crypto: arm64/aes-neon - fix for big endian
authorArd Biesheuvel <ard.biesheuvel@linaro.org>
Tue, 11 Oct 2016 18:15:18 +0000 (19:15 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 12 Jan 2017 10:22:50 +0000 (11:22 +0100)
commit83f8a200af458ef65ee3205e000e9581194f33e7
tree46d9be846180509e758186b81ef0362a165b4d7a
parentbd17e2d366154b884abfce23f136e9b74c2d0c75
crypto: arm64/aes-neon - fix for big endian

commit a2c435cc99862fd3d165e1b66bf48ac72c839c62 upstream.

The AES implementation using pure NEON instructions relies on the generic
AES key schedule generation routines, which store the round keys as arrays
of 32-bit quantities stored in memory using native endianness. This means
we should refer to these round keys using 4x4 loads rather than 16x1 loads.
In addition, the ShiftRows tables are loading using a single scalar load,
which is also affected by endianness, so emit these tables in the correct
order depending on whether we are building for big endian or not.

Fixes: 49788fe2a128 ("arm64/crypto: AES-ECB/CBC/CTR/XTS using ARMv8 NEON and Crypto Extensions")
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/arm64/crypto/aes-neon.S