diff options
Diffstat (limited to 'arch/x86/crypto/camellia-aesni-avx2-asm_64.S')
| -rw-r--r-- | arch/x86/crypto/camellia-aesni-avx2-asm_64.S | 1047 |
1 files changed, 1047 insertions, 0 deletions
diff --git a/arch/x86/crypto/camellia-aesni-avx2-asm_64.S b/arch/x86/crypto/camellia-aesni-avx2-asm_64.S new file mode 100644 index 0000000000..a0eb94e53b --- /dev/null +++ b/arch/x86/crypto/camellia-aesni-avx2-asm_64.S @@ -0,0 +1,1047 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * x86_64/AVX2/AES-NI assembler implementation of Camellia + * + * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@iki.fi> + */ + +#include <linux/linkage.h> +#include <asm/frame.h> + +#define CAMELLIA_TABLE_BYTE_LEN 272 + +/* struct camellia_ctx: */ +#define key_table 0 +#define key_length CAMELLIA_TABLE_BYTE_LEN + +/* register macros */ +#define CTX %rdi +#define RIO %r8 + +/********************************************************************** + helper macros + **********************************************************************/ +#define filter_8bit(x, lo_t, hi_t, mask4bit, tmp0) \ + vpand x, mask4bit, tmp0; \ + vpandn x, mask4bit, x; \ + vpsrld $4, x, x; \ + \ + vpshufb tmp0, lo_t, tmp0; \ + vpshufb x, hi_t, x; \ + vpxor tmp0, x, x; + +#define ymm0_x xmm0 +#define ymm1_x xmm1 +#define ymm2_x xmm2 +#define ymm3_x xmm3 +#define ymm4_x xmm4 +#define ymm5_x xmm5 +#define ymm6_x xmm6 +#define ymm7_x xmm7 +#define ymm8_x xmm8 +#define ymm9_x xmm9 +#define ymm10_x xmm10 +#define ymm11_x xmm11 +#define ymm12_x xmm12 +#define ymm13_x xmm13 +#define ymm14_x xmm14 +#define ymm15_x xmm15 + +/********************************************************************** + 32-way camellia + **********************************************************************/ + +/* + * IN: + * x0..x7: byte-sliced AB state + * mem_cd: register pointer storing CD state + * key: index for key material + * OUT: + * x0..x7: new byte-sliced CD state + */ +#define roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, t0, t1, t2, t3, t4, t5, t6, \ + t7, mem_cd, key) \ + /* \ + * S-function with AES subbytes \ + */ \ + vbroadcasti128 .Linv_shift_row(%rip), t4; \ + vpbroadcastd .L0f0f0f0f(%rip), t7; \ + vbroadcasti128 .Lpre_tf_lo_s1(%rip), t5; \ + vbroadcasti128 .Lpre_tf_hi_s1(%rip), t6; \ + vbroadcasti128 .Lpre_tf_lo_s4(%rip), t2; \ + vbroadcasti128 .Lpre_tf_hi_s4(%rip), t3; \ + \ + /* AES inverse shift rows */ \ + vpshufb t4, x0, x0; \ + vpshufb t4, x7, x7; \ + vpshufb t4, x3, x3; \ + vpshufb t4, x6, x6; \ + vpshufb t4, x2, x2; \ + vpshufb t4, x5, x5; \ + vpshufb t4, x1, x1; \ + vpshufb t4, x4, x4; \ + \ + /* prefilter sboxes 1, 2 and 3 */ \ + /* prefilter sbox 4 */ \ + filter_8bit(x0, t5, t6, t7, t4); \ + filter_8bit(x7, t5, t6, t7, t4); \ + vextracti128 $1, x0, t0##_x; \ + vextracti128 $1, x7, t1##_x; \ + filter_8bit(x3, t2, t3, t7, t4); \ + filter_8bit(x6, t2, t3, t7, t4); \ + vextracti128 $1, x3, t3##_x; \ + vextracti128 $1, x6, t2##_x; \ + filter_8bit(x2, t5, t6, t7, t4); \ + filter_8bit(x5, t5, t6, t7, t4); \ + filter_8bit(x1, t5, t6, t7, t4); \ + filter_8bit(x4, t5, t6, t7, t4); \ + \ + vpxor t4##_x, t4##_x, t4##_x; \ + \ + /* AES subbytes + AES shift rows */ \ + vextracti128 $1, x2, t6##_x; \ + vextracti128 $1, x5, t5##_x; \ + vaesenclast t4##_x, x0##_x, x0##_x; \ + vaesenclast t4##_x, t0##_x, t0##_x; \ + vinserti128 $1, t0##_x, x0, x0; \ + vaesenclast t4##_x, x7##_x, x7##_x; \ + vaesenclast