**** BEGIN LOGGING AT Fri Nov 11 02:59:57 2011 Nov 11 07:54:07 sgw: ping Nov 11 08:21:27 jama: here but about the head to sleep Nov 11 08:22:18 sgw: ok, it can wait.. I have fix for libsdl issue (will send to ML) and wanted to ask about your circular dependency with that base.bbclass change for subversion-native Nov 11 08:24:20 jama: yes, please resend the libsdl, I will review and test tomorrow. Regarding the circular dependency issue, I had something similar for updating gnu-config with git-native and had to add ASSUME_PROVIDED, I don't remember the details of your base.bbclass change other than it caused problems Nov 11 08:25:55 ok I'll resend that too (as the old version doesn't apply after last bb.data.*Var* changes Nov 11 08:27:17 BTW, I will be queuing the ARM changes for tomorrow (monday pull). Nov 11 08:27:54 those thumb issues? Nov 11 08:28:06 yes Nov 11 08:28:21 Are they still valid, or do you have a different solution Nov 11 08:28:23 do you want me to resend with PR bumps and added aspell? Nov 11 08:28:47 they are still needed to build on armv4t Nov 11 08:29:17 and nobody sent newer pulseaudio or libatomic-ops upgrade.., so yes they are still valid Nov 11 08:29:53 Yes please, that will delay them until Monday probably, unless RP__ pulls directly. Nov 11 08:30:08 I will not do anything with them tonight. Nov 11 08:30:50 ok.. I'll send it all in one bigger pull request Nov 11 08:31:02 JaMa: Ok thanks. Nov 11 09:07:32 morning all Nov 11 09:08:19 morning Nov 11 10:35:48 morning all Nov 11 10:40:56 good morning Nov 11 10:45:01 RP__: morning Nov 11 10:45:12 RP__: did you read my message about fixup? Nov 11 11:08:05 good morning. bluelightning told me yesterday it's not easy to get list of packages to be installed. What about after image was created. Is there a way to list all the packages inside the image without running it? Nov 11 11:26:05 jkprg: there's the manifest post rootfs hook Nov 11 11:26:53 otavio: oh really? Nov 11 11:27:03 otavio: which message? Nov 11 11:29:20 [08:26] < otavio> [23:33:18] RP__: I am getting a failure of operation denied Nov 11 11:29:20 [08:26] < otavio> [23:34:18] RP__: http://paste.debian.net/144630/ Nov 11 11:29:20 [08:26] < otavio> [23:34:37] RP__: do you have any clue about it? Nov 11 11:29:47 RP__: it happens in populate_sysroot in a cross package Nov 11 11:33:05 otavio: something doesn't add up here since fixup_perms is part of do_package, not do_populate_sysroot Nov 11 11:33:13 and do_package doesn't run for a cross recipe Nov 11 11:41:27 RP__: will do a clean build and paste the log for you to look Nov 11 11:54:37 RP__: it fails in do_package sorry Nov 11 11:54:40 RP__: http://paste.debian.net/144753/ Nov 11 12:10:53 otavio: same was happening to me while experimenting with klcc-cross Nov 11 12:11:28 when I was trying to package the cross-recipe Nov 11 12:11:57 and eventualy forcing install in sysroot Nov 11 12:15:49 in fact, it means you're doing something forbidden Nov 11 12:31:45 ant_work: do you remember what you were doing wrong? Nov 11 12:46:38 otavio: iirc it was when I insisted declaring PACKAGES Nov 11 12:47:39 ant_work: humm Nov 11 12:47:46 if it is a cross recipe there are no packages for target Nov 11 12:47:55 (well, klcc is a diferent, border case) Nov 11 12:47:56 Do the arm targets support hardfp ? Nov 11 12:48:27 was thinking about an armv7hl build for a Trimslice Nov 11 12:50:43 otavio: I think the packaging code explodes when trying to access sysroot Nov 11 12:52:02 otavio: are you converting an oe-classic recipe? Nov 11 12:54:17 ant_work: no Nov 11 12:54:32 ant_work: we are adding chicken as compiler Nov 11 12:54:37 ant_work: and it used to work Nov 11 13:11:01 slaine: I believe so... Nov 11 14:01:33 HI. I want to modify qemux86-64 machine for my test. So I decided to do it in my own empty layer to see how things work. I copied machine conf and its includes and specified layers "meta meta-yocto meta-test" in build directory. But I got an error Nothing RPROVIDES 'v86d'. do you know why? thx Nov 11 14:16:53 v86d is defined in MACHINE_ESSENTIAL_EXTRA_RDEPENDS in my test layer but I don't know why BB can't find it in meta/recipes-bsp/v86d/ Nov 11 14:17:49 make sure your bbfiles in your layer.conf is correct Nov 11 14:19:39 kergoth: I don't have any bbfiles in my layer. I thought BB gets BBFILES also from underlying layers Nov 11 14:19:49 no.. Nov 11 14:19:55 layer.conf adds the layer to bbpath and bbfiles Nov 11 14:20:03 every layer must have a layer.conf Nov 11 14:20:07 there's no magic or implicit search here Nov 11 14:20:29 putting a layer in BBLAYERS just means that layer's layer.conf is parsed. its up to the layer.conf to make sure the variables bitbake actually obeys to find its files are defined Nov 11 14:20:42 see all the existing layer.conf files in all the existing layers.. Nov 11 14:26:55 well I specified BBLAYERS but I also have my meta-test/conf/layer.conf that loads all bb files from my "empty" meta-test