On Mon, 29 Jul 2019 at 12:12, Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> wrote: > > Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx> wrote: > > The generic aegis128 driver has been updated to support using SIMD > > intrinsics to implement the core AES based transform, and this has > > been wired up for ARM and arm64, which both provide a simd.h header. > > > > As it turns out, most architectures don't provide this header, even > > though a version of it exists in include/asm-generic, and this is > > not taken into account by the aegis128 driver, resulting in build > > failures on those architectures. > > > > So update the aegis128 code to only import simd.h (and the related > > header in internal/crypto) if the SIMD functionality is enabled for > > this driver. > > > > Reported-by: Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> > > Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx> > > --- > > crypto/aegis128-core.c | 22 +++++++++++++++++----- > > 1 file changed, 17 insertions(+), 5 deletions(-) > > I think we should dig a little deeper into why asm-generic isn't > working for this case. AFAICS we rely on the same mechanism for > errno.h on m68k and that obviously works. > It is simply a matter of adding simd.h to the various arch/<...>/include/asm/Kbuild files, but we'd have to do that for all architectures.