t4##_x, t1##_x, t1##_x; \ + vinserti128 $1, t1##_x, x7, x7; \ + vaesenclast t4##_x, x3##_x, x3##_x; \ + vaesenclast t4##_x, t3##_x, t3##_x; \ + vinserti128 $1, t3##_x, x3, x3; \ + vaesenclast t4##_x, x6##_x, x6##_x; \ + vaesenclast t4##_x, t2##_x, t2##_x; \ + vinserti128 $1, t2##_x, x6, x6; \ + vextracti128 $1, x1, t3##_x; \ + vextracti128 $1, x4, t2##_x; \ + vbroadcasti128 .Lpost_tf_lo_s1(%rip), t0; \ + vbroadcasti128 .Lpost_tf_hi_s1(%rip), t1; \ + vaesenclast t4##_x, x2##_x, x2##_x; \ + vaesenclast t4##_x, t6##_x, t6##_x; \ + vinserti128 $1, t6##_x, x2, x2; \ + vaesenclast t4##_x, x5##_x, x5##_x; \ + vaesenclast t4##_x, t5##_x, t5##_x; \ + vinserti128 $1, t5##_x, x5, x5; \ + vaesenclast t4##_x, x1##_x, x1##_x; \ + vaesenclast t4##_x, t3##_x, t3##_x; \ + vinserti128 $1, t3##_x, x1, x1; \ + vaesenclast t4##_x, x4##_x, x4##_x; \ + vaesenclast t4##_x, t2##_x, t2##_x; \ + vinserti128 $1, t2##_x, x4, x4; \ + \ + /* postfilter sboxes 1 and 4 */ \ + vbroadcasti128 .Lpost_tf_lo_s3(%rip), t2; \ + vbroadcasti128 .Lpost_tf_hi_s3(%rip), t3; \ + filter_8bit(x0, t0, t1, t7, t6); \ + filter_8bit(x7, t0, t1, t7, t6); \ + filter_8bit(x3, t0, t1, t7, t6); \ + filter_8bit(x6, t0, t1, t7, t6); \ + \ + /* postfilter sbox 3 */ \ + vbroadcasti128 .Lpost_tf_lo_s2(%rip), t4; \ + vbroadcasti128 .Lpost_tf_hi_s2(%rip), t5; \ + filter_8bit(x2, t2, t3, t7, t6); \ + filter_8bit(x5, t2, t3, t7, t6); \ + \ + vpbroadcastq key, t0; /* higher 64-bit duplicate ignored */ \ + \ + /* postfilter sbox 2 */ \ + filter_8bit(x1, t4, t5, t7, t2); \ + filter_8bit(x4, t4, t5, t7, t2); \ + vpxor t7, t7, t7; \ + \ + vpsrldq $1, t0, t1; \ + vpsrldq $2, t0, t2; \ + vpshufb t7, t1, t1; \ + vpsrldq $3, t0, t3; \ + \ + /* P-function */ \ + vpxor x5, x0, x0; \ + vpxor x6, x1, x1; \ + vpxor x7, x2, x2; \ + vpxor x4, x3, x3; \ + \ + vpshufb t7, t2, t2; \ + vpsrldq $4, t0, t4; \ + vpshufb t7, t3, t3; \ + vpsrldq $5, t0, t5; \ + vpshufb t7, t4, t4; \ + \ + vpxor x2, x4, x4; \ + vpxor x3, x5, x5; \ + vpxor x0, x6, x6; \ + vpxor x1, x7, x7; \ + \ + vpsrldq $6, t0, t6; \ + vpshufb t7, t5, t5; \ + vpshufb t7, t6, t6; \ + \ + vpxor x7, x0, x0; \ + vpxor x4, x1, x1; \ + vpxor x5, x2, x2; \ + vpxor x6, x3, x3; \ + \ + vpxor x3, x4, x4; \ + vpxor x0, x5, x5; \ + vpxor x1, x6, x6; \ + vpxor x2, x7, x7; /* note: high and low parts swapped */ \ + \ + /* Add key material and result to CD (x becomes new CD) */ \ + \ + vpxor t6, x1, x1; \ + vpxor 5 * 32(mem_cd), x1, x1; \ + \ + vpsrldq $7, t0, t6; \ + vpshufb t7, t0, t0; \ + vpshufb t7, t6, t7; \ + \ + vpxor t7, x0, x0; \ + vpxor 4 * 32(mem_cd), x0, x0; \ + \ + vpxor t5, x2, x2; \ + vpxor 6 * 32(mem_cd), x2, x2; \ + \ + vpxor t4, x3, x3; \ + vpxor 7 * 32(mem_cd), x3, x3; \ + \ + vpxor t3, x4, x4; \ + vpxor 0 * 32(mem_cd), x4, x4; \ + \ + vpxor t2, x5, x5; \ + vpxor 1 * 32(mem_cd), x5, x5; \ + \ + vpxor t1, x6, x6; \ + vpxor 2 * 32(mem_cd), x6, x6; \ + \ + vpxor t0, x7, x7; \ + vpxor 3 * 32(mem_cd), x7, x7; + +/* + * Size optimization... with inlined roundsm32 binary would be over 5 times + * larger and would only marginally faster. + */ +SYM_FUNC_START_LOCAL(roundsm32_x0_x1_x2_x3_x4_x5_x6_x7_y0_y1_y2_y3_y4_y5_y6_y7_cd) + roundsm32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, %ymm15, + %rcx, (%r9)); + RET; +SYM_FUNC_END(roundsm32_x0_x1_x2_x3_x4_x5_x6_x7_y0_y1_y2_y3_y4_y5_y6_y7_cd) + +SYM_FUNC_START_LOCAL(roundsm32_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab) + roundsm32(%ymm4, %ymm5, %ymm6, %ymm7, %ymm0, %ymm1, %ymm2, %ymm3, + %ymm12, %ymm13, %ymm14, %ymm15, %ymm8, %ymm9, %ymm10, %ymm11, + %rax, (%r9)); + RET; +SYM_FUNC_END(roundsm32_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab) + +/* + * IN/OUT: + * x0..x7: byte-sliced AB state preloaded + * mem_ab: byte-sliced AB state in memory + * mem_cb: byte-sliced CD state in memory + */ +#define two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, i, dir, store_ab) \ + leaq (key_table + (i) * 8)(CTX), %r9; \ + call roundsm32_x0_x1_x2_x3_x4_x5_x6_x7_y0_y1_y2_y3_y4_y5_y6_y7_cd; \ + \ + vmovdqu x0, 4 * 32(mem_cd); \ + vmovdqu x1, 5 * 32(mem_cd); \ + vmovdqu x2, 6 * 32(mem_cd); \ + vmovdqu x3, 7 * 32(mem_cd); \ + vmovdqu x4, 0 * 32(mem_cd); \ + vmovdqu x5, 1 * 32(mem_cd); \ + vmovdqu x6, 2 * 32(mem_cd); \ + vmovdqu x7, 3 * 32(mem_cd); \ + \ + leaq (key_table + ((i) + (dir)) * 8)(CTX), %r9; \ + call roundsm32_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab; \ + \ + store_ab(x0, x1, x2, x3, x4, x5, x6, x7, mem_ab); + +#define dummy_store(x0, x1, x2, x3, x4, x5, x6, x7, mem_ab) /* do nothing */ + +#define store_ab_state(x0, x1, x2, x3, x4, x5, x6, x7, mem_ab) \ + /* Store new AB state */ \ + vmovdqu x4, 4 * 32(mem_ab); \ + vmovdqu x5, 5 * 32(mem_ab); \ + vmovdqu x6, 6 * 32(mem_ab); \ + vmovdqu x7, 7 * 32(mem_ab); \ + vmovdqu x0, 0 * 32(mem_ab); \ + vmovdqu x1, 1 * 32(mem_ab); \ + vmovdqu x2, 2 * 32(mem_ab); \ + vmovdqu x3, 3 * 32(mem_ab); + +#define enc_rounds32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, i) \ + two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, (i) + 2, 1, store_ab_state); \ + two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, (i) + 4, 1, store_ab_state); \ + two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, (i) + 6, 1, dummy_store); + +#define dec_rounds32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, i) \ + two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, (i) + 7, -1, store_ab_state); \ + two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, (i) + 5, -1, store_ab_state); \ + two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd, (i) + 3, -1, dummy_store); + +/* + * IN: + * v0..3: byte-sliced 32-bit integers + * OUT: + * v0..3: (IN <<< 1) + */ +#define rol32_1_32(v0, v1, v2, v3, t0, t1, t2, zero) \ + vpcmpgtb v0, zero, t0; \ + vpaddb v0, v0, v0; \ + vpabsb t0, t0; \ + \ + vpcmpgtb v1, zero, t1; \ + vpaddb v1, v1, v1; \ + vpabsb t1, t1; \ + \ + vpcmpgtb v2, zero, t2; \ + vpaddb v2, v2, v2; \ + vpabsb t2, t2; \ + \ + vpor t0, v1, v1; \ + \ + vpcmpgtb v3, zero, t0; \ + vpaddb v3, v3, v3; \ + vpabsb t0, t0; \ + \ + vpor t1, v2, v2; \ + vpor t2, v3, v3; \ + vpor t0, v0, v0; + +/* + * IN: + * r: byte-sliced AB state in memory + * l: byte-sliced CD state in memory + * OUT: + * x0..x7: new byte-sliced CD state + */ +#define fls32(l, l0, l1, l2, l3, l4, l5, l6, l7, r, t0, t1, t2, t3, tt0, \ + tt1, tt2, tt3, kll, klr, krl, krr) \ + /* \ + * t0 = kll; \ + * t0 &= ll; \ + * lr ^= rol32(t0, 1); \ + */ \ + vpbroadcastd kll, t0; /* only lowest 32-bit used */ \ + vpxor tt0, tt0, tt0; \ + vpshufb tt0, t0, t3; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t2; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t1; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t0; \ + \ + vpand l0, t0, t0; \ + vpand l1, t1, t1; \ + vpand l2, t2, t2; \ + vpand l3, t3, t3; \ + \ + rol32_1_32(t3, t2, t1, t0, tt1, tt2, tt3, tt0); \ + \ + vpxor l4, t0, l4; \ + vpbroadcastd krr, t0; /* only lowest 32-bit used */ \ + vmovdqu l4, 4 * 32(l); \ + vpxor l5, t1, l5; \ + vmovdqu l5, 5 * 32(l); \ + vpxor l6, t2, l6; \ + vmovdqu l6, 6 * 32(l); \ + vpxor l7, t3, l7; \ + vmovdqu l7, 7 * 32(l); \ + \ + /* \ + * t2 = krr; \ + * t2 |= rr; \ + * rl ^= t2; \ + */ \ + \ + vpshufb tt0, t0, t3; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t2; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t1; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t0; \ + \ + vpor 4 * 32(r), t0, t0; \ + vpor 5 * 32(r), t1, t1; \ + vpor 6 * 32(r), t2, t2; \ + vpor 7 * 32(r), t3, t3; \ + \ + vpxor 0 * 32(r), t0, t0; \ + vpxor 1 * 32(r), t1, t1; \ + vpxor 2 * 32(r), t2, t2; \ + vpxor 3 * 32(r), t3, t3; \ + vmovdqu t0, 0 * 32(r); \ + vpbroadcastd krl, t0; /* only lowest 32-bit used */ \ + vmovdqu t1, 1 * 32(r); \ + vmovdqu t2, 2 * 32(r); \ + vmovdqu t3, 3 * 32(r); \ + \ + /* \ + * t2 = krl; \ + * t2 &= rl; \ + * rr ^= rol32(t2, 1); \ + */ \ + vpshufb tt0, t0, t3; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t2; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t1; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t0; \ + \ + vpand 0 * 32(r), t0, t0; \ + vpand 1 * 32(r), t1, t1; \ + vpand 2 * 32(r), t2, t2; \ + vpand 3 * 32(r), t3, t3; \ + \ + rol32_1_32(t3, t2, t1, t0, tt1, tt2, tt3, tt0); \ + \ + vpxor 4 * 32(r), t0, t0; \ + vpxor 5 * 32(r), t1, t1; \ + vpxor 6 * 32(r), t2, t2; \ + vpxor 7 * 32(r), t3, t3; \ + vmovdqu t0, 4 * 32(r); \ + vpbroadcastd klr, t0; /* only lowest 32-bit used */ \ + vmovdqu t1, 5 * 32(r); \ + vmovdqu t2, 6 * 32(r); \ + vmovdqu t3, 7 * 32(r); \ + \ + /* \ + * t0 = klr; \ + * t0 |= lr; \ + * ll ^= t0; \ + */ \ + \ + vpshufb tt0, t0, t3; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t2; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t1; \ + vpsrldq $1, t0, t0; \ + vpshufb tt0, t0, t0; \ + \ + vpor l4, t0, t0; \ + vpor l5, t1, t1; \ + vpor l6, t2, t2; \ + vpor l7, t3, t3; \ + \ + vpxor l0, t0, l0; \ + vmovdqu l0, 0 * 32(l); \ + vpxor l1, t1, l1; \ + vmovdqu l1, 1 * 32(l); \ + vpxor l2, t2, l2; \ + vmovdqu l2, 2 * 32(l); \ + vpxor l3, t3, l3; \ + vmovdqu l3, 3 * 32(l); + +#define transpose_4x4(x0, x1, x2, x3, t1, t2) \ + vpunpckhdq x1, x0, t2; \ + vpunpckldq x1, x0, x0; \ + \ + vpunpckldq x3, x2, t1; \ + vpunpckhdq x3, x2, x2; \ + \ + vpunpckhqdq t1, x0, x1; \ + vpunpcklqdq t1, x0, x0; \ + \ + vpunpckhqdq x2, t2, x3; \ + vpunpcklqdq x2, t2, x2; + +#define byteslice_16x16b_fast(a0, b0, c0, d0, a1, b1, c1, d1, a2, b2, c2, d2, \ + a3, b3, c3, d3, st0, st1) \ + vmovdqu d2, st0; \ + vmovdqu d3, st1; \ + transpose_4x4(a0, a1, a2, a3, d2, d3); \ + transpose_4x4(b0, b1, b2, b3, d2, d3); \ + vmovdqu st0, d2; \ + vmovdqu st1, d3; \ + \ + vmovdqu a0, st0; \ + vmovdqu a1, st1; \ + transpose_4x4(c0, c1, c2, c3, a0, a1); \ + transpose_4x4(d0, d1, d2, d3, a0, a1); \ + \ + vbroadcasti128 .Lshufb_16x16b(%rip), a0; \ + vmovdqu st1, a1; \ + vpshufb a0, a2, a2; \ + vpshufb a0, a3, a3; \ + vpshufb a0, b0, b0; \ + vpshufb a0, b1, b1; \ + vpshufb a0, b2, b2; \ + vpshufb a0, b3, b3; \ + vpshufb a0, a1, a1; \ + vpshufb a0, c0, c0; \ + vpshufb a0, c1, c1; \ + vpshufb a0, c2, c2; \ + vpshufb a0, c3, c3; \ + vpshufb a0, d0, d0; \ + vpshufb a0, d1, d1; \ + vpshufb a0, d2, d2; \ + vpshufb a0, d3, d3; \ + vmovdqu d3, st1; \ + vmovdqu st0, d3; \ + vpshufb a0, d3, a0; \ + vmovdqu