**** BEGIN LOGGING AT Wed Sep 09 02:59:59 2015 Sep 09 06:19:04 Hi yocto.. In my target some dev-pkgs are installed not all. Sep 09 06:19:35 But i didn't enable dev-pkgs by default in local.conf Sep 09 06:20:14 And there is no new changes in recipes in my image. Sep 09 06:20:37 I m wondering how they came into target. Sep 09 06:20:49 Any ideas? Sep 09 07:28:00 hi guys Sep 09 07:40:54 good morning Sep 09 10:59:45 is there any difference between doing "variable ?= defaultvalue" and "variable ??= defaultvalue" I have seen booth ways of assigning a default value, and I'm wondering which is the difference if any between "?=" and "??=" if any Sep 09 11:15:29 clopez: that's in the documentation - ??= is weaker than ?= Sep 09 11:16:56 i see, thanks :) Sep 09 13:24:09 I have a directory that a recipe (package) creates during initial installation during the image build. But, when I install an update to that package, it removes that directory. How can I mark the directory as a something that should not be removed and the contents are kept? Sep 09 13:24:30 I think I saw that in the manual someplace but cannot find it again. Sep 09 13:58:46 <_valle_> Hi, I have a problem with some includes missing. http://pastebin.com/FSS3CDuy Sep 09 13:59:22 Hi all: What's the best way to configure which interface the default tty console is attached to? I want my image w/ systemd to default to a UART. Sep 09 14:09:09 os there any way to append a value to a variable only if that value is not already appended? or a way to remove duplicate words from a variable? Sep 09 14:25:33 yookid: no lcd console? or just want to have login on serial tty? Sep 09 14:57:04 T0mW: Yes, I just want to be able to log in and get a shell from a serial TTY over UART. Sep 09 14:58:52 Hi folks, anyone have any experience with linux-yocto-custom.bb? Sep 09 14:59:24 I've been banging my head against a "Could not locate meta series" error for a few days Sep 09 15:00:12 Using Yocto 1.8, trying to pull the 3.18.11 kernel from linux-stable.git Sep 09 15:01:59 When it gets to the do_patch step, it calls patchme, and that script seems to need a yocto-ized kernel. Sep 09 15:03:07 | + do_patch | + cd /sandboxes2/yocto_1.8/poky/build/tmp/work-shared/tmru-ct/kernel-source | + patchme tmru-ct | meta_dir set to .meta | ERROR. Could not locate meta series for tmru-ct Sep 09 15:03:16 darthduck: maybe just throw the recipe-so-far into a pastebin? Sep 09 15:03:31 LetoThe2nd: good idea Sep 09 15:04:03 It will take a bit... corporate policy blocked pastebin at the firewall Sep 09 15:04:09 darthduck: my ideas only know three levels. good - excellent - outright fantastic. Sep 09 15:04:14 *coughcough* Sep 09 15:04:18 hehe Sep 09 15:04:26 SCNR Sep 09 15:05:36 Argh... can't ssh to home server... home server down... Sep 09 15:06:09 I suppose I should put the recipe on the mail list at this point Sep 09 15:07:23 oooh, maybe I can use pastebin on my phone... Sep 09 15:13:15 LetoThe2nd: http://pastebin.com/2v1S3aMR Sep 09 15:15:16 LetoThe2nd: I have upgraded your idea to fantastic. :) Sep 09 15:15:30 darthduck: hrhr thanks Sep 09 15:16:09 darthduck: only thing that meet my eye at the moment is the name= directive in SRC_URI.. what do you need it for? Sep 09 15:18:38 LetoThe2nd: likely a cut-and-paste from linux-yocto-custom.bb. There it had "name=machine" Sep 09 15:19:46 I did a search-replace for "machine" with that recipe and put tmru-ct in it's place, figuring that's what was needed for my machine layer. Sep 09 15:19:53 darthduck: its not my core field of expertise, so i'm mostly poking randomly. but basically i'd suggest stripping away everything and then build up later again. Sep 09 15:20:26 darthduck. if you can get me a copy of that recipe, I can debug it for you. (bruce.ashfield@windriver.com). Sep 09 15:20:43 darthduck: and maybe have a look at the kernel.bbclass, as well as trying to remove the linux-yocto.inc Sep 09 15:20:49 LetoThe2nd: I stripped the "name" from the SRC_URI, same result. Sep 09 15:22:38 LetoThe2nd: didn't think about