**** BEGIN LOGGING AT Wed Sep 24 03:00:00 2014 Sep 24 05:48:57 hi guys Sep 24 05:49:00 I get this error Sep 24 05:49:01 "QtQuick.Controls" is not installed Sep 24 05:49:16 where exactly in qt is that control? Sep 24 05:49:18 which lib? Sep 24 06:37:14 tried making qemu arm image in Build Appliance, i get warning Nothing PROVIDES syslinux ... incompatible with host arm-poky-.... . So there is no bootloader for arm here ? Sep 24 06:42:28 I need some help for in yocto Sep 24 06:42:31 anyone there ? Sep 24 06:44:57 I created a layer for my project and kept kernel patches in that layer Sep 24 06:47:00 I created a layer for my project and kept kernel patch in that. Kernel patches are getting applied. Now i want to run a shell script to apply patch on other component. How to do that ? Sep 24 06:53:44 after I get Hob Found an error, in the Hob cannot build your image top window, I cannot click Edit Image, Open Log, or File a bug (want that one...) buttons Sep 24 06:55:40 and after you click Close on that "hob found and error' window, the top window with buttons is also gone, so , no way to proceed from this really Sep 24 07:24:01 morning all Sep 24 07:34:21 how do i look inside packagegroup ? Sep 24 07:34:35 i'm in Hob, Step 1 of 2 : Edit recipes Sep 24 07:34:47 want to see what them put into packageroup-core-boot Sep 24 07:40:01 gees, must be early .. yocto guys sleeping ? Sep 24 07:45:27 oh wow, nightly-qa-systemd message .. Sep 24 07:52:07 hello everybody Sep 24 07:52:17 I've got a question: I have two recipes for a package with different versions (e.g. mylib_1.0 and mylib_2.0) Sep 24 07:52:53 I want to use a specific version in my image (1.0). how can I do that? Sep 24 07:53:26 I don't want to do it in machine.conf as both versions are for that machine Sep 24 07:53:56 and i just want to create one image with 1.0 and one with 2.0 Sep 24 08:05:10 how much fs space realistically i need for using Yocto? Sep 24 08:11:07 i got an Issue warning ! :P building nativesdk-eglibc-2.19-r0 Sep 24 08:11:24 nscd has relocation in .text Sep 24 08:25:36 hi guys Sep 24 08:27:15 I am trying to register for the Yocto Project Developer Day in Dusseldorf. Do you know if there are no more places left or what's the problem? If I try via linuxfoundation.org, I can't do it, neither via Yocto web-site. Sep 24 08:59:40 ivanstojanovic, is there also a user day ? Sep 24 08:59:54 havent seen any yocto events for october there ... Sep 24 09:00:09 https://www.yoctoproject.org/tools-resources/events/yocto-project-developer-day-elce-2014 Sep 24 09:01:09 I see you are from USA...this is in Europe, Dusseldorf Sep 24 09:01:09 october 16, ah Sep 24 09:01:13 yeap Sep 24 09:01:22 i 'm thinking i'm in october already Sep 24 09:01:30 hey, no i'm in Europe Sep 24 09:01:48 :) sorry...your info says Chicago :) Sep 24 09:02:32 do you think that event is good for starters ? is it worth a trip ? Sep 24 09:02:41 i'm evaluating Yocto for our future projects Sep 24 09:02:54 before now, i was building all these small images from scratch Sep 24 09:04:06 according to the schedule, there are few trainings for the beginners in the forenoon and a bit more advance in the afternoon Sep 24 09:04:53 i should go through some docs and play roud before that, so want to do both Sep 24 09:04:59 begin and advance Sep 24 09:05:11 in the afternoon, two main topics are "toaster" and "coping with build errors" Sep 24 09:05:29 my friend summed up Yocto for me as "just another abstraction layer" Sep 24 09:05:34 i want to see more of it Sep 24 09:06:16 during the ELCE 2014 (13-15. October), there will be a lot of lectures about Yocto, so you might check it as well Sep 24 09:06:35 "coping with build errors" - hmmm... Sep 24 09:06:52 :) Sep 24 09:06:54 I'm trying to build a Gumstix image through Yocto but making life harder by building offline. Most of it seems to work, but I'm seeing the fetcher trying to access git rather than my mirror for prelink-cross. Sep 24 09:06:59 I can't fathom why it's ignoring what appears