[yocto] crownbay-noemgd poky-image-minimal fails to build

[Bruce Ashfield]

On 07/29/11 10:51, Malcolm Crossley wrote:
> On 29/07/2011 15:33, Bruce Ashfield wrote:
>> On 07/29/11 10:11, Malcolm Crossley wrote:
>>> On 29/07/2011 15:02, Bruce Ashfield wrote:
>>>> On 07/29/11 09:57, Malcolm Crossley wrote:
>>>>> On 29/07/2011 14:12, Bruce Ashfield wrote:
>>>>>> On 07/29/11 06:05, Andre Haupt wrote:
>>>>>>> On Fri, Jul 29, 2011 at 09:23:59AM +0200, Andre Haupt wrote:
>>>>>>>> Hi all,
>>>>>>>>
>>>>>>>> I am trying to build a poky bernard image for crownbay-noemgd.
>>>>>>>> For this i checked out the bernard branch of both poky and
>>>>>>>> meta-intel
>>>>>>>> from git.yoctoproject.org.
>>>>>>>>
>>>>>>>> Bitbake fails during the kernels compile_perf task complaining that
>>>>>>>> it can not
>>>>>>>> find libdw.h and bfd.h and that using /usr/local/include as include
>>>>>>>> location is unsafe for cross compilation (full log attached).
>>>>>>>>
>>>>>>>> What do i miss here?
>>>>>>>
>>>>>>> I did work around this with
>>>>>>>
>>>>>>> diff --git a/meta/recipes-kernel/linux/linux-yocto-stable_git.bb
>>>>>>> b/meta/recipes-
>>>>>>> index 3c1e669..85caa92 100644
>>>>>>> --- a/meta/recipes-kernel/linux/linux-yocto-stable_git.bb
>>>>>>> +++ b/meta/recipes-kernel/linux/linux-yocto-stable_git.bb
>>>>>>> @@ -42,4 +42,4 @@ KERNEL_FEATURES=features/netfilter
>>>>>>> addtask kernel_link_vmlinux after do_compile before do_install
>>>>>>> addtask validate_branches before do_patch after do_kernel_checkout
>>>>>>>
>>>>>>> -require linux-tools.inc
>>>>>>> +#require linux-tools.inc
>>>>>>>
>>>>>>>
>>>>>>> not sure that this is the right thing to do though.
>>>>>>
>>>>>> It's an acceptable work around to keep you building on your
>>>>>> local machine, assuming that you don't need perf. But the
>>>>>> linux-yocto recipes always want to provide perf, and hence
>>>>>> have that include.
>>>>>>
>>>>>> I'm trying to dust off my memory here. I know that this has
>>>>>> been seen, and I also know that this built fine in the bernard
>>>>>> release.
>>>>>>
>>>>>> The first part of the issue is the:
>>>>>>
>>>>>> Makefile:508: No libdw.h found or old libdw.h found, disables dwarf
>>>>>> support. Please install elfutils-devel/elfutils-dev
>>>>>>
>>>>>> We may have a missing host dependency in the checks. Have you
>>>>>> tried installing the relevant package for F13 ?
>>>>>>
>>>>>> As for the second, there's a patch in the 2.6.34 kernel tree
>>>>>> that is dealing with it:
>>>>>>
>>>>>> http://git.pokylinux.org/cgit/cgit.cgi/linux-yocto-2.6.34/commit/?h=standard&id=72ca49ab08b8eb475cec82a10049503602325791
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> It sounds like there may be a SRCREV problem for the board that
>>>>>> isn't picking up that change. Can you confirm that this commit
>>>>>> is in your board branch ? You can check in your build directory
>>>>>> for linux.
>>>>>
>>>>> I have found that stable branch 2.6.34 contains a bug when trying to
>>>>> build the perf tools. I have a patch but I've not been able to
>>>>> upstream
>>>>> it yet, I've copy pasted it inline here, hopefully it doesn't wrap too
>>>>> badly:
>>>>
>>>> Just so I've got all the context here. This failure is
>>>> on top of the one you were asking about ? (the unsafe
>>>> / includes). And this is unique to the 2.6.34-stable
>>>> perf variant (i.e. no other newer kernel release has
>>>> the issue) ?
>>> The patch is required on 2.6.34-stable but not on 2.6.34 vanilla. I have
>>> not tested newer kernels but I believe there would be some compliants if
>>> it was still broken. So I think there was some specific backporting
>>> breakage with 2.6.34-stable. Without the patch I got this error on
>>> 2.6.34.6:
>>
>> odd. Our autobuilders and local builds should have shown this,
>> since we are already past 2.6.34.6 for those boards.
>>
>> We'll have to look into this one a bit more.
>>
>
> I saw this problem using Openembedded stable 2009 branch, so yocto could
> be fine due to different environment setup. I was posting the patch in
> case it helped with this issue because it looked similar.

