**** BEGIN LOGGING AT Thu May 26 02:59:58 2016 May 26 06:57:10 good morning May 26 07:06:26 good morning to you too , mckoan May 26 09:21:14 hi I was trying to build nginx package using local source code May 26 09:21:34 so I have replaced its SRC_URI url with May 26 09:21:35 ${COREBASE}/../nginx-1.9.4 May 26 09:21:56 but getting an error:expression was ${@bb.fetch.get_checksum_file_list(d)} ${@get_lic_checksum_file_list(d)} which triggered exception MalformedUrl May 26 09:22:10 Failure expanding variable do_fetch[file-checksums] May 26 09:51:16 niteshnarayanlal: thats because your SRC_URI was malformed (you didn't put file:// at the begining to make it a url) May 26 09:51:37 niteshnarayanlal: but if you're doing that for local hacking then just use the externalsrc class or devtool instead May 26 12:20:14 hello May 26 12:21:20 I'm building yocto for aarch64, with most of the userspace as aarch64 except for one package (and dependency libraries) that are built as lib32-package with multilib May 26 12:22:25 it runs OK but when I try to gdb the 32-bit process, gdb warns of different architectures, and then gives an internal error May 26 12:22:50 does anyone have experience on such multilib environments? how would you debug a 32-bit process? May 26 12:27:55 wouldn't you need lib32-gdb? May 26 12:29:25 wouldn't lib32-gdb fail to install at the same time as normal gdb? May 26 12:29:47 (I'm bitbaking with that suggestion, starting now...) May 26 12:30:24 yeah it would May 26 12:30:35 (no idea, never tried gdb on arm with multilib) May 26 12:32:22 I can always put lib32-gdb in the sdk and say only remote debugging would be supported. This is assuming 64-bit gdbserver would not mind 32-bit processes. May 26 12:32:27 I'll give it a go May 26 12:42:14 fredcadete, which hw is that on ? May 26 12:42:44 rcar-gen3 May 26 12:44:15 rob_w: I'm also building to see if I reproduce this on qemuarm64 May 26 12:44:37 ok May 26 13:10:09 # Hi everyone, I have a question about the support of hardfloat and softfloat on arm target. I would to build libgcc with both softfloat and hardfloat, first to compile initrd binaries with klibc and the second for the real rootfs. So, what is it the best way to do that ? May 26 13:22:42 hi, I have a problem getting a custom kernel module (out of tree) to compile with a yocto recipe.. builds fine with SDK but cant get it to work in a recipe.. says it cant find certain include files (even when adding an extra -I flag) May 26 13:22:53 can anybody help out? im probably overlooking something.. May 26 13:24:18 I can try... May 26 13:25:41 Primordus: There is an example of kernel module into meta-skeleton (meta-skeleton/recipes-kernel/hello-mod/hello-mod_0.1.bb) May 26 13:25:42 Primordus: what is the exact issue? have you seen the example recipe in meta-skeleton/recipes-kernel/hello-mod ? May 26 13:29:09 t0mmy, fredcadete : I started from module.bbclass also, error is as follows: | cc1: error: arch/arm/xenomai/include: No such file or directory [-Werror] May 26 13:29:36 (I checked if the files also exist and seems to be the case) May 26 13:37:20 Primordus: is that a relative path? the kernel modules are built out-of-tree (that's why the example recipe does make -C $(KERNEL_SRC) ) May 26 13:40:38 fredcadete: it is, and I added an absolute path with -I to EXTRA_CFLAGS pointing to the parent directory of arch/ (which is build/tmp/work-shared/MACHINE/kernel-source), idk if this is correct..? compiler flag seems to be ignored? May 26 13:42:15 Primordus: if in doubt about the final compile flags, check the actual compiler line that fails (it is probably in log.do_compile) May 26 13:46:22 fredcadete: only shows make -C ... (sorry if this