to be a good wget from the mirror and then doing git branch ..., git remote rm origin, git remote add --mirror=fetch orgin git://... Sep 24 09:07:05 I'm getting no login shells on my virtual console (ctrl+alt+f2), what did I do wrong? Sep 24 09:07:39 Any suggestions on how I debug this? I've got into the python for the fetcher but am struggling to see what happens after the wget succeeds for the git stuff to be necessary. I'm not getting any checksum errors that would cause this and am not sure what to look for next. Sep 24 09:10:10 @stumble: so, you changed prelink recipe and you want it to use your git mirror now? Sep 24 09:12:44 I need to start a getty client that listens to /dev/tty1 right Sep 24 09:14:15 @ivansojanvic: no, I just mirrored everything I thought I needed (then grabbed the stuff I forgot) onto a local server, pointed the local.conf at it using PREMIRRORS_prepend and BB_FETCH_PREMIRRORONLY. Everything seems to be building fine grabbing what it needs from the mirror until it gets to prelink-native which sort of ignores the rule to convert git:// to http:// and grab from the mirror. I've not changed the prelink recipe Sep 24 09:21:24 Crofton: in the matchbox-terminal readme? its so feature-free does it need documentation :) Sep 24 09:24:47 @stumble: you might try changing it manually to use your mirror and see if that will help Sep 24 09:28:59 ah, that's not a bad plan to isolate things. I'll give that a go. Sep 24 10:41:01 ivanstojanovic, hmm, don't see many, just some had yocto mentioned in topics .. Sep 24 10:42:05 where BUIDLDIR is defined at top level ? Sep 24 10:42:32 BUILDDIR variable is defined where in yocto ? Sep 24 13:04:45 think I might have narrowed my problem down a little. anyone know what _contains_ref in bitbake/lib/bb/fetch2/git.py is supposed to do? Sep 24 13:06:16 rburton, hi, if you remember i had trouble with glic with eGtouchD, the manufacturer finaly sent me a correct version (eglibc compatible) and its working. Thank you for your help Sep 24 14:20:06 can i call bb.fatal from a shell function in a recipe? Sep 24 14:23:05 j8: no Sep 24 14:23:08 you can exit 1 Sep 24 14:33:59 hi everybody Sep 24 14:34:14 anoyone played around with qt and qtwayland from yocto? Sep 24 16:15:57 rburton, I have a marketing guy wanting to create compelling demos breathing down my neck Sep 24 16:16:09 I suppuse I need to install xterm Sep 24 17:34:03 hi Sep 24 17:34:55 it seems like BSPs for older versions of Atom processors are old and don't build properly anymore Sep 24 17:35:23 is it "correct" to use newer BSPs to build for older Atom processors? Sep 24 17:36:07 for example does Valley Sand BSP (E38XX) work for Cedar Trail (N26XX)??? Sep 24 18:15:54 RP, zeddii tells me you are talking to him about the basename failure. Sep 24 18:16:15 here is what it looks like for me, with verbose/debug: http://pastebin.com/Sgr2U6Dq Sep 24 18:16:20 paulg: I was asking if it was an initramfs do_rootfs or a main image Sep 24 18:16:33 main image in my case I think. Sep 24 18:16:43 I've got a layer which is building both Sep 24 18:17:25 RP, as an AB member, I'm not sure how I feel about the Architect doing wheelies on a fast looking motorcycle :) Sep 24 18:17:36 the interesting part in this case, which may help solve it, is that I'm using zeddii linux-dev kernel, and yet the log shows it still looking at/for linux-3.14 stuff Sep 24 18:17:38 but, I guess it keeps you sane Sep 24 18:17:58 better wheelies than stoppies. Sep 24 18:18:15 very true Sep 24 18:18:19 Crofton: The picture was an exception rather than every lap if it helps Sep 24 18:18:32 at least not like that one anyhow... Sep 24 18:18:54 paulg: that helps a lot actually Sep 24 18:19:00 sweet! Sep 24 18:19:05 I should have a demo for the YP booth at ELCE Sep 24 18:19:24 Crofton, a wheelie demo? That would be cool. ;) Sep 24 18:19:34 paulg: did you get a ton of warnings some time earlier about conflicting sysroot files? Sep 24 18:19:46 not that I recall. Sep 24 18:19:58 paulg: have a