[Buildroot] [autobuild.buildroot.net] Build results for 2015-07-22

[Romain Naour <romain.naour@openwide.fr>]

Hi Thomas,

Le 23/07/2015 22:11, Thomas Petazzoni a ?crit :
> Dear Romain Naour,
> 
> On Thu, 23 Jul 2015 19:51:00 +0200, Romain Naour wrote:
>> Hi Alexey,
>>
>> Le 23/07/2015 08:30, Thomas Petazzoni a ?crit :
>>> Detail of failures
>>> ===================
>>
>>>       xtensa |         host-gcc-initial-4.9.3 | NOK | http://autobuild.buildroot.net/results/56de85f9bc0ba44248003212b5dc4cb8dcde19c5/
>>
>> I noticed a new build failure in autobuilders due to your gcc package patch:
>> http://patchwork.ozlabs.org/patch/453849/
> 
> Did you notice xtensa or other platforms as well? Before applying
> Alexey's patch I tested an ARM toolchain build with -O2 and -Os and it
> built fine.

No, only on xtensa.

Here is the content of CFLAGS_FOR_TARGET and CXXFLAGS_FOR_TARGET:

CFLAGS_FOR_TARGET="-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE
-D_FILE_OFFSET_BITS=64 -mlongcalls -mtext-section-literals  -Os "

CXXFLAGS_FOR_TARGET="-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE
-D_FILE_OFFSET_BITS=64 -mlongcalls -mtext-section-literals  -Os "

Probably -mtext-section-literals is the culprit...

Ah! did you notice the typo in Alexey's patch ?
+# Propagate options used for target software building to GCC target libs
+HOST_GCC_COMMON_CONF_ENV += CFLAGS_FOR_TARGET="$(GCC_COMMON_TARGET_CXXFLAGS)"
			     ^^^				    ^^^
I'll send a patch for that ;-)

Best regards,
Romain

> 
> Though it's indeed a problem if it causes a regression on other
> architectures. Alexey, Max, can you have a look at least for the Xtensa
> case? If there is no feedback soon, I'll revert Alexey's patch in the
> mean time.
> 
> Thanks!
> 
> Thomas
>