**** BEGIN LOGGING AT Thu Jan 26 02:59:57 2012 **** ENDING LOGGING AT Thu Jan 26 03:37:48 2012 **** BEGIN LOGGING AT Thu Jan 26 03:55:12 2012 Jan 26 04:45:48 Crofton|work: hello Jan 26 06:40:32 is it possible to override rm_work for a specific recipe Jan 26 06:40:53 biab Jan 26 06:42:05 better Jan 26 08:08:15 JaMa: are you familiar with kernel recipes with initramfs? Jan 26 08:26:45 eFfeM_work: sort of Jan 26 08:28:16 JaMa: I'm running into the issue how to specify dependency between kernel and image, did this: Jan 26 08:28:18 do_configure[depends] += "syrcxx-bootmode-image:do_rootfs" Jan 26 08:28:30 that works, but only if you do not use rm_work Jan 26 08:28:53 the do_configure is in my kernel recipe Jan 26 08:29:17 so I am looking for a better way to specify the dependency Jan 26 08:30:08 BTW is it possible to inhibit rm_work for a single recipe ? Jan 26 08:30:12 or override ? Jan 26 08:31:20 check linux-kexecboot Jan 26 08:31:31 was doing so Jan 26 08:31:33 and imho it's not posible to inihibit it per recipe Jan 26 08:31:46 rm_work doesn't work well in oe-core anyway Jan 26 08:32:01 I was hoping it would be possible to override rm_work in the recipe Jan 26 08:33:27 i'm still on oe classic, issue is that we run regression and release builds on an autobuilder and without rm work I use 12 GB, with only 2; with 20 projects or so, it makes quite a difference Jan 26 08:47:48 hiho Jan 26 08:47:51 hm good news Jan 26 08:48:10 a german court ruled that e.v. can have ther G.A. in chatrooms Jan 26 08:54:17 woglinde: interesting Jan 26 08:54:19 morning all Jan 26 08:57:56 bl yes Jan 26 09:34:20 good morning Jan 26 09:34:27 hi mckoan Jan 26 09:46:35 hi folks! Jan 26 09:50:26 i am having a bit of a hardtime getting a kernel rebuilt with small changes. it seems like the chain: bitbake virtual/kernel -c clean : cp backup/linux-2.6.29* downloads ; bitbake virtual/kernel -c configure ; bitbake virtual/kernel -f -c compile ; bitbake virtual/kernel -f -c compile_kernelmodules ; bitbake virtual/kernel -c deploy ; bitbake core-image-minimal Jan 26 09:50:48 does not necessary lead to the new kernel beeing integrated into the core-image-minimal Jan 26 09:50:52 what am i missing? Jan 26 09:51:25 btw. iam allready, i guess in userland :) Kernel panic - not syncing: Attempted to kill init! . Jan 26 09:52:25 how do oe hardcore kernelcompilers compile their kernels and how would i force to get it into a new jffs image Jan 26 09:53:36 i followed more wor less: http://www.slimlogic.co.uk/2011/05/openembeddedangstrom-kernel-workflow/ , but that does not really cover the "system image bakery" Jan 26 09:54:00 find -name "*defconfig*" Jan 26 09:54:33 good morning Jan 26 09:54:38 hi florian Jan 26 09:54:47 hi woglinde, florian. Jan 26 09:55:39 well, is or isnt there a way in oe to do "fast" kernel recompiles and "automagic" deployment into a filesystem image? Jan 26 09:56:08 maybe its just not made for that. no idea. ive been wron on what and how oe is ued and what for in the past. Jan 26 09:59:07 edit the defconfig Jan 26 09:59:13 bump the pr of your kernel recipe Jan 26 09:59:34 woglinde: ohg sorry, havent recognized you where talink to me. thought it was a classical case of "wrong terminal" Jan 26 10:01:58 ipaqlor: yes, you landed in userland finally :) now you seem suffer from getty/inittab problems Jan 26 10:02:23 check the tty* Jan 26 10:03:16 and as woglinde said, just increase PR of the kernel recipe (for the moment it's easier, otherwise just rerun the task populating /boot) Jan 26 10:03:28 mhm Jan 26 10:04:15 