No worries! thanks for the report and the post, I'm
going to keep this on my radar going forward.

Cheers,

Bruce

>
> Regards
>
> Malcolm
>>
>>>
>>> NOTE: make -j 2
>>> DESTDIR=/home/gefanuc/auto/main/build/work/ge-sbc612-angstrom-linux/linux-ge-2.6.x+22411-r0/image
>>>
>>> CC=powerpc-angstrom-linux-gcc LD=powerpc-angstrom-linux-ld
>>> Makefile:508: No libdw.h found or old libdw.h found, disables dwarf
>>> support. Please install elfutils-devel/elfutils-dev
>>> touch .perf.dev.null
>>> CC perf.o
>>> CC bench/sched-messaging.o
>>> bench/sched-messaging.c:34: error: expected '=', ',', ';', 'asm' or
>>> '__attribute__' before 'use_pipes'
>>> bench/sched-messaging.c:36: error: expected '=', ',', ';', 'asm' or
>>> '__attribute__' before 'thread_mode'
>>> bench/sched-messaging.c: In function 'fdpair':
>>> bench/sched-messaging.c:61: error: 'use_pipes' undeclared (first use in
>>> this function)
>>> bench/sched-messaging.c:61: error: (Each undeclared identifier is
>>> reported only once
>>> bench/sched-messaging.c:61: error: for each function it appears in.)
>>> bench/sched-messaging.c: In function 'receiver':
>>> bench/sched-messaging.c:120: error: 'thread_mode' undeclared (first use
>>> in this function)
>>> bench/sched-messaging.c: In function 'create_worker':
>>> bench/sched-messaging.c:149: error: 'thread_mode' undeclared (first use
>>> in this function)
>>> bench/sched-messaging.c: In function 'reap_worker':
>>> bench/sched-messaging.c:189: error: 'thread_mode' undeclared (first use
>>> in this function)
>>> bench/sched-messaging.c: In function 'group':
>>> bench/sched-messaging.c:232: error: 'thread_mode' undeclared (first use
>>> in this function)
>>> bench/sched-messaging.c: At top level:
>>> bench/sched-messaging.c:255: error: 'use_pipes' undeclared here (not in
>>> a function)
>>> bench/sched-messaging.c:257: error: 'thread_mode' undeclared here (not
>>> in a function)
>>> make: *** [bench/sched-messaging.o] Error 1
>>> make: *** Waiting for unfinished jobs....
>>>
>>> Regards
>>>
>>> Malcolm
>>>
>>>
>>>>
>>>> with those answers, I can appropriately deal with it
>>>> and vector it to the right places.
>>>>
>>>> Cheers,
>>>>
>>>> Bruce
>>>>
>>>>>
>>>>> Author: Malcolm Crossley <malcolm.crossley at ge.com>
>>>>> Date: Fri Jul 22 22:44:53 2011 +0100
>>>>>
>>>>> Kernel version 2.6.34.6. has a bug with regards to it's perf tool
>>>>> source
>>>>> files. This patch adds a missing include directive.
>>>>>
>>>>> Signed-off-by: Malcolm Crossley <malcolm.crossley at ge.com>
>>>>>
>>>>> diff --git a/tools/perf/perf.h b/tools/perf/perf.h
>>>>> index 6fb379b..3c497c5 100644
>>>>> --- a/tools/perf/perf.h
>>>>> +++ b/tools/perf/perf.h
>>>>> @@ -73,6 +73,7 @@
>>>>> #include <unistd.h>
>>>>> #include <sys/types.h>
>>>>> #include <sys/syscall.h>
>>>>> +#include <stdbool.h>
>>>>>
>>>>> #include "../../include/linux/perf_event.h"
>>>>> #include "util/types.h"
>>>>>
>>>>>
>>>>> Regards
>>>>>
>>>>> Malcolm
>>>>>
>>>>>>
>>>>>> Bruce
>>>>>>
>>>>>>>
>>>>>>> cheers,
>>>>>>>
>>>>>>> Andre
>>>>>>> _______________________________________________
>>>>>>> yocto mailing list
>>>>>>> yocto at yoctoproject.org
>>>>>>> https://lists.yoctoproject.org/listinfo/yocto
>>>>>>
>>>>>> _______________________________________________
>>>>>> yocto mailing list
>>>>>> yocto at yoctoproject.org
>>>>>> https://lists.yoctoproject.org/listinfo/yocto
>>>>>
>>>>> _______________________________________________
>>>>> yocto mailing list
>>>>> yocto at yoctoproject.org
>>>>> https://lists.yoctoproject.org/listinfo/yocto
>>>>
>>>
>>
>