nov 27 13:23:03 but there where also errors Nov 27 13:23:14 i build the xserver-xorg package, but i dont have neither a mouse module nor a keyboard module Nov 27 13:23:19 is this the normal behavior? Nov 27 13:23:23 the dbus packages makes errors Nov 27 13:23:25 nik0n: because xorg is modular now Nov 27 13:23:46 hrw: wre these module sincluded in OE? Nov 27 13:23:48 urmel: first rule: if you want to complain about errors show them Nov 27 13:23:56 nik0n: yes, you just need to build them Nov 27 13:24:07 nik0n: look into packages/xorg-drivers Nov 27 13:24:16 paste the log here? Nov 27 13:24:26 ~pastebin Nov 27 13:24:27 [~pastebin] A "pastebin" is a web-based service where you can paste anything over 3 lines without flooding the channel. Here are links to a few : http://www.pastebin.com , http://pastebin.ca , http://channels.debian.net/paste , http://paste.lisp.org , http://www.rafb.net/paste Nov 27 13:24:32 ok Nov 27 13:24:42 urmel: if you want to be shot then here - but it will be pain Nov 27 13:25:33 hrw: got it .. thanks Nov 27 13:25:46 last time i sed xserver is like hmm ... 5 years ago or so Nov 27 13:27:02 RP: ping Nov 27 13:27:59 mickeyl: http://bugs.openembedded.org/show_bug.cgi?id=3376 Nov 27 13:31:04 mickeyl: RP is lunching now Nov 27 13:32:57 k Nov 27 13:33:02 there's a bogus line in meta-toolchain.bb Nov 27 13:33:19 the .pc files are changed to point to a wrong path Nov 27 13:33:39 * mickeyl tries a fix Nov 27 13:33:56 Is linux.inc the style-compliant way of doing kernel recipes now? Nov 27 13:34:17 rwhitby: linux.inc content waits for being incldued in kernel.bbclass Nov 27 13:34:32 so I would use it Nov 27 13:34:53 hrw: ok, good. Does that mean that building u-boot-mkimage-openmoko will be in kernel.bbclass? Nov 27 13:35:19 that part I would leave for kernel recipes rather Nov 27 13:35:33 hrw i clean all, update OE and start build, gnu-config give me and error http://pastebin.ca/799241 Nov 27 13:35:52 BabelO: gnu-config-native? Nov 27 13:35:55 hrw quilt is not good :) Nov 27 13:36:08 yes Nov 27 13:36:09 hrw: why is that line in linux.inc anyway? Not everyone's kernel requires uboot-mkimage ... Nov 27 13:36:34 BabelO: bitbake quilt-native Nov 27 13:37:00 rwhitby: we had over 10 lines in a way "DEPENDS_machine_append = "u-boot-mkimage-openmoko" Nov 27 13:37:22 rwhitby: and one day we just changed it to one line as more and more u-boot powered devices started to use this recipe Nov 27 13:37:22 shouldn't that be in a machine.inc file? Nov 27 13:37:41 MACHINE_FEATURES += "uboot" Nov 27 13:37:43 rwhitby: machine does not needs it - kernel needs it Nov 27 13:38:05 patches welcome to make it dependent of MACHINE_FEATURE or KERNEL_IMAGETYPE Nov 27 13:38:07 kernel only needs it for those machines, not all. Nov 27 13:38:32 ok, MACHINE_FEATURE I agree with. Nov 27 13:39:41 hrw: ok but that what i do but is it not supposed to be a depend package then ? Nov 27 13:39:45 So, that requires a little bit of python to make a variable definition dependent on the value of another variable, right? Nov 27 13:39:59 rwhitby: basically yes Nov 27 13:40:16 rwhitby: look into wpa-supplicant and madwifi/pci check Nov 27 13:40:22 (and that will make Koen's eyes bleed, apparently, even though bitbake is written in python, and python is the only supported extension mechanism in .bb files ...) Nov 27 13:40:43 rwhitby: ignore him Nov 27 13:41:12 hrw: that's probably the best advice I've been given all