**** BEGIN LOGGING AT Fri Aug 08 02:59:59 2014 Aug 08 07:32:06 Hi yocto while compiling at-spi2-core_2.12.0.bb file i am getting pastebin error http://pastebin.com/p0k320Su please help me how to rectify this . Aug 08 07:33:05 kindly help me, If any one on the line.. Aug 08 07:36:56 missing gettext? Aug 08 07:45:56 embedd: try adding gettext to the inherit line Aug 08 08:23:27 morning all Aug 08 08:50:45 how do I install qemu image to rootfs system? Aug 08 08:58:24 *I mean to say how do I install qemu image to rootfs image? Aug 08 09:23:39 good morning Aug 08 09:26:46 I am trying to get flexcan support in fsl-community-bsp (daisy), but can't either see the driver (using dmesg) or the devices? Aug 08 09:27:21 BTW I am using a wandboard (imx6) Aug 08 09:28:23 I would appreciate if someone could help Aug 08 09:28:59 ant_work: Thank you very much no it's working. Aug 08 09:29:08 ant_work: Thank you very much now it's working. Aug 08 09:36:15 embed: pls send the patch to the ML Aug 08 09:37:49 ant_work: are you mean .bb file Aug 08 09:39:21 I mean ideally you should do a commit and push the obtained patch Aug 08 09:40:00 ..but I see there is an upgrade -> at-spi2-atk_2.12.1 in master-next Aug 08 09:42:28 sigh..I flushed many mails, that one included Aug 08 09:42:59 yay for flushing! Aug 08 09:43:47 I can't easily reply ;) Aug 08 09:48:20 looking here it could need xz as well.. Aug 08 09:48:22 http://braumeister.org/formula/at-spi2-core Aug 08 09:48:40 JaMa: ^ Aug 08 09:52:13 what? Aug 08 09:52:55 the recipe seems missing inherit gettext Aug 08 09:53:26 rburton: ^ Aug 08 09:53:39 I don't maintain at-spi2-core and it's not even in meta-oe Aug 08 09:53:53 yes, is oe-core Aug 08 09:55:55 fwiw I've found such snippets 'at-spi2-core-devel BuildRequires: gettext' Aug 08 09:56:45 sorry but I can't check here (plus, I did ignore existence of that recipe until this morning;) Aug 08 10:23:35 hello, I got stuck with debugging. I wonder if someone could help me with this issue? http://paste.kde.org/pgsvuolol Aug 08 10:24:24 I believe this would be the culprit: Cannot access memory at address 0x0 Aug 08 10:24:25 [Inferior 1 (process 9079) exited with code 01] Aug 08 10:24:31 not sure why that would be happening. Aug 08 11:16:57 lpapp: sorry, can't help you there. Thanks for your help yesterday, i'm one step closer to my fully customized distro. The next step is to make xmms2 work. I have created a recipe and it builds and runs fine, i can even play music. But there is a problem when reading id3 tags. XMMS2 uses glib to convert from/to utf8 (g_convert function) which fails to convert anything. I suspect that iconv support is missing in glib, but have no clue how to fix this. Any ideas? Aug 08 11:18:21 no idea, I dislike glib/gtk/gnome :-) Aug 08 11:19:02 but you could have a .bbappend for glib, no? Aug 08 11:19:32 auke-: I'm not sure, this would be an ideal question for rburton Aug 08 11:20:16 auke-: have you seen this? http://cgit.openembedded.org/cgit.cgi/openembedded-core/tree/meta/recipes-core/glib-2.0/glib.inc#n18 Aug 08 11:23:24 hi Aug 08 11:24:10 can anybody help me to build linux distribution for odroid - U3 board Aug 08 11:25:03 its BSP is not available on yocto project, Aug 08 11:25:31 how can I build it Aug 08 11:25:40 anybody thr ? Aug 08 11:25:48 you are a newcomer to Yocto, are you Aug 08 11:25:53 please mail me at just4tech83@gmail.com Aug 08 11:26:04 yes this is my firsttime Aug 08 11:26:13 on IRC chat Aug 08 11:26:21 Bryan__: I doubt people will email you, you could start with the quick start manual. Aug 08 11:26:26 but* Aug 08 11:26:35 i have built image for qemuX86 , qemuarm and beagleboard Aug 08 11:26:36 you need to get familiar with the basic concepts. Aug 08 11:26:44 but those BSP are available Aug 08 11:27:21 what is wrong about meta-odroid? Aug 08 11:27:40 https://github.com/timmy-toner/meta-odroid Aug 08 11:27:48 (it is not on the layer indexer, apparently) Aug 08 11:27:59 actually i am using meta-odroid from https://github.com/timmy-toner/meta-odroid/tree/master/recipes-bsp/u-boot-odroid Aug 08 11:28:02 this link Aug 08 11:28:16 its for adroid -U2 Aug 08 11:28:24 