[PATCH] x86/build: Do not allow X86_NATIVE_CPU to be enabled on non-x86_64 hosts

[PATCH] x86/build: Do not allow X86_NATIVE_CPU to be enabled on non-x86_64 hosts

[Nathan Chancellor]

When cross compiling x86_64 from an aarch64 machine using clang with
CONFIG_X86_NATIVE_CPU enabled, there may be a slew of warnings from each
.c file, which makes it nearly impossible to find real issues.

  '+crc' is not a recognized feature for this target (ignoring feature)
  '+fp-armv8' is not a recognized feature for this target (ignoring feature)
  '+lse' is not a recognized feature for this target (ignoring feature)
  '+neon' is not a recognized feature for this target (ignoring feature)
  '+sha2' is not a recognized feature for this target (ignoring feature)
  '+sha3' is not a recognized feature for this target (ignoring feature)
  '-rand' is not a recognized feature for this target (ignoring feature)
  '-sme' is not a recognized feature for this target (ignoring feature)
  '-sme2' is not a recognized feature for this target (ignoring feature)
  '-sve' is not a recognized feature for this target (ignoring feature)

Internally, clang expands '-march=native' into a list of target features
to enable based on the detected CPU. The x86 part of the clang driver
notes

  // FIXME: Reject attempts to use -march=native unless the target matches
  // the host.

which was first added by the initial '-march=native' support in clang in
2009 [1] (other architecture drivers appear to have a similar problem).

Due to the lack of validation that the host CPU is an x86_64 CPU,
'-march=native' is expanded using the aarch64 host CPU's target
features, which are unrecognized by the x86 backend.

While this should obviously be fixed in the compiler, this does not help
currently released toolchains that have this issue. As a simple
workaround, only allow CONFIG_X86_NATIVE_CPU to be enabled when
targeting and running on an x86_64 host.

Cc: stable@vger.kernel.org
Fixes: ea1dcca1de12 ("x86/kbuild/64: Add the CONFIG_X86_NATIVE_CPU option to locally optimize the kernel with '-march=native'")
Link: https://github.com/llvm/llvm-project/commit/e13ada6a3b236f505b24501d456dec2518e4944c [1]
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 arch/x86/Kconfig.cpu | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/Kconfig.cpu b/arch/x86/Kconfig.cpu
index 6e7a366f0798..4043b7227e36 100644
--- a/arch/x86/Kconfig.cpu
+++ b/arch/x86/Kconfig.cpu
@@ -216,7 +216,7 @@ config RUSTC_HAS_APXF
 
 config X86_NATIVE_CPU
 	bool "Build and optimize for local/native CPU"
-	depends on X86_64
+	depends on X86_64 && $(success,test "$(shell,uname -m)" = "x86_64")
 	depends on CC_HAS_MARCH_NATIVE
 	depends on !RUST || RUSTC_HAS_APXF
 	help

---
base-commit: 6628e7d4c6b25c8d18f1345e3a20d21829aae7e4
base-tree: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
change-id: 20261113-x86-no-march-native-on-non-x86_64-88233b48a7d7

Best regards,
--  
Cheers,
Nathan

Re: [PATCH] x86/build: Do not allow X86_NATIVE_CPU to be enabled on non-x86_64 hosts

[Borislav Petkov]