week ... Nov 27 13:41:25 heh Nov 27 13:41:27 ;DD Nov 27 13:42:08 * rwhitby continues his quest to remove the slugos-ness from ixp4xx stuff in OE. Nov 27 13:42:52 * cbrake returns Nov 27 13:42:57 is the new kernel naming convention linux-.inc (like linux-openmoko and linux-bfin) ? Nov 27 13:43:09 nud: no, I have not ever tried PCI Nov 27 13:43:22 oh my bad Nov 27 13:43:27 So a rewritten ixp4xx-kernel should be called linux-ixp4xx ? Nov 27 13:43:29 nud: there is a good chance my patchset in OE is not complete for PCI Nov 27 13:43:42 cbrake: your lot of patches doesn't include cm-x270-pci.c Nov 27 13:43:48 rwhitby: linux-ixp4xx sounds better for me Nov 27 13:44:09 rwhitby: but does it is a policy.. Nov 27 13:44:32 but I've seen that some guy from compulab wrote a patch for an older version of the kernel Nov 27 13:44:51 the same guy that did a em-x270 arch in OE Nov 27 13:45:18 cbrake: have you used it as a base or so ? Nov 27 13:45:30 If one of my ixp4xx kernel modules (zd1211rw for example) requires a companion package to be in the image (zd1211-firmware for example), then do I just DEPENDS on the zd1211-firmware package and RDEPENDS_kernel-module-zd1211rw on zd1211-firmware too? Nov 27 13:45:40 nud: Mike (from compulab) just posted a new patch to the LAK list Nov 27 13:45:50 rwhitby: (naming) yes, i think that sounds sensible Nov 27 13:45:58 nud: I started with his patch from 2.6.16, but have never synchronized since Nov 27 13:46:08 rwhitby: (DEPENDS) yes, that sounds good as well Nov 27 13:46:12 nud: it would probably make sense to try to get his latest patch into OE Nov 27 13:46:20 cbrake: ok Nov 27 13:46:34 I was wondering because I'm in a hurry to get some video input working for next week Nov 27 13:46:35 mickeyl: and would those DEPENDS and RDEPENDS lines go in linux-ixp4xx.inc ? Nov 27 13:46:48 or in linux.inc ? Nov 27 13:47:10 I'll end up being fired because of that crappy card with a frame grabber not supported on linux Nov 27 13:47:12 nud: It should apply cleanly to 2.6.23 as his latest patch is probably against the git HEAD Nov 27 13:47:18 (I don't see any others in linux.inc, and I can't believe this is the first one ever needed) Nov 27 13:47:24 cbrake: do you have a pointer ? Nov 27 13:47:29 rwhitby: module RDEPEND on firmware but not viceversa Nov 27 13:47:56 what does LAK mean ? Linux ... Kernel ? Nov 27 13:47:59 ARM Nov 27 13:48:15 ah, of course Nov 27 13:48:20 hrw: right, module RDEPEND on firmware, but firmware also has to be built so that it's available to get in the image, so a DEPENDS on firmware is required too, no? Nov 27 13:48:21 nud: looking for the patch ... Nov 27 13:48:22 rwhitby: go into kernel.bbclass rather Nov 27 13:49:23 rwhitby: or we will add DEPENDS = "1-firmware 2-firmware 3-firmware" into kernel.bbclass or kernel recipes Nov 27 13:49:53 I guess you only want DEPENDS_kernel-module-foo = firmware Nov 27 13:50:12 so it is not always built if the image doesn't include the kernel module Nov 27 13:52:48 nud: cm-x270 support is now in mainline -- let me know if I can find a way to extract the relevant patches from git Nov 27 13:54:07 nud: if 2.6.24 was released, we would be all set ... Nov 27 13:54:13 cbrake: you would save me once again \o/ Nov 27 13:54:22 I guess I'll have to update my OE tree though Nov 27 13:54:35 or just backport the new kernel... seems safer to me Nov 27 13:55:08 cbrake: native java support for OE is in the making :) Nov 27 13:55:18 