[oe] Can't build libcgroup

Jason Kridner jkridner at beagleboard.org
Fri Oct 14 04:37:23 UTC 2011


On Thu, Oct 13, 2011 at 5:12 PM, Gary Thomas <gary at mlbassoc.com> wrote:
> On 2011-10-13 14:58, Andreas Müller wrote:
>>
>> On Thursday, October 13, 2011 10:46:48 PM Gary Thomas wrote:
>>>
>>> On 2011-10-13 14:44, Khem Raj wrote:
>>>>
>>>> On Thu, Oct 13, 2011 at 1:42 PM, Gary Thomas<gary at mlbassoc.com>   wrote:
>>>>>
>>>>> On 2011-10-13 14:33, Khem Raj wrote:
>>>>>>
>>>>>> On Thu, Oct 13, 2011 at 11:57 AM, Gary Thomas<gary at mlbassoc.com>
>>>>>> wrote:
>>>>>>>
>>>>>>> On 2011-10-13 12:48, Khem Raj wrote:
>>>>>>>>
>>>>>>>> On Thu, Oct 13, 2011 at 7:28 AM, Gary Thomas<gary at mlbassoc.com>
>>>>>>>>   wrote:
>>>>>>>>>
>>>>>>>>> On 2011-10-13 08:16, Kridner, Jason wrote:
>>>>>>>>>>
>>>>>>>>>> On Oct 13, 2011, at 8:39 AM, "Gary Thomas"<gary at mlbassoc.com>
>>>>>>>>>>   wrote:
>>>>>>>>>>
>>>>>>>>>>> Trying to follow the Angstrom instructions directly.
>>>>>>>>>>>
>>>>>>>>>>> OE Build Configuration:
>>>>>>>>>>> BB_VERSION        = "1.13.3"
>>>>>>>>>>> TARGET_ARCH       = "arm"
>>>>>>>>>>> TARGET_OS         = "linux-gnueabi"
>>>>>>>>>>> MACHINE           = "beagleboard"
>>>>>>>>>>> DISTRO            = "angstrom"
>>>>>>>>>>> DISTRO_VERSION    = "v2011.10-core"
>>>>>>>>>>> TUNE_FEATURES     = "armv7a vfp neon cortexa8"
>>>>>>>>>>> TARGET_FPU        = "vfp-neon"
>>>>>>>>>>> meta-angstrom     =
>>>>>>>>>>> "master:66263650f8548d6b0b5f593ab728f536d5e05a9f"
>>>>>>>>>>> meta-oe
>>>>>>>>>>> meta-efl
>>>>>>>>>>> meta-gpe
>>>>>>>>>>> meta-gnome
>>>>>>>>>>> meta-xfce         =
>>>>>>>>>>> "master:1e5d73ec4f44200104de085838f04e31a49435da"
>>>>>>>>>>> meta-ti           =
>>>>>>>>>>> "master:3530f1025efc3792334b897a4e70c6d5822ef56d"
>>>>>>>>>>> meta              =
>>>>>>>>>>> "master:5ad1ca59dea6d5045f252ed7b786ad193faced64"
>>>>>>>>>>>
>>>>>>>>>>> Fails with this error:
>>>>>>>>>>> | ccache arm-angstrom-linux-gnueabi-gcc  -march=armv7-a
>>>>>>>>>>> -fno-tree-vectorize      -mthumb-interwork -mfloat-abi=softfp
>>>>>>>>>>> -mfpu=neon
>>>>>>>>>>> -mtune=cortex-a8 -mthumb-interwork -mno-thumb
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> --sysroot=/local/new-angstrom/build/tmp-angstrom_2010_x-eglibc/sysroots/beagleboard
>>>>>>>>>>> -DHAVE_CONFIG_H -I. -I.. -I../include    -O2 -pipe -g
>>>>>>>>>>> -feliminate-unused-debug-types -Wall -c walk_test.c
>>>>>>>>>>> | libcg_ba.cpp:18:18: fatal error: string: No such file or
>>>>>>>>>>> directory
>>>>>>>>
>>>>>>>> Does this happen with lastest angstrom on oe-core ? or classic
>>>>>>>> oe.dev
>>>>>>>> I am unable to reproduce it on my end
>>>>>>>
>>>>>>> Latest angstrom with layers as above.
>>>>>>
>>>>>> strange, I have it working well for qemuarm and I cant imagine how C++
>>>>>> headers could be broken for beagle
>>>>>
>>>>> Maybe you could try it using the Angstrom tools?  Here's the
>>>>> [documented]
>>>>> sequence I used:
>>>>>   % MACHINE="beagleboard" ./oebb.sh config beagleboard
>>>>>   ... adjust conf/site.conf for my local DL_DIR
>>>>>   ... adjust conf/local.conf to include MACHINE=beagleboard, set
>>>>> parallel
>>>>> build flags, etc
>>>>>   ... removed all meta-intel repos from conf/bblayers.conf (to avoid
>>>>> xorg-dri
>>>>> issues)
>>>>>   % . ~/.oe/environment-oecore
>>>>>   % bitbake systemd-image
>>>>>
>>>>
>>>> needless to say thats what I have done
>>>> but for qemuarm
>>>
>>> Just making sure :-)  I'll try qemuarm here with the same setup.
>>>
>>>
>> Hmm few days ago I had unexplainable effects about iostream [1] which
>> could be cleaned by rebuild (and I swear I have seen same for string before)
>>
>> Some race / parallel install issue?
>>
>> [1]
>> http://lists.linuxtogo.org/pipermail/openembedded-core/2011-October/010869.html
>
> If it fails for me again using qemuarm, I'll try a fresh build
> where I don't try to run full-out -I normally run with:
>  BB_NUMBER_THREADS = "4"
>  PARALLEL_MAKE = "-j 4"
> on a 4-core machine.
>

I started over and reduced my BB_NUMBER_THREADS to 2, but left my
PARALLEL_MAKE at -j24 and the issue went away.  I had it set for 8
threads before when it failed.




More information about the Openembedded-devel mailing list