On 14 January 2018 at 16:41, Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx> wrote: > Add an implementation of SHA3 to arm64 using the new special instructions, > and another one using scalar instructions but coded in assembler (#2) > > In preparation of that, fix a bug in the SHA3 (#1) and add some new test > vectors to get better test coverage (#3). > > v2: Drop generic SHA3 as a fallback for the arm64 module. Instead, provide > a special arm64 version to use as a fallback when the instructions are > not available or when executing in a context that does not allow SIMD > > Drop patches that simplify the generic SHA3 and make it reusable by > other modules. > > Ard Biesheuvel (3): > crypto/generic: sha3 - fixes for alignment and big endian operation > crypto/arm64: sha3 - new scalar + v8.2 Crypto Extensions > implementation > crypto/testmgr: sha3 - add new testcases > > arch/arm64/crypto/Kconfig | 4 + > arch/arm64/crypto/Makefile | 3 + > arch/arm64/crypto/sha3-arm64-core.S | 512 ++++++++++++++++++ > arch/arm64/crypto/sha3-arm64-glue.c | 192 +++++++ > crypto/sha3_generic.c | 5 +- > crypto/testmgr.h | 550 ++++++++++++++++++++ > 6 files changed, 1264 insertions(+), 2 deletions(-) > create mode 100644 arch/arm64/crypto/sha3-arm64-core.S > create mode 100644 arch/arm64/crypto/sha3-arm64-glue.c > Herbert, Could you hold off on the SHA-3 patches for a little while? With the performance fix for the generic code, it may no longer be worthwhile to have a special arm64 implementation as well. I will respin a series containing everything I think is needed. The SM3 patch is independent, and is good to go IMO (with Steve's Tested-by) Thanks, Ard.