rwhitby: bitbake does not know which packages will be created from kernel recipe Nov 27 13:55:35 cbrake: complete with self-compiled ecj Nov 27 13:55:54 nud: an easier option would probably be to just add a linux_2.6.23+2.6.24-rcx recipe to OE Nov 27 13:56:10 hrw: so lines 207 and 208 of kernel.bbclass are bogus? Nov 27 13:56:18 yes maybe Nov 27 13:56:24 nud: pre-release kernels are plenty stable for ARM in my experience, and it would be a lot easier than backporting patches to 2.6.23 Nov 27 13:56:36 cbrake: all that kernel stuff has always frightened me Nov 27 13:56:46 cbrake: what should I do to achieve that ? Nov 27 13:57:07 rwhitby: would have to check Nov 27 13:58:01 rwhitby: hm. it works for package. but that workaround is also in task-base iirc Nov 27 13:58:34 nud: look at the linux_2.6.22+2.6.23-rc5.bb and create a linux_2.6.23+2.6.24-rc3.bb Nov 27 13:58:55 nud: about all you should have to do is create a defconfig and possible add a few of my patches that are not in mainstream Nov 27 13:59:33 nud: like ramdisk-load, and nand-simplify name Nov 27 14:00:13 ok Nov 27 14:00:20 thanks Nov 27 14:00:23 I'll try Nov 27 14:00:47 rschuster: great, looking forward to seeing how it works out! Nov 27 14:01:40 rschuster: I'm still using jamvm + your classpath recipe, so I need to move forward one of these days. Nov 27 14:14:58 do you have replacement for db3-native if download failed ? Nov 27 14:17:48 use db not db3 Nov 27 14:18:26 hrw ok so i try to build db then Nov 27 14:20:05 BabelO: use angstrom as distro Nov 27 14:20:10 it makes life easier Nov 27 14:21:10 hrw: yes i can, i just need to have a clean system like greenphone system to be able to build additionnal package for gp system Nov 27 14:43:37 does it harm any kittens if "make -C somedir sometarget" is called from within a recipe or should I use oe_runmake instead? Nov 27 14:44:04 oe_runmake Nov 27 14:44:05 can anyone out there tell me how the bluez-utils is setup...that is, it doesn't look like any of the FILES_ sections are being installed into the rootfs for things like l2ping etc. Is there a reason why those aren't installed? Nov 27 14:47:44 hrw: ok, thanks Nov 27 15:10:06 hrw: do you happen to know when I look at the line PACKAGES =+ "${PN}-compat ${PN}-alsa" in bluez-utils_3.17.bb, should that be including FILES_${PN}-compat = " \ into the fs ? Nov 27 15:11:07 fs as rootfs? Nov 27 15:11:42 yes, rootfs. that is, I see other files installed, but not those Nov 27 15:12:15 your image does not get bluez-utils-compat installed Nov 27 15:13:23 NOTE: package gcc-cross-kernel-3.3.4-3.3.4-r5: task do_fetch: started Nov 27 15:13:24 NOTE: The MD5Sums did not match. Wanted: 'a1c267b34f05c8660b24251865614d8b' and Got: '2e8f7ac3bd1c6c543476f73d439935d9' Nov 27 15:13:26 NOTE: Task failed: Checksum of 'ftp://ftp.gnu.org/gnu/gcc/releases/gcc-3.3.4/gcc-3.3.4.tar.bz2' failed Nov 27 15:13:30 NOTE: package gcc-cross-kernel-3.3.4-3.3.4-r5: task do_fetch: failed Nov 27 15:13:32 anyone seen it? Nov 27 15:15:32 hrw: ok. I'll look and see what is going in. thanks Nov 27 15:16:03 otavio: No such directory `gnu/gcc/releases/gcc-3.3.4'. Nov 27 15:16:22 otavio: give me url to 3.3.4 and I will check its md5 Nov 27 15:16:41 hrw: this is the URL that bb gives me Nov 27 15:16:52 hrw: i hadn't change it Nov 27 15:17:03 and its wrong - this show howunused it is ;) Nov 27 15:17:11 hrw: Nov 27 15:17:14 hrw: ftp://gcc.gnu.org/pub/gcc/releases/gcc-3.3.4/gcc-3.3.4.tar.bz2 