**** BEGIN LOGGING AT Thu Jun 05 02:59:58 2014 Jun 05 06:53:51 Hm. What decides the version of gstreamer libs? Compiling gstreamer_0.10.36 in 1.4/1.6 gives me different "so-versions", and they are newer on 1.4 Jun 05 06:53:58 also, good morning Jun 05 08:08:16 good morning Jun 05 08:08:31 good moaning Jun 05 08:08:35 satisfy_dependencies_for: Cannot satisfy the following dependencies for metergateway: Jun 05 08:08:35 * libllvm3.4-lto * Jun 05 08:08:38 getting that error Jun 05 08:09:00 but I dont any idea where this "*libllvm3.4-lto*" is Jun 05 08:09:06 where does it come from? Jun 05 08:09:11 bizarre Jun 05 08:09:15 do_rootfs fail Jun 05 08:09:41 opkg_install_cmd fail Jun 05 08:09:58 I have a recipy called llvm-mono Jun 05 08:10:04 that builds mono's version of llvm Jun 05 08:10:10 it does not have such a file Jun 05 08:10:39 there is no such library or combenation of lib + app + version + -lto?? anywhere Jun 05 08:25:23 pompomJuice: -c package -f and look into the package and packages-split dir Jun 05 08:25:41 ok Jun 05 08:25:56 maybe I should ask somethig else Jun 05 08:26:00 llvm takes ages to build Jun 05 08:26:14 and it's makefiles cannot detect a build that was alread built Jun 05 08:26:35 how can I make bitbake go from say do_compile to do_install? Jun 05 08:26:43 without passsing do_compile Jun 05 08:26:58 withtout having to insert a exit 0 into do_compile() Jun 05 08:27:31 is there some sort of bitbake syntax that allows sstate manipulation? Jun 05 08:28:05 it would be of tremendous benifit if I could do that Jun 05 08:28:13 force it into a state Jun 05 09:06:30 morning all Jun 05 09:08:55 yebo Jun 05 09:11:00 hi bluelightning Jun 05 09:11:56 koen: https://github.com/Angstrom-distribution/meta-ti/blob/master/classes/sdcard_image.bbclass Jun 05 09:12:00 can I use that? Jun 05 09:12:08 does it just work? Jun 05 09:12:23 seems like we have an older version of that... Jun 05 09:12:33 it stopped working in yocto 1.6 Jun 05 09:12:57 hi woglinde, pompomJuice Jun 05 09:13:37 because uboot files I have no idea about (uEnv.txt) are now missing Jun 05 09:28:14 anyone know where I can find the genext2fs-native stuff? Jun 05 09:28:20 depricated? Jun 05 09:29:42 pompomJuice: genext2fs is no longer used, we have a more reliable method for producing ext2/3/4 images now Jun 05 09:30:06 (basically built-in support in mke2fs) Jun 05 09:30:07 gief Jun 05 09:30:12 ok Jun 05 09:30:15 so I can just replace Jun 05 09:30:23 thought it was doing something special Jun 05 09:30:30 awesome thanks Jun 05 09:31:00 well, special only in that at the time it was one of the only ways to create an ext2 filesystem without being root Jun 05 09:31:16 aah Jun 05 09:31:39 one of genext2fs' serious limitations, apart from being unmaintained upstream, was that it only supports ext2 Jun 05 09:31:52 so you had to then convert the ext2 partition to ext4 if that's what you wanted Jun 05 09:32:17 and of course that means ext4 partitions produced by our build system couldn't take advantage of all ext4 features Jun 05 09:32:23 for small values of "convert" Jun 05 09:32:31 right, what bluelightning said :) Jun 05 09:32:46 trying to replace this command Jun 05 09:32:59 ext4) Jun 05 09:33:00 #genext2fs -z -N $FS_NUM_INODES -b $FS_SIZE_BLOCKS -d ${IMAGE_ROOTFS} ${LOOPDEV_FS} Jun 05 09:33:00 #tune2fs -L ${IMAGE_NAME} -j -O extents,uninit_bg,dir_index ${LOOPDEV_FS} Jun 05 09:33:27 mkfs.ext4 ${LOOPDEV_FS}? Jun 05 09:33:57 not sure what that loop maneuver is Jun 05 09:35:35 pompomJuice: look at how oe_mkext234fs is implemented in meta/classes/image_types.bbclass Jun 05 09:35:47 thanks Jun 05 09:35:48 maybe even use that function itself if appropriate Jun 05 09:35:50 thats what I want Jun 05 09:39:02 yes Jun 05 09:39:04 its gonna work Jun 05 09:39:07 awesome Jun 05 09:48:08 trying to adapt this recipy koen wrote for ti Jun 05 09:48:21 it has some sort of /dev/loop1 tricks going on, Jun 05 09:48:42 sure its some convention driven thing... if only I could get the convention Jun 05 09:51:10 ok im reading up on how this works. Jun 05 10:02:25 the loop stuff shouldn't be needed anymore with the oe_mkfs stuff Jun 05 10:02:31 but I haven't tried it myself Jun 05 11:41:41 aah ok, thanks koen. I will fix. Just need some pointers Jun 05 11:41:48 needed* Jun 05 11:46:47 anyone using the meta-xilinx layer (daisy version)? I got a working system running on linux-xlnx 3.10, but when i switch to linux-xlnx 3.14, i don't even get "Uncompressing Linux...". Same DTS, same boot-loader etc Jun 05 12:09:15 You'll need to update the dts Jun 05 12:09:29 Xilinx is