diff options
| author |   Ard Biesheuvel <ard.biesheuvel@linaro.org> | 2018-09-10 16:41:14 +0200 |
|---|---|---|
| committer |   Herbert Xu <herbert@gondor.apana.org.au> | 2018-09-21 13:24:50 +0800 |
| commit | dd597fb33ff0d2a158d7dd098be6dc849b7c2bcc (patch) | |
| tree | b3f3a1c2f95f4e5a087cabc435479b55a28bd4ab /arch/arm64/crypto/aes-modes.S | |
| parent | crypto: arm64/aes-blk - revert NEON yield for skciphers (diff) | |
| download | linux-dev-dd597fb33ff0d2a158d7dd098be6dc849b7c2bcc.tar.xz linux-dev-dd597fb33ff0d2a158d7dd098be6dc849b7c2bcc.zip | |
crypto: arm64/aes-blk - add support for CTS-CBC mode
Currently, we rely on the generic CTS chaining mode wrapper to
instantiate the cts(cbc(aes)) skcipher. Due to the high performance
of the ARMv8 Crypto Extensions AES instructions (~1 cycles per byte),
any overhead in the chaining mode layers is amplified, and so it pays
off considerably to fold the CTS handling into the SIMD routines.
On Cortex-A53, this results in a ~50% speedup for smaller input sizes.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Diffstat (limited to '')
| -rw-r--r-- | arch/arm64/crypto/aes-modes.S | 79 |
1 files changed, 78 insertions, 1 deletions
diff --git a/arch/arm64/crypto/aes-modes.S b/arch/arm64/crypto/aes-modes.S index 35632d11200f..9697eda3b4d1 100644 --- a/arch/arm64/crypto/aes-modes.S +++ b/arch/arm64/crypto/aes-modes.S @@ -171,6 +171,84 @@ AES_ENDPROC(aes_cbc_decrypt) /* + * aes_cbc_cts_encrypt(u8 out[], u8 const in[], u32 const rk[], + * int rounds, int bytes, u8 const iv[]) + * aes_cbc_cts_decrypt(u8 out[], u8 const in[], u32 const rk[], + * int rounds, int bytes, u8 const iv[]) + */ + +AES_ENTRY(aes_cbc_cts_encrypt) + adr_l x8, .Lcts_permute_table + sub x4, x4, #16 + add x9, x8, #32 + add x8, x8, x4 + sub x9, x9, x4 + ld1 {v3.16b}, [x8] + ld1 {v4.16b}, [x9] + + ld1 {v0.16b}, [x1], x4 /* overlapping loads */ + ld1 {v1.16b}, [x1] + + ld1 {v5.16b}, [x5] /* get iv */ + enc_prepare w3, x2, x6 + + eor v0.16b, v0.16b, v5.16b /* xor with iv */ + tbl v1.16b, {v1.16b}, v4.16b + encrypt_block v0, w3, x2, x6, w7 + + eor v1.16b, v1.16b, v0.16b + tbl v0.16b, {v0.16b}, v3.16b + encrypt_block v1, w3, x2, x6, w7 + + add x4, x0, x4 + st1 {v0.16b}, [x4] /* overlapping stores */ + st1 {v1.16b}, [x0] + ret +AES_ENDPROC(aes_cbc_cts_encrypt) + +AES_ENTRY(aes_cbc_cts_decrypt) + adr_l x8, .Lcts_permute_table + sub x4, x4, #16 + add x9, x8, #32 + add x8, x8, x4 + sub x9, x9, x4 + ld1 {v3.16b}, [x8] + ld1 {v4.16b}, [x9] + + ld1 {v0.16b}, [x1], x4 /* overlapping loads */ + ld1 {v1.16b}, [x1] + + ld1 {v5.16b}, [x5] /* get iv */ + dec_prepare w3, x2, x6 + + tbl v2.16b, {v1.16b}, v4.16b + decrypt_block v0, w3, x2, x6, w7 + eor v2.16b, v2.16b, v0.16b + + tbx v0.16b, {v1.16b}, v4.16b + tbl v2.16b, {v2.16b}, v3.16b + decrypt_block v0, w3, x2, x6, w7 + eor v0.16b, v0.16b, v5.16b /* xor with iv */ + + add x4, x0, x4 + st1 {v2.16b}, [x4] /* overlapping stores */ + st1 {v0.16b}, [x0] + ret +AES_ENDPROC(aes_cbc_cts_decrypt) + + .section ".rodata", "a" + .align 6 +.Lcts_permute_table: + .byte 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff + .byte 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff + .byte 0x0, 0x1, 0x2, 0x3, 0x4, 0x5, 0x6, 0x7 + .byte 0x8, 0x9, 0xa, 0xb, 0xc, 0xd, 0xe, 0xf + .byte 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff + .byte 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff + .previous + + + /* * aes_ctr_encrypt(u8 out[], u8 const in[], u8 const rk[], int rounds, * int blocks, u8 ctr[]) */ @@ -253,7 +331,6 @@ AES_ENTRY(aes_ctr_encrypt) ins v4.d[0], x7 b .Lctrcarrydone AES_ENDPROC(aes_ctr_encrypt) - .ltorg /* |