Nov 27 15:17:19 hrw: it's getting here Nov 27 15:17:38 ftp://ftp.gnu.org/gnu/gcc/gcc-3.3.4/gcc-3.3.4.tar.bz2 Nov 27 15:17:48 a1c267b34f05c8660b24251865614d8b gcc-3.3.4.tar.bz2 Nov 27 15:17:57 just fetched from GNU site Nov 27 15:18:13 otavio: rm gcc-3.3.4.tar.bz2 and fetch again Nov 27 15:18:43 hrw: doing it Nov 27 15:28:53 cbrake: did you already experiment with 2.6.24 actually ? Nov 27 15:33:25 nud: no, I have not Nov 27 15:46:21 hrw: i ve restarted build from scratch without tmp directory, and angstrom-console-image ... same error with quilt and db3 :( i build quilt by hand, and now i start db-native-4.3.29 by hand too Nov 27 15:47:09 db!=db3 Nov 27 15:48:29 hrw: yes, maybe bu i use bitbake angstrom-console-image, so i miis something to tell bitbake to use db instead of db3 Nov 27 15:52:13 distro config should do that Nov 27 15:52:30 BabelO: you need to set DISTRO to angstrom version Nov 27 16:29:35 * BabelO open the window and jump :( Nov 27 16:34:02 * mwester adds BabelO to the list of developers who have perished due to OE ;) Nov 27 16:36:12 mwester yes , i try since yesturday to build a OE and restart from scratch each time with different try... still fail in glib-2.0 build and say error about libiconv , don't know what i can do with that :( Nov 27 17:18:05 * * OE Bug 3377 has been created by autobuild(AT)openembedded.org Nov 27 17:18:06 * * contacts-0.7-r1-do_compile Nov 27 17:18:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3377 Nov 27 17:18:43 ok last wuestion else i jump again :) how can i solve that gconvert.c:47:2: error: #error GNU libiconv not in use but included iconv.h is from libiconv Nov 27 17:20:33 bye all Nov 27 17:20:51 BabelO: use angstrom-2007.1 distro Nov 27 17:21:45 hrw|gone: it is set in local.conf ? Nov 27 17:23:49 hello, i'm back with my at91sam9263ek problem. log is at http://pastebin.ca/799439 Nov 27 18:05:50 03florian 07org.oe.dev * re43e0563... 10/ (1 packages/libical packages/libical/libical_0.27.bb): libical: New recipe: iCal (RFC 2445) C library Nov 27 18:05:55 03florian 07org.oe.dev * ref601489... 10/ (4 files in 3 dirs): libical: Fix linking bug Nov 27 18:06:00 03florian 07org.oe.dev * r5ec0cd5f... 10/ (1 packages/gpephone/libcalenabler2_svn.bb): libcalenabler2: Add svn recipe Nov 27 18:06:07 03florian 07org.oe.dev * r058f820e... 10/ (4 files in 3 dirs): libiac: Add patch to work around i18n build breakage. Nov 27 18:06:15 03florian 07org.oe.dev * r362860dc... 10/ (5 files in 2 dirs): ptim: Add 1.0 ptim release set of components. Nov 27 18:06:23 03florian 07org.oe.dev * r7ee08c19... 10/ (1 packages/gpephone/libabenabler2_svn.bb): libabenabler2: Add svn recipe Nov 27 18:06:33 03florian 07org.oe.dev * r1c609444... 10/ (4 files in 3 dirs): libabenabler2: Major update to make it usable. Nov 27 18:06:41 03florian 07org.oe.dev * r2cdc353e... 10/ (1 packages/gpephone/contact_svn.bb): contact: New svn recipe for new G(PE)^2 contact application. Nov 27 18:06:48 03florian 07org.oe.dev * r0b14e939... 10/ (1 conf/checksums.ini): checksums.ini: Remove some G(PE)^2 bits to work around a stupid package "update". Nov 27 18:06:56 03florian 07org.oe.dev * r76520665... 10/ (1 packages/tinymail/tmut_svn.bb): tmut_svn: Update Nov 27 18:07:05 03florian 07org.oe.dev * ra9270d8c... 10/ (1 packages/tinymail/libtinymail_svn.bb): libtinymail_svn: Update version number Nov 27 18:07:14 03florian 07org.oe.dev * rb38c38c0... 