**** BEGIN LOGGING AT Fri Jan 06 02:59:57 2012 Jan 06 03:31:35 how do folks do automated build and build version control with oe? is there a doc/howto on that? Jan 06 03:55:28 kmacleod: you can use buildbot jenkins hudson etc. Jan 06 03:55:54 and what do u mean by build version control ? Jan 06 03:58:02 khem, capturing downloaded source/patches, keeping built packages and ensuring they don't rebuild without manual (local) change, integrating local applications into the build process Jan 06 04:00:37 kmacleod: some distro's using oe maintain feeds for packages Jan 06 04:00:46 the biggest problems I'm seeing is that there doesn't seem to be a way to 1) capture the downloaded sources+patches and make sure they are the only ones used in subsequent builds unless someone approves new downloads, and 2) capturing built packages so they don't have to be rebuilt every night for automated builds Jan 06 04:00:58 I dont think there is a single document that would explain all the above Jan 06 04:02:19 you can always set own mirrors Jan 06 04:06:25 and it does do checksums so even if it downloaded the tarball it has to be good otherwise fetcher will complain Jan 06 04:07:01 I've looked into the build/download directories but it doesn't seem obvious how to build a mirror from that Jan 06 04:08:21 have them somewhere on your network under same dir Jan 06 04:09:49 and then set SOURCE_MIRROR_URL = and INHERIT own-mirrors in your local.conf Jan 06 04:14:59 khem, I missed what you said earlier: some distro's are maintaining feeds, which ones? Jan 06 04:19:11 I've looked into MV, WR, and DENX and they all seem focused on providing the "distribution building kit" rather than a distro Jan 06 04:41:45 kmacleod: angstrom e.g. Jan 06 04:42:09 all of you said use oe Jan 06 06:13:08 hello all Jan 06 06:13:26 I m getting this error while building strace..... QA Issue with strace: No GNU_HASH in the elf binary : Jan 06 06:55:33 I am getting uclibc compilation errors ,uclibc version 0.9.30.2.-35-r1 , http://pastebin.com/V23BfkHM Jan 06 06:56:00 khem ,if your awake plz see the uclibc errors Jan 06 07:05:58 erwt: show me the patch you backported Jan 06 07:06:54 I guess its happening after you have backported some patches right Jan 06 07:07:21 khem,ya Jan 06 07:07:49 these were the 2 patches i had applied Jan 06 07:08:22 actually there is a file missing libc/string/strverscmp.c Jan 06 07:08:51 this file strverscmp.c is not there in 0.9.30.2.-r35-1 Jan 06 07:09:15 i unpacked the 0.9.31 source Jan 06 07:10:49 khem, now i get this error http://pastebin.com/TdPrnuq7 Jan 06 07:10:51 ok so you copied the file over Jan 06 07:11:00 ya Jan 06 07:11:39 add libc_hidden_proto(strverscmp) as the end in include/string.h Jan 06 07:11:49 that should fix it Jan 06 07:12:01 I already did that Jan 06 07:14:09 khem http://pastebin.com/Tu1rEc6d Jan 06 07:14:35 you need to add Jan 06 07:14:36 extern int strverscmp (__const char *__s1, __const char *__s2) __THROW __attribute_pure__ __nonnull ((1, 2)); Jan 06 07:14:40 libc_hidden_proto(strverscmp) Jan 06 07:15:17 and add it after line 76 Jan 06 07:15:59 or 156 Jan 06 07:19:12 khem ,again got the same error as http://pastebin.com/TdPrnuq7 Jan 06 07:26:55 khem ,i think only headers are defined not the function strverscmp Jan 06 07:27:46 does libc/string/strverscmp.c has libc_hidden_def(strverscmp) at the end Jan 06 07:29:24 i solved it Jan 06 07:29:51 there was no file strverscmp.c in the package Jan 06 07:30:26 i copied it from 0.9.31 now it suceeds Jan 06 07:30:32 ok Jan 06 07:30:43 khem ,thanks for your help Jan 06 07:30:56 when you are done please submit the patches to 2011.03 branch Jan 06 07:31:03 so others can use it too Jan 06 07:31:06 np Jan 06 07:31:59 khem, ya definitely ,i will ping u when i am ready Jan 