**** BEGIN LOGGING AT Thu May 02 02:59:58 2013
May 02 04:09:53 <Circuitsoft> Crofton|work - Works for me. Maybe network issues. Point me to a pastebin that works for you, and I'll post it there.
May 02 07:17:56 <mckoan> good morning
May 02 08:40:32 <Net147> how do I install to usr/local/bin? it says Files/directories were installed but not shipped when I try build image with custom recipe
May 02 08:42:23 <bluelightning> Net147: typically you wouldn't do that
May 02 08:42:57 <Net147> any particular reason?
May 02 08:43:02 <bluelightning> Net147: for autotools we set bindir so that /usr/bin is used
May 02 08:43:23 <bluelightning> it's the standard I guess
May 02 08:44:14 <bluelightning> if the software your recipe is building is already installing stuff into /usr/local/bin and you want to keep it that way you can just add that directory to FILES_${PN} within the recipe
May 02 08:44:28 <Net147> bluelightning: thanks
May 02 09:10:28 <rburton> Garibaldi: --module is for building plugins, not libraries
May 02 11:04:07 <Net147> is hdddirect image supposed to include rootfs.img as well in the fat partition in addition to the ext3 root partition?
May 02 11:20:35 <Net147> I guess not...
May 02 11:34:44 * Net147 investigates conflict between bootimg and boot-directdisk
May 02 11:41:55 <Net147> I think I see it..
May 02 12:09:32 <Net147> if two different tasks inside a single image recipe set are executing at same time and set HDDDIR variable, is the HDDDIR variable shared between the two tasks or separate?
May 02 12:11:28 <Net147> it looks like do_bootdirectdisk is setting HDDDIR variable while do_bootimg is executing and overwriting what do_bootimg initially set HDDDIR to
May 02 13:23:53 <Net147> what's this d variable in the bbclass files?
May 02 13:25:06 <bluelightning> Net147: d is the data dictionary, it allows access to variables among other things
May 02 13:25:31 <bluelightning> Net147: the class is defined in bitbake/lib/bb/data_smart.py if you want to look at the code
May 02 13:26:37 <Net147> bluelightning: okay, thanks. I'm slowly working on solving many issues I have found with image generation.
May 02 13:32:00 <Net147> bluelightning: does ${SOMEVAR} in a somefunction() { implicitly use access d?
May 02 13:35:02 <Net147> bluelightning: SOMEVAR = 123 equivalent to d.setVar('SOMEVAR', '123') and ${SOMEVAR} equivalent to d.getVar('SOMEVAR') ?
May 02 13:35:19 <bluelightning> Net147: kind of, it's more a case of d expanding those references
May 02 13:36:35 <bluelightning> Net147: yes
May 02 13:36:57 <bluelightning> Net147: well actually the latter is equivalent to d.getVar('SOMEVAR', True)
May 02 13:37:18 <bluelightning> Net147: that will ensure that any variable references within the value are fully expanded
May 02 13:37:26 <B4gder> roxell: wohoo, fancy hostmask there =)
May 02 13:45:05 <Net147> bluelightning: can functions have arguments?
May 02 13:47:51 <Net147> bluelightning: the problem i'm having is image bbclasses clobbering each other's variables inside the same image recipe
May 02 13:55:02 <Net147> is there a bitbake language reference manual?
May 02 13:58:01 <bluelightning> Net147: there is a bitbake manual, it's in the process of being updated
May 02 13:59:32 <bluelightning> Net147: bitbake functions can't have arguments, no
May 02 14:38:43 <mcfrisk> how to build depend on opengl stuff with cmake? compile is failing at cmake opengl check.
May 02 14:39:22 <rburton> assuming you have depends = virtual/libgl, then your problem is with the cmake script
May 02 14:40:42 <mcfrisk> yes, have that in DEPENDS, trying to compile openscenegraph
May 02 14:51:14 <b737800> hi all
May 02 14:51:51 <b737800> i'v downloaded a toolchain from yocto nightly build serevr
May 02 14:52:38 <b737800> and i foudna .debug directory under the /lib directory with some debug libraries but nothing under /usr/lib
May 02 14:53:20 <b737800> why do we need debug symboles on a SDK ? and why i cannot find teh debug sumbols of the others libraries under /usr/lib ?