I'm not sure that counts as a "fast" recompile. :-} Jan 26 10:04:42 pb_: hey, thanks, your config made me see the light :) Jan 26 10:04:52 hi pb Jan 26 10:05:05 oh, very good. what was the problem? Jan 26 10:05:52 pb_: well first those SA1100 specific option had to be turned on, all ;) Jan 26 10:06:40 pb_: CONFIG_MTD_SA1100=y is a really hit for example :) Jan 26 10:08:14 pb_: know of a faster way to do kernel recompiles and image integration with oe? Jan 26 10:08:40 well, the easiest way to do kernel recompiles is just to open a devshell and type "make". Jan 26 10:08:46 also, is there any option to makes me clean everything but the doenloaded files belonging to that package? Jan 26 10:08:54 if you need it integrated into a jffs2 then that's harder. I can't think offhand of any easy and quick way to do that. Jan 26 10:09:08 scp ;-) Jan 26 10:09:22 heh Jan 26 10:09:25 pb_: yes. the integration is the problem, as right now i can only bvoot jffs2 images. Jan 26 10:09:40 I guess you could make a script that mounted your jffs2 on a loop device, copied the kernel in, and unmounted it. Jan 26 10:09:50 but, fwiw, crl bootldr can boot kernels from anywhere, they don't have to be in jffs2 Jan 26 10:10:02 pb: yeah, i looked into this oscenity, chaining to modules together Jan 26 10:10:12 I'm pretty sure you can have them as loose files on vfat, or just loaded straight into RAM. Jan 26 10:10:21 pb_: well, i havent fiigured out yet how that is doen. Jan 26 10:11:39 pb_: my nex goal is to get a kexecbootkernel running. so that i am not bound to some bootloader for bootstrapping a kernel and am also able to boot from cf Jan 26 10:12:12 ... Jan 26 10:12:51 you should be able to boot from cf already. but, yes, using kexec is not a bad idea in any case. Jan 26 10:13:23 but the download cleanup really is anoying. this is really unneccessary trafic that wastes not only time (due to slow kernel.org for example) but energy bandwith and nerves :) Jan 26 10:13:26 ipaqlor: your machine.conf has MACHINE_ESSENTIAL_EXTRA_RDEPENDS = "kernel ipaq-boot-params" Jan 26 10:13:49 what do you mean by "download cleanup"? Jan 26 10:14:12 maybe -c cleansstate ? Jan 26 10:14:22 pb_: i mean that when i clean virtual/kernel the kernel tarball gets erased :) thats why i back it up and copy it back after the cleanup Jan 26 10:14:38 oh. well, that shouldn't be happening, and it doesn't happen for me. Jan 26 10:14:54 pb_: one second, i try again. Jan 26 10:14:56 only -c cleanall would do Jan 26 10:15:36 ant_work: a yes, cleanall, Jan 26 10:15:43 ^_^ Jan 26 10:15:54 oh, right. well, don't do that. Jan 26 10:16:02 don't do it, unless you want to redownload all ;) Jan 26 10:16:13 ok, but i did the cleanall because the clean didt do me much good as far as i know :) Jan 26 10:16:24 I think "cleansstate" is probably what you want. Jan 26 10:16:40 the only difference is that cleanall removes the downloaded files too. if that isn't what you want, don't run it. Jan 26 10:16:42 pb_: and where do i inert it? Jan 26 10:16:49 pb_: after the clean, or before? Jan 26 10:18:18 hmm, bitbake virtual/kernel -c clean , ls downloads/linux-2.6.29* Jan 26 10:18:19 ls: cannot access downloads/linux-2.6.29*: No such file or directory Jan 26 10:18:24 well. no idea Jan 26 10:18:36 i rebuild and try it again afterwards Jan 26 10:21:55 ipaqlor: once you have a valid defconfig you could try to run the task do_savedefconfig (maybe is default, can't remember) Jan 26 10:22:13 this will give you a shrinked defconfig Jan 26 10:22:33 which you can test against 3.x