**** BEGIN LOGGING AT Mon Jun 16 02:59:59 2014 Jun 16 06:26:21 is there any exmaple for how to write a Makefile to fit yocto/oe build system? for example upstream only provide C source code.. Jun 16 06:43:00 good morning Jun 16 06:43:22 dlan: http://www.yoctoproject.org/docs/1.5.1/dev-manual/dev-manual.html#usingpoky-extend-addpkg-makefile Jun 16 06:46:09 mckoan: thanks, will look Jun 16 08:33:47 hello everybody Jun 16 08:34:08 i am trying to boot my kernel image over tftp Jun 16 08:34:20 receiving a "Bad data CRC" in uboot Jun 16 08:34:44 load address and size are verified to be correct Jun 16 08:37:54 morning all Jun 16 08:52:04 old 2.6 kernel works. new 3.10 from yocto CRC error Jun 16 08:52:33 system is a powerpc Jun 16 08:53:45 petero_: explain details and how did you build the old an the new Jun 16 08:54:06 i did not build the old one, i got that as legacy software Jun 16 08:54:30 i guess even if the new kernel would be misconfigured the CRC should still be ok, right? Jun 16 08:54:44 bad CRC is from uboot Jun 16 08:54:52 the kernel is not even launched yet Jun 16 08:55:29 so based on the fact that i can boot the complete system with TFTP/NFS with old legacy SW i assume there is nothing wrong with TFTP/NFS Jun 16 08:57:15 yocto kernel is gzip compress, like odld working kernel Jun 16 08:57:50 however, i use uboot "U-Boot 2010.06-00073-g54841ab-dirty" Jun 16 09:38:51 solved Jun 16 09:39:10 image load address was too low and the kernel image too big so it overwrote itself Jun 16 10:55:24 I'm getting "Fetcher failure errors " while running bitbake, any clue on what may be issue? Jun 16 10:56:02 the rest of the log will tell you what the failures were Jun 16 10:58:33 it is unable to resolve only certain host address Jun 16 10:59:35 do you need a proxy to reach the internet? Jun 16 11:00:58 no, no issues with internet Jun 16 11:01:34 i suggest just trying again, and seeing if the same names resolve outside of bitbake Jun 16 11:02:17 ok, i will try to resolve outside of bitbake, Thanks **** BEGIN LOGGING AT Mon Jun 16 11:51:49 2014 Jun 16 12:09:01 I'm still fiddling around with the SDK and I get some strange behavior: I once tried setting TOOLCHAIN_HOST_TASK="" and now it stays empty even tough I've set TOOLCHAIN_HOST_TASK back to its default value again. "bitbake myimage -c clean -c cleansstate" somehow doesn't help. Any ideas? Jun 16 12:10:09 And by "it stays empty" I mean the sysroots/host folder contains nothing. The TOOLCHAIN_HOST_TASK is fine when I look at it in bitbake -e Jun 16 12:10:19 So I suppose it's some kind of caching issue? Jun 16 13:13:48 hi all. I have to move a build environment to a new server at another location. The env is 444GB in size which would take way too long to copy. Am I correct in thinking that I can delete the tmp/ folder and copy everything else including the sstate-cache/ directory and PR-Server DB and I will be able to build identical packages on the new server? Or will that not work? Jun 16 13:17:03 hsychla: that should work fine yes Jun 16 13:19:12 bluelightning, great, thanks! Jun 16 13:35:01 "ERROR: QA Issue: Recipe /home/projects/.jenkins/workspace/CCP/CCP_02.xx/POKY_CCP_MINIMAL_LATEST/oe/build/../meta-ccp/recipes-kernel/linux/linux_3.2.bb has PN of "linux" which is in OVERRIDES, this can result in unexpected behaviour.", can this be avoided without renaming the linux recipe? Jun 16 13:40:14 Denwid: it's not advisable... Jun 16 13:41:14 it can be called "linux-", just not "linux" Jun 16 13:43:19 Thanks, bluelightning! Jun 16 13:59:02 hi Jun 16 13:59:36 i have a src_uri question. Jun 16 14:00:07 gotan_: hi, please ask your question Jun 16 14:01:18 I'm forward porting from 1.2 to 1.6, and openjade-native is failing for no reason. After digging into it, it seems that the SRC_URI defined in the recipe is being filtered, such that two patchfiles (which are available) are not copied to the work folder before compilation. Jun 16 14:02:00 so the build fails because the patches are not copied. this is a yocto 1.6 Jun 16 14:02:47 whic btw is also complaining about my FC19 build host, Jun 16 14:02:58 gotan_: that sounds strange, I'm not quite sure how that would happen... Jun 16 14:03:52 gotan_: as of 1.6 that recipe has 5 patches - are the other three there? Jun 16 14:04:55 they are. but just verbalising the problem made me check something - it may be my build environment is broken. Jun 16 14:05:11 it would explain some of the other things. Jun 16 16:21:24 bluelightning: do you have a second? Jun 16 16:39:53 zecke: you are one of the Osmocom guys? Jun 16 16:45:00 volker-: I wear that head from time to time Jun 16 16:59:02 Is there a full listing of the config fragments included with linux-yocto? I'm trying to figure out how to use KERNEL_FEATURES Jun 16 17:00:50 zecke: hi, what's up? Jun 16 17:04:41 bluelightning: Did you consider supporting travis-ci in some way? Jun 16 17:04:54 bluelightning: e.g. by