06 09:15:47 morning all Jan 06 09:21:49 mornin Jan 06 09:26:51 hi Jin^eLD Jan 06 09:26:57 Jin^eLD: did you get in touch with nitin? Jan 06 09:28:47 bluelightning: yes, I did, thanks Jan 06 09:28:54 cool Jan 06 09:29:16 he's in the pacific time zone btw Jan 06 09:49:26 gm Jan 06 09:49:46 hey likewise Jan 06 09:50:52 Jin^eLD: hey, how are you? Jan 06 09:51:32 fine, thanks, lots of work Jan 06 09:51:35 you? Jan 06 10:24:35 why were the ncurses .so devel files moved to /lib? this directory is not not searched by ld and the change breaks build Jan 06 10:43:34 hey guies i have compiled a x11-image using uclibc ,but my task fails at the last task package x11-image-1.0-r0: task do_rootfs: Failed Jan 06 10:44:05 can anyone tell me whts the problem with the last task http://pastebin.com/APxke9Dc Jan 06 10:45:09 Jin^eLD: making myself angry over a i.mx53 quick start board Jan 06 10:45:25 :) Jan 06 10:49:07 I have compiled the kernal for at91sam9260k, When loading I am geting error," devtmpfs: error mounting -2 ". pls help me! Jan 06 12:26:07 Basith whats your kernel version? Jan 06 13:28:34 can i disable the license check in my meta-data, or is there a LICENSE which does not require LIC_FILES_CHKSUM? Jan 06 13:29:25 likewise: afaik "CLOSED" Jan 06 13:29:51 which would probably be a hack if its a meta package... Jan 06 13:31:19 Jin^eLD: I have some old recipies that have file:// URLs, but LIC_FILES_CHKSUM does not work on these files. Jan 06 13:34:38 likewise: LICENSE = "CLOSED" will not ask for a checksum, afaik its used for binary/firmware stuff Jan 06 13:35:02 Jin^eLD: yes, seems to work. Jan 06 13:35:04 Jin^eLD: tnx Jan 06 13:35:21 Jin^eLD: we should have the counter part: public domain Jan 06 13:35:41 or OPEN :) Jan 06 13:35:56 :) Jan 06 13:36:10 well, another trick that I do sometimes to get around the license stuff is LIC_FILES_CHKSUM = "file://${COMMON_LICENSE_DIR}/GPL-2.0;md5=801f80980d171dd6425610833a22dbe6" Jan 06 13:36:49 surely this is not what was intended by the LIC_FILES_CHKSUM feature, but the world is not perfect ;> Jan 06 13:44:36 Jin^eLD: well, the build system has to rely on what we put into the metadata anyway (LICENSE is leading). Jan 06 13:45:04 Jin^eLD: So any pointer to a license should suffice. LIC_FILES_CHKSUM is meant to track license *changes* AFAIK. Jan 06 13:46:10 likewise: yes, but for meta packages this is quite a pain Jan 06 13:46:21 because LIC_FILES_CHKSUM seems to be mandatory Jan 06 14:09:57 Jin^eLD: I've thought about trying to let it be set to N/A or similar for recipes that don't install any files Jan 06 14:10:09 the latter would need to be verified though Jan 06 14:10:27 bluelightning: it would be cool to have at least some accepted workaround Jan 06 14:35:59 woglinde, hi. Is "EXTRA_OMAKE..." not typo in the https://github.com/woglinde/meta-java/commit/7b845e3f4de24678fffd1d2d748cbcf5ca3b64d9 Jan 06 14:36:01 ? Jan 06 14:39:29 jekhor args thanks Jan 06 14:39:37 it build anyway Jan 06 14:40:48 :) Jan 06 14:41:11 hm okay will fix it now Jan 06 14:41:47 it builds if you have ecj & classpath in your host system :) Jan 06 14:42:50 jekhor and I wondered why I didnt saw the extra parameters Jan 06 14:49:08 jekhor fix pushed Jan 06 15:10:26 can anyone tell me how can i disable bluez4 package during compilation Jan 06 15:15:18 remove bt feature from your machine dev Jan 06 15:19:14 woglinde, u mean OEBASE_Uclibc/openembedded/conf/machine/my .conf Jan 06 15:29:13 woglinde, a bit specific Jan 06 15:37:03 Hi, I want to prevent generation of locales ipk Jan 06 15:37:23 It takes extra time and I want to prevent it Jan 06 15:37:46 Is ENABLE_BINARY_LOCALE_GENERATION variable true variable to prevent locale generation? Jan 06 15:39:03 erwt search your machine file Jan 06 15:39:10 dont know