**** BEGIN LOGGING AT Tue Sep 22 02:59:59 2015 Sep 22 07:48:50 May I know why do we use LAYERVERSION in layer.conf? Sep 22 07:49:50 so that we can do inter-layer dependencies Sep 22 07:49:52 https://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#var-LAYERVERSION Sep 22 07:49:52 There are layers which have LAYERVERSION implemented like for example in meta, or meta-yocto etc. But at the same time I couldn't find LAYERVERSION in meta-qt5 Sep 22 07:50:42 joshuagl: Couldn't we have inter-layer dependencies with LAYERDEPENDS ? Sep 22 07:52:14 joshuagl: I went through that link, but I couldn't get the real picture. So I thought of asking here. :) Sep 22 07:54:09 sujith_h: LAYERVERSION can be used with LAYERDEPENDS if you want to depend on a specific version of a layer, rather than any version of it Sep 22 07:56:21 joshuagl: I see Sep 22 07:57:32 joshuagl: so which means, every layer can have same or different version, right? Sep 22 07:58:30 joshuagl: I mean I saw that in meta-opembedded: meta-python and meta-networking have LAYERVERSION as "1" Sep 22 08:13:26 yes, layer version are independent Sep 22 08:27:26 morning all Sep 22 08:45:03 morning all Sep 22 12:16:24 * opkg_install_cmd: Cannot install package linux-firmware-iwlwifi-135-6. Sep 22 12:16:24 * opkg_install_pkg: Package run-postinsts md5sum mismatch. Either the opkg or the package index are corrupt. Try 'opkg update'. Sep 22 12:16:45 do rootfs failed. Sep 22 13:35:43 seebs: there? Sep 22 13:49:59 kergoth: https://autobuilder.yoctoproject.org/main/builders/build-appliance/builds/488/steps/BuildImages_1/logs/stdio :/ Sep 22 14:21:56 odd - I did a build core-image-minimal .... Sep 22 14:22:19 then I did a bitbake -c devshell virtual/kernel Sep 22 14:22:32 then needed to checkout the kernel, patch etc etc Sep 22 14:22:46 I would have expected that it did all that to build core-image-minimal .... Sep 22 14:38:09 I am trying to build yocto to run on a wandboard. I have followed the instructions found at http://wiki.wandboard.org/index.php/Getting_started_with_Yocto_on_Wandboard and I am getting to the point where I try to boot from my sdcard Sep 22 14:39:42 I see the U-Boot display screen and after going through the CPU, Board, DRAM info it says: "Can't find PMIC:PFUZE100" and then "initcall sequence 4ff95c90 failed at call 1780300c (err=-19)" Sep 22 14:40:37 googling for err=-19 shows that 19==ENODEV Sep 22 14:41:53 any hints on how to resolve this or what it means? Sep 22 14:58:34 further googling suggests that I need to set MACHINE="wandboard-quad" Sep 22 15:00:13 but bitbake doesn't like this setting and says that I need to set a valid MACHINE in my local.conf or environment Sep 22 15:02:18 RP: huh, sounds like it was a build done without BB_GENERATE_MIRROR_TARBALLS, and the tarball it had to begin with was out of date, so it was removed. I thought it was clean and straightforward to do it that way, but it does mean we end up with no tarball rather than an out of date tarball in a non-BB_GENERATE_MIRROR_TARBALLS situation. If we want to keep around such an out of date tarball, we could presumably write an additional file out, ind Sep 22 15:02:18 icating that it needs updating, for the later build with BB_GENERATE_MIRROR_TARBALLS=1 to use Sep 22 15:02:30 but i haven't had any caffeine yet, just woke up Sep 22 15:02:40 * kergoth ponders Sep 22 15:03:47 the fetching stuff gives me a headache Sep 22 15:06:46 looks like MACHINE="wandboard-quad" isn't supported but MACHINE="wandboard" is. I am trying that now. Sep 22 15:06:59 kergoth: it looks like the main kernel repo broke badly and lost data. We then wiped out the mirror tarballs in a effort to fix it :/ Sep 22 15:08:17 kergoth: I'm only just realising how much breakage there is :( Sep 22 15:08:21 * RP needs zeddii Sep 22 15:17:01 RP: Why is yocto skipping version number 1.9 ? Sep 22 15:17:28 it's not 'skipping'... 