d2, st0; \ + \ + transpose_4x4(a0, b0, c0, d0, d2, d3); \ + transpose_4x4(a1, b1, c1, d1, d2, d3); \ + vmovdqu st0, d2; \ + vmovdqu st1, d3; \ + \ + vmovdqu b0, st0; \ + vmovdqu b1, st1; \ + transpose_4x4(a2, b2, c2, d2, b0, b1); \ + transpose_4x4(a3, b3, c3, d3, b0, b1); \ + vmovdqu st0, b0; \ + vmovdqu st1, b1; \ + /* does not adjust output bytes inside vectors */ + +/* load blocks to registers and apply pre-whitening */ +#define inpack32_pre(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, rio, key) \ + vpbroadcastq key, x0; \ + vpshufb .Lpack_bswap(%rip), x0, x0; \ + \ + vpxor 0 * 32(rio), x0, y7; \ + vpxor 1 * 32(rio), x0, y6; \ + vpxor 2 * 32(rio), x0, y5; \ + vpxor 3 * 32(rio), x0, y4; \ + vpxor 4 * 32(rio), x0, y3; \ + vpxor 5 * 32(rio), x0, y2; \ + vpxor 6 * 32(rio), x0, y1; \ + vpxor 7 * 32(rio), x0, y0; \ + vpxor 8 * 32(rio), x0, x7; \ + vpxor 9 * 32(rio), x0, x6; \ + vpxor 10 * 32(rio), x0, x5; \ + vpxor 11 * 32(rio), x0, x4; \ + vpxor 12 * 32(rio), x0, x3; \ + vpxor 13 * 32(rio), x0, x2; \ + vpxor 14 * 32(rio), x0, x1; \ + vpxor 15 * 32(rio), x0, x0; + +/* byteslice pre-whitened blocks and store to temporary memory */ +#define inpack32_post(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, mem_ab, mem_cd) \ + byteslice_16x16b_fast(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, \ + y4, y5, y6, y7, (mem_ab), (mem_cd)); \ + \ + vmovdqu x0, 0 * 32(mem_ab); \ + vmovdqu x1, 1 * 32(mem_ab); \ + vmovdqu x2, 2 * 32(mem_ab); \ + vmovdqu x3, 3 * 32(mem_ab); \ + vmovdqu x4, 4 * 32(mem_ab); \ + vmovdqu x5, 5 * 32(mem_ab); \ + vmovdqu x6, 6 * 32(mem_ab); \ + vmovdqu x7, 7 * 32(mem_ab); \ + vmovdqu y0, 0 * 32(mem_cd); \ + vmovdqu y1, 1 * 32(mem_cd); \ + vmovdqu y2, 2 * 32(mem_cd); \ + vmovdqu y3, 3 * 32(mem_cd); \ + vmovdqu y4, 4 * 32(mem_cd); \ + vmovdqu y5, 5 * 32(mem_cd); \ + vmovdqu y6, 6 * 32(mem_cd); \ + vmovdqu y7, 7 * 32(mem_cd); + +/* de-byteslice, apply post-whitening and store blocks */ +#define outunpack32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, \ + y5, y6, y7, key, stack_tmp0, stack_tmp1) \ + byteslice_16x16b_fast(y0, y4, x0, x4, y1, y5, x1, x5, y2, y6, x2, x6, \ + y3, y7, x3, x7, stack_tmp0, stack_tmp1); \ + \ + vmovdqu x0, stack_tmp0; \ + \ + vpbroadcastq key, x0; \ + vpshufb .Lpack_bswap(%rip), x0, x0; \ + \ + vpxor x0, y7, y7; \ + vpxor x0, y6, y6; \ + vpxor x0, y5, y5; \ + vpxor x0, y4, y4; \ + vpxor x0, y3, y3; \ + vpxor x0, y2, y2; \ + vpxor x0, y1, y1; \ + vpxor x0, y0, y0; \ + vpxor x0, x7, x7; \ + vpxor x0, x6, x6; \ + vpxor x0, x5, x5; \ + vpxor x0, x4, x4; \ + vpxor x0, x3, x3; \ + vpxor x0, x2, x2; \ + vpxor x0, x1, x1; \ + vpxor stack_tmp0, x0, x0; + +#define write_output(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \ + y6, y7, rio) \ + vmovdqu x0, 0 * 32(rio); \ + vmovdqu x1, 1 * 32(rio); \ + vmovdqu x2, 2 * 32(rio); \ + vmovdqu x3, 3 * 32(rio); \ + vmovdqu x4, 4 * 32(rio); \ + vmovdqu x5, 5 * 32(rio); \ + vmovdqu x6, 6 * 32(rio); \ + vmovdqu x7, 7 * 32(rio); \ + vmovdqu y0, 8 * 32(rio); \ + vmovdqu y1, 9 * 32(rio); \ + vmovdqu y2, 10 * 32(rio); \ + vmovdqu y3, 11 * 32(rio); \ + vmovdqu y4, 12 * 32(rio); \ + vmovdqu y5, 13 * 32(rio); \ + vmovdqu y6, 14 * 32(rio); \ + vmovdqu y7, 15 * 32(rio); + + +.section .rodata.cst32.shufb_16x16b, "aM", @progbits, 32 +.align 32 +#define SHUFB_BYTES(idx) \ + 0 + (idx), 4 + (idx), 8 + (idx), 12 + (idx) +.Lshufb_16x16b: + .byte