10/ (1 conf/distro/include/sane-srcrevs.inc): sane-srcrevs.inc: Update some G(PE)^2 related versions and tmut Nov 27 18:16:11 Once again, I have problem with an at91sam9263ek: '/bin/sh: kinds.h: No such file or directory' => http://pastebin.ca/799439 Nov 27 18:17:21 hrw|gone: ok, i stop .... now it fail with libtool with angstrom-2007.1 Nov 27 18:26:26 hrw|gone: PING Nov 27 18:53:24 Does anybody know of any machines other than poodle that use pxafb on a pxa25x? I am trying to see if the screen corruption on resume is poodle specific or not. Nov 27 18:55:04 hvontres|work: look at handhelds.org, for machine htc universal and htc magician Nov 27 18:55:08 they use pxafb Nov 27 18:55:21 hvontres|work: it is pxa27x Nov 27 18:56:35 BabelO: hmm..so are sl-CXXXX zarus... I think the problem is pxa25x specific tho... Nov 27 18:57:14 hvontres|work: wait i look in cvs, did ou look at other pxa devices ? Nov 27 18:57:45 BabelO: What other ones are there? Nov 27 19:00:02 hvontres|work: ok, i have poodle :) , roverp1, spitz,trizep4, palmld , palmt3, lubbock, lpd270, looxc550, hx2750, htc alpine , htc apache, h4000, h3900, h1900,eseries , asus devices... Nov 27 19:00:10 ok i ll look which one is pxa25x Nov 27 19:01:21 hvontres|work: ok ipaq are pxa25x some asus axim too Nov 27 19:01:30 and same for rover Nov 27 19:02:16 03florian 07org.oe.dev * r840894a6... 10/ (1 packages/xorg-lib/libx11-native_1.1.1.bb): libx11-native: Add 1.1.1 - courtesy by Arne Zellentin Nov 27 19:03:05 BabelO: Thanks. I'll take a look at those later tonight then :) Nov 27 19:03:28 hvontres|work: i know that h4000 works perfect ;) Nov 27 19:03:49 hvontres|work: maybe you have to check your LCCRx register Nov 27 19:04:32 BabelO: Ok..that gives me something to strart with... I have been trying to figure this thing out since 2.6.19 :( Nov 27 19:05:02 * hvontres|work gets back to actual day job Nov 27 19:07:22 hi greentux Nov 27 19:09:49 hi florian, nice ideas around this deviced thingies Nov 27 19:10:28 greentux: eh which one? Nov 27 19:12:17 florian: http://fl0rian.wordpress.com/2006/05/29/platform-integration-for-mobile-linux-devices/ Nov 27 19:14:47 ah Nov 27 19:15:40 greentux: yes right, I am about to start with this topic again. Nov 27 19:16:25 greentux: I just started to investigate what the existign bits are able to do and what we can (re-)use them for. Nov 27 19:17:45 florian: yeah thats fine... nils is in china these days? Nov 27 19:20:52 greentux: He is going to fly on friday.... Nov 27 19:21:29 florian: ah ok Nov 27 19:24:39 Hi, I have a problem to compile angstroem for the at91sam9263ek. Nov 27 19:26:54 * florian lacks such a cute piece of hardware Nov 27 19:27:58 :-( I get this error messages: http://pastebin.ca/799439 Nov 27 19:35:54 urmelZAUS: That is the old fortran bug. Update your meta-data and rebuild Nov 27 19:36:25 which metadata? Nov 27 19:36:38 OE Nov 27 19:36:39 the whole openembedded? Nov 27 19:36:55 mtn pull && mtn update Nov 27 19:36:55 mtn update should do the job? Nov 27 19:37:02 thanks Nov 27 19:37:06 np Nov 27 19:43:33 hvontres|work: ping Nov 27 19:47:11 CoreDump: pong Nov 27 19:47:44 CoreDump: It is ok for a rebuild to just make "bitbake ..." ? Nov 27 19:48:18 urmelZAUS: In theory that should pick up the updated gcc* reciepes, yes. Nov 27 20:45:18 what "distro" is a good starting point for a system without display (only network services). wrt54oe? Nov 27 20:46:19 there is a wrt54oe? wow, I didn't even know Nov 27 20:48:23 