kernels maybe... Jan 26 10:22:54 best is to enter devshell and do 'make oldconfig' by hand in that cases, though Jan 26 10:23:22 but you see the path Jan 26 10:23:30 ant_work: hehe, well, thanks, no, im rather "clueless" and feeling itchy allready with the "every 10 minutes" a new kernel programm i am into. for 2.6.29 that is :) Jan 26 10:24:13 :) Jan 26 10:24:25 ant_work: right now i am manually replacing the .config after runnung bitbake -c configure Jan 26 10:24:31 but please don't forget to send patches or your effort will be lost Jan 26 10:25:16 ant_work: shure, by the time this makes sense i may know how to do that for all those gitjunkys out their. normal diffs arent going to be enough i guess :) Jan 26 10:25:55 see, it depends on your lazyness... Jan 26 10:26:10 at least pl send a tarball of your modified files Jan 26 10:26:24 ant_work: haha, lazyness. you are funny :) i am unemplyed and sitting 15 hours a day on this computer for a week now. Jan 26 10:26:52 just fiddling around with this ipaqstuff, bitbake, git, crazy documentation, lost data.... Jan 26 10:28:28 and still got no init started! Jan 26 10:28:48 i am not lazy, im near to "micro burnout tm" Jan 26 10:29:04 sorry, no offensive intentions here ;) Jan 26 10:29:31 remember nobody is payed for work on those old devices Jan 26 10:30:38 ant_work: well, some people for shure. but thats not the point. i am not taliking about money, money is not interesting, for i have non anyways :) luckily i found someone who had the serial cable and gave it to me fro free. Jan 26 10:31:28 that's the spirit we have in OE Jan 26 10:40:19 yeah Jan 26 10:40:53 * Jay7 looked over set of Zauruses.. Jan 26 10:41:13 how retro ;) Jan 26 10:42:39 come on lets search the web via hotbot, altavista and fireball, get out our "dino"Zauruses. and watch the SUN go down ;) Jan 26 10:44:42 :-) Jan 26 10:49:28 I'm thinking of setting up an oe-core layer for Kirkwood devices (e.g. Sheevaplug). It seems to have not been carried over to the new oe-core layers. Has anyone already done this? Jan 26 10:50:06 hi Jan 26 10:50:28 has anybody had problems pulling the meta-smartphone ? Jan 26 10:52:26 is it possible to ignore it? (assuming it has smartphone specific stuff Jan 26 11:03:43 icanicant: I did a provisional layer yes, all worked well Jan 26 11:04:34 icanicant: I can't really maintain it myself though as I don't own a sheevaplug / other kirkwood device Jan 26 11:04:49 icanicant: I can publish the files though if you would like Jan 26 11:05:07 perhaps you could maintain such a layer? ;) Jan 26 11:05:36 bluelightning: that was my next question :) Would be interested in seeing it - I have a Sheevaplug so should be able to maintain. Jan 26 11:05:47 great Jan 26 11:05:59 there wasn't much work required IIRC Jan 26 11:06:10 bluelightning: is it just a case of copying the machine conf files over from oe classic? Jan 26 11:11:28 ipaqlor: I'm thinking about doing an Xmame device for my kid to destroy once he's done with iphone. The hope is he'll keep fingers away from my next-to-be-bought ipad3... Jan 26 11:11:39 ... but I know this won't work :/ Jan 26 11:13:04 icanicant: actually I just checked... the machine conf files I used were the same files from OE unmodified Jan 26 11:13:07 icanicant: so yes :) Jan 26 11:13:25 icanicant: all you need is conf/machine/sheevaplug.conf and conf/machine/include/kirkwood.conf Jan 26 11:13:49 maybe add the formfactor stuff Jan 26 11:14:08 ant_work: it's not so important at least for sheevaplug since it has no display :) Jan 26 11:14:28 