is a stupid question, first time I had this problem) May 26 13:55:15 rob_w, rburton: on qemuarm64 of current poky master, 64-bit gdb debugs 32-bit binaries just fine May 26 13:55:44 so now I'll search for the differences to my rcar-gen3 build May 26 13:55:49 thanks anyway May 26 14:06:55 Primordus: for the kernel's buildsystem, you can add VERBOSE=1 when invoking make and you will get the full command line for every invocation May 26 14:07:19 assuming you are using the kernel's buildsystem... May 26 14:24:52 fredcadete: found the problem, ended up being an error in my Makefile.. thanks for helping out! May 26 14:25:32 Primordus: no problem, rubber-duck programming FTW! May 26 14:30:50 :) May 26 15:03:58 fredcadete, that gives me hope, when i will be needed to touch the rpi3 May 26 15:05:03 rob_w: in fact, poky master has very recently upgraded to GDB 7.11. This is the version that adds support for aarch64/arm multilib May 26 15:05:16 check, thx May 26 15:05:22 np May 26 15:54:08 Does sato provide a way to disable/enable the screen/display blanking behavior? May 26 15:55:08 kergoth: yes does now May 26 15:55:28 ah, where is it? i couldnt' find it when i was looking the other day May 26 15:55:32 kergoth: oe-core 32ac0bfb0167cb350be5d4b8faca490956dd7d93 May 26 15:55:52 * kergoth checks it out May 26 15:55:52 thanks May 26 15:56:25 ah, nice May 26 18:24:39 what can I do with errors like this when I run bitbake on Arch? https://gist.github.com/rpannek/bf6ca48cb4bce41b40006e6845cd13ac#file-gistfile1-txt-L91 May 26 18:26:06 I am setting the gcc version to 5 already in the local.conf May 26 19:10:00 Jeena: its invoking "gcc" so i doubt you did, can you share your local.conf changes? May 26 19:56:54 I'm confused about how virtual/kernel works, is there a good resource that explains that well? May 26 19:57:33 its a "virtual name" (thus the virtual bit) which is resolved to a real recipe by bitbake May 26 19:57:40 all the kernels say they PROVIDE = virtual/kernel May 26 19:58:05 and somewhere in your configuration (and the default configuration) there's a PREFERRED_PROVIDER assignment saying that the preferred recipe for virtual/kernel is May 26 19:58:16 ie linux-yocto by default May 26 20:01:31 rburton: Thank! That does help. May 26 20:01:55 rburton, you where right I reploaded another error, and the local.conf too May 26 20:02:22 under the same url https://gist.github.com/rpannek/bf6ca48cb4bce41b40006e6845cd13ac May 26 20:02:42 the local.conf is at the bottom May 26 20:17:25 Jeena: how strange. worked here! do bitbake -e elfutils-native and look for the BUILD_CC section, check that it has the value you epect May 26 21:25:30 rburton, good idea, will do that tomorrow morning when I'm back at work! May 26 22:44:21 Hey folks, I've been struggling for a while with a build issue and would appreciate some pointers May 26 22:45:15 I've got the reproducing case down to a freshly checked-out copy of poky, default local.conf and bblayers.conf except setting the MACHINE to qemuarm May 26 22:45:29 https://gist.github.com/henryhallam/2ab385ba50bdd4ea8078bd4d1c54136b May 26 22:46:30 bitbaking core-image-minimal runs fine until it gets to perl, which compiles but fails the stripping / debug symbol copying phase May 26 22:47:19 on my original failure case rather than this cut-down reproduction, 'runstrip' would fail rather than objcopy, but same deal. May 26 22:47:41 this is on 'jethro' BTW - rolling back to one of the tagged jethro releases didn't help May 26 23:13:21 same result on krogoth, FWIW. I think this must be some issue with a dependency on the debian stretch host, but I don't know how to track that down. **** ENDING LOGGING AT Fri May 27 02:59:58 2016