**** BEGIN LOGGING AT Tue Jul 15 02:59:59 2014
Jul 15 07:38:12 <mckoan> good morning
Jul 15 07:39:18 <woglinde> hi mckoan
Jul 15 07:40:12 <mckoan> hi woglinde 
Jul 15 07:45:17 <bluelightning> morning mckoan, woglinde, all
Jul 15 07:45:53 <mckoan> hi bluelightning 
Jul 15 07:49:05 <woglinde> hi bluelightning
Jul 15 07:53:17 <pompomJuice> any linux experts in the house?
Jul 15 07:55:26 <pompomJuice> I just cannot understand by what mechanic systemd knows about a start-stop-daemon process that I have launch from postinst.service... how does it know
Jul 15 08:05:21 <woglinde> pompomJuice what=
Jul 15 08:05:22 <woglinde> ?
Jul 15 08:05:48 <pompomJuice> http://hastebin.com/mupuvoqoqe.bash
Jul 15 08:05:53 <pompomJuice> look at that script
Jul 15 08:07:20 <pompomJuice> are you familier with the opkg postinstall mechanism?
Jul 15 08:08:01 <woglinde> pompomJuice none
Jul 15 08:08:13 <woglinde> systemd knows only service files
Jul 15 08:08:42 <pompomJuice> ok
Jul 15 08:08:58 <pompomJuice> when baking an image using bitbake
Jul 15 08:09:21 <pompomJuice> bitbake uses opkg install to create the image during the rootfs stage
Jul 15 08:09:29 <pompomJuice> the problem is
Jul 15 08:09:38 <pompomJuice> package sometimes have a postinstall script that needs to run
Jul 15 08:09:40 <pompomJuice> or
Jul 15 08:10:04 <pompomJuice> inside the recipies you can spesify a script that needs to run after opkg has installed (copied) the ipk's files
Jul 15 08:10:13 <pompomJuice> it looks like this inside the recipy
Jul 15 08:10:45 <pompomJuice> pkg_postinst_${PN} () {
Jul 15 08:11:09 <pompomJuice> whatever you put inside that begin end bracket is the script that gets executed after opkg installed the files of an package
Jul 15 08:11:28 <pompomJuice> sometimes these scripts don't work when you are baking an image
Jul 15 08:11:41 <pompomJuice> they need to execute on a running system for whatever reason
Jul 15 08:12:56 <pompomJuice> the way that this pkg_postinst_${PN} thing handles that is when it runs the script during image baking and the script returns non zero, the opkg postinstall framework deferres the script to run on first boot
Jul 15 08:13:56 <woglinde> yes
Jul 15 08:14:18 <pompomJuice> ok
Jul 15 08:14:30 <pompomJuice> to be more spesific
Jul 15 08:14:33 <ant_work> there are valid reasons to defer ;)
Jul 15 08:14:51 <pompomJuice> there is a service called run-postinst.service that was part of systemd
Jul 15 08:14:55 <pompomJuice> exactly
Jul 15 08:15:21 <pompomJuice> from 1.5 -> 1.6 this run-postinst.service was moved to it's own package called run-postinst
Jul 15 08:15:27 <pompomJuice> but that is not the issue here
Jul 15 08:15:31 <pompomJuice> just for some background
Jul 15 08:15:34 <woglinde> yes
Jul 15 08:15:37 <woglinde> okay
Jul 15 08:15:41 <woglinde> whats the issue now?