Zero_Chaos: #@DESCRIPTION: Distribution configuration for the WRT54 Linux Nov 27 20:49:17 neato Nov 27 20:51:25 Zero_Chaos: neato is not an "oe distro" ?!?! Nov 27 20:51:53 neato means "I think that is neat (cool)" Nov 27 20:52:47 ;-) sorry ... some times ago, there was a linux distribution with this name Nov 27 21:26:01 urmelZAUS: choose Angstrom, and a machine which does not have the 'screen' feature set. Nov 27 21:26:37 rwhitby: thanks Nov 27 21:27:37 Hmm - Angstrom console image is too big for NSLU2 :-( Nov 27 21:29:14 how do I use arm-kernel-shim ? Nov 27 21:29:37 chouimat|work: what do you want to use it for? Nov 27 21:29:53 rwhitby: I can't use apex on a 64MB slug Nov 27 21:30:05 chouimat|work: why not? Nov 27 21:30:16 rwhitby: no serial port Nov 27 21:30:40 hang on, you soldered on ram chips, but didn't solder on a four pin header for serial? Nov 27 21:30:40 rwhitby: I'm just looking at all possibilities ... Nov 27 21:30:57 you don't need serial to run apex Nov 27 21:30:57 rwhitby: I didn't soldered the ram chips I got it like this Nov 27 21:31:22 slugos builds in OE and runs out of the box with apex without serial. Nov 27 21:31:46 k Nov 27 21:31:49 but if you're going to do anything with a fatslug I would recommend serial. Nov 27 21:32:29 anyway, arm-kernel-shim won't help with extra memory. you need apex (this is known to work) or a recompiled redboot (this has never been tested) to support more memory. Nov 27 21:32:46 (cause the bootloader has to change the dram controller) Nov 27 21:32:46 rwhitby: ok thanks Nov 27 21:32:57 np Nov 27 21:34:24 chouimat|work: I'm happy to help you get it working. Let me know if slugos doesn't work out of the box. Once you're booted with 32MB, you need to run apex-env from userland to change apex to look for the 64MB and tell the kernel it's got 64MB. Nov 27 21:35:12 rwhitby: I have to replace redboot with apex ? Nov 27 21:36:02 No, we use apex as a second-stage loader. No danger to your slug, you can always use reset button upgrade mode to recover. Nov 27 21:36:21 rwhitby: cool I never tried this one Nov 27 21:36:23 Hmm - Angstrom minimalist image is also to big for the NSLU2 :-( Nov 27 21:37:13 Hmm - looks like a kernel build problem - the kernel is bigger than the rootfs! Nov 27 21:39:54 how big is rootfs on the NSLU2? Nov 27 21:41:39 8mb IIRC Nov 27 21:43:12 oh wow... Nov 27 21:43:22 rwhitby: I have copied the at91sam9263ek.conf and removed the "screen" from MACHINE_FEATURES. Is "bitbake bootstrap-image" the next step? Nov 27 21:53:12 * rwhitby has been modifying the ixp4xx kernel recipe, so it's probably just an error I've injected. Nov 27 22:31:06 I get an error during unpack from linux-2.6.21: cp: cannot stat `/mnt/sda2/global/tmp/buildroot/oe/org.openembedded.dev/packages/linux/linux-2.6.21/defconfig': No such file or directory Nov 27 22:31:26 There is only a "defconfig.patch" Nov 27 22:43:38 you need a .config (defconfig) for your target Nov 27 22:46:29 Ohh. I have changed the MACHINE to disable the "screen" property. Thanks ... Nov 27 22:47:29 np Nov 27 22:49:22 I had to change SRCREV_pn-python-gsmd = "${AUTOREV}" to SRCREV_pn-python-gsmd = "103" to get my bb working Nov 27 22:49:40 else the svn checkout line will look like: Nov 27 22:49:55 svn co -r ${AUTOREV} Nov 27 22:49:59 and produce an error Nov 27 22:50:20 what do I do to get always the latest revision? Nov 27 22:51:07 svn co Nov 27 22:51:11 svn up Nov 27 22:58:01 any chance