whats your machine is Jan 06 15:39:29 mgundes_ yes Jan 06 15:39:34 set it to 0 Jan 06 15:39:52 ok thanks Jan 06 15:44:19 Can someone try to build udisks on oe-c Jan 06 15:44:26 Can someone try to build udisks on oe-core/meta-oe? Jan 06 15:51:46 otavio: sure Jan 06 15:51:55 otavio: is yesterdays snapshot ok? Jan 06 15:52:09 otavio: BTW: https://www.freescale.com/webapp/Download?colCode=IMX53_LINUX_1109_BSP_PATCH Jan 06 15:52:39 otavio: I'm merging the Freescale 2011.12 patches into my (newly cloned) GIT over the next week. Jan 06 15:54:28 otavio: ping Jan 06 15:54:30 otavio: it built Jan 06 15:55:05 likewise: version 1.0.4? Jan 06 15:55:07 yes Jan 06 15:55:11 humm Jan 06 15:55:19 * otavio will check why Jan 06 15:55:19 got issues with that version? Jan 06 15:55:23 yes Jan 06 15:55:25 03Dongxiao Xu  07master * r8ee9c6cc47 10bitbake.git/lib/bb/ (command.py cooker.py): Jan 06 15:55:26 command.py: add initCooker API Jan 06 15:55:26 initCooker is to set the cooker to the initial state with nothing Jan 06 15:55:26 parsed. Jan 06 15:55:26 Signed-off-by: Dongxiao Xu Jan 06 15:55:26 Signed-off-by: Richard Purdie Jan 06 15:55:27 03Dongxiao Xu  07master * r5124351d6c 10bitbake.git/lib/bb/server/xmlrpc.py: Jan 06 15:55:27 xmlrpc: Change BitbakeServerInfo init function Jan 06 15:55:27 Pass host and port to BitbakeServerInfo class instead of the "server" Jan 06 15:55:28 instance. With this change, GUI can connect with server individually Jan 06 15:55:28 by host address and port. Jan 06 15:55:28 Signed-off-by: Dongxiao Xu Jan 06 15:55:29 Signed-off-by: Richard Purdie Jan 06 15:55:29 03Dongxiao Xu  07master * r92fec2788e 10bitbake.git/lib/bb/cooker.py: (log message trimmed) Jan 06 15:55:31 cooker: remove command import in cooker.py Jan 06 15:55:31 There is no direct use of command in cooker.py, and it is using Jan 06 15:56:22 bitbake: add a new option "--server-only" Jan 06 15:56:22 Create a new option "--server-only" for bitbake command, which allows Jan 06 15:56:22 bitbake runs as a server, and let frontend connect the server itself. Jan 06 15:56:22 "--server-only" should work with "-t xmlrpc", or bitbake will exit. Jan 06 15:56:22 bitbake --server-only -t xmlrpc will print out the server address and Jan 06 15:56:23 port information. Jan 06 15:56:24 03Dongxiao Xu  07master * r3ef73dee8b 10bitbake.git/lib/bb/command.py: Jan 06 15:56:24 (13 lines omitted) Jan 06 15:56:36 likewise: did you the the changes I've done in my tree? Jan 06 15:56:51 likewise: I did a good amount of clean up on the meta-freescale repo Jan 06 15:57:11 woglinde, even I have added ENABLE_BINARY_LOCALE_GENERATION to my local.conf, bitbake still generates locale packages Jan 06 15:57:27 http://pastebin.com/ATPjDDbA Jan 06 15:57:37 otavio: I threw away my repo, recloned it yesterday Jan 06 15:57:40 from you Jan 06 15:57:42 I copied my local conf and generated packages outputs to there Jan 06 16:00:43 likewise: I didn't check the changes from 2011.12 yet Jan 06 16:00:49 likewise: many new things? Jan 06 16:02:23 otavio: see PDF link in PM Jan 06 16:04:13 otavio: my biggest concern is that it is impossible to build the stuff that is on the supplied SD Card (Ubuntu based) from source. Sources do not match the versions on the SD card, and much is missing. Gstreamer plug-in performance and codecs are different etc. Jan 06 16:04:45 likewise: yes; I fully agree Jan 06 16:05:04 likewise: even worse, if you check the git repository doesn't has those patches Jan 06 16:05:16 otavio: exactly, see my email (you on CC) Jan 06 16:05:39 otavio: I'm going to also take the LTIB route, and see if I can spot differences Jan 06 16:08:22 otavio: on IMX53? Jan 06 16:08:40 likewise: I'll give a call to my Freescale contact and see if he can talk to his director to speed up git