look at the log/cooker/* files Sep 24 18:20:03 ok Sep 24 18:20:06 grep for WARNING Sep 24 18:22:15 bunch of license warnings about LGPL and GPL but nothing about sysroot files. Sep 24 18:22:30 * zeddii calls the license police on paulg Sep 24 18:22:55 not my infraction. I'm just an unfortunate observer. Sep 24 18:23:03 paulg: I'm 99% sure this is from switching the kernel providers around and confusing the sysroot Sep 24 18:23:10 it should have threw warnings though :/ Sep 24 18:24:07 RP, I switched to linux-dev a while ago, and since had successful builds many times since then. Sep 24 18:24:17 this only showed up in the last couple of days. Sep 24 18:24:36 paulg: machine you didn't build for a while? Sep 24 18:24:57 at most, maybe a 3-4 day lapse. Sep 24 18:25:14 there may be a way to tell actually Sep 24 18:25:23 I regularly build generic-x86-64 and corei7 from dvh's intel layer. Sep 24 18:25:28 I just need to remember a command I posted to the mailing list a while ago :/ Sep 24 18:25:43 well, I can tell from my log files I guess. Sep 24 18:29:27 paulg: basically, can you grep the files in tmp/sstate-control and see if you can see any duplicates? Sep 24 18:29:50 hard to tell from the dates of the logs actually; seems I was also getting undiagnosed failures in license_create_manifest prior to this basename fail Sep 24 18:29:50 paulg: I thought I could find the shell magic command to do it but my searching isn't working Sep 24 18:30:00 will go look now... Sep 24 18:30:30 paulg: those files each contain a list of files they installed (ignore directories) Sep 24 18:30:46 paulg: We're looking for overlap where two kernel recipes installed the same files Sep 24 18:30:54 oh, and btw, this configuration enables multilib, in case that is relevant. Sep 24 18:33:48 RP, interesting... I'm only seeing a manifest for linux-yocto Sep 24 18:34:10 zeddii, the dev one should have "-dev" suffix, right? Sep 24 18:34:29 * zeddii nods. Sep 24 18:34:40 manifest-genericx86-64-linux-yocto.packagedata:/home/paul/poky/build/tmp/sysroots/genericx86-64/pkgdata/runtime-reverse/kernel-module-ipt-masquerade Sep 24 18:34:48 there is the line that screws me Sep 24 18:34:57 the dev kernel doesn't build that module. Sep 24 18:37:02 I'll try a force-clean on linux-yocto, in case the lingering turd from pre-conversion to -dev is causing it. Sep 24 18:37:17 there _is_ one there, showing when I switched over to -dev Sep 24 18:37:20 drwxrwxr-x 3 paul paul 4096 Jul 11 09:32 linux-yocto Sep 24 18:37:29 is there a DESCRIPTION_$X or SUMMARY_$X? Sep 24 18:37:33 for split packages Sep 24 18:38:05 paulg: you'll probably have to clean linux-yocto *and* linux-yocto-dev, then it will work Sep 24 18:38:42 oooh. OK. I'll try both, if just linux-yocto doesn't cut it. Sep 24 18:38:51 paulg: what I want to understand is why the code in sstate.bbclass (see the "The recipe is trying to install files into a shared area" part) didn't kicj in Sep 24 18:39:25 paulg: think about this for a minute - when you clean linux-yocto, its going to wipe files in the sysroot that are shared with linux-yocto-dev Sep 24 18:39:41 if you clean them both, the -dev files will get restored from sstate Sep 24 18:40:18 there is an open bug about the fact things can break like this Sep 24 18:40:26 it has been assumed there is warning when it happens though Sep 24 18:41:06 * RP back in a short while Sep 24 18:41:07 ok, so the clean of linux clobbered the 3.14 manifest files. Sep 24 18:42:29 no rush. do_rootfs takes ~20m for this configuration. :( Sep 24 18:49:17 oooh. I get a new error now. Sep 24 18:49:20 | sed: can't read /home/paul/poky/build/tmp/sysroots/genericx86-64/pkgdata/runtime-reverse/kernel-3.17.0-rc6-yocto-standard: No such file or directory Sep 24 18:50:26 cleanall of linux-dev and subsequent kernel rpm pkging and do_rootfs will take another while... Sep 24 18:50:43 at least is has abandoned 3.14-x Sep 24 19:03:18 paulg: a clean is fine, don't need cleanall Sep 24 19:05:31 too late. :) Sep 24 19:07:30 