Jul 15 08:15:45 <pompomJuice> ok
Jul 15 08:15:46 <pompomJuice> the issue is
Jul 15 08:16:01 <pompomJuice> my postinstall script needs to run after android runs
Jul 15 08:16:23 <pompomJuice> but run-posinst.service is run really really early during bootup
Jul 15 08:16:28 <pompomJuice> which makes sense
Jul 15 08:16:52 <woglinde> hm
Jul 15 08:16:58 <pompomJuice> what I am trying to achive is to create a postinstall script that runs and immediately exits, but not before it spawned a daemon that waits for android to start
Jul 15 08:17:12 <pompomJuice> this has failed
Jul 15 08:17:16 <pompomJuice> I have tried everything
Jul 15 08:17:43 <pompomJuice> somehow run-postinstall (or systemd) sees my daemon process and waits for it to exit before moving on
Jul 15 08:17:50 <pompomJuice> my question is
Jul 15 08:17:53 <pompomJuice> how does it know
Jul 15 08:17:53 <woglinde> cannt you write a systemd service file which waits for andriod and configures somehting and check for some variables/files if has already configured?
Jul 15 08:18:07 <woglinde> downside is it will be run every boot
Jul 15 08:18:08 <pompomJuice> I could do that
Jul 15 08:18:12 <pompomJuice> yes
Jul 15 08:18:14 <pompomJuice> I thought of that
Jul 15 08:18:18 <woglinde> but that should marginal
Jul 15 08:18:21 <pompomJuice> but you can delete the service when it ran once
Jul 15 08:18:23 <woglinde> for boot time
Jul 15 08:18:31 <pompomJuice> a type=oneshot service
Jul 15 08:18:58 <pompomJuice> the reason why I am upset is I cannot figure out how systemd knows about my daemon
Jul 15 08:19:00 <pompomJuice> how does it know
Jul 15 08:19:21 <pompomJuice> is it a linux security mechanism where a parent can always control any child processes it spawns 
Jul 15 08:19:48 <pompomJuice> if you look at my hastebin
Jul 15 08:19:56 <pompomJuice> I try very hard to hide the daemon from systemd
Jul 15 08:19:58 <pompomJuice> but I cant
Jul 15 08:20:04 <pompomJuice> why?
Jul 15 08:20:26 <pompomJuice> does anyone know
Jul 15 08:22:00 <woglinde> did you install it in etc/init.d/ ?
Jul 15 08:22:10 <pompomJuice> no
Jul 15 08:22:19 <woglinde> sorry I forgot to say systemd looks there and starts die iniscripts from there
Jul 15 08:22:26 <woglinde> if they are not masked
Jul 15 08:22:35 <pompomJuice> aah
Jul 15 08:22:52 <pompomJuice> that also changed in 1.6
Jul 15 08:23:02 <pompomJuice> 1.6 sysvinit seems to be disabled
Jul 15 08:23:07 <pompomJuice> I am guessing here
Jul 15 08:23:07 <bluelightning> pompomJuice: it runs services within a cgroup, thus any process that a service starts is also contained within that cgroup and considered part of the service
Jul 15 08:23:17 <pompomJuice> aah
Jul 15 08:23:27 <pompomJuice> that is 
Jul 15 08:23:39 <woglinde> ah okay thanks bluelightning
Jul 15 08:23:42 <pompomJuice> that sounds like the something I dont understand
Jul 15 08:23:58 <pompomJuice> thanks man
Jul 15 08:24:06 <pompomJuice> it is driving me crazy
Jul 15 08:25:13 <pompomJuice> I will take woglinde's advice then
Jul 15 08:25:29 <pompomJuice> create a oneshot service instead that destroys/digests itself after running
Jul 15 08:42:21 <woglinde> hm pompomJuice oneshot only means it will be run one time for startup
Jul 15 08:42:48 <woglinde> so you will need some glue to figure out if it is configured or not
Jul 15 08:43:05 <woglinde> or it does not matter and you will run the config a every startup
Jul 15 08:46:11 <pompomJuice> I will have to look at that yes
Jul 15 08:46:22 <pompomJuice> at least I wont be banging  my head against cgroups
Jul 15 08:51:40 <heeen> is there a way to make bitbake continue with as many tasks as it can even if one fails
Jul 15 08:52:03 <heeen> likle fetch fails because one source is down but it can still compile the rest that does not depend
Jul 15 08:56:23 <woglinde> -k
Jul 15 10:32:12 <Shawn286> Hello
Jul 15 10:32:44 <Shawn286> hi iWalrus 
Jul 15 10:33:15 <Shawn286> phew what a powerfull command
Jul 15 10:33:48 <Shawn286> I can speak in every channel from one channel :D
Jul 15 10:34:20 <Shawn286> /notice ## boo
Jul 15 11:18:39 <kroon> I have a q about the environment script in the SDK that one should source. When I use the SDK, and run autogen.sh for my project, I see perl warnings that "Setting locale failed"
Jul 15 11:19:05 <kroon> LC_ALL and LANGUAGE appears to be unset
Jul 15 11:19:41 <kroon> Would it make sense if the environment script also set LC_ALL=C for the user ?