2.0 doesn't come after 1.9 Sep 22 15:17:40 https://wiki.yoctoproject.org/wiki/Releases Sep 22 15:17:44 Why is 1.9 not listed here? Sep 22 15:17:59 the 1.x era is done, 2.x has begun.. some of it is marketing, some of it is that so many changes have happened sonce 1.0 that a 2.0 is about time Sep 22 15:20:28 RP: ugh, sorry for my part in that, thought this made sense. https://gist.github.com/c216b46e506a6e5dd7a8 is what i was thinking as an alternate (untested) method of handling the out of date tarball thing. of course, it'd be nice to have better handling of corrupted clones in general.. Sep 22 15:20:31 * kergoth gets food Sep 22 15:28:16 fray: ugh Sep 22 15:28:40 it's just a version number.. we could call it 'fred' and it wouldn't change the contents Sep 22 15:42:16 fray: I'll tell people the decision was made to align more closely with Microsoft. They skipped Windows 9 in favor of 10. Everyone will understand. Sep 22 15:55:04 my recipe linux-wandboard_3.14.18.bb is failing in do_compile Sep 22 15:57:31 the function return_address in /fsl-community-bsp/build/tmp/work-shared/wandboard/kernel-source/arch/arm/kernel/return_address.c is multiply defined Sep 22 16:03:55 Ulf` seems like the 9 is being avoided for release numbers Sep 22 17:21:29 should I be getting build errors when running yocto for the wandboard or do you think my configuration is wrong? Sep 22 17:25:31 who can I ask for help? Sep 22 17:27:02 the crowd in this room. Sep 22 17:27:43 I'm trying to do that lol Sep 22 17:28:02 getting build errors when trying to make a yocto image for the wandboard Sep 22 17:28:31 not sure if that means I am configured improperly or if the code actually has errors in it Sep 22 17:32:41 Mike__NewbNeedsH, what is the error you are getting ? Sep 22 17:33:04 functions are multiply defined Sep 22 17:33:43 try pastebin it Sep 22 17:33:57 the first one was return_address and I fixed that by changing the declaration extern inline void *return_address(..) to extern void *return_address(..) Sep 22 17:34:26 ok. I'll pastebin it Sep 22 17:37:44 Does anyone have some advice or pointers to documentation on how best to handle a single SDK and package source (opkg) for supporting multiple machine types or is this something that's totally unsupported? Sep 22 17:39:26 http://pastebin.com/iPpB9Y1r Sep 22 17:39:33 that is the error I am getting... http://pastebin.com/iPpB9Y1r Sep 22 17:39:47 it is the log file being output by yocto Sep 22 17:50:32 any thoughts about what I am doing wrong? Sep 22 18:08:43 unfortunately no... Sep 22 18:10:28 I'm going to delete the tmp/ directory and start again. I was initially using the wrong machine because the instructions on wandboard.org are incorrect Sep 22 18:31:53 deleting tmp/ and starting over didn't help. I get to the same error as before. return_address is defined a second time in wandboard/kernel-source/arch/arm/include/linux/ftrace.h Sep 22 18:32:22 if I fix that I will undoubtedly come to the errors I put in pastebin Sep 22 18:47:58 Hi,I'm trying to use the ft5x06_ts driver and I'm having issue with the IRQ (on nitrogen6 max) that I'm using. The driver detect that I'm using irq #176 but the interrupt doesn't shows up in /proc/interrupts nor in /proc/irq/xxx Sep 22 19:09:40 "By default, packages produced for the target are marked as being specific to the architecture of the target machine ... However, if the recipe configures the software to be built specific to the target machine (...), then you should mark the packages produced as being machine-specific by adding the following to the recipe: PACKAGE_ARCH = "${MACHINE_ARCH}"" Sep 22 19:10:01 what's the point of adding it if it's already the default? :P Sep 22 19:12:01 MACHINE is the -board-, while the 'architecture of the target machine' is more generic.. Sep 22 19:12:15 one architecture may satisfy many boards.. but the 'MACHINE' is only one board Sep 22 19:13:00 (default in the PACKAGE_ARCH = "${TUNE_ARCH_${DEFAULTTUNE}}" MACHINE_ARCH is something that is specific to that particular board Sep 22 19:14:55 ok, i think i get it. it could probably be put in a clearer way in the manual though. Sep 22 19:14:58 thanks Sep 22 19:16:46 file a bug and it will get some attention Sep 22 19:18:34 "by default, packages apply to any machine with the same architecture as the target machine. when a recipe produces packages that are machine-specific, they should be marked as such by adding the following to the recipe: ..." Sep 22 19:18:38 or something like that perhaps Sep 22 19:19:04 the manual seems to like long sentences ;) Sep 22 20:08:42 Really dumb question. How do I mount a usb thumb drive. It shows up as /dev/sda1 but when I do "mount /dev/sda1 /tmp/usbkey" it says mounting failed: Invalid argument. The directory is there. Sep 22 20:15:28 CharlieDay: check if there's anything at the end of dmesg Sep 22 20:20:25 If I check dmesg after inserting the usb thumb drive it shows up. the model is recognized. write protect is off and it is attached as sda1 Sep 22 20:46:31 Does the meta-linaro-toolchain layer work? With fido? It's not clear to me if it's maintained or not. Sep 22 20:51:38 CharlieDay: i meant after the mount. the arguments look fine at least, provided you didn't mess them up somehow. Sep 22 21:35:38 hey guys i am trying to compile a kernel without bitbake Sep 22 21:36:00 by directly using the sources and running make commands Sep 22 21:36:17 but after i give the cross compiling command, i get the following error Sep 22 21:36:26 make: *** [drivers] Error 2 Sep 22 21:36:38 LD drivers/net/built-in.o Sep 22 21:36:38 make: *** [drivers] Error 2 Sep 22 21:36:58 any ideas on whats going on? Sep 22 21:37:12 none of those lines are the actual error message. if you want help, you'll need to post the entire log somehwere (ideally a pastebin). do *not* paste that here Sep 22 21:37:24 also, this is off topic here. this channel is for use of yocto / oe / bitbake Sep 22 21:40:07 kergoth, not keen on seeing a screen full of crap showing x86 gcc trying to build an ARM kernel? :) Sep 22 21:42:25 kergoth: okay can you tell me how to change the configs files so that bitbake makes a kernel with networking enabled? Sep 22 21:42:55 kergoth: the reason i am compiling from source on my own without bitbake is because i dont know how to tell bitbake to enable networking on the kernel images and swap Sep 22 21:43:53 bitbake -c menuconfig virtual/kernel -> enable the changes you want, bitbake virtual/kernel, will work temporarily. if you want to keep those changes, then run bitbake -c diffconfig virtual/kernel after that and copy the .cfg it gives you into your layer, create a bbappend, and add that .cfg to your SRC_URI Sep 22 21:43:59 fairly sure that's in the yocto project documentation Sep 22 21:44:32 i've never seen a oe/yocto bsp that didn't have networking enabled by default, that seems tremendously unlikely. perhaps you meant some particular network driver? Sep 22 21:44:54 http://www.lick.moe/paste/d4314f8a53b16c25bd59cf96e392bb5e here is the gcc error messages Sep 22 21:47:04 kergoth: i will try those changes out hold on Sep 22 21:48:45 seems like your kernel source was never tested using the option -Werror=date-time and it doesn't compile. You're non-bitbake build is probably using different compiler flags. Sep 22 21:50:07 don't know enough about bitbake to know where overall and kernel specific additional compiler options are set. Sep 22 21:52:14 * paulg would fix the crimes in the source instead... Sep 22 21:52:21 I mean who does that? Sep 22 21:53:47 yeah, fixing the driver would probably be the better thing, but the reason bitbake works while manual doesn't, assuming one is using the same source code, is probably different compiler flags, no? Sep 22 21:55:22 most likely Sep 22 21:56:16 xyzzy42: is