SHUFB_BYTES(0), SHUFB_BYTES(1), SHUFB_BYTES(2), SHUFB_BYTES(3) + .byte SHUFB_BYTES(0), SHUFB_BYTES(1), SHUFB_BYTES(2), SHUFB_BYTES(3) + +.section .rodata.cst32.pack_bswap, "aM", @progbits, 32 +.align 32 +.Lpack_bswap: + .long 0x00010203, 0x04050607, 0x80808080, 0x80808080 + .long 0x00010203, 0x04050607, 0x80808080, 0x80808080 + +/* NB: section is mergeable, all elements must be aligned 16-byte blocks */ +.section .rodata.cst16, "aM", @progbits, 16 +.align 16 + +/* + * pre-SubByte transform + * + * pre-lookup for sbox1, sbox2, sbox3: + * swap_bitendianness( + * isom_map_camellia_to_aes( + * camellia_f( + * swap_bitendianess(in) + * ) + * ) + * ) + * + * (note: '⊕ 0xc5' inside camellia_f()) + */ +.Lpre_tf_lo_s1: + .byte 0x45, 0xe8, 0x40, 0xed, 0x2e, 0x83, 0x2b, 0x86 + .byte 0x4b, 0xe6, 0x4e, 0xe3, 0x20, 0x8d, 0x25, 0x88 +.Lpre_tf_hi_s1: + .byte 0x00, 0x51, 0xf1, 0xa0, 0x8a, 0xdb, 0x7b, 0x2a + .byte 0x09, 0x58, 0xf8, 0xa9, 0x83, 0xd2, 0x72, 0x23 + +/* + * pre-SubByte transform + * + * pre-lookup for sbox4: + * swap_bitendianness( + * isom_map_camellia_to_aes( + * camellia_f( + * swap_bitendianess(in <<< 1) + * ) + * ) + * ) + * + * (note: '⊕ 0xc5' inside camellia_f()) + */ +.Lpre_tf_lo_s4: + .byte 0x45, 0x40, 0x2e, 0x2b, 0x4b, 0x4e, 0x20, 0x25 + .byte 0x14, 0x11, 0x7f, 0x7a, 0x1a, 0x1f, 0x71, 0x74 +.Lpre_tf_hi_s4: + .byte 0x00, 0xf1, 0x8a, 0x7b, 0x09, 0xf8, 0x83, 0x72 + .byte 0xad, 0x5c, 0x27, 0xd6, 0xa4, 0x55, 0x2e, 0xdf + +/* + * post-SubByte transform + * + * post-lookup for sbox1, sbox4: + * swap_bitendianness( + * camellia_h( + * isom_map_aes_to_camellia( + * swap_bitendianness( + * aes_inverse_affine_transform(in) + * ) + * ) + * ) + * ) + * + * (note: '⊕ 0x6e' inside camellia_h()) + */ +.Lpost_tf_lo_s1: + .byte 0x3c, 0xcc, 0xcf, 0x3f, 0x32, 0xc2, 0xc1, 0x31 + .byte 0xdc, 0x2c, 0x2f, 0xdf, 0xd2, 0x22, 0x21, 0xd1 +.Lpost_tf_hi_s1: + .byte 0x00, 0xf9, 0x86, 0x7f, 0xd7, 0x2e, 0x51, 0xa8 + .byte 0xa4, 0x5d, 0x22, 0xdb, 0x73, 0x8a, 0xf5, 0x0c + +/* + * post-SubByte transform + * + * post-lookup for sbox2: + * swap_bitendianness( + * camellia_h( + * isom_map_aes_to_camellia( + * swap_bitendianness( + * aes_inverse_affine_transform(in) + * ) + * ) + * ) + * ) <<< 1 + * + * (note: '⊕ 0x6e' inside camellia_h()) + */ +.Lpost_tf_lo_s2: + .byte 0x78, 0x99, 0x9f, 0x7e, 0x64, 0x85, 0x83, 0x62 + .byte 0xb9, 0x58, 0x5e, 0xbf, 0xa5, 0x44, 0x42, 0xa3 +.Lpost_tf_hi_s2: + .byte 0x00, 0xf3, 0x0d, 0xfe, 0xaf, 0x5c, 0xa2, 0x51 + .byte 0x49, 0xba, 0x44, 0xb7, 0xe6, 0x15, 0xeb, 0x18 + +/* + * post-SubByte transform + * + * post-lookup for sbox3: + * swap_bitendianness( + * camellia_h( + * isom_map_aes_to_camellia( + * swap_bitendianness( + * aes_inverse_affine_transform(in) + * ) + * ) + * ) + * ) >>> 1 + * + * (note: '⊕ 0x6e' inside camellia_h()) + */ +.Lpost_tf_lo_s3: + .byte 0x1e, 0x66, 0xe7, 0x9f, 0x19, 0x61, 0xe0, 0x98 + .byte 0x6e, 0x16, 0x97, 0xef, 0x69, 0x11, 0x90, 0xe8 +.Lpost_tf_hi_s3: + .byte 0x00, 0xfc, 0x43, 0xbf, 0xeb, 0x17, 0xa8, 0x54 + .byte 0x52, 0xae, 0x11, 0xed, 0xb9, 0x45, 0xfa, 0x06 + +/* For isolating SubBytes from AESENCLAST, inverse shift row */ +.Linv_shift_row: + .byte 0x00, 0x0d, 0x0a, 0x07, 0x04, 0x01, 0x0e, 0x0b + .byte 0x08, 0x05, 0x02, 0x0f, 0x0c, 0x09, 0x06, 0x03 + +.section .rodata.cst4.L0f0f0f0f, "aM", @progbits, 4 +.align 4 +/* 4-bit mask */ +.L0f0f0f0f: + .long 0x0f0f0f0f + +.text + +SYM_FUNC_START_LOCAL(__camellia_enc_blk32) + /* input: + * %rdi: ctx, CTX + * %rax: temporary storage, 512 bytes + * %ymm0..