On Sat, 2024-05-11 at 09:49 +0800, Kai Kang via lists.openembedded.org wrote:
> From: Kai Kang <[email protected]>
> 
> The header files conflic when multilib enabled for arm:
> 
> > Error: Transaction test error:
> >   file /usr/include/finclude/math-vector-fortran.h conflicts between 
> > attempted installs of
>     lib32-libc6-dev-2.39+git0+312e159626-r0.armv7at2hf_neon and 
> libc6-dev-2.39+git0+312e159626-r0.cortexa72
> 
> Invoke function oe_multilib_header to resolve it.
> 
> But gfortran-cross can NOT recognize macros such as `#ifdef` as
> expected, so pass option `-nostdinc` to gfortran-cross-aarch64 that not
> preinclude multilib specific math-vector-fortran.h [1].
> 
> [1]: 
> https://gcc.gnu.org/git/?p=gcc.git;a=blob;f=gcc/config/gnu-user.h;h=f7eefdafe8b330787c412ad950675d7a2d595e61;hb=HEAD#l158
> 
> Signed-off-by: Kai Kang <[email protected]>
> ---
>  meta/conf/bitbake.conf                    | 5 ++++-
>  meta/recipes-core/glibc/glibc-package.inc | 2 ++
>  2 files changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/conf/bitbake.conf b/meta/conf/bitbake.conf
> index b2c500d873..262c3e32ee 100644
> --- a/meta/conf/bitbake.conf
> +++ b/meta/conf/bitbake.conf
> @@ -555,11 +555,14 @@ HOSTTOOLS_NONFATAL += "git-lfs"
>  
>  CCACHE ??= ""
>  
> +FC_OPTIONS ??= ""
> +FC_OPTIONS:append:aarch64 = " -nostdinc"
> +
>  TOOLCHAIN_OPTIONS = " --sysroot=${STAGING_DIR_TARGET}"
>  

Does nostdinc not have other fairly nasty side effects such as
requiring application level changes to ensure headers are included?

A lot of these fixes feel like the solution is worse and more risky
than the original problem :(.

Cheers,

Richard


-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#199205): 
https://lists.openembedded.org/g/openembedded-core/message/199205
Mute This Topic: https://lists.openembedded.org/mt/106033238/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to