From: Arnd Bergmann <arnd@arndb.de>

The headercheck tries to call clang with a mix of compiler arguments
that don't include the target architecture. When building e.g. x86
headers on arm64, this produces a warning like

   clang: warning: unknown platform, assuming -mfloat-abi=soft

Add in the CLANG_FLAGS, which contain the target, in order to make it
build properly.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 usr/include/Makefile | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/usr/include/Makefile b/usr/include/Makefile
index 58a9cbe4eba4..99f9e904adcd 100644
--- a/usr/include/Makefile
+++ b/usr/include/Makefile
@@ -10,7 +10,7 @@ UAPI_CFLAGS := -std=c90 -Wall -Werror=implicit-function-declaration
 
 # In theory, we do not care -m32 or -m64 for header compile tests.
 # It is here just because CONFIG_CC_CAN_LINK is tested with -m32 or -m64.
-UAPI_CFLAGS += $(filter -m32 -m64 --target=%, $(KBUILD_CFLAGS))
+UAPI_CFLAGS += $(filter -m32 -m64 --target=%, $(KBUILD_CFLAGS) $(CLANG_FLAGS))
 
 # USERCFLAGS might contain sysroot location for CC.
 UAPI_CFLAGS += $(USERCFLAGS)
-- 
2.39.5

On Fri, Feb 21, 2025 at 10:20:07PM +0100, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@arndb.de>
> 
> The headercheck tries to call clang with a mix of compiler arguments
> that don't include the target architecture. When building e.g. x86
> headers on arm64, this produces a warning like
> 
>    clang: warning: unknown platform, assuming -mfloat-abi=soft
> 
> Add in the CLANG_FLAGS, which contain the target, in order to make it
> build properly.
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

This looks like a similar problem as the one fixed by
commit 1b71c2fb04e7 ("kbuild: userprogs: fix bitsize and target
detection on clang"), should it be addressed in the same manner
(especially since I think the Fixes commit would be the same)? Given
that the filter expression is the same, maybe it would be possible to
unify them but that could happen as a follow up.

> ---
>  usr/include/Makefile | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/usr/include/Makefile b/usr/include/Makefile
> index 58a9cbe4eba4..99f9e904adcd 100644
> --- a/usr/include/Makefile
> +++ b/usr/include/Makefile
> @@ -10,7 +10,7 @@ UAPI_CFLAGS := -std=c90 -Wall -Werror=implicit-function-declaration
>  
>  # In theory, we do not care -m32 or -m64 for header compile tests.
>  # It is here just because CONFIG_CC_CAN_LINK is tested with -m32 or -m64.
> -UAPI_CFLAGS += $(filter -m32 -m64 --target=%, $(KBUILD_CFLAGS))
> +UAPI_CFLAGS += $(filter -m32 -m64 --target=%, $(KBUILD_CFLAGS) $(CLANG_FLAGS))
>  
>  # USERCFLAGS might contain sysroot location for CC.
>  UAPI_CFLAGS += $(USERCFLAGS)
> -- 
> 2.39.5
>

On Fri, Feb 21, 2025, at 22:26, Nathan Chancellor wrote:
> On Fri, Feb 21, 2025 at 10:20:07PM +0100, Arnd Bergmann wrote:
>> From: Arnd Bergmann <arnd@arndb.de>
>> 
>> The headercheck tries to call clang with a mix of compiler arguments
>> that don't include the target architecture. When building e.g. x86
>> headers on arm64, this produces a warning like
>> 
>>    clang: warning: unknown platform, assuming -mfloat-abi=soft
>> 
>> Add in the CLANG_FLAGS, which contain the target, in order to make it
>> build properly.
>> 
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>
> This looks like a similar problem as the one fixed by
> commit 1b71c2fb04e7 ("kbuild: userprogs: fix bitsize and target
> detection on clang"), should it be addressed in the same manner
> (especially since I think the Fixes commit would be the same)? Given
> that the filter expression is the same, maybe it would be possible to
> unify them but that could happen as a follow up.

My bad, I actually had an old patch that addressed both issues
and then dropped it when the Thomas' patch made it into linux-next
but ended up picking my old change for usr/include/Makefile instead
of trying to understand the difference.

I'll send a v2 after some more build testing.

     Arnd