working on getting the dt mapping upstream and this is leading to churn Jun 05 12:09:40 I have 3.14 based on their recipe working Jun 05 12:09:54 do you know what dts bindings have changed since 3.10? Jun 05 12:10:02 read the sample file Jun 05 12:10:05 I forget the details Jun 05 12:10:15 yes, it is annoying, but it is getting better Jun 05 12:10:24 the sample file in meta-xilinx? or is there one in the linux-xlnx tree? Jun 05 12:11:27 I would look at the on ein linux-xlnx Jun 05 12:14:20 thanks Jun 05 12:17:02 Crofton|work, if you use the linux-xlnx_3.14 recipe from meta-xilinx (daisy), it will still rely on the same DTS base (meta-xilinx/conf/machine/boards/common/zynq7-base.dtsi) .. so I guess this means that the 3.14 is broken in daisy? I guess then you do not include this base-DTS in your devicetree, but rather provide something from scratch? Jun 05 12:17:29 right Jun 05 12:17:45 I suppose I should check 3.14 on my zc702 Jun 05 12:18:03 I did build 3.14 outside OE and run it on a zc702 Jun 05 12:18:49 perhaps i should just abandon linux-xlnx altogether and go for a linux-yocto-custom derivative pointing at the linux-xlnx git repo with a DTS from scratch Jun 05 12:19:04 bitch on meta-xilinx Jun 05 12:19:19 sorry :) Jun 05 12:20:13 I'm serious Jun 05 12:20:22 I am annoyed they are reinventing wheels Jun 05 12:21:07 I think there is some merit to things they do,m but they should be working upstream, not making custom inc files to modify behavior Jun 05 12:21:50 * Crofton|work is cranky. The coffee is starting to kick in Jun 05 12:31:45 i see what you mean with curn Crofton|work. A lot of diff 3.10 -> 3.14 and then again a lot of diff -> current Jun 05 12:31:59 yeah Jun 05 12:32:10 this should calm down as stuff goes upstream Jun 05 12:32:27 there were a number of drivers they did that duped existing in kernel drivers Jun 05 12:32:39 perhaps i should just stay at 3.10 then :) next release of meta-xilinx should carry most of these changes i think Jun 05 12:32:43 th emultiple soc vendors writing drivers for the same IP block Jun 05 12:33:13 then you lose the benfit of bug fixes unrelated to zynq Jun 05 12:33:29 chose your poison Jun 05 12:33:53 yeah, i was quite confused to see that my network was working even though I had disabled CONFIG_MACB (GEM) in my config. turns out xemacsps kicked in and took over. not sure why there are multiple kernel drivers for the same mac .. but since MACB is broken in 3.10, i'm happy there are alternatives.. Jun 05 12:34:19 yeah, macb is one driver Jun 05 12:34:27 but xilinx wrot their own Jun 05 12:34:39 but it iwll never go upstream because first in kernel wins Jun 05 12:34:50 xilnx is now working to get the in kernel one running Jun 05 12:35:02 there are lessons for any soc vendors listening :) Jun 05 12:35:12 check the existing drivers before re-inventing the wheel Jun 05 12:35:17 yeh, saw _alot_ of bugfix commits in it when reviewing the logs for >3.10 Jun 05 12:35:50 at least xilinx is merging mainline into the vendor kernel Jun 05 12:35:58 so we are not stuck in some ancient kernel Jun 05 12:36:03 like some other vendors Jun 05 12:36:06 you seem to like xilinx ;-p Jun 05 12:36:18 well, I ahve to use their stuff Jun 05 12:36:27 and parts of xilnx are trying hard Jun 05 12:42:30 jo jkridner Jun 05 14:43:49 ok Jun 05 14:43:58 ${ROOTFS} has been removed? Jun 05 14:44:20 that variable? Jun 05 14:44:24 in image.do_rootfs? Jun 05 14:45:21 in 1.6 Jun 05 15:01:52 pompomJuice: it's only defined when live or vmdk are in IMAGE_FSTYPES Jun 05 15:01:57 I don't think that is different to 1.5 Jun 05 15:02:07 basically Jun 05 15:02:16 where are you trying to use it? Jun 05 15:02:34 I had a foo_sdcard_image.bbclass that refferred to ${ROOTFS} Jun 05 15:02:53 that file inherits from image.bbclass Jun 05 15:03:10 there are also references to ${IMAGE_ROOTFS} Jun 05 15:03:12 referred to in what context? how is it expecting to use it? Jun 05 15:03:16 most are IMAGE_ROOTFS Jun 05 15:03:21 but this one is ROOTFS Jun 05 15:03:21 IMAGE_ROOTFS is always defined Jun 05 15:16:14 ok so it was legacy references to ${ROOTFS} lying around Jun 05 16:07:10 Hi, I am getting this error, when I compiling for raspberrypi. "ERROR: unifdef-native: chrpath command failed with exit code 7:". Can someone share pointers to fix this? Jun 05 19:50:38 ka6sox: ping Jun 05 19:51:21 ka6sox: please add http://downloads.sourceforge.net/project/mikmod/libmikmod/3.3.3/libmikmod-3.3.3.tar.gz to sources.oe.org, I can provide the tarball if you cannot find it anywhere **** ENDING LOGGING AT Fri Jun 06 03:00:00 2014