#include <linux/linkage.h>
#define CTX %rdi /* arg1 */
#define BUF %rsi /* arg2 */
#define CNT %rdx /* arg3 */
#define REG_A %ecx
#define REG_B %esi
#define REG_C %edi
#define REG_D %eax
#define REG_E %edx
#define REG_TB %ebx
#define REG_TA %r12d
#define REG_RA %rcx
#define REG_RB %rsi
#define REG_RC %rdi
#define REG_RD %rax
#define REG_RE %rdx
#define REG_RTA %r12
#define REG_RTB %rbx
#define REG_T1 %r11d
#define xmm_mov vmovups
#define avx2_zeroupper vzeroupper
#define RND_F1 1
#define RND_F2 2
#define RND_F3 3
.macro REGALLOC
.set A, REG_A
.set B, REG_B
.set C, REG_C
.set D, REG_D
.set E, REG_E
.set TB, REG_TB
.set TA, REG_TA
.set RA, REG_RA
.set RB, REG_RB
.set RC, REG_RC
.set RD, REG_RD
.set RE, REG_RE
.set RTA, REG_RTA
.set RTB, REG_RTB
.set T1, REG_T1
.endm
#define HASH_PTR %r9
#define BLOCKS_CTR %r8
#define BUFFER_PTR %r10
#define BUFFER_PTR2 %r13
#define PRECALC_BUF %r14
#define WK_BUF %r15
#define W_TMP %xmm0
#define WY_TMP %ymm0
#define WY_TMP2 %ymm9
# AVX2 variables
#define WY0 %ymm3
#define WY4 %ymm5
#define WY08 %ymm7
#define WY12 %ymm8
#define WY16 %ymm12
#define WY20 %ymm13
#define WY24 %ymm14
#define WY28 %ymm15
#define YMM_SHUFB_BSWAP %ymm10
#define W_SIZE (80*2*2 +16)
#define WK(t) ((((t) % 80) / 4)*32 + ( (t) % 4)*4 + ((t)/80)*16 )(WK_BUF)
#define PRECALC_WK(t) ((t)*2*2)(PRECALC_BUF)
.macro UPDATE_HASH hash, val
add \hash, \val
mov \val, \hash
.endm
.macro PRECALC_RESET_WY
.set WY_00, WY0
.set WY_04, WY4
.set WY_08, WY08
.set WY_12, WY12
.set WY_16, WY16
.set WY_20, WY20
.set WY_24, WY24
.set WY_28, WY28
.set WY_32, WY_00
.endm
.macro PRECALC_ROTATE_WY
.set WY_32, WY_28
.set WY_28, WY_24
.set WY_24, WY_20
.set WY_20, WY_16
.set WY_16, WY_12
.set WY_12, WY_08
.set WY_08, WY_04
.set WY_04, WY_00
.set WY_00, WY_32
.set WY, WY_00
.set WY_minus_04, WY_04
.set WY_minus_08, WY_08
.set WY_minus_12, WY_12
.set WY_minus_16, WY_16
.set WY_minus_20, WY_20
.set WY_minus_24, WY_24
.set WY_minus_28, WY_28
.set WY_minus_32, WY
.endm
.macro PRECALC_00_15
.if (i == 0) # Initialize and rotate registers
PRECALC_RESET_WY
PRECALC_ROTATE_WY
.endif
.if ((i & 7) == 0)
vmovdqu (i * 2)(BUFFER_PTR), W_TMP
.elseif ((i & 7) == 1)
vinsertf128 $1, ((i-1) * 2)(BUFFER_PTR2),\
WY_TMP, WY_TMP
.elseif ((i & 7) == 2)
vpshufb YMM_SHUFB_BSWAP, WY_TMP, WY
.elseif ((i & 7) == 4)
vpaddd K_XMM + K_XMM_AR(%rip), WY, WY_TMP
.elseif ((i & 7) == 7)
vmovdqu WY_TMP, PRECALC_WK(i&~7)
PRECALC_ROTATE_WY
.endif
.endm
.macro PRECALC_16_31
.if ((i & 7) == 0)
vpalignr $8, WY_minus_16, WY_minus_12, WY
vpsrldq $4, WY_minus_04, WY_TMP
.elseif ((i & 7) == 1)
vpxor WY_minus_08, WY, WY
vpxor WY_minus_16, WY_TMP, WY_TMP