unfortunately that means I'm waiting on do_fetch still. Sep 24 19:11:17 hey I'm new to yocto, where can I find the meaning of the different branches? Sep 24 19:13:29 ah nevermind I found a table of the releases Sep 24 19:35:29 ancky: https://wiki.yoctoproject.org/wiki/Releases? Sep 24 19:37:02 otavio: sigh that would be why the patch didn't apply. my log grep-fu was failing me obviously! Sep 24 19:45:29 RP, NOTE: Tasks Summary: Attempted 5341 tasks of which 5314 didn't need to be rerun and all succeeded. Sep 24 19:45:31 woot! Sep 24 19:56:23 strange, why do we run configure in glibc's do_populate_sysroot? Sep 24 19:56:46 glibc/2.20-r0/temp/log.do_populate_sysroot.38824: configure: error: C compiler cannot create executables Sep 24 19:58:13 whaaaa Sep 24 19:59:14 JaMa|Off: its to generate the site cache files, its not a full configure run Sep 24 19:59:23 JaMa|Off: its actually a very small different configure.ac Sep 24 19:59:45 paulg: great. the question then becomes why the warnings didn't trigger :/ Sep 24 20:00:20 RP: ah thanks, now I also see why it was failing in my setup Sep 24 20:04:58 RP, yeah -- I've NFI why. But in theory we have an easy reproducer now. Build default kernel and then switch to dev one. Sep 24 20:09:29 paulg: right Sep 24 20:10:59 I don't see dvh here or on OFTC; would be interesting to know if that matches how he triggered a similar failure. Sep 24 20:11:41 paulg: I have a suspicion he switched from systemd to udev (or vice versa) Sep 24 20:11:55 paulg: no proof as yet mind Sep 24 20:12:23 the question is how does the build know that this is problematic... Sep 24 20:13:11 presence of manifest files that aren't matching any pkgs to be installed? Sep 24 20:13:40 or are they supposed to be left there, but simply not processed? Sep 24 20:14:48 my bitbake foo is minimal, so I'll leave that to folks who understand it better. At least we understand the problem better and have a workaround now. Sep 24 20:15:20 and I can runtime test my build output again! Thanks for that. Sep 24 20:17:15 I doubt I would have figured out to go sniffing at the manifest files on my own. Sep 24 20:18:09 paulg: just glad we got some handle on it Sep 24 20:18:18 cat * | sort | uniq -d | grep /$ -v was the command I was looking for earlier Sep 24 20:19:10 rburton: thanks! Sep 24 20:27:00 rburton: :) np Sep 24 21:08:36 paulg: FWIW I tried switching kernels on master and I did get a WARNING Sep 24 21:09:07 RP, interesting. Sep 24 21:09:57 as I'd mentioned earlier, I'd switched ages ago (well 1~2 mo) and it was working fine for a while, and then it started failing (w/o warning) Sep 24 21:10:31 tmp/log/cooker/qemux86/20140924202656.log:WARNING: The recipe linux-yocto-dev is trying to install files into a shared area when those files already exist. Those files and their manifest location are: Sep 24 21:51:13 RP, interesting. I did get those warnings on older builds, but then they _stopped_ Sep 24 21:51:19 -------------------- Sep 24 21:51:20 paul@yow-pgortmak-d4:~/poky/build/tmp/log/cooker$ grep 'already exist' */* | grep yocto-dev | sed 's/:.*//' Sep 24 21:51:20 genericx86-64/20140730153224.log Sep 24 21:51:20 genericx86-64/20140730153224.log Sep 24 21:51:20 genericx86-64/20140801145019.log Sep 24 21:51:21 genericx86-64/20140801145019.log Sep 24 21:51:23 genericx86-64/20140822234245.log Sep 24 21:51:25 intel-corei7-64/20140729185640.log Sep 24 21:51:27 paul@yow-pgortmak-d4:~/poky/build/tmp/log/cooker$ Sep 24 21:51:29 -------------- Sep 24 21:51:48 got them in july and august but never for any of my september builds. Go figure. Sep 24 21:52:04 anyway, time to go enjoy what is left of the sun... Sep 24 21:55:47 Sorry. How do I identify myself? Sep 24 21:57:22 Guest27586: /nick myname Sep 24 21:57:43 Test Sep 24 21:58:15 oh you entered as calvin Sep 24 21:58:30 maybe that's a reserved name and you didn't authenticate, pick something else Sep 24 22:00:46 Test. Sorry guys. Sep 24 22:08:57 I have a question. I've added a patch in files directory and applied