On Fri, Nov 13, 2026 at 05:31:34PM -0700, Nathan Chancellor wrote:
> When cross compiling x86_64 from an aarch64 machine using clang with
> CONFIG_X86_NATIVE_CPU enabled, there may be a slew of warnings from each
> .c file, which makes it nearly impossible to find real issues.
> 
>   '+crc' is not a recognized feature for this target (ignoring feature)
>   '+fp-armv8' is not a recognized feature for this target (ignoring feature)
>   '+lse' is not a recognized feature for this target (ignoring feature)
>   '+neon' is not a recognized feature for this target (ignoring feature)
>   '+sha2' is not a recognized feature for this target (ignoring feature)
>   '+sha3' is not a recognized feature for this target (ignoring feature)
>   '-rand' is not a recognized feature for this target (ignoring feature)
>   '-sme' is not a recognized feature for this target (ignoring feature)
>   '-sme2' is not a recognized feature for this target (ignoring feature)
>   '-sve' is not a recognized feature for this target (ignoring feature)
> 
> Internally, clang expands '-march=native' into a list of target features
> to enable based on the detected CPU. The x86 part of the clang driver
> notes
> 
>   // FIXME: Reject attempts to use -march=native unless the target matches
>   // the host.
> 
> which was first added by the initial '-march=native' support in clang in
> 2009 [1] (other architecture drivers appear to have a similar problem).
> 
> Due to the lack of validation that the host CPU is an x86_64 CPU,
> '-march=native' is expanded using the aarch64 host CPU's target
> features, which are unrecognized by the x86 backend.
> 
> While this should obviously be fixed in the compiler, this does not help
> currently released toolchains that have this issue. As a simple
> workaround, only allow CONFIG_X86_NATIVE_CPU to be enabled when
> targeting and running on an x86_64 host.
> 
> Cc: stable@vger.kernel.org
> Fixes: ea1dcca1de12 ("x86/kbuild/64: Add the CONFIG_X86_NATIVE_CPU option to locally optimize the kernel with '-march=native'")
> Link: https://github.com/llvm/llvm-project/commit/e13ada6a3b236f505b24501d456dec2518e4944c [1]
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
>  arch/x86/Kconfig.cpu | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/Kconfig.cpu b/arch/x86/Kconfig.cpu
> index 6e7a366f0798..4043b7227e36 100644
> --- a/arch/x86/Kconfig.cpu
> +++ b/arch/x86/Kconfig.cpu
> @@ -216,7 +216,7 @@ config RUSTC_HAS_APXF
>  
>  config X86_NATIVE_CPU
>  	bool "Build and optimize for local/native CPU"
> -	depends on X86_64
> +	depends on X86_64 && $(success,test "$(shell,uname -m)" = "x86_64")
>  	depends on CC_HAS_MARCH_NATIVE
>  	depends on !RUST || RUSTC_HAS_APXF
>  	help

This looks like a whack-a-mole-y fix to me. Especially if the other arches
would have the same problem.

So wouldn't a more generic fix checking

	native arch == target arch

somewhere in the makefile be better?

Thx. 

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH] x86/build: Do not allow X86_NATIVE_CPU to be enabled on non-x86_64 hosts

[Nathan Chancellor]

On Sat, Nov 14, 2026 at 12:25:24PM -0800, Borislav Petkov wrote:
> This looks like a whack-a-mole-y fix to me. Especially if the other arches
> would have the same problem.

I made that comment more about the different architecture
implementations in the clang driver but I guess it could apply to the
kernel as well. However...

> So wouldn't a more generic fix checking
> 
> 	native arch == target arch
> 
> somewhere in the makefile be better?

No other architectures provide an option to turn on '-march=native' like
x86 does. It all lives within arch/x86, so I would just be adding a
check to arch/x86/Makefile instead of arch/x86/Kconfig.cpu. Making the
option unselectable seems like a better option. If other architectures
did want to add this option, I agree generically abstacting this check
would be reasonable.

-- 
Cheers,
Nathan

Re: [PATCH] x86/build: Do not allow X86_NATIVE_CPU to be enabled on non-x86_64 hosts

[Borislav Petkov]

On Sat, Nov 14, 2026 at 01:46:05PM -0700, Nathan Chancellor wrote:
> No other architectures provide an option to turn on '-march=native' like
> x86 does. It all lives within arch/x86, so I would just be adding a
> check to arch/x86/Makefile instead of arch/x86/Kconfig.cpu. Making the
> option unselectable seems like a better option. If other architectures
> did want to add this option, I agree generically abstacting this check
> would be reasonable.

Are you going to move it somewhere generic when other arches decide all of
a sudden they love the native thing?

(If it were me, I'd simply whack that native thing. It is such a waste of
energy).

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette