[Buildroot] [PATCH 6/6] liburcu: disable build with gcc-4.8.x on ARM

Samuel Martin s.martin49 at gmail.com
Fri Mar 7 12:59:29 UTC 2014


Hi Thomas, all,

On Thu, Mar 6, 2014 at 11:17 PM, Thomas Petazzoni
<thomas.petazzoni at free-electrons.com> wrote:
> Dear Samuel Martin,
>
> On Sun,  2 Mar 2014 00:39:53 +0100, Samuel Martin wrote:
>
>> +     # liburcu/gcc-4.8.x bug
>> +     depends on !((BR2_arm || BR2_armeb) && (BR2_GCC_VERSION_4_8_X || \
>> +             BR2_TOOLCHAIN_EXTERNAL_LINARO_2013_11 || \
>> +             BR2_TOOLCHAIN_EXTERNAL_LINARO_2013_10 || \
>> +             BR2_TOOLCHAIN_EXTERNAL_LINARO_2013_09 || \
>> +             BR2_TOOLCHAIN_EXTERNAL_CODESOURCERY_ARM201311))
>
> The exclusion of BR2_GCC_VERSION_4_8_X is wrong, because we have in
> Buildroot the patch fixing the
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58854 gcc bug.
>
> But generally, I'm not really happy with this approach, as we would
> have to update these dependencies every time we bump/change the
> external toolchain Kconfig options... Unfortunately I don't have a
> better suggestion.

I did this because of the patch in liburcu checking for the compiler version.
I did not notice BR carries the patch fixing this bug.

>
> Maybe I would prefer something like:
>
> config BR2_TOOLCHAIN_HAS_GCC_BUG_58854
>         bool
>
> And then the toolchain impacted by the bug would select it. For
> example, are you sure the Linaro toolchains are affected by the
> problem? Most likely they have backported the change that is needed to
> fix this problem.

Indeed, it sounds like a less restrictive way of handling this (even
if it implies more testing ;-])

Regards,

-- 
Samuel



More information about the buildroot mailing list