May 02 14:57:45 <mcfrisk> is there a good example how to do_configure/do_compile with cmake?
May 02 15:02:44 <bluelightning> mcfrisk: cmake.bbclass should be handling that already if you "inherit cmake" ...
May 02 15:05:42 <mcfrisk> bluelightning: yep, noticed but there are still some conflicts left in openscenegraph, an internal version of openthreads for example
May 02 15:09:36 <b737800> any one for my question regarding debug symbols ?
May 02 16:05:50 <Circuitsoft> Hi - I'm working on a crownbay-based system with Yocto 1.4, Kernel 3.8.4, and getting compile errors on emgd.
May 02 16:06:14 <Circuitsoft> warning: ignoring return value of 'mutex_lock_interruptible', declared with attribute warn_unused_result
May 02 16:18:15 <nitink> Circuitsoft: We have built it many times without issues, give me more details of your issue
May 02 16:19:17 <Circuitsoft> /export/home/aaustin/<customer>/yocto/build/tmp/work/<customer>-poky-linux/linux-yocto/3.8.4+gitAUTOINC+2a6d36e75ca0a121570a389d7bab76ec240cbfda_47aed0c17c1c55988198ad39f86ae88894c8e0a4-r4.1.1/linux/drivers/gpu/drm/emgd/emgd/display/dsp/tnc/dsp_tnc.c:304:2: warning: initialization from incompatible pointer type [enabled by default]
May 02 16:19:34 <Circuitsoft> Same error on several lines and files, all in emgd.
May 02 16:22:55 <Circuitsoft> https://gist.github.com/Circuitsoft/451f09d3f5ccf9419348
May 02 16:27:54 <Circuitsoft> Gist updated with logs
May 02 16:34:18 <nitink> Circuitsoft: what git commits of yocto project layers are you using?
May 02 16:35:17 <Circuitsoft> Poky is 789b2b7
May 02 16:35:32 <Circuitsoft> Meta-intel is b237e2a
May 02 16:36:55 <nitink> Circuitsoft: I am surprised to see your failure, we have validated all this things many times
May 02 16:37:49 <nitink> Circuitsoft: do you nay any local changes?
May 02 16:37:52 <Circuitsoft> I suspect a problem with my bbappend, but I don't know what the problem could be, other than that the config fragments I'm pulling in were originally written for k3.4.38
May 02 16:38:18 <Circuitsoft> No local changes to poky or meta-intel
May 02 16:38:29 <Circuitsoft> I posted my kernel bbappend and build log in the gist above.
May 02 16:38:56 <Circuitsoft> *3.4.36, not 3.4.38
May 02 16:39:10 <nitink> Circuitsoft: BTW you are not using latest 1.4 git HEAD for meta-intel
May 02 16:39:59 <Circuitsoft> Is 6147ba0e better?
May 02 16:40:20 <nitink> Circuitsoft: that is the one validated
May 02 16:40:20 <Circuitsoft> I'm using latest dylan.
May 02 16:42:00 <nitink> Circuitsoft: for poky you are using the latest dylan, but for meta-intel, you are not using the latest dylan,
May 02 16:42:00 <Circuitsoft> Just did a cleansstate. I'll try the build again and see what happens, if something changed after pulling 6147ba
May 02 16:42:42 <Circuitsoft> Kind of odd that I got b237e since I cloned on Tuesday.
May 02 16:44:38 <Circuitsoft> | ERROR: "cfb_fillrect" [drivers/gpu/drm/emgd/emgd.ko] undefined! | ERROR: "cfb_imageblit" [drivers/gpu/drm/emgd/emgd.ko] undefined! | ERROR: "cfb_copyarea" [drivers/gpu/drm/emgd/emgd.ko] undefined!