heh Jan 26 11:15:18 icanicant: then you can just steal the layer conf file from another layer, add a README and COPYING.MIT, and you have a layer :) Jan 26 11:15:35 (well, you'll need to trivially modify the layer config) Jan 26 11:15:45 bluelightning: great, thanks. Sounds rather simple :) Jan 26 11:16:06 icanicant: yeah I think for this machine it's pretty straightforward, there wasn't really any bitrot :) Jan 26 11:16:28 ant_work: Cheers, hadn't read about formfactor - that just took me to the Poky ref manual which looks quite useful (and up to date for oe-core) Jan 26 11:16:44 icanicant: let me know when you publish it and I will add it to the layer index page on the wiki Jan 26 11:19:09 bluelightning: thanks for the hints, much appreciated. I'll let you know when it's ready. Jan 26 11:20:04 icanicant: awesome, thanks. I was hoping someone would come along and take care of sheevaplug :) Jan 26 11:28:41 JaMa: thanks for the kexec pointer, it lead me to kernel.bbclass and INITRAMFS_IMAGE Jan 26 11:28:44 seems to work now Jan 26 12:03:26 ant_work: mame, well, on what device? the ipaq for instance sucks as a gaming device because of the cheap controls. Jan 26 12:03:47 1 button, and 4 copper meander.... really sucks Jan 26 12:17:18 hi, all! Jan 26 12:17:24 Marex, Hi Jan 26 12:17:39 privet slapin Jan 26 12:18:25 I have trouble building python lately, it seems to be OK but some (most of) builtin modules aren't built Jan 26 12:18:39 Jin^eLD: privet Jan 26 12:21:22 e.g. ImportError: No module named IN Jan 26 12:22:08 I don't see any errors in python build log. Jan 26 12:22:40 * Jin^eLD has not updated OE recently Jan 26 12:26:23 slapin: long time not seen :) Jan 26 12:26:36 Jin^eLD: recently = last 2 months? Jan 26 12:26:42 Jay7: hi! Jan 26 12:28:26 moment, hab grad anruf von telekabel, sorry Jan 26 12:28:36 oops wrong window Jan 26 12:32:08 does anyone here use ubi? and does it make sense on 32mb mtd devices? Jan 26 12:32:37 ipaqlor: ubi is much faster on mount Jan 26 12:32:43 than jffs2 Jan 26 12:32:50 about python: files are at http://ossfans.org/py/python-configure http://ossfans.org/py/python-compile Jan 26 12:33:10 could somebody look and tell what is wrong? Jan 26 12:33:45 python just doesn't build most of system-specific modules silently Jan 26 12:33:51 Jay7: thats good. is it as "stable" as jffs2? Jan 26 12:34:15 I'm not too aware of python build system and what could make it go haywire Jan 26 12:34:37 ipaqlor: I haven't used ubi yet :) Jan 26 12:34:47 but lot of devices have it as default fs on mtd Jan 26 12:34:52 on nand even Jan 26 12:35:51 jay7: ok :) i just seen the cerner config option so many times i thought id ask. right now im more or less bound to jffs2 so im asking rather theoretical here. Jan 26 12:36:40 we (Zaurus owners) are waiting for ubi support in klibc to grab it into kexecboot to be able to boot from ubi NAND :) Jan 26 12:36:43 jay7: it does more writes to flash as far as i found out and saves ram. Jan 26 12:37:24 hi. Is there way for use 'require ' directive where is in a separated layer? Jan 26 12:45:00 do anybody aware of OE python stuff? Jan 26 12:45:39 jekhor: usually you only include/require .inc files Jan 26 12:46:37 what that http://pastebin.com/31yp8mSz means? Jan 26 12:47:17 slapin: that's normal that some modules fail Jan 26 12:49:01 ant_work, we create custom toolchain based on meta-toolchain. I have resolved problem --- we forgot to add .bb path in require directive :) Jan 26 12:49:39 yes, respect the path Jan 26 12:50:10 JaMa: but ncurses? and no