%ymm15: 32 plaintext blocks + * output: + * %ymm0..%ymm15: 32 encrypted blocks, order swapped: + * 7, 8, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8 + */ + FRAME_BEGIN + + leaq 8 * 32(%rax), %rcx; + + inpack32_post(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx); + + enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 0); + + fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, + ((key_table + (8) * 8) + 0)(CTX), + ((key_table + (8) * 8) + 4)(CTX), + ((key_table + (8) * 8) + 8)(CTX), + ((key_table + (8) * 8) + 12)(CTX)); + + enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 8); + + fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, + ((key_table + (16) * 8) + 0)(CTX), + ((key_table + (16) * 8) + 4)(CTX), + ((key_table + (16) * 8) + 8)(CTX), + ((key_table + (16) * 8) + 12)(CTX)); + + enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 16); + + movl $24, %r8d; + cmpl $16, key_length(CTX); + jne .Lenc_max32; + +.Lenc_done: + /* load CD for output */ + vmovdqu 0 * 32(%rcx), %ymm8; + vmovdqu 1 * 32(%rcx), %ymm9; + vmovdqu 2 * 32(%rcx), %ymm10; + vmovdqu 3 * 32(%rcx), %ymm11; + vmovdqu 4 * 32(%rcx), %ymm12; + vmovdqu 5 * 32(%rcx), %ymm13; + vmovdqu 6 * 32(%rcx), %ymm14; + vmovdqu 7 * 32(%rcx), %ymm15; + + outunpack32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, (key_table)(CTX, %r8, 8), (%rax), 1 * 32(%rax)); + + FRAME_END + RET; + +.align 8 +.Lenc_max32: + movl $32, %r8d; + + fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, + ((key_table + (24) * 8) + 0)(CTX), + ((key_table + (24) * 8) + 4)(CTX), + ((key_table + (24) * 8) + 8)(CTX), + ((key_table + (24) * 8) + 12)(CTX)); + + enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 24); + + jmp .Lenc_done; +SYM_FUNC_END(__camellia_enc_blk32) + +SYM_FUNC_START_LOCAL(__camellia_dec_blk32) + /* input: + * %rdi: ctx, CTX + * %rax: temporary storage, 512 bytes + * %r8d: 24 for 16 byte key, 32 for larger + * %ymm0..%ymm15: 16 encrypted blocks + * output: + * %ymm0..%ymm15: 16 plaintext blocks, order swapped: + * 7, 8, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8 + */ + FRAME_BEGIN + + leaq 8 * 32(%rax), %rcx; + + inpack32_post(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx); + + cmpl $32, %r8d; + je .Ldec_max32; + +.Ldec_max24: + dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 16); + + fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, + ((key_table + (16) * 8) + 8)(CTX), + ((key_table + (16) * 8) + 12)(CTX), + ((key_table + (16) * 8) + 0)(CTX), + ((key_table + (16) * 8) + 4)(CTX)); + + dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 8); + + fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, + ((key_table + (8) * 8) + 8)(CTX), + ((key_table + (8) * 8) + 12)(CTX), + ((key_table + (8) * 8) + 0)(CTX), + ((key_table + (8) * 8) + 4)(CTX)); + + dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 0); + + /* load CD for output */ + vmovdqu 0 * 32(%rcx), %ymm8; + vmovdqu 1 * 32(%rcx), %ymm9; + vmovdqu 2 * 32(%rcx), %ymm10; + vmovdqu 3 * 32(%rcx), %ymm11; + vmovdqu 4 * 32(%rcx), %ymm12; + vmovdqu 5 * 32(%rcx), %ymm13; + vmovdqu 6 * 32(%rcx), %ymm14; + vmovdqu 7 * 32(%rcx), %ymm15; + + outunpack32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, (key_table)(CTX), (%rax), 1 * 32(%rax)); + + FRAME_END + RET; + +.align 8 +.Ldec_max32: + dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rax, %rcx, 24); + + fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, + ((key_table + (24) * 8) + 8)(CTX), + ((key_table + (24) * 8) + 12)(CTX), + ((key_table + (24) * 8) + 0)(CTX), + ((key_table + (24) * 8) + 4)(CTX)); + + jmp .Ldec_max24; +SYM_FUNC_END(__camellia_dec_blk32) + +SYM_FUNC_START(camellia_ecb_enc_32way) + /* input: + * %rdi: ctx, CTX + * %rsi: dst (32 blocks) + * %rdx: src (32 blocks) + */ + FRAME_BEGIN + + vzeroupper; + + inpack32_pre(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rdx, (key_table)(CTX)); + + /* now dst can be used as temporary buffer (even in src == dst case) */ + movq %rsi, %rax; + + call __camellia_enc_blk32; + + write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0, + %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9, + %ymm8, %rsi); + + vzeroupper; + + FRAME_END + RET; +SYM_FUNC_END(camellia_ecb_enc_32way) + +SYM_FUNC_START(camellia_ecb_dec_32way) + /* input: + * %rdi: ctx, CTX + * %rsi: dst (32 blocks) + * %rdx: src (32 blocks) + */ + FRAME_BEGIN + + vzeroupper; + + cmpl $16, key_length(CTX); + movl $32, %r8d; + movl $24, %eax; + cmovel %eax, %r8d; /* max */ + + inpack32_pre(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rdx, (key_table)(CTX, %r8, 8)); + + /* now dst can be used as temporary buffer (even in src == dst case) */ + movq %rsi, %rax; + + call __camellia_dec_blk32; + + write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0, + %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9, + %ymm8, %rsi); + + vzeroupper; + + FRAME_END + RET; +SYM_FUNC_END(camellia_ecb_dec_32way) + +SYM_FUNC_START(camellia_cbc_dec_32way) + /* input: + * %rdi: ctx, CTX + * %rsi: dst (32 blocks) + * %rdx: src (32 blocks) + */ + FRAME_BEGIN + subq $(16 * 32), %rsp; + + vzeroupper; + + cmpl $16, key_length(CTX); + movl $32, %r8d; + movl $24, %eax; + cmovel %eax, %r8d; /* max */ + + inpack32_pre(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7, + %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, + %ymm15, %rdx, (key_table)(CTX, %r8, 8)); + + cmpq %rsi, %rdx; + je .Lcbc_dec_use_stack; + + /* dst can be used as temporary storage, src is not overwritten. */ + movq %rsi, %rax; + jmp .Lcbc_dec_continue; + +.Lcbc_dec_use_stack: + /* + * dst still in-use (because dst == src), so use stack for temporary + * storage. + */ + movq %rsp, %rax; + +.Lcbc_dec_continue: + call __camellia_dec_blk32; + + vmovdqu %ymm7, (%rax); + vpxor %ymm7, %ymm7, %ymm7; + vinserti128 $1, (%rdx), %ymm7, %ymm7; + vpxor (%rax), %ymm7, %ymm7; + vpxor (0 * 32 + 16)(%rdx), %ymm6, %ymm6; + vpxor (1 * 32 + 16)(%rdx), %ymm5, %ymm5; + vpxor (2 * 32 + 16)(%rdx), %ymm4, %ymm4; + vpxor (3 * 32 + 16)(%rdx), %ymm3, %ymm3; + vpxor (4 * 32 + 16)(%rdx), %ymm2, %ymm2; + vpxor (5 * 32 + 16)(%rdx), %ymm1, %ymm1; + vpxor (6 * 32 + 16)(%rdx), %ymm0, %ymm0; + vpxor (7 * 32 + 16)(%rdx), %ymm15, %ymm15; + vpxor (8 * 32 + 16)(%rdx), %ymm14, %ymm14; + vpxor (9 * 32 + 16)(%rdx), %ymm13, %ymm13; + vpxor (10 * 32 + 16)(%rdx), %ymm12, %ymm12; + vpxor (11 * 32 + 16)(%rdx), %ymm11, %ymm11; + vpxor (12 * 32 + 16)(%rdx), %ymm10, %ymm10; + vpxor (13 * 32 + 16)(%rdx), %ymm9, %ymm9; + vpxor (14 * 32 + 16)(%rdx), %ymm8, %ymm8; + write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0, + %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9, + %ymm8, %rsi); + + vzeroupper; + + addq $(16 * 32), %rsp; + FRAME_END + RET; +SYM_FUNC_END(camellia_cbc_dec_32way) |