there any way to fix this issue, so that i can just compile manually? Sep 22 21:59:15 I'd guess something like adding EXTRA_CFLAGS="-Wno-error=date-time" to the kernel make command. I don't know where you'd look in the bb recipes to find what bitbake is really doing to copy it, but would like to. Sep 22 22:36:46 kergoth: you there? Sep 22 22:37:18 kergoth: i did up to bitbake -c diffconfig virtual/kernel Sep 22 22:37:28 kergoth: but i dont know where this .cfg file is Sep 22 22:41:46 it prints the path to it Sep 22 22:41:51 look more closely at the output Sep 22 22:42:46 kergoth: it said task do_diffconfig does not exist for target virtual/kernel Sep 22 22:43:13 diffconfig should exist for all the kernels, unless you're on a really old version, afaik Sep 22 22:44:54 kergoth: is there some other way to write the configuration settings? Sep 22 22:45:34 kergoth: i got the BSP from intel's website Sep 22 22:55:45 hey if i was compiling the source from scratch on the intel galileo itself Sep 22 22:55:53 would i get the same use flag errors Sep 22 22:56:01 xyzzy42: what do you think? Sep 22 22:57:24 If I have a .bb file that's got "X = foo", and I want to know if X had a different value beforehand, would there be some sort of debug statement that I could add to print it? I.e., in a Makefile one could write $(warning X was $(X)). There a way to do that in a bb file? Sep 22 22:58:55 dasabhi, only thing that should matter is what version of gcc you are using. I think that error is newish, so an older gcc would not trigger it. But does adding the EXTRA_CFLAGS not work? Sep 22 23:01:40 xyzzy42: well my exact compile commands are the following Sep 22 23:01:44 xyzzy42: make oldconfig Sep 22 23:01:49 xyzzy42: make menuconfig Sep 22 23:01:58 xyzzy42: bitbake -e will show full variable historyi nformation on all variables, including every time it was set, appendeded, etc Sep 22 23:02:01 xyzzy42: this is where i tell it shit Sep 22 23:02:07 xyzzy42: bitbake -e in this case Sep 22 23:02:31 at least the bitbake in the past couple yocto project releases will Sep 22 23:02:57 I'm trying daisy now. It seemed like the linaro toolchain layer didn't work with fido Sep 22 23:02:57 xyzzy42: and then i run ARCH=i386 LOCALVERSION= CROSS_COMPILE=i586-poky-linux- make -j8 Sep 22 23:03:39 xyzzy42: and i added the toolchain path to my path variable Sep 22 23:04:08 xyzzy42: how would i add this EXTRA_CFLAGS thing to that make command? like what would it be exactly? Sep 22 23:05:17 dasabhi: Usually you put those variables after the make command, not before. EXTRA_CFLAGS="-Wno-error=date-time" is exactly what I would add Sep 22 23:07:28 xyzzy42: okay i will try it out Sep 22 23:08:50 xyzzy42: its doing something... Sep 22 23:11:58 xyzzy42: HOLLY SHIT IT WORKED Sep 22 23:12:29 xyzzy42: i fucking love you Sep 22 23:12:41 xyzzy42: looks like i didnt need bitbake after all Sep 22 23:14:38 Thanks. if only I could fix what *I'm* doing! Sep 22 23:56:08 So if anyone is willing to bite, I have a gumstix question: I have an old gumstix overo water before the 1GB NAND upgrade (so 512MB NAND) that I'm trying to build an sdk_populated console image for. I have built everything on the host system just fine, and the image will boot on the gumstix if I replace the MLO to an older one. But when I try to upgrade smart or install the package-core-debug-tools, I get a "Fixing recursive fault but reboo Sep 22 23:56:09 t is needed!"...some time pass...then it starts up again and hangs. I *think* I need to create a custom MLO AND u-boot.img under the old overo water config (which I don't know how to do and would appreciate any pointers to a place online I can go to to learn how) After some google, yahoo, and old nabble trolling to no avail, I figured I would check here if anyone had experienced this or knew a work around. Thanks in advance! **** ENDING LOGGING AT Wed Sep 23 02:59:59 2015