IN module? Jan 26 12:52:20 slapin: /home/build1/slave/angstrom_build/build/build/tmp-angstrom_2010_x-eglibc/sysroots/x86_64-linux/usr/include/ncurses.h:41:23: fatal error: curses-32.h: No such file or directory Jan 26 12:52:30 that looks like problem with your curses headers Jan 26 13:01:19 JaMa: so it depends on -native headers? Jan 26 13:36:07 | arm-angstrom-linux-gnueabi-libtool: link: `/home/balister/src/git/oe-core/build/tmp-angstrom_2010_x-eglibc/sysroots/usrp-e1xx/usr/lib/libsystemd-login.la' is not a valid libtool archive Jan 26 13:36:16 anyone recognize this class of error? Jan 26 13:49:20 Crofton|work: there has been an awful lot of systemd commits in the last days Jan 26 13:50:05 I'd start from there (meta-oe) Jan 26 13:52:34 yeah, it is a libtool like issue though Jan 26 14:28:34 ant_work, seems like I need a linux media server so I can get my photos off this machine Jan 26 14:28:44 disk full during build really messes things up :) Jan 26 14:46:55 mhm, my kernel is too silent. to most information i get is when i enable jffs2 debugging for i ca see the last file/dir handling operations Jan 26 14:48:15 is tehre anything to a "init" process except it has not to terminate? would it be feasable to copile a small programm that does nothing but loops for ever? Jan 26 14:52:05 yes, or you could use the shell, or anything else you want Jan 26 14:52:20 russ nelson always had this plan to use python as /sbin/init. I'm not quite sure whether he got a working system that way. Jan 26 14:52:44 pb_: yes, i used /bib/ash and ture, and whatnot, but its allways the same Jan 26 14:52:55 true would be no good, since it exits immediately Jan 26 14:53:08 pb_: ture :) but i tried it anyways :) Jan 26 14:53:34 does your filesystem work in qemu? Jan 26 14:54:19 pb_: i have no idea. havent thought about emulation much. its just another construction place :) Jan 26 14:55:22 the filesystem layout is definitly ok, everything a little busybox sys v init needs in my opinion. Jan 26 14:56:44 so im trying to see if i can get anything compiled that the kernel can load Jan 26 14:58:22 for(; ; ) { nop } ... or something like that. Jan 26 15:02:28 ok, lets see if that makes a fidderence :) main() { for(; ; ) { printf(X \n"); }} Jan 26 15:09:22 cu later Jan 26 15:19:44 Crofton|work: pong Jan 26 15:23:28 khem, I was thinking we could have an oe toolchain layer, and a straight linaro layer Jan 26 15:24:07 Crofton|work: I dont want to duplicate the work Jan 26 15:24:26 of cuorse Jan 26 15:24:27 since distros want to use same compiler for non arm Jan 26 15:24:39 using plain linaro might not be sufficient Jan 26 15:25:00 I know, but it would be helpful for reporting problems Jan 26 15:25:04 shr and angstrom uses the same compiler across the board and linaro is arm centric Jan 26 15:25:29 khem: hi, there are still some issues with gcc and multimachine builds :/ Jan 26 15:25:31 its a lot of work though Jan 26 15:25:50 JaMa: the one with libstdc++ libtool ? Jan 26 15:25:52 or more Jan 26 15:25:52 khem, we would want help from the linaro guys Jan 26 15:26:16 Crofton|work: that same they will know that the new layer has their patches Jan 26 15:26:31 I will be at Linaro connect so I will discuss it there Jan 26 15:26:46 great! Jan 26 15:26:53 I am happy with current scenario though Jan 26 15:27:10 still I want the toolchain layer to be separate Jan 26 15:27:24 since I want to bring in gdb binutils improvements Jan 26 15:27:25 khem: no, something like this http://paste.pocoo.org/show/541082/ Jan 26 15:27:36 basicaly, I would