stripping the linux-yocto.inc... that's another excellent idea. Sep 09 15:23:02 zeddii: I'll send you the recipe over email if that's most convenient Sep 09 15:23:43 works for me. I'll launch a build and see what blows up. Sep 09 15:31:46 zeddii: my 2 year old is obsessed with putting 'z' in front of everything... I should try having her say "jedi" when I get home today! Sep 09 15:31:59 heh Sep 09 15:32:22 I have the email. switching to 1.8 and launching a build. Sep 09 15:32:38 * darthduck crossing fingers Sep 09 15:36:24 Is there a step before the patchme script is called that will (or should) create a .meta directory in the source? Sep 09 15:59:46 * darthduck looks at createme script, how it handles no meta branch case... Sep 09 16:04:00 darthduck. I routinely test against korg trees, no meta branch, defconfigs, etc. that's why it is strange yours is breaking. The error messages are more informative in master, but in fido, I need to see it break here .. and it worked fine in my test run. Sep 09 16:05:43 zeddii: good! I'm guessing it has something to do with my environment then. Sep 09 16:07:53 zeddii: running a bitbake -c cleanall linux-yocto-tmru-ct Sep 09 16:13:50 http://errors.yoctoproject.org/Errors/Build/8617/ and http://errors.yoctoproject.org/Errors/Build/8618/ thats world build with clang Sep 09 16:18:06 zeddii: cleanall/rebuild didn't work. I'm going to try changing the machine to qemux86-64 to see if I can reproduce your success... Sep 09 16:19:43 ack'd. We'll figure it out. Sep 09 16:31:26 zeddii: with qemux86-64, the recipe gets past the do_patch step successfully. Only edits I did was to local.conf (changed MACHINE), and added COMPATIBLE_MACHINE_qemux86-64 = "qemux86-64" to the bb recipe. Sep 09 16:33:36 aha. good data point. Sep 09 16:34:01 do you have a machine definition for the one you were using that I can use ? I changed it, simply because I didn't have that machine defined locally. Sep 09 16:35:02 I'll see what I can package up without sending you the whole half-baked bag. :) Sep 09 16:42:39 zeddii: will send a tarball shortly. I had to add a tune file to get avx support, so a file is needed to be added down in meta-intel as well. Sep 09 16:43:16 zeddii: likely there is a better way, but my time was short... Sep 09 16:47:57 np. you should see the crimes I commit in my local layers :) Sep 09 17:20:29 diffconfig task is so very handy Sep 09 17:21:04 should add a recipetool/devtool version which puts the generated fragment in the specified layer and adds it to SRC_URI in a bbappend Sep 09 17:21:06 * kergoth yawns Sep 09 17:23:03 Hmm, should make do_ optional in the task portion of a recipe:task style target Sep 09 17:30:36 I'm getting this - WARNING: .../sstate-control/manifest---cross.populate_sysroot not found Sep 09 17:31:07 anything I can do to "fix" this warning? Sep 09 17:33:24 I'm assuming it's due to http://cgit.openembedded.org/openembedded-core/tree/meta/classes/cross.bbclass#n62 Sep 09 17:56:08 RP: yt ? Sep 09 17:56:16 RP: I have updated https://github.com/kraj/openembedded-core/tree/kraj/systemd-225 Sep 09 17:56:23 is there anything pending on that ? Sep 09 17:56:27 let me know Sep 09 17:56:56 some of the patches from that got merged Sep 09 17:57:11 I have refreshed it on top of latest master Sep 09 17:58:14 https://github.com/kraj/openembedded-core/commit/e4c70967dd05d254bd59bd4f8a49836c3d499bfb is missing in master-next but that can follow up as well Sep 09 18:14:16 the way the wks is found for the new wic image type is just stupid Sep 09 18:14:33 hardcoding the path to be next to hte image filename? how exactly does that work with our layering mechanism? Sep 09 18:14:40 i'm not shoving files into oe-core or creating custom images for everything Sep 09 18:14:57 either use file:// to find it, or at the very least put the path in a variable we can overrdie Sep 09 18:14:59 or better yet, both Sep 09 18:17:02 zeddii: wow... it's working for me now. Sep 09 18:18:11 Before I tarr'ed up the meta-tmru-ct layer, I cleaned up a few files in the files dir. Those files weren't mentioned in the SRC_URI, but apparently something picked them up before. Sep 09 18:18:28 Now that they're gone, do_patch passes Sep 09 18:18:43 There were a few .cfg and .scc files in there that I wasn't using Sep 09 18:27:15 Just following up on my question from earlier -- any ideas on how to get a TTY/shell/login console over UART by default? Sep 09 18:29:13 darthduck. very odd. I'm going to see if I can create some randomness and break it here.' Sep 09 18:32:18 zeddii: I'll svn revert to get back the files and send you a tarball. Sep 09 18:34:51 zeddii: actually, I'll revert them one-at-a-time and see which one causes the failure Sep 09 18:41:51 yookid: we're using core-image-minimal with x86-64, serial console seems to be on by default there. Sep 09 19:31:57 darthduck: Thanks for your response. Is it exposed via USB or UART? I have the console on USB; I want it to work off of a UART peripheral instead. Not sure how to remap it... Sep 09 20:07:13 yookid: we're using a UART, /dev/ttyS0 Sep 09 20:08:27 yookid: not sure if this affects the getty for logins, but to get kernel console output, we have a bit in our kernel command line. Give me a second and I'll dig it out.... Sep 09 20:10:49 yookid: console=ttyS0,115200n8 Sep 09 20:17:00 hey guys, strange problem with my yocto sdk....if I source the env script and run cmake from the command line on a project that requires QT4, it works fine....if the same command is invoked from a script it fails with the following error: Sep 09 20:17:03 " /opt/poky/1.7.2/sysroots/x86_64-pokysdk-linux/usr/bin/qmake reported Sep 09 20:17:03 QT_INSTALL_LIBS as "/opt/poky/1.7.2/sysroots/x86_64-pokysdk-linux/usr/lib" Sep 09 20:17:03 but QtCore could not be found there. Qt is NOT installed correctly for the Sep 09 20:17:03 target build environment. Sep 09 20:17:03 " Sep 09 20:17:49 AFAIK, the env is identical in each instance Sep 09 20:18:40 Hm. It seems like one of my recipes is trying to use the wrong compiler. Any tips for how to troubleshoot this? Sep 09 20:21:18 xulfer: probably some makefile that thinks using "gcc" is always right, you'll have to read the compile log to find out what bit of the build is broken Sep 09 20:21:59 Hm I'll try that. It works on some machines / docker. No idea why it doesn't on this one. Sep 09 20:22:16 I'm sure this will be tons of fun since it's using GYP + Ninja, heh Sep 09 20:23:46 zeddii: The tmru-ct-preempt-rt.scc is the key to the failure. If it's present, and you're performing a clean build, do_patch fails. Sep 09 20:24:48 zeddii: If I bitbake -c cleanall linux-yocto-tmru-ct, remove the .scc file, and re-start the bitbake build, it's fine. Sep 09 20:25:28 rburton: Could this be something I'm doing wrong in Yocto? The first thing I notice is that it's in an i586-poky-linux work directory. Sep 09 20:26:30 Oh wow. I figured it out already. Someone changed the MACHINE line in local.conf Sep 09 20:33:11 khem`: the issue with systemd 225 is the date problem. I haven't had a chance to figure that out :( Sep 09 20:44:49 fixed it....aliases aren't expanded in non-interactive scripts by default....so the alias in the cmake env script does nothing Sep 09 22:34:48 RP: did you read my email ? Sep 09 22:34:56 explaining the date problem Sep 09 22:35:05 and how its happening ? Sep 09 22:37:00 RP: http://lists.openembedded.org/pipermail/openembedded-core/2015-September/110202.html Sep 09 22:39:07 khem`: I hadn't when I replied earlier but I have now. Ross' version of the patch is in testing Sep 09 22:40:14 link ? Sep 09 22:44:55 khem`: http://git.yoctoproject.org/cgit.cgi/poky/commit/?h=master-next&id=d8f47ac5bb8ff6c3e24f46f1dfd0b7f4183b78b9 Sep 10 00:20:39 halstead: how can I make the light theme of git.yp.org as my default Sep 10 00:22:06 khem`: I don't think there is a way currently. You can bookmark that version though. Sep 10 00:23:28 OK I guess I reference it enough so chrome indexes it higher Sep 10 00:23:40 is a viable solution **** ENDING LOGGING AT Thu Sep 10 02:59:58 2015