layer so I don't touch BBFILES :-) Nov 11 14:28:17 I read meta-yocto/conf/layer.conf and it just adds recipes from its layer. So if I don't touch BBFILE it should contain files from underlying layers, right? Nov 11 14:28:40 huh? Nov 11 14:29:08 you created your layer, therefore a file you created touches bbfiles Nov 11 14:29:12 this isn't that hard to understand Nov 11 14:29:15 every layer.conf includes something like this: BBFILES := "${BBFILES} ${LAYERDIR}/recipes-*/*/*.bb Nov 11 14:29:16 read the layer.conf and make sure its correct Nov 11 14:29:31 no, every layer does not do the same thing. it depends on the layout of that layer Nov 11 14:34:24 strange thing is if I don't specify BBFILES in my layer.conf BB still finds task-core-boot.bb and core-image-sato.bb but not v86d... Nov 11 15:06:46 ok. I've found the problem. As I defined new machine but v86d is compatible only with qemux86 … anyway thx for effort Nov 11 15:26:11 Is it possible to override a target in a .bbappend file? Nov 11 15:29:48 ollebull: a target? Nov 11 15:30:35 Well, maybe it's not what it is called, anyway I want to override the do_install_append in a .bb Nov 11 15:31:36 Seems like creating the same label in my .bbappend file doesn't replace the one in the .bb file Nov 11 15:34:33 bluelightning any idea? Nov 11 15:35:51 ollebull: hmm, I'm not sure you can override it, maybe kergoth can suggest something Nov 11 15:37:53 ollebull: _append is an operation, like +=, not a part of the variable name, so one can't override another, its cumulative. you could try overriding the entire do_install with an override, but I think nowadays appends apply to overrides to. could try using an anonymous python snippet, think those run after the appends are applied Nov 11 15:37:59 Ok bluelightning... Maybe there is another way. Of course I could copy the recipe and modify it, but seems like that's seen as an uggly way of doing things. Nov 11 15:38:53 kergoth_ ok Nov 11 15:39:45 think I'll modify the recipe directly then, seems like the easiest way Nov 11 15:58:48 xxiao: yes, in the case of initramfs you have to mount devtmpfs /dev Nov 11 16:03:33 xxiao: see http://cateee.net/lkddb/web-lkddb/DEVTMPFS_MOUNT.html Nov 11 16:08:19 details here: http://comments.gmane.org/gmane.comp.lib.uclibc.buildroot/32038 Nov 11 17:39:50 JaMa: I've just sent out a patch which fixes the opkg permissions issue Nov 11 17:40:02 s/permissions/ownership/ Nov 11 19:38:56 question: where are packages specified for build. For example, when I 'bitbake meta-toolchain-sdk', it is building packages I dont want or need, which I'd like to remove/override. Nov 11 19:39:18 read the recipe. Nov 11 19:39:35 i did Nov 11 19:39:50 for sh4, meta-toolchain-sdk builds libatomics-ops which is required by pulseaudio. libatomics-ops does not build for sh4, but I dont need it anyways Nov 11 19:40:00 i'd like to disable pulseaudio and anything that requires it Nov 11 20:07:55 RP__: great Nov 11 20:08:09 will try it with dbus later tomorrow Nov 11 20:31:42 fray: what stuff do I need to pull into edison to test the ppc64 fix? Nov 11 20:31:53 just a cross prelink update? Nov 11 20:39:53 msm, any update on the sh4 issue I sent to the list? Looks like it is in both oe-core and poky and has a simple fix. Nov 11 20:44:10 mebrown: you need to port this patch from oe to oe-core yocto Nov 11 20:44:13 +@@ -2321,11 +2321,12 @@ Nov 11 20:44:13 commit 3aa8afe97e9cf1340feb9c4442a6ed88b7e32c96 Nov 11 20:44:13 Author: Khem Raj Nov 11 20:44:13 Date: Sun Jan 9 13:36:19 2011 -0800 Nov 11 20:44:14 gcc-4.5: Fix toolchain builds for SH4/SH3 Nov 11 20:44:14 Nov 11 20:44:15 Signed-off-by: Khem Raj Nov 11 20:45:12 Only one line is required for 4.6 Nov 11 20:45:23 EXTRA_OECONF_append_sh4 = " --with-multilib-list= --enable-incomplete-targets " Nov 11 20:45:34 the changes to gcc/config.gcc are present in 4.6 already Nov 11 20:45:49 So, in essence, it is a one line patch for current yocto Nov 11 20:52:30 msm, since yocto has both 4.5 and 4.6, do you want fixes for both? or would fix for 4.6 be sufficient? Nov 11 20:52:53 i'd ask others, im not the expert here Nov 11 20:52:59 submit a patch fixing both Nov 11 20:53:03 and see what the commentary is Nov 11 20:56:40 I pushed commit 7e08dd4765db27f721f1a8c861f8d131a0f7ee7f to https://github.com/superchalupa/oe-core Nov 11 20:56:59 https://github.com/superchalupa/oe-core/commit/7e08dd4765db27f721f1a8c861f8d131a0f7ee7f Nov 11 20:57:44 Where do I send an email? It applies to both yocto and oe-core. If I send to yocto mailing list, will it get picked up for both? Nov 11 20:59:07 standard procedure is to send to oe-core Nov 11 20:59:13 and it gets pushed to yocto trees afterwards Nov 11 20:59:31 yam. yet another mailing list. Nov 11 20:59:40 https://wiki.yoctoproject.org/wiki/Contribution_Guidelines Nov 11 20:59:56 these lists can be fun - because sometimes they only let subscribers send Nov 11 21:16:59 patch sent **** ENDING LOGGING AT Sat Nov 12 02:59:56 2011