though it has started building Aug 08 11:28:25 another: https://github.com/siretart/meta-odroid Aug 08 11:29:08 i think its the same only Aug 08 11:29:59 in its conf file it has support for machine:@NAME: ODROID-U2 (Exynos 4412) Aug 08 11:30:00 another: https://github.com/MarkusEich/meta-odroid Aug 08 11:30:06 (which also has a fork for XU) Aug 08 11:30:09 and and using tune-cortexa8.inc Aug 08 11:30:46 but Odroid U3 use cortex a9 Aug 08 11:30:51 well, if it does not work with those, you need to ask specific questions, what you are struggling with to get u3 working. Aug 08 11:32:08 can u please tell me how can i modify odroid-u2.conf to build for U3 board Aug 08 11:33:26 I think it requires domain specific knowledge. Aug 08 11:35:03 can u please tell me any link or document where i can help to customize BSP Aug 08 11:35:18 *get help Aug 08 11:37:48 Bryan__: http://www.yoctoproject.org/docs/current/bsp-guide/bsp-guide.html Aug 08 11:39:09 lpapp: could eglibc-gconv-* have something to do with it? Aug 08 11:42:29 auke-: not sure. Aug 08 11:43:42 bluelightning: scott may wish to get more information from you here, https://bugzilla.yoctoproject.org/show_bug.cgi?id=6620 Aug 08 11:43:44 Bug 6620: normal, Undecided, ---, scott.m.rifenbark, NEW , Add a quick example about .gdbinit Aug 08 11:51:21 lpapp: yes, that did the trick! I have added eglibc-gconv-iso8859-1 and eglibc-gconv-utf-16. Suppose i want all eglibc-gconv packages, do i have to list them all or can i use a wildcard? Aug 08 11:53:10 not sure, but I would list them explicitly. Aug 08 11:59:02 okay... thanks Aug 08 13:43:14 how is the yocto gdb guru here? Aug 08 14:06:15 JaMa: actually these days it seems like you can do -fuse-ld=bfd on the gcc command line Aug 08 14:07:03 cool Aug 08 14:07:11 ah, figured out. Aug 08 14:15:51 what does Yocto do to build the -dbg packages? It runs some special target like make debug, or how? Aug 08 14:36:55 I am asking this because my dbg package is empty, so something must have gone wrong. Aug 08 15:42:56 is yocto adding -O1 to the builds automatically? I do not see it in our makefiles, yet it is passed to my binary build. Aug 08 15:47:03 lpapp: this? meta/conf/bitbake.conf:export TARGET_LDFLAGS = "-Wl,-O1 ${TARGET_LINK_HASH_STYLE}" Aug 08 15:48:11 ndec: hmm, I see, thanks. Is it really a good idea btw? Aug 08 15:48:42 I would imagine either -O0 or -O2, but -O1 does not seem to be completely good from either directin. Aug 08 15:48:45 direction.* Aug 08 15:50:23 I guess -O0 should be used for the debug package, but that is empty in my case for some reason... I wonder why that can be happening. Got a clue? Aug 08 15:51:47 it's -Wl,O1, so it's the linker O1 optimization, not the compiler. Aug 08 15:52:37 yet, my binary is optimized even with -ggdb. Aug 08 15:52:52 which means I cannot get the value of optimized variables, etc, when debugging with gdbserver. Aug 08 15:53:07 I think the right way would be to figure out why my dbg package is empty. Aug 08 15:53:17 but I do not know what Yocto does for creating those type of packages. :/ Aug 08 15:54:21 the source code and the non striped binaries are put in -dbg. Aug 08 15:54:42 non-stripped is not enough Aug 08 15:54:52 it also has to be -ggdb and -O0 Aug 08 15:55:38 ok, in that case the -dbg package is not what I need since I do not need the source files on the target when using gdbserver Aug 08 15:56:19 I am actually pretty surprised that I even need unstripped -ggdb -O0 with gdbserver... I thought gdbserver is for being able to debug stripped binaries on the target with all the symbols and no-optimization on the host? Aug 08 15:56:23 well, the -dbg have been enough for my uses case. i generally use gdb on the target Aug 08 15:56:39 i haven't used gdb-server with oe. Aug 08 15:56:53 yeah, I understand. We cannot have that luxurious environment here, unfortunately. Aug 08 15:57:45 but my question is important IMHO; shouldn't gdbserver allow me to have a stripped and optimized binary on the target, whereas the host can be bloated with the non-stripped and non-optimized binary in turn? Aug 08 15:58:17 I thought that is pretty much the only use case behind gdbserver :-) Aug 08 15:58:35 (and that it is smaller than its full-blown gdb counterpart) Aug 08 16:01:20 hmm, apparently, I was wrong... the binaries have to match 100% Aug 08 16:01:33 how can I debug why the -dbg package is not populated? Aug 08 16:05:50 "you need to have a copy of the program you want to debug. gdbserver does not need your program's symbol table, so you can strip the program if necessary to save space. GDB on the host system does all the symbol handling." Aug 08 16:14:43 ndec: do you know if the -dbg package is made with zero optimization and -g? Aug 08 16:15:09 there is nothing rebuilt for -gdb. the component is built only once. Aug 08 16:15:48 hmm, that is a serious flaw, I would say. Aug 08 16:16:06 IMHO, the -dbg package should be built with -g and -O0 to be a proper debug package. Aug 08 16:16:22 and the "real" package with -O2 or above and without -g (and stripped for sure). Aug 08 16:16:45 if -dbg works like that, I will not be able to debug any binaries of mine with it ... :-) Aug 08 16:16:49 :-(* Aug 08 16:17:11 is there any way, even if just workaround, to supply extra debug arguments to be built with? Aug 08 16:36:16 I know that some people from Intel work here. We have some bigger issues with the BIOS of the Atom NUC, we also tried the latest BIOS and it makes things worse. Aug 08 16:36:50 Does anyone has a contact to someone from Intel here? Field Application Engineer or someone like that? Aug 08 16:38:11 volker: is it Yocto related? Don't you have contact details for your product? Aug 08 16:38:52 lpapp: no, not yocto related (but we run with yocto on that device). Aug 08 16:39:42 lpapp: kernel crash, usb issues or disk issues with yocto since we updated the bios. with the old bios version it was fine. So I would call it a BIOS issue and not Yocto issue. I just know that some people from Intel work on yocto and are here. Aug 08 16:40:19 (and the old bios version has its own flaw: not able to enable eMMC or only under awkward circumstances which we couldn't streamline yet) Aug 08 16:41:24 volker: cannot you contact the person who gave you the new bios? Aug 08 16:41:42 or contact them via their contact form? Aug 08 16:43:02 lpapp: yes, that is another way we will go. Sometimes people have different contacts, so I thought it doesn't hurt to ask here :) Aug 08 16:43:13 the new BIOS is on the intel.com website ;-) Aug 08 16:43:57 yeah, that is why I would use the contact form. Do you have issues outside Yocto, too? Aug 08 16:44:46 we only run Yocto on the NUC. but there are multiple bugs in the latest bios (to us it seems more as more bugs as with the stock BIOS it came with) Aug 08 16:48:03 ndec: hmm, I was talking nonsense, I am afraid... about the flags, so my issue is just to figure out why the -dbg package is empty. Aug 08 16:49:41 volker: I work for Intel, but unfortunately I'm not much better placed than you are with regard to that side of things Aug 08 16:52:45 ./tmp/work/armv5te-foo-linux-gnueabi/nic-git/6.0-r0/image/usr/bin/foo: setuid ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.14, not stripped Aug 08 16:52:54 hmm, this is interesting, the binary put into the image is non-stripped? Aug 08 16:54:24 bluelightning: that's ok. doesn't hurt to ask :) Aug 08 16:54:31 does anyone else uses NUCs here with Yocto? Aug 08 17:16:55 lpapp: something in your config? Aug 08 17:18:26 nerdboy: what is? Aug 08 17:18:52 your missing symbols? Aug 08 17:19:24 nerdboy: I am not missing symbols. :D Aug 08 17:19:46 empty dbg-pkg comment above? Aug 08 17:19:51 I am missing the -dbg package content, and the strip activity on files in the image/ "sysroot". Aug 08 17:20:00 nerdboy: ah, you mean that, sure. Aug 08 17:20:23 what is in your EXTRA_IMAGE_FEATURES ? Aug 08 17:21:11 I am not building an image currently, just packages. Aug 08 17:21:30 yeah, but it's still part of your config Aug 08 17:22:02 default should include debug-tweaks unless you changed it Aug 08 17:22:22 nerdboy: bitbake -e | grep ^EXTRA_IMAGE_FEATURES Aug 08 17:22:22 EXTRA_IMAGE_FEATURES="debug-tweaks" Aug 08 17:24:44 nerdboy: but actually, since I use gdbserver