cbrake is alive? Nov 27 22:58:43 email it is Nov 27 23:01:38 Fique: I'm just heading out of the office -- feel free to leave me a message and I'll try to check back later this evening Nov 27 23:02:26 i'll just email you... I have a lot of stuff to send you Nov 27 23:02:33 Fique: ok Nov 27 23:02:37 well... not a lot, but a lengthy description Nov 27 23:02:48 btw i'm the person who contacted you earlier about debugging pci for the compulab Nov 27 23:02:54 and thanks again, Nov 27 23:03:23 Fique: ahh, ok. Catch you later then. Nov 27 23:09:01 re crofton Nov 27 23:13:35 Zero_Chaos, substituting = with ?= fixed my problem Nov 27 23:13:55 josch: why you tell me? Nov 27 23:14:07 well you tried to help :-) Nov 27 23:19:34 oh, heh Nov 27 23:24:05 re Nov 27 23:42:35 is there anywhere a good tutorial on how to create a bitbake recipe? Nov 27 23:43:54 openembedded.org has a good manual Nov 27 23:47:32 thx found it Nov 27 23:51:40 Hmm - OE just built me a kernel image which is 3.2GB (yes gigabytes). Nov 27 23:52:22 at least the compressed version is only 4MB ... Nov 27 23:53:04 and it only happens in the Angstrom distro, not the slugos distro. Nov 27 23:53:15 rwhitby: that has to be an official record Nov 27 23:53:16 ;) Nov 27 23:54:27 anyone know enough about how a kernel is built to tell me where to start looking to debug why the Image file is 3GB ? Nov 27 23:54:52 rwhitby: 2.6.23? Nov 27 23:55:12 pH5: yes Nov 27 23:55:29 rwhitby: two possible patches :) Nov 27 23:55:30 rwhitby: look at linux/linux-ezx-2.6.23/patches/binutils-buildid-arm.patch Nov 27 23:56:01 zecke_: what is the other one? :) Nov 27 23:57:13 pH5: haha, I think I only know this one. The one that went into the kernel is probably different. But I lacked the time to look into it :} Nov 27 23:57:47 zecke_: would it already be in 2.6.23.9 ? Nov 27 23:57:56 i think not Nov 27 23:58:18 better shuts up and emulates to work :} Nov 27 23:59:48 pH5: trying that now, thanks for the pointer. Nov 28 00:01:57 :) good night Nov 28 00:18:04 * * OE Bug 3378 has been created by andrea.adami(AT)gmail.com Nov 28 00:18:06 * * gmp-native-4.2.1-r2 fails to compile Nov 28 00:18:08 * * http://bugs.openembedded.org/show_bug.cgi?id=3378 Nov 28 00:18:13 I'm trying to compile wesnoth within an openembedded overlay and I managed to automatically call the autogen script after svn checkout but got stuck with the configure option for the host architecture Nov 28 00:18:31 to test I specified armv4t but get this error: Nov 28 00:18:38 | checking host system type... Invalid configuration `armv4t': machine `armv4t' not recognized Nov 28 00:18:58 what could be the right option? Nov 28 00:25:52 it is armv4t-angstrom-linux-gnueabi Nov 28 00:29:23 pH5's patch fixed the kernel size problem on 2.6.23 Nov 28 00:30:58 rwhitby: what machine were you building 2.6.23 for? Nov 28 00:31:09 ixp4xx Nov 28 00:31:54 rwhitby: was the bug, machine, distro or linux 2.6.23 specific? Nov 28 00:32:38 reason I ask is cause i've been builing 2.6.23 for the Compulab Platform with no issues Nov 28 00:43:59 Fique: seems to be linux-2.6.23 on arm Nov 28 00:44:23 and I think it may be gcc version or oabi/eabi specific Nov 28 00:44:47 cool Nov 28 00:45:03 I think i'll take a deeper look at my setup and see if there may be some danger areas Nov 28 00:47:32 I still get the error | /media/data/moko/build/tmp/staging/x86_64-linux/bin/python: can't open file 'setup.py': [Errno 2] No such file or