Prior to [1], all non-GPL modules were able to make use of SIMD on x86 by making use of the __kernel_fpu_* API. Given that __kernel_fpu_* were both EXPORT_SYMBOL'd and kernel_fpu_* are such trivial wrappers around the now-static __kernel_fpu_*, it seems to me that there is no reason to have different licensing rules for them.
In the case of OpenZFS, the lack of SIMD on newer Linux kernels has caused significant performance problems (since ZFS uses SIMD for calculation of blkptr checksums as well as raidz calculations). [1]: commit 12209993e98c ("x86/fpu: Don't export __kernel_fpu_{begin,end}()") Cc: "Jason A. Donenfeld" <ja...@zx2c4.com> Signed-off-by: Aleksa Sarai <cyp...@cyphar.com> --- arch/x86/kernel/fpu/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c index 2e5003fef51a..8de5687a470d 100644 --- a/arch/x86/kernel/fpu/core.c +++ b/arch/x86/kernel/fpu/core.c @@ -127,14 +127,14 @@ void kernel_fpu_begin(void) preempt_disable(); __kernel_fpu_begin(); } -EXPORT_SYMBOL_GPL(kernel_fpu_begin); +EXPORT_SYMBOL(kernel_fpu_begin); void kernel_fpu_end(void) { __kernel_fpu_end(); preempt_enable(); } -EXPORT_SYMBOL_GPL(kernel_fpu_end); +EXPORT_SYMBOL(kernel_fpu_end); /* * Save the FPU state (mark it for reload if necessary): -- 2.21.0