pushing of changes Jan 06 16:08:48 likewise: yes Jan 06 16:09:16 likewise: 2.6.35 also doesn't works for my need as I need systemd and thus I need 2.6.39 or newer Jan 06 16:10:02 You might want to check the Linaro kernel for imx53, it's fully mainline (but you lose all the binary-infested drivers)... Jan 06 16:21:11 likewise: Ok; call done ;-) Jan 06 17:18:03 hi, I keep having that: Jan 06 17:18:05 | /home/gnutoo/embedded/oe/oe-core/angstrom/oetmps/angstrom/work-shared/gcc-4.5-r45+svnr181733/gcc-4_5-branch/gcc/cp/cfns.gperf:101:1: error: 'gnu_inline' attribute present on 'libc_name_p' Jan 06 17:18:24 it's gcc-cross Jan 06 17:18:34 what did I forget to cleansstate? Jan 06 17:18:41 or maybe I'll retry with one bbthread Jan 06 17:20:53 hmmm gperf Jan 06 17:21:06 I've cleaned the kernel libc headers already Jan 06 17:31:14 I'll clean gperf-native and retry Jan 06 17:49:48 hi Jan 06 17:50:21 gi Jan 06 17:51:10 I'm trying to get a kernel image with included initramfs-initfs, but I haven't fully figured out the post-processing of the resulting initramfs*.ipk file Jan 06 17:51:35 i suppose i cannot include that into the kernel as is Jan 06 17:51:44 do i have to unpack it Jan 06 17:51:46 ? Jan 06 17:52:04 or set a different option in the initramfs .bb Jan 06 17:52:23 hm? Jan 06 17:52:51 its maybee possible Jan 06 17:52:52 i have a kernel and bitbaked an initramfs Jan 06 17:53:06 but didnt looked it yet up Jan 06 17:53:07 now i want to include it with menuconfig Jan 06 17:53:14 :p ;) Jan 06 17:53:36 you have to build kernel again with initrd file poiting to the generated initrd Jan 06 17:53:41 and im not sure the path to the ipk is the right thing to go into the config Jan 06 17:53:54 I know someone (on oe-core) was looking into it.. but I don't know what resolution, if any they had for the kernel build.. Jan 06 17:54:17 fray the initramfs image in oe-core needs to be fixed anyway Jan 06 17:54:35 ya... I know it's iffy.. Jan 06 17:55:28 so i shall try and error (im not flashing to any chips so should be np) Jan 06 17:55:46 what I (vaguely) remember being discussed is initramfs being an image.. and how a kernel can require an image.. Jan 06 17:56:06 yes i can select the path to an image in menuconfig Jan 06 17:56:18 but i douobt .ipk is an image format Jan 06 17:56:23 ;) Jan 06 17:57:06 but perhaps it contains one *whistle* Jan 06 18:01:59 nope it doesnt it contains some tar.gz with an initscript Jan 06 18:02:01 -.- Jan 06 18:03:08 what would be an acceptable image format to include in a kernel config? Jan 06 18:03:23 i mean for the initramfs Jan 06 18:03:40 cpio? Jan 06 18:04:02 Does way to define arch-specific SRC_URIs exists in OE? Jan 06 18:07:55 still the same error.... Jan 06 18:07:56 sigh Jan 06 18:08:03 I'll recompile from scratch Jan 06 18:11:10 yeah cpio just realised i baked the wrong recipe, initramfs-image.bb in images/ subfolder actually bakes an image -.- Jan 06 18:32:44 ok so i made a menuconfig and tried to recompile with bitbake -f -c compile virtual kernel, but it seems the initramfs is ignored as the resulting kernel is about 1 mb (.,.) and the initramfs is 7 mb Jan 06 18:32:56 virtual/kernekl Jan 06 18:33:57 is that not the right way to compile after making a new config? Jan 06 18:58:44 same error Jan 06 18:58:53 | /home/gnutoo/embedded/oe/oe-core/angstrom/oetmps/angstrom/work-shared/gcc-4.5-r45+svnr181733/gcc-4_5-branch/gcc/cp/cfns.gperf:101:1: error: 'gnu_inline' attribute present on 'libc_name_p' Jan 06 18:59:00 I've to wait for khem I guess Jan 06 18:59:01 bbl Jan 06 19:00:12 denix: did you open a channel yet? if not, ##oe-ev-board Jan 06 19:01:47 bitbake menuconfig doesnt work (or saves the file to some occult location where it isnt found), got to edit the config manually Jan 06 19:02:57 and