like it to be easy to switch from the working layer to the test layer :) Jan 26 15:30:01 khem: and this http://paste.pocoo.org/show/541085/ (I have seen this one on previous gcc upgrade too) Jan 26 15:39:24 JaMa: thats metadata problem not gcc problem Jan 26 15:39:45 JaMa: can you post me the do.run_configure script of before and after ? Jan 26 15:40:26 Crofton|work: dont worry it will be a small switch for you just add another layer and it would work as it does Jan 26 15:40:49 Crofton|work: I want to add experimental gcc 4.7 there too Jan 26 15:40:53 for the bravehearts Jan 26 15:46:14 JaMa: the second issue is weard Jan 26 15:46:32 JaMa: do you have enable-plugins in binutils ? Jan 26 15:49:35 khem: no.. default binutils from oe-core Jan 26 15:49:54 ok the first problem is something to do with the metadata Jan 26 15:49:56 but grep -R plugins doesn't show them explicitly enabled or disabled Jan 26 15:50:01 second could be a lto issue Jan 26 15:50:05 but I am not sure yet Jan 26 15:50:26 and here are logs http://build.shr-project.org/tests/jama/gcc-issue/gcc-cross-intermediate-4.6.2+svnr181430-r27-temp/ Jan 26 15:51:02 to be clear, this happens only on 2nd build of same gcc version (which shouldn't be rebuild afaik) Jan 26 15:55:18 JaMa: for the first one I need to look into temp of gcc-cross-intermediate Jan 26 15:55:39 ah thats what you gave ok Jan 26 16:03:23 JaMa: for first issue I will ask you do something. Compile the first time and make a copy of gcc sources under work-shared then compile for armv7 and make yet another copy Jan 26 16:03:29 and then diff the gcc source trees Jan 26 16:03:40 and let me know if something in the source tree changes Jan 26 16:03:57 we do lot of munging Jan 26 16:04:01 more than needed Jan 26 16:04:12 and something might be biting in the backside Jan 26 16:04:54 OK.. I'll next time as it takes couple of hours on that box.. **** ENDING LOGGING AT Thu Jan 26 17:46:57 2012 **** BEGIN LOGGING AT Thu Jan 26 17:48:19 2012 Jan 26 18:28:51 xf86-video-fbdev-2_0.4.2-r17.0/xf86-video-fbdev-0.4.2/configure: line 15925: `XORG_DRIVER_CHECK_EXT(RANDR, randrproto)' Jan 26 18:28:52 bah Jan 26 20:58:25 i neeed to know where is the compilation script i mean img_iso one.. Jan 26 20:58:54 im rebuilding system.img and system.sfs and dont works :/ Jan 26 20:59:40 i know how to open and mount it but i need to learn how to build by hand Jan 26 20:59:44 rebuild* Jan 26 21:00:09 some guru whaching? Jan 26 21:09:06 ALoGeNo: I suspect you're in the wrong channel (at least system.sfs does not really ring a bell for me). Jan 26 21:09:44 there is no compilation script or so in openembedded, if you want to rebuild things look e.g. at www.openembedded.org for info (or the yocto website) Jan 26 21:28:58 lol Jan 26 21:29:16 eFfeM, of course yep ;) sorry Jan 26 21:29:28 np Jan 26 21:29:30 wrong channel :) Jan 26 21:29:44 Jan 26 21:29:49 but this is a nice one to ask that too hehe Jan 26 21:30:05 is just a squash question :) Jan 26 21:30:07 ALoGeNo: may I ask which build system this question was for ? Jan 26 21:30:26 yeah np, this is for android-x86 ;) Jan 26 21:31:14 i have 3.2 fullworking in my eeepc t101mt and i want to rebuild the system to use like instalation one ;) Jan 26 21:31:35 atm.. Jan 26 21:37:53 ah ok, guess that is why sfs vaguely rang a bell Jan 26 21:38:04 cool Jan 26 22:33:51 hi Jan 26 22:34:42 i'm trying to build console-image for a beagleboard. had to change sources/layers.txt to point to a different meta-smartphone repository, because it was failing to find it Jan 26 