anyway, I am happy with not having -dbg packages working. Aug 08 17:25:12 it would be nice to get it working, but it is not pressing. Aug 08 17:25:28 let me know if you have an idea that I can double check. Aug 08 17:27:50 INHIBIT_PACKAGE_STRIP and INHIBIT_PACKAGE_DEBUG_SPLIT should both default to zero Aug 08 17:28:41 so unless the package build setup itself is weird, you should have them Aug 08 17:31:28 bitbake + grep for those returns nothing for those. Aug 08 17:31:39 - for those Aug 08 17:31:55 looks like ptest will enable that Aug 08 17:32:41 otherwise it's only enabled for allarch packages Aug 08 17:32:57 then my recipe is broken I guess. Aug 08 17:33:04 or in specific recipes, ie, toolchain stuff, kernel modules Aug 08 17:33:37 that seems kinda weird Aug 08 17:33:57 you don't have ptest enabled, correct? Aug 08 17:34:01 correct Aug 08 17:34:12 then back to weird... Aug 08 17:34:26 oh, ah, we have this in our recipe: INHIBIT_PACKAGE_DEBUG_SPLIT = "1" Aug 08 17:35:32 doh! Aug 08 17:35:37 looks like cargo cult programming to me. Aug 08 17:36:52 I am still surprised bitbake -e | grep ^INHIBIT_PACKAGE_DEBUG_SPLIT did not return it Aug 08 17:37:10 thanks btw! :) Aug 08 17:37:35 * nerdboy is pretty good at questions Aug 08 17:37:44 sometimes even answers Aug 08 17:37:52 my colleague seems to have had issues with it: debugsources.list: No such file or directory (do_package) Aug 08 17:38:00 and that is why he has put it in there... does that sound any sane? Aug 08 17:38:11 I cannot reproduce that debugsources.list issue what-so-ever. Aug 08 17:40:03 * nerdboy successfully created many empty packages with classic... Aug 08 17:40:23 and several wtf moments Aug 08 17:48:52 nerdboy: thanks again, g2g. Aug 08 18:21:49 Did someone mention NUC firmware issue here? Aug 08 18:23:32 sgw_ : ^ Aug 08 18:24:46 Nitin: volker was asking about NUC firmware earlier Aug 08 18:25:15 volker: Nitin is an Intel Engineer who might be able to help you (if you are still around) Aug 08 18:33:24 volker: at this link the BIOS for various NUC platforms is available: https://downloadcenter.intel.com/SearchResult.aspx?lang=&ProductID=3743&ProdId=3743 Aug 08 18:33:57 volker: make sure the BIOS is made your the model of your NUC Aug 08 18:37:42 volker: you may want to look here for support: http://www.intel.com/p/en_US/support/highlights/dsktpboards/db-d54250wyb Aug 08 18:38:40 Nitin: yes, we verified the correct version. The 0030 one seems to be faulty but the latest we found when searching for our board Aug 08 18:38:58 volker: I am not a firmware guy, I work on Yocto Project and will be able to help you out on the Yocto Side, but for the firmware issues you will have to go through the channel available on intel.com Aug 08 18:39:22 volker: http://www.intel.com/p/en_US/support/contactsupport?group=softwareservices Aug 08 18:40:08 Nitin: ok, thanks. thats what I thought. But asking never hurts :) Aug 08 18:40:38 volker: Wish you best ! :) Aug 08 18:41:06 Nitin: yeah, we need that. I don't know what Intel does with the NUC bioses, all different NUC models we had made trouble with the BIOS Aug 08 18:42:32 volker: I am bit surprised with your findings. the NUC family is popular if it is common issue it should have been escalated fast Aug 08 18:42:59 volker: Are you have right settings in the BIOS? may be try resetting all the settings to default Aug 08 18:44:20 volker: sorry for my bad English here, I think I am typing too fast Aug 08 18:45:30 volker: Try going through the support route, they will open a case for your issue Aug 08 19:38:35 Nitin: resetting should be the case when the BIOS is flashed the first time. So, no that is not the issue. It goes down to the kernel level. And I have to disagree, the forum is full with the first versions of the NUC with BIOS issues. You couldn't even install windows on the first ones ;-) Aug 08 20:40:44 * darknighte shares with khem that his legs are sufficiently stretched now Aug 08 20:44:14 darknighte: tmi dude! Aug 08 20:45:57 sgw_: get your mind out of the gutter! :) Aug 08 22:59:13 darknighte: that was a long stretch ;) **** ENDING LOGGING AT Sat Aug 09 02:59:59 2014