Jul 15 11:19:52 <kroon> At least this gets rid of the perl warnings
Jul 15 11:26:43 <kroon> Or perhaps emit it conditionally based on the value of SDKUSE_NLS
Jul 15 11:27:05 <kroon> My Linux locale skills are limited :-(
Jul 15 12:55:37 <pompomJuice> angstrom has cpu-ondemand.service enable by default
Jul 15 12:55:50 <pompomJuice> what is the correct way to disable services that comes with my stock distro?
Jul 15 13:42:23 <woglinde> pompomJuice search for systemctl mask
Jul 15 13:42:37 <pompomJuice> ok
Jul 15 13:42:45 <pompomJuice> I understand that part
Jul 15 13:42:53 <pompomJuice> aah I see
Jul 15 13:43:01 <woglinde> I use it in config packages or post_image
Jul 15 13:43:09 <pompomJuice> I should add some recipy that creates those simlinks
Jul 15 13:43:19 <pompomJuice> aah ok
Jul 15 13:43:20 <pompomJuice> thanks
Jul 15 13:43:23 <woglinde> depends on the use case
Jul 15 13:43:34 <pompomJuice> makes sense
Jul 15 13:43:47 <woglinde> I have one project where I disable gpsd because I have to start 2 other gpsd
Jul 15 13:43:49 <pompomJuice> I will create a meter configuration package
Jul 15 13:44:38 <woglinde> and I have another project where I mask services wether systemd is installed or not wihtin image creation
Jul 15 13:44:47 <woglinde> got to go now
Jul 15 13:44:50 <woglinde> till later
Jul 15 14:16:03 <Guest83683> how can i get rid of this error : ./configure: /tmp/ffconf.XXNQGlls.sh: /bin/sh: bad interpreter: Permission denied | Sanity test failed.
Jul 15 14:55:14 <dolo> hi guys. In c++, if I allocate a memory inside of a function, do I have to delete it at the end of the function or is it deleted automatically since it is in the scope of the function?
Jul 15 14:57:00 <JaMa> someone seeing huge memory usage while building gdb? right now it's eating 13GB/14.5GB RES/VIRT
Jul 15 14:57:10 <JaMa> 1 cc1 process
Jul 15 14:59:35 <dolo> no one?
Jul 15 15:02:07 <kroon> dolo, C++ related questiones are better asked in #c++
Jul 15 15:03:34 <dolo> kroon, i didn't know the existence of such group
Jul 15 15:03:36 <dolo> thanks
Jul 15 15:03:54 <dolo> however, it goes to overflow
Jul 15 15:04:20 <dolo> just a simple question, coulnd't you please answer it shortly like yes or no?
