**** BEGIN LOGGING AT Tue May 19 02:59:59 2015 May 19 03:42:59 khem`: Looks like there some issues with musl and the changes that happened with regards to using ld=gold. I am getting lots of packages failing with 'copy reloc against protected `stderr' is invalid' May 19 03:43:30 this sort of thing seems to rememdy some 'LDFLAGS_append_pn-expat = " -fuse-ld=gold"' May 19 03:49:34 hmm May 19 03:49:48 I use gold so natually I dont see it May 19 03:49:56 DISTRO_FEATURES = "ld-is-gold" May 19 03:50:14 what arch are you on ? May 19 03:51:14 trying qemuarm at the moment so armv5 May 19 03:59:21 yer looks like setting it to use gold resolves the linking issues May 19 04:05:52 but thats not good May 19 04:06:24 are you using gcc-5 ? May 19 04:14:12 i tried 4.9 and 5.1, both had the same result May 19 04:21:46 nrossi: can you backport https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=f74839702efcea048ac61374b1539ae81e7c86cc May 19 04:21:56 and see if ld.bfd is happier May 19 04:23:29 will give it a try and let you know May 19 06:11:31 khem`: Good news, that patch fixed up the linker issues. Bad news, microblaze binaries dont run :( May 19 06:13:07 I sort of expected that May 19 06:13:17 yer me too, but one can hope :) May 19 06:13:18 you need a gcc + glibc patch too May 19 06:14:51 its x86 specific completed toolchain wife May 19 06:14:53 wide May 19 06:15:07 other architectures may be not so much May 19 06:15:48 see https://gcc.gnu.org/ml/gcc-patches/2015-03/msg00257.html May 19 06:15:54 so you need gcc as well as glibc patch May 19 06:16:08 and unfortuantely gcc patch is only for gcc-5 May 19 06:16:21 and no chance of it coming back to older gcc May 19 06:17:41 bit lost, your saying that its the patch that is addressing the protected linker stuff that is causing the issue on microblaze? May 19 06:19:27 no, I am just completing the set May 19 06:19:52 was microblaze working before this patch ? May 19 06:20:00 the patch sort of changes ABI May 19 06:20:19 w.r.t how protected symbols are handled May 19 06:20:43 i had not tested May 19 06:21:23 hmm ok so may be you should do that firs t May 19 06:21:48 I assume you are using gcc-4.9 + glibc 2.21 + binutils 2.25 here May 19 06:21:52 for microblaze May 19 06:22:47 oh sorry no i mean with musl-libc for microblaze May 19 06:38:10 so i removed that binutils patch, it appears that its not needed for microblaze, so we will see how with booting once its done May 19 07:16:16 gm May 19 07:43:07 good morning May 19 08:04:37 morning all May 19 09:06:45 what is the proper way of overriding a recipe SRC_URI in a bbappend for just one machine? if I create a .bbappendfile and prepend a new FILESPATHEXTRA where i put my overridden file, it will be applied to all machines. Should I only prepend for my machine, or do I add a new SRC_URI and then append to the do_install task so that it overwrites the previously installed (default) file with my new one? May 19 09:08:21 mago_: it depends - if the file is already in SRC_URI, you can just prepend FILESEXTRAPATHS and put the file in a subdirectory under the prepended directory named with the machine May 19 09:08:35 mago_: if it isn't already in SRC_URI, do a conditional prepend May 19 09:23:53 bluelightning: oh, right. machine subfolder May 19 12:11:53 anyone from Toradex around? May 19 12:17:37 crofton dont think so May 19 12:19:54 We've got a customer having usb performance issues with one oftheir tegra 3 boards May 19 12:21:40 hm May 19 12:22:08 there was some discussion on the arm linux ml about dma stuff May 19 12:22:31 in tegra but only for upstream where toradex is mainly using the downstream kernel May 19 12:23:01 typical May 19 12:27:08 hm yes still 3.1.10 with own patches for tegra colibri boards May 19 12:44:00 Crofton|work: you can probably send question to stefan agnier May 19 12:46:08 abelloni hm hm difficult May 19 12:46:15 that is more on nvidia directly May 19 12:46:34 and with service contract May 19 12:47:43 still, it doesn't cost much asking stefan :) May 19 12:53:34 yeah, trying to get the customer to engage them more :) May 19 16:52:14 hm, how is the offical way to set variables in a custom-image file and use them in a snippet within POSTPROCESS_COMMAND May 19 16:52:38 getEnv setEnv? May 19 16:53:20 uhm I meant python getVar May 19 16:56:35 woglinde: as of a few versions ago, functions specified in *_POSTPROCESS_COMMAND can be python functions May 19 16:57:19 but naturally you can also use variable references in shell functions called via there May 19 16:59:33 I need to change the PACKAGES variable based on PACKAGECONFIG setting, any ideas? May 19 17:02:16 use anonymous python May 19 17:02:20 er, inline, that is May 19 17:04:37 hm I have made a snippet set entry in etc/resolv.conf now I am inherit it in my image and want to set the variables for the dns stuff there May 19 17:05:34 bluelightning it has do work on dora to May 19 17:06:22 hm okay I require the snippets od course May 19 17:07:17 woglinde: right, it was daisy we made python functions callable from *_POSTPROCESS_COMMAND, but it seems to me that probably isn't required for this case May 19 17:08:24 Crofton|work: right, what kergoth said... ${@bb.utils.contains('PACKAGECONFIG', 'feature', '... May 19 17:09:06 thanks May 19 17:09:15 hm I have DNSSERVER = "${@d.getVar('DNS_SERVER', True) or ""}" in the .inc file May 19 17:09:36 and DNS_SERVER = "127.0.0.1" in the image file May 19 17:09:40 maybe the order matters May 19 17:11:31 where's the inc file being included? May 19 17:12:57 I tried it after setting the vars May 19 17:13:09 now I am trying it putting it before May 19 17:13:22 uhm May 19 17:13:23 args May 19 17:13:24 if you mean in the image file, shouldn't make a difference May 19 17:13:29 okay May 19 17:13:35 but it should work? May 19 17:14:01 I can't see why it shouldn't based on what you've said here May 19 17:14:08 hm no May 19 17:15:09 hm okay May 19 17:15:16 btw, you probably want or ''}" not or ""}" May 19 17:15:19 looks like my shell foo is wrong May 19 17:15:52 (single quotes within the outer double quotes - not that that's necessarily going to break things but I'd advise doing it that way) **** ENDING LOGGING AT Wed May 20 02:59:58 2015