it by adding "SRC_URI += file://mypatch.patch" to the .bb in parent directory. The build finishes successfully but the patch wasn't applied. build/tmp/work///version/temp/log.do_patch shows this error at the end: fatal: corrupt patch at line 119 / DEBUG: Shell function do_patch finished. How do I diagnose why the fatal error didn't raise a flag to sto Sep 24 22:10:05 Also, is there a way to see more information about the fatal error? I've tried bitbake -D" Sep 24 22:21:49 Calvin_: oh that's not good Sep 24 22:22:15 :( Sep 24 22:22:29 the words "fatal" come from patch itself, but the calling code didn't handle the error code Sep 24 22:22:33 can you file a bug? Sep 24 22:23:09 you should be able to verify the patch is corrupt by applying it manually in the work directory Sep 24 22:23:43 also make sure your patch returned a non-zero exit code when it saw it as corrupt Sep 24 22:23:46 just to be safe.. :) Sep 24 22:23:55 I would love to file a bug and get further help but unfortunately this is on company's code. I'm sorry about that. Let me try applying it manually Sep 24 22:24:02 patch binary, that is Sep 24 22:24:35 Is "patch -p1 < myfile.patch" the correct way to simulate what bitbake does? Sep 24 22:27:30 kergoth: I'll check for "patch" tool's return code after a manual patch attempt Sep 24 22:35:18 Calvin_: close enough Sep 24 22:35:43 The patch was actually successful when I applied it manually... Sep 24 22:35:51 $ patch -p1 < ../mypatch.patch Sep 24 22:35:53 patching file /.c Sep 24 22:35:55 patch unexpectedly ends in middle of line Sep 24 22:35:58 Hunk #9 succeeded at 3303 with fuzz 1 (offset -5 lines). Sep 24 22:36:00 $ echo $? Sep 24 22:36:01 0 Sep 24 22:39:33 huh Sep 24 22:39:43 Looking at the patch, line 119 is the end of the patch and there's no newline at the end. Does a patch need a newline? Sep 24 22:40:05 generally they end in a newline yeah Sep 24 22:40:22 note that typically IIRC its quilt that does the apply, and that behaves slightly differently Sep 24 22:40:23 I'll try that and report back. Thank you for your help Sep 24 22:56:33 rburton: Adding the newline at the end of the patch made quilt happy. The patch was successfully applied via bitbake. I'm still not sure why the failure didn't cause the build to stop, but I'm happy for now. Thank you for your help! Sep 24 22:57:40 good stuff Sep 24 23:08:21 Calvin_: trying to replicate the problem and struggling :( Sep 24 23:10:01 rburton: Actually the successful result was on a small scale "bitbake -D -f -c do_patch ". I'm doing a full build without sstate-cache so I'll report back. I've also had bitbake failure upon a patch failure so I don't understand why the patching didn't stop the build this time. Sep 24 23:10:35 i wouldn't be shocked if quilt had an error situation that didn't result in the right error code being emitted Sep 24 23:11:45 i'm deliberately corrupting a patch and its still applying fine! Sep 24 23:12:08 Try not having a newline at the end :) Sep 24 23:12:15 that's what i've tried Sep 24 23:12:19 for both context and the patch Sep 24 23:12:40 oh maybe emacs is adding one for me :) Sep 24 23:13:03 no Sep 24 23:13:14 You can also try garbage text in the patch to make it fail Sep 24 23:17:33 no joy :/ Sep 24 23:27:46 Calvin_: presumably you can't share the patch Sep 24 23:35:06 rburton: The patch was successfully applied on the full build :) Sep 24 23:35:40 Right. Sorry about that :( I really wish I can Sep 24 23:40:52 rburton: Well, the code will eventually become open source, so I'll submit a bug when it's published Sep 24 23:41:43 I'm going to sign off. Thanks again. Sep 24 23:59:10 Anyone had success with Archlinux and use of yocto? Sep 25 00:03:02 I do not think we specifically test on Arch, but it should be OK as long as it isn't really old Sep 25 00:03:09 is the general answer Sep 25 01:26:37 Hello, I have a question, if I want to add a ssh server or a software tool in a image, then do I need create a layer? **** ENDING LOGGING AT Thu Sep 25 03:00:01 2014