May 02 16:44:48 <nitink> Circuitsoft: can you also try by removing your ucvideo.cfg from SRC_URI
May 02 16:48:33 <nitink> Circuitsoft: you are mssing this in your kernel config: CONFIG_FB_CFB_COPYAREA
May 02 16:50:57 <nitink> Circuitsoft: your bbappend is removing some of the needed parts from the kernel config
May 02 16:51:25 <Circuitsoft> Oh. How do I make sure I'm only adding and not removing things?
May 02 16:52:02 <Circuitsoft> I figured since the SRC_URI gets appended to, it would still keep anything in the meta-intel bbappend.
May 02 16:52:33 <nitink> Circuitsoft: that depends on kernel's config dependencies
May 02 16:53:06 <nitink> Circuitsoft: Try building the kernel with out your bbappend, and save the kernel's .config file somewhere
May 02 16:53:23 <nitink> Circuitsoft: then check what is happening to .config with your changes
May 02 16:55:12 <nitink> Circuitsoft: as some of the kernel configs are exclusive, they disable other options
May 02 16:56:09 <Circuitsoft> I suppose one thing to ask is why the existing upstream kernel disables input_evdev?
May 02 16:56:47 <Circuitsoft> Doing so seems to break X, especially in the case of a touchscreen.
May 02 16:59:01 <nitink> Circuitsoft: I think that is unrelated thing to this issue
May 02 16:59:34 <Circuitsoft> True, but that is one thing I had to fix with these config chunks.
May 02 17:00:57 <Circuitsoft> Anyway, built with default and captured config, trying my config now.
May 02 17:03:38 <b737800> hello again
May 02 17:04:38 <Circuitsoft> Lotsa big differences, many that I haven't come near touching.
May 02 17:04:53 <Circuitsoft> So, clearly the default configs are not getting pulled in and merged with my fragments.
May 02 17:07:20 <nitink> Circuitsoft: try with an empty .cfg file in SRC_URI, it should give you unchanged .config
May 02 17:10:30 <Circuitsoft> One odd thing, If I comment out most-all of my bbappend, it builds 3.8.4+gitAUTOINC+27b63fdbd25ad1a37bacc05f49a205c150d21779_d9a45e3325030f7bd6f37947a7a0b12da7f602c3-r4.1.1
May 02 17:10:58 <Circuitsoft> If I uncomment all of my bbappend (which specifies same SRCREVs as the meta-crownbay one), it builds 3.8.4+gitAUTOINC+2a6d36e75ca0a121570a389d7bab76ec240cbfda_47aed0c17c1c55988198ad39f86ae88894c8e0a4-r4.1.1
May 02 17:12:52 <Circuitsoft> I'm seeing all the same differences when I reference only an empty .cfg file.
May 02 17:14:23 <nitink> Circuitsoft: can you try "bitbake -f -c cleansstate" for trying each of these
May 02 17:14:37 <nitink> Circuitsoft: just to take any sstate related stuff out
May 02 17:14:56 <Circuitsoft> Is "bitbake linux-yocto -c cleansstate" enough?
May 02 17:15:27 <Circuitsoft> We have an actual shared sstate in use, so I don't want to slow other people's builds too much.
May 02 17:21:30 <Circuitsoft> Same.
May 02 17:21:49 <nitink> Circuitsoft: there must be something in your bbappend to change the git commit ids
May 02 17:22:25 <nitink> Circuitsoft: I think the metal intel does not cover your machine name, so it is getting commits from the poky
May 02 17:22:49 <nitink> Circuitsoft: so keep your bbappend, just disable the SRC_URI changes
May 02 17:44:35 <Circuitsoft> nitink: Even no SRC_URI spec in the bbappend still has huge config changes.
May 02 17:45:07 <nitink> Circuitsoft: humm
May 02 17:45:29 <nitink> Circuitsoft: definitely somethings is missing or extra in your bbappend then
May 02 17:45:53 <nitink> Circuitsoft: you need to fix SRC_URI_crownbay
May 02 17:46:12 <nitink> Circuitsoft: it should of your <customer> stuff
May 02 17:53:10 <Circuitsoft> I figured that still needed to be crownbay since I have KMACHINE_<cust> = "crownbay"
May 02 17:56:31 <Circuitsoft_> Well, that mostly fixed it, but now my configs aren't applied. I tried appending my .cfg files to SRC_URI_<customer> instead of SRC_URI. I'll see how that works in a second.
May 02 17:57:24 <evanp> my automated build machines occassionally bomb with the error: CalledProcessError: Command 'tar -cf - -C /home/blah/blah/sysroot-destdir/ -ps . | tar -xf - -C /home/blah/blah/tmp/sysroots/machinename/' returned non-zero exit status 2 with output tar: ./usr/lib64/random_file.so: Cannot stat: No such file or directory
May 02 17:57:45 <evanp> does that ring any bells for anyone? this is Yocto 1.2.
May 02 17:58:13 <Circuitsoft_> evanp: Any chance somebody else is trying to delete files at the same time tar is trying to add them?
May 02 17:58:15 <bluelightning> evanp: what version of tar is installed on the host?
May 02 18:00:26 <bluelightning> evanp: if it's 1.2.3 or older check if tar-replacement-native got built
May 02 18:00:32 <evanp> bluelightning: looks like 1.22; they are Fedora 13.
May 02 18:00:45 <bluelightning> evanp: there was a bug in tar 1.2.3 or older that did the wrong thing when replacing symlinks
May 02 18:01:00 <bluelightning> not sure if that's what's happening here but it sounds like it could be
May 02 18:01:28 <bluelightning> we're supposed to build tar-replacement-native at the start and thus avoid the issue in the case where the host's tar is too old
May 02 18:01:59 <evanp> bluelightning: I see "package tar-replacement-native-1.26-r1: task do_populate_sysroot_setscene: Succeeded" very far before the error
May 02 18:02:15 <bluelightning> hmm, so it did build...
May 02 18:02:15 <evanp> so, not that I guess
May 02 18:02:39 <bluelightning> are the files it complains about symlinks at all?
May 02 18:06:57 <sgw1> khem: ping regarding gcc4.8 and binutils status
May 02 18:07:37 <evanp> bluelightning: if I'm reading the makefile right, they are...now looking for more direct confirmation
May 02 18:10:16 <sgw1> RP: you done with pulls?  Shall I fire master on AB, will do so if I don't here from you in 10 or so minutes
May 02 19:15:27 <Circuitsoft_> nitink: So, I'm finding the missing configs in tmp/sysroots/crownbay/usr/src/kernel/meta/patches/standard/base/links/kernel-cache/cfg/
May 02 19:15:32 <Circuitsoft_> Where do those come from?
May 02 19:16:04 <nitink> zeddii: Can you take a look at the Circuitsoft_ kernel config issue here
May 02 19:21:28 <zeddii> can you send me an email with all the details ?
May 02 19:21:41 <zeddii> trolling back through IRC and trying to piece it together isn't what I'd like to do :)
May 02 19:21:59 <zeddii> or call it a bug, and put it in bugzilla, that works too. and I can mark it as fixed if it really isn't a problem.
May 02 19:22:00 <Circuitsoft> zedii: Start here: https://gist.github.com/Circuitsoft/451f09d3f5ccf9419348
May 02 19:22:06 * zeddii repeats
May 02 19:22:07 <zeddii> no IRC
May 02 19:22:39 <Circuitsoft> Want to /msg me your email address? I can try to put it all together there.
May 02 19:23:00 <zeddii> nitink, can hook us up too. but I don't mind broadcasting it. bruce.ashfield@windriver.com
May 02 19:23:18 <zeddii> I can fix it with a simple enough configure.
May 02 19:23:33 <zeddii> I'll just work on it from home tonight, and I leave my IRC window here @work
May 02 19:52:18 <kergoth> wtf, bitbake is selecting this recipe as a runtime provider of something that isn't in its PACKAGES or RPROVIDES or PACKAGES_DYNAMIC anywhere, as far as i can see
May 02 20:49:14 <nitink> zeddii: all the details are here: https://gist.github.com/Circuitsoft/451f09d3f5ccf9419348
May 02 20:49:35 <nitink> zeddii: he is adding a bbappend for his new machine based on crownbay
May 02 20:50:01 <nitink> Circuitsoft: you can cc me on the email details: nitin.a.kamble@intel.com
May 02 20:52:41 <Net147> sgw1: I resubmitted the patch. let me know if you have any other issues.
May 02 21:43:27 <Circuitsoft> nitink: zeddii was able to help me through it - I just forwarded the email chain in case you'd like to see the fix.
May 02 21:43:30 <Circuitsoft> Thanks for the introduction.
May 02 22:02:16 <Garibaldi> Question for you all:  Say I have package A that profiles some header /usr/include/foo.h, and I have package B that needs foo.h, do I DEPEND="A-native" or DEPEND="A" or something else?  When I try to build B, it can't seem to find foo.h
May 02 22:02:33 <Garibaldi> err, I bet a need A-dev
May 02 22:08:30 <Garibaldi> or...
May 02 22:08:45 <Garibaldi> how does package separation work for BBCLASSEXTEND = "native" packages?
May 02 22:09:02 <kergoth> DEPENDS is automatically adusted where appropriate. if B depends on A, B-native will depend on A-native.
May 02 22:09:57 <Garibaldi> ah
May 02 22:09:59 <Garibaldi> make sense
May 02 22:11:19 <Garibaldi> so here B is non-native, but needs A-dev to compile
May 02 22:12:31 <Garibaldi> eh, I'm too tired to think straight.  I'll look at this again tomorrow
May 02 22:12:38 <Garibaldi> kergoth: thanks
May 02 22:12:55 <kergoth> A-dev is a runtime binary package, it's not relevant here
May 02 22:13:02 <kergoth> build dependencies are expressed against recipes, in DEPENDS
May 02 22:13:10 <kergoth> DEPENDS += "A" in B.bb is all you should need
May 02 22:13:16 <Garibaldi> ah
May 02 22:15:12 <Garibaldi> so when I'm compiling B where B DEPENDS on A, where would B expect to pick up /usr/include/foo.h?
May 02 22:15:34 <Garibaldi> foo.h provided by A
May 02 22:15:57 <kergoth> in the default search paths . you don't have to do anything as long as it was installed in the correct place
May 02 22:16:21 <kergoth> the files installed by do_install are automaitcally installed into the sysroot, and oe passes the correct args in the CC variable to get the files there
May 02 22:16:43 <Garibaldi> gah, I see the problem
May 02 22:16:47 <kergoth> if it's not working, then likely you just need to get the buildsystem obeying our CC..
May 02 22:16:49 <kergoth> & friends
May 02 22:16:54 <kergoth> or your do_install in A is broken
May 02 22:16:54 <Garibaldi> it's not in /usr/include/foo.h, it's in /usr/include/A/foo.h
May 02 22:17:32 <Garibaldi> would the install of A automatically move things to /usr/include/A/ ?
May 02 22:18:39 <Garibaldi> I'll need to recheck my A recipe to make sure I'm not doing something wrong
May 02 22:19:02 <Garibaldi> kergoth: again, thanks a lot for your help
May 02 22:20:06 <kergoth> you can always CFLAGS += "-I ${STAGING_INCDIR}/A" in B, but i doubt that's the preferred method
May 03 01:55:19 <Circuitsoft> Does anyone know how to add more python standard modules to my build?
May 03 01:56:28 <Circuitsoft> I've tried this python bbappend, but it's not helping: https://gist.github.com/Circuitsoft/ba0e9301f7fec5790595
May 03 01:56:41 <Circuitsoft> The EXTA_OECONF is applied, but the extra modules aren't.
**** ENDING LOGGING AT Fri May 03 02:59:58 2013