directory Nov 28 00:47:43 when compiling my code with bitbake Nov 28 00:47:46 hello, i am getting this error when i run bitbake, does anyone know how to fix it? http://rafb.net/p/UwSRQQ26.html Nov 28 00:47:58 but there is no reference to any setup.py in my sourcecode Nov 28 00:52:32 there is a setup.py in my bitbake folder - does it want to have this one? what am I doing wrong?? Nov 28 00:55:56 lemnisca: I'm no expert... but maybe an incompatible python interpreter? Nov 28 00:56:10 hmm, which one should i be using then? Nov 28 00:56:19 i'm trying to find what I am using right now Nov 28 00:56:45 ok, thanks Nov 28 00:57:23 the website just says python 2.4.0 or later Nov 28 00:57:26 Lemnisca: i'm using version 2.5.1, while it looks like you may be using 2.4 Nov 28 00:57:48 ok, well, i will try upgrading it and see it that helps Nov 28 00:57:51 Lemnisca: like I said, i'm no expert... but its my first guess on wierd syntax issues with python itself Nov 28 00:57:58 nod, ok Nov 28 00:58:02 i'll give it a go, thanks Nov 28 00:58:46 Josch: which task is failing when looking for setup.py? Nov 28 00:58:53 do_compile? Nov 28 00:58:57 jop Nov 28 01:00:20 and I do not know where this comes from Nov 28 01:07:45 I know that bitbake uses a setup.py but why can't it be found? Nov 28 01:07:58 thats whats confusing Nov 28 01:08:12 you have your environment variables set, otherwise bitbake wouldnt execute Nov 28 01:08:28 and I can execute any other bb files without errors Nov 28 01:08:29 but its looking for random stuff in the staging area, which is out of whack Nov 28 01:09:25 josch: which .bb file is it failing on? Nov 28 01:09:46 I created a custom one - will post it on pastebin Nov 28 01:10:12 only thing I can think of is that the compile for your custom package is trying to execute its own python scripts Nov 28 01:10:28 and may be looking for something thats not setup in your staging Nov 28 01:10:37 http://pastebin.ca/800017 Nov 28 01:12:16 I'm new to bitbake recipes and successfully used this bitbake file: http://pastebin.ca/800021 from which I copied everything Nov 28 01:12:44 josch: # Nov 28 01:12:54 hvontres|work, ?? Nov 28 01:13:05 heh is this game good? Nov 28 01:13:07 inherit distutils is what is pulling in setup.py I think Nov 28 01:13:15 ah thx Nov 28 01:13:27 gota run Nov 28 01:17:09 wohoo! this was it! Nov 28 01:17:31 now another strange issue Nov 28 01:17:59 the bb will only work if I set SRCREV_pn-python-gsmd ?= "21988" instead of SRCREV_pn-python-gsmd ?= "${AUTOREV}" Nov 28 01:18:13 autorev worked in my other bb file - why not there? Nov 28 01:32:03 is there anything important missing: http://pastebin.ca/800037 Nov 28 01:32:35 log.do_configure just sais: nothing to configure Nov 28 01:32:55 s/sais/says/ Nov 28 01:33:43 do_compile takes no time and does also nothing Nov 28 01:34:26 shouldn't bitbake just run ./configure, make and make install? Nov 28 01:35:05 why does it say "nothing to compile/configure" in the log? Nov 28 01:35:37 do_install even says: *** No rule to make target `install' Nov 28 01:36:53 Fique, yeah this game is great and since the neo1973 has 640x480 resolution and the game is playable with only the left mouse button even touchscreen gaming might be possible Nov 28 01:37:35 I'm just stuck with bitbake atm Nov 28 02:14:46 ok, i have upgraded python and still the same error Nov 28 02:15:52 http://rafb.net/p/wN6zFK77.html **** ENDING LOGGING AT Wed Nov 28 02:59:56 2007