Jul 15 15:04:46 <kroon> dolo, you need to register a freenode nickname i think
Jul 15 15:05:17 <kroon> dolo, new/malloc:ed memory should be freed after its not needed anymore
Jul 15 15:05:45 <dolo> even if it is allocated in the function
Jul 15 15:05:49 <dolo> kroon thanks a lot
Jul 15 15:06:09 <kroon> dolo, C++ as it is does no automatic garbage collection for you
Jul 15 15:07:04 <dolo> krron, yes but I know that the variables allocated in the function is only valid as long as the function is alive. so I thought it is gone after the function ends
Jul 15 15:07:05 <kroon> JaMa, just tried it here, no apparent spike. gdb 7.7, MACHINE=wandboard-solo
Jul 15 15:09:14 <JaMa> kroon: thanks
Jul 15 15:09:41 <JaMa> I'll keep it swapping for a while to see if it is successful in the end
Jul 15 15:09:45 <kroon> dolo, the memory for automatic variables are allocated on the stack, that memory is freed once the function returns. memory allocated by new/malloc are allocated from the heap, and need to be returned once its not used anymore
Jul 15 15:53:30 <khem> if you allocate from heap you have to give it back at some point. It doesnt have to be end of function 
Jul 15 15:54:22 <khem> JaMa: is it gdb or gdb-cross ?
Jul 15 16:26:52 <nerdboy> JaMa: run out of gremlin juice?
Jul 15 16:42:35 <JaMa> khem: target gdb
Jul 15 16:42:49 <JaMa> khem: and it's still running with peak over 20GB
Jul 15 16:43:15 <JaMa> strace showing just bunch of mmap, then brk, then munmap
Jul 15 16:43:48 <JaMa> cc1 running for 2 hours already
Jul 15 16:50:57 <khem> JaMa: gcc 4.9
Jul 15 16:52:05 <khem> arm ?
Jul 15 17:10:54 <JaMa> gcc-4.9 + arm920t, but I've seen it with other arch as well
Jul 15 17:12:01 <JaMa> but qemux86-64 build was OK
Jul 15 17:30:14 <JaMa> khem: and here is htop http://bpaste.net/show/465784/
Jul 15 17:42:47 <challinan> khem: I hit this bug trying to build meta-xilinx/daisy today:
Jul 15 17:42:59 <challinan> khem: https://bugs.launchpad.net/gcc-linaro/+bug/1312931
Jul 15 17:43:07 <challinan> khem: did you ever find a resolution?
Jul 15 17:43:45 * challinan hopes the answer isn't to move to another toolchain ;)
Jul 15 18:23:58 <JaMa> khem: interesting.. removing -mthumb fixes the memory consumption and hanging build
Jul 15 18:40:32 <koen> JaMa: 4.9 or linaro-4.9?
Jul 15 18:50:28 <JaMa> 4.9 from oe-core
Jul 15 18:51:51 * JaMa waiting for first successful build since daisy release :/
Jul 15 18:55:33 <khem> challinan: thats an issue IIRC wrong arguments to assembler
Jul 15 21:14:07 <JaMa> oh..
Jul 15 21:14:08 <JaMa> | /OE/build/shr-core/tmp-eglibc/work/arm920tt-oe-linux-gnueabi/icu/53.1-r0/icu/source/common/ubidiwrt.c: In function 'ubidi_writeReordered_53':
Jul 15 21:14:11 <JaMa> | /OE/build/shr-core/tmp-eglibc/work/arm920tt-oe-linux-gnueabi/icu/53.1-r0/icu/source/common/ubidiwrt.c:643:1: internal compiler error: in patch_jump_insn, at cfgrtl.c:1275
Jul 15 21:14:14 <JaMa> |  }
Jul 15 21:14:42 <ant_home> o,O
Jul 15 21:15:48 <nerdboy> JaMa: anything more than that?
Jul 15 21:16:00 <nerdboy> hardware problem?
Jul 15 21:29:46 <khem> JaMa: thumb1 is a depreecated so I expect more and more problems with it 
Jul 15 21:30:42 <khem> JaMa: that gcc ICE again seems to be thumb relared to me 
Jul 15 21:31:02 <khem> since its not able to create appropriate jump veneers
**** ENDING LOGGING AT Wed Jul 16 02:59:59 2014