.elseif ((i & 7) == 2)
vpxor WY_TMP, WY, WY
vpslldq $12, WY, WY_TMP2
.elseif ((i & 7) == 3)
vpslld $1, WY, WY_TMP
vpsrld $31, WY, WY
.elseif ((i & 7) == 4)
vpor WY, WY_TMP, WY_TMP
vpslld $2, WY_TMP2, WY
.elseif ((i & 7) == 5)
vpsrld $30, WY_TMP2, WY_TMP2
vpxor WY, WY_TMP, WY_TMP
.elseif ((i & 7) == 7)
vpxor WY_TMP2, WY_TMP, WY
vpaddd K_XMM + K_XMM_AR(%rip), WY, WY_TMP
vmovdqu WY_TMP, PRECALC_WK(i&~7)
PRECALC_ROTATE_WY
.endif
.endm
.macro PRECALC_32_79
.if ((i & 7) == 0)
vpalignr $8, WY_minus_08, WY_minus_04, WY_TMP
.elseif ((i & 7) == 1)
vpxor WY_minus_28, WY, WY
.elseif ((i & 7) == 2)
vpxor WY_minus_16, WY_TMP, WY_TMP
.elseif ((i & 7) == 3)
vpxor WY_TMP, WY, WY
.elseif ((i & 7) == 4)
vpslld $2, WY, WY_TMP
.elseif ((i & 7) == 5)
vpsrld $30, WY, WY
vpor WY, WY_TMP, WY
.elseif ((i & 7) == 7)
vpaddd K_XMM + K_XMM_AR(%rip), WY, WY_TMP
vmovdqu WY_TMP, PRECALC_WK(i&~7)
PRECALC_ROTATE_WY
.endif
.endm
.macro PRECALC r, s
.set i, \r
.if (i < 40)
.set K_XMM, 32*0
.elseif (i < 80)
.set K_XMM, 32*1
.elseif (i < 120)
.set K_XMM, 32*2
.else
.set K_XMM, 32*3
.endif
.if (i<32)
PRECALC_00_15 \s
.elseif (i<64)
PRECALC_16_31 \s
.elseif (i < 160)
PRECALC_32_79 \s
.endif
.endm
.macro ROTATE_STATE
.set T_REG, E
.set E, D
.set D, C
.set C, B
.set B, TB
.set TB, A
.set A, T_REG
.set T_REG, RE
.set RE, RD
.set RD, RC
.set RC, RB
.set RB, RTB
.set RTB, RA
.set RA, T_REG
.endm
.macro RND_FUN f, r
.if (\f == RND_F1)
ROUND_F1 \r
.elseif (\f == RND_F2)
ROUND_F2 \r
.elseif (\f == RND_F3)
ROUND_F3 \r
.endif
.endm
.macro RR r
.set round_id, (\r % 80)
.if (round_id == 0)
.set ROUND_FUNC, RND_F1
mov B, TB
rorx $(32-30), B, B
andn D, TB, T1
and C, TB
xor T1, TB
.endif
RND_FUN ROUND_FUNC, \r
ROTATE_STATE
.if (round_id == 18)
.set ROUND_FUNC, RND_F2
.elseif (round_id == 38)
.set ROUND_FUNC, RND_F3
.elseif (round_id == 58)
.set ROUND_FUNC, RND_F2
.endif
.set round_id, ( (\r+1) % 80)
RND_FUN ROUND_FUNC, (\r+1)
ROTATE_STATE
.endm
.macro ROUND_F1 r
add WK(\r), E
andn C, A, T1
lea (RE,RTB), E
rorx $(32-5), A, TA
rorx $(32-30),A, TB
PRECALC (\r)
and B, A
xor T1, A
lea (RE,RTA), E
.endm
.macro ROUND_F2 r
add WK(\r), E
lea (RE,RTB), E
rorx $(32-5), A, TA
.if ((round_id) < 79)
rorx $(32-30), A, TB
.endif
PRECALC (\r)
.if ((round_id) < 79)
xor B, A
.endif
add TA, E
.if ((round_id) < 79)
xor C, A
.endif
.endm
.macro ROUND_F3 r
add WK(\r), E
PRECALC (\r)
lea (RE,RTB), E
mov B, T1
or A, T1
rorx $(32-5), A, TA
rorx $(32-30), A, TB
and C, T1
and B, A
or T1, A
add TA, E
.endm
.macro ADD_IF_GE a, b, c, d
mov \a, RTA
add $\d, RTA
cmp $\c, \b
cmovge RTA, \a
.endm
.macro SHA1_PIPELINED_MAIN_BODY
REGALLOC
mov (HASH_PTR), A
mov 4(HASH_PTR