even that leaves me with exactly the same kernel as before Jan 06 19:04:34 any tips? Jan 06 19:06:59 re Jan 06 19:07:05 ah, here he is Jan 06 19:07:09 haayZ^^ Jan 06 19:07:27 yeah sorry, had to relocate to get power and net Jan 06 19:07:43 np Jan 06 19:07:52 see mail for channel Jan 06 19:08:04 or join #oe-board ;) Jan 06 19:12:51 cs_nbp: try bitbake -f -c compile virtual/kernel; bitbake virtual/kernel Jan 06 19:13:22 cs_nbp: not sure, but after compile, the second command makes sure that all subsequent tasks also run (after just compiling the kernel). Jan 06 19:13:36 cs_nbp: I mean, I am not sure if that integrates the initramfs Jan 06 19:13:47 gotta go now. Jan 06 19:13:49 bye all Jan 06 19:22:02 he mickeyl hny Jan 06 19:35:15 happy new year to you and the family woglinde_ ! Jan 06 19:47:46 anyone else seen issues with libiconv_1.14 when building? Jan 06 19:48:15 no Jan 06 19:48:36 I'm getting "| collect2: ld returned 1 exit status Jan 06 19:48:55 not here Jan 06 19:49:08 I got it build with gcc and uclibc Jan 06 19:50:17 when bugger! Jan 06 19:50:19 -when Jan 06 19:51:16 today Jan 06 19:52:07 I meant ignore "when" Jan 06 19:52:10 still Jan 06 19:53:32 ? Jan 06 19:53:52 http://pastebin.com/itRadPJj Jan 06 19:53:57 that's basically what I'm looking at Jan 06 19:55:06 everything's been running fine up to here (save for some site mirror configs that had to be readjusted) Jan 06 19:55:47 hm armv5 Jan 06 19:55:50 I used armv7 Jan 06 19:56:38 well, it's a slightly older arm, heh Jan 06 19:59:10 yeah it shouldnt have to do something with it Jan 06 19:59:14 but it somehow has Jan 06 19:59:38 could you try to build with machine qemu-arm Jan 06 19:59:38 which is weird because I know others have compiled successfuly Jan 06 19:59:45 well, shit, not recently though Jan 06 20:00:05 I guess Jan 06 20:02:38 what's a good test target? Jan 06 20:03:00 qemuarm Jan 06 20:03:11 that's the specified machine Jan 06 20:03:18 so I bitbake qemuarm? Jan 06 20:04:24 let's try this one.. Jan 06 20:04:30 no Jan 06 20:04:37 set machine qemuarm Jan 06 20:04:42 yeah, it is Jan 06 20:05:24 ok, it's going for qemuarm Jan 06 20:05:26 armv5 Jan 06 20:05:36 lessee how it goes. Jan 06 20:10:56 building core-image-minimal Jan 06 20:11:13 try building libiconv only Jan 06 20:11:21 should be much faster to get there Jan 06 20:12:04 bitbake libiconv? Jan 06 20:13:02 yes Jan 06 20:13:25 oh goddamnit more broken kernel.org mirrors. Jan 06 20:22:28 lovely. So I have to use kernel.org proper for part of the process, because mirrors are outdated, and then switch to a mirror because kernel.org is missing files. Jan 06 21:16:43 woglinde_: Same issue with qemuarm Jan 06 21:16:51 strange Jan 06 21:17:03 you are on eglibc right= Jan 06 21:17:05 ? Jan 06 21:17:11 yeah Jan 06 21:17:19 the previous was uclibc Jan 06 21:17:25 but qemuarm was eglibc Jan 06 21:17:40 strange Jan 06 21:17:51 can you try a armv7 target like beagleboard? Jan 06 21:18:02 it fails on both when building for armv5te Jan 06 21:18:05 I could I guess Jan 06 21:21:04 hold on, gotta add the layer for beagleboard Jan 06 21:21:29 I think Jan 06 21:21:34 hm I will be in bed Jan 06 21:21:38 when its ready Jan 06 21:21:46 but muriani write it ot the oe-core ml Jan 06 21:21:53 ok **** ENDING LOGGING AT Fri Jan 06 22:06:39 2012 **** BEGIN LOGGING AT Fri Jan 06 22:07:27 2012 **** ENDING LOGGING AT Fri Jan 06 22:08:09 2012 **** BEGIN LOGGING AT Fri Jan 06 22:08:41 2012 Jan 06 23:23:54 btw theres a working booting full-of-modules 3.1 kernel for the netbook-pro, even prepacked in a nBkProOs.img Jan 06 23:58:38 cs_nbp: o really? Jan 06 23:58:52 with the xepson stuff? Jan 07 00:15:11 muriani jumped on the netbook-pro 3.1 kernel train :D Jan 07 01:25:57 bbZ **** ENDING LOGGING AT Sat Jan 07 02:59:57 2012