providing a sstate that can be easily used by travis build jobs? Jun 16 17:15:11 zecke: great job :) Jun 16 17:16:49 zecke: I have looked briefly at travis-ci, but I don't have any experience with it Jun 16 17:17:12 zecke: I think we do export our sstate from the autobuilder already; let me just check Jun 16 17:17:51 bluelightning: the main issue is that the build time is 'capped'. So if the base system is provided is pre-built then I could probably use travis-ci Jun 16 17:34:49 zecke: hmm... I can't tell if we are exporting sstate or not Jun 16 17:34:56 at any rate, something to think about... Jun 16 17:42:29 ca-certificates has 20130610 as version... the hash used is also from that date. Jun 16 17:48:39 yocto meta contains recipes-support. Now I try to file a bug but don't find "defects in meta-support" nor anything that sounds like support Jun 16 18:15:46 Idle question: Does anyone happen to know of pseudo-related problems that were pending on me having time to look at things? I might be able to get some time, but I don't remember any outstanding issues I haven't got fixes for in the 1.6 tree. Jun 16 18:15:50 Which is not the same as there not being any. Jun 16 19:19:32 I'm trying to cross compile a driver, but I get an error scripts/mod/modpost: no such file or directory. So I go to the kernel/scripts/mod/ and make modpost. But the I have the error elfconfig.h no such file or directory. Anyone knows how I should fix that? I,m pretty sure copying the elfconfig.h from my host linux is not the way to go Jun 16 19:23:13 sounds like you didn't properly point it to STAGING_DIR_KERNEL Jun 16 19:23:31 our kernel builds put a minimized kernel source tree there specifically for use by out of tree modules Jun 16 19:27:36 @kergoth: where is supposed to be the STAGING_DIR_KERNEL^ Jun 16 19:28:01 I don't understand the question Jun 16 19:28:07 it's a bitbake variable holding the path to the kernel source tree Jun 16 19:28:09 read some recipes Jun 16 19:29:21 kergoth: Ok so if I move the kernel source tree on my host after I ran bitbake I can have issues? Jun 16 19:29:42 no, no, no Jun 16 19:29:53 i just told you, bitbake builds its own kernel, and puts the sources in that directory Jun 16 19:30:01 all you have to do is tell your module build to look there for the kenrel Jun 16 19:32:30 kergoth: That's what I did. I used the command make arch=arm CROSS_COMPILE=/path/to/arm-poky-linux-gnueabi- KERNELDIR=~/yocto/build/tmp/sysroots/nitroen6x/usr/src/kernel/ Jun 16 19:33:24 its extremely bad behavior to be poking at hte poky sysroot from outside of poky, you should be creating a recipe for your kernel module instead Jun 16 19:33:50 that said, scripts/mod/modpost should be in tmp/sysroots/nitrogen6x/usr/src/kernel/ already, is it not? Jun 16 19:34:41 It's not, I only have the source file, modpost.c Jun 16 19:34:51 hello. not sure if right place, but i'm trying to get debian installed on a via amos-820 from scratch. anyone have experience with the board? Jun 16 19:35:25 mtetreault: you should really just create a recipe and use the module bbclass :) Jun 16 19:36:27 kergoth: but if I'm currently working a driver, I will have to re-build my image every time. Jun 16 20:14:46 Can anyone explain what KMETA signifies in kernel recipes? Jun 16 20:15:31 It appears to be different from KMACHINE Jun 16 20:15:42 and both are needed when checking out from a local git repository Jun 16 20:15:47 but I can’t find documentation on what it actually does Jun 16 20:23:18 i assume you read the yocto kernel development manual? Jun 16 20:24:10 I’ve looked through the development manuel and the reference manuel Jun 16 20:24:27 and the kernel lab Jun 16 20:24:40 let me see if there a separate kernel development manuel Jun 16 20:25:00 ah, there it is Jun 16 20:25:01 thanks Jun 16 20:25:27 np Jun 16 20:52:55 * dvhart owes kergoth a drink Jun 16 20:53:22 oh? I won't complain, but why? :) Jun 16 20:53:35 fielding a "read the kernel dev manual" for me ;-) Jun 16 20:53:47 yocti should start doing that I guess Jun 16 20:53:48 dvhart: Error: "should" is not a valid command. Jun 16 20:53:54 there are so many manuels to keep track of… :/ Jun 16 20:54:04 mkeeter, not a criticism of you Jun 16 20:54:09 so far i’ve saved links to quick start, dev, bsp, reference, and kernel dev Jun 16 20:54:12 mkeeter, just a pretty common event here Jun 16 20:54:19 yeah, I can imagine Jun 16 20:54:34 mkeeter, ideas on how we can make that more self discoverable? Jun 16 20:54:36 open to suggestions Jun 16 20:54:58 dvhart, for some reason it’s not well-indexed by google Jun 16 20:55:13 or rather, patches with those particular keywords end up coming up first if you search for ‘yocto KMETA’ Jun 16 20:55:23 hrm Jun 16 20:55:33 so a little SEO to do... Jun 16 20:55:37 I'll let our web folks know Jun 16 20:55:52 yeah, that would be a huge help for discoverability **** ENDING LOGGING AT Tue Jun 17 02:59:58 2014