22:35:18 is this beagleboard specific or is there something wrong with my oe setup? Jan 26 22:47:50 evening Jan 27 00:00:21 rlrosa: sometimes git servers might be down Jan 27 00:00:26 always retry Jan 27 00:43:50 is there a neat way to write a machine layer without providing a kernel recipe? for example making it pull the defconfig from kernel sources. meta-fsl-ppc looks quite nice (they have made their own mechanism for copying the kernel defconfig file) but still uses a kernel recipe. Jan 27 00:44:42 no one says you have to provide a custom recipe. you're free to bbappend an existing one Jan 27 00:48:27 kergoth: thanks for feedback, haven't looked into bbappend. does that still sit in a recipes-kernel/linux folder in the machine meta-layer, or is there an easy way of just using the default "linux" recipe and just providing the machine .conf file? Jan 27 00:49:22 you still probably need to provide a defconfig at a minimum to configure the kernel properly Jan 27 00:49:28 and you still have to use a bbappend to add to the search path for that file Jan 27 00:51:36 icanicant: to answer your question, traditionally yes (and it's a good idea to use the same structure) although nothing mandates that as long as the name matches Jan 27 00:51:39 kergoth`: thanks I'll look into that. Jan 27 00:55:54 bluelightning: thanks, i'll go with the standard naming conventions. i was just curious whether any recipe was needed at all - all of the meta layers i've looked at provide a kernel recipe but it struck me that you only really need a defconfig (and even that might be fetchable from the kernel sources). that way only a machine .conf is required for the hardware layer. Jan 27 00:58:03 icanicant: right... so in this case, you just need a bbappend which does FILESEXTRAPATHS_prepend (see other examples) and SRC_URI += "file://defconfig" Jan 27 01:02:40 bluelightning: and that goes in something like recipes-kernel/linux with the defconfig delivered as part of the layer (rather than fetched from the kernel sources)? so the layer contains two folders, conf/machine and recipes-kernel/linux Jan 27 01:03:20 with only a .bbappend and a defconfig in recipes-kernel/linux Jan 27 01:04:00 icanicant: so, it would be possible to copy the defconfig from the kernel sources; but my understanding was linus is waging war on defconfigs within kernel sources anyway Jan 27 01:05:06 icanicant: typically files relating to a recipe are in a subdir named with the recipe name (${PN}) Jan 27 01:05:18 e.g. recipes-kernel/linux/linux Jan 27 01:06:41 icanicant: FYI, oe-core does not provide an old-style OE kernel recipe (i.e. linux.inc) but instead one which fetches from the linux-yocto git tree Jan 27 01:07:21 meta-oe does contain linux.inc but then you'd still need to provide a full recipe Jan 27 01:08:01 so this is all well and good but there's the question of which kernel recipe you would be bbappending... Jan 27 01:08:34 I gotta get some sleep now but will be on tomorrow Jan 27 01:08:50 gn8 Jan 27 01:09:23 bluelightning: thanks for the hints, sounds like a clean way of doing it. this is for a kirkwood layer btw. Jan 27 01:10:11 icanicant: yeah, I figured... Jan 27 01:10:33 np Jan 27 01:10:34 bluelightning: will let you sleep, cheers :) Jan 27 01:10:35 night all Jan 27 01:15:35 pixman-0.24.2/0008-Generic-C-implementation-of-pixman_blt-with-overlapp.patch isn't found allthough meta-oe is in my bblayers Jan 27 01:20:34 ah sry forgot to git pull meta-oe after oe-core update, works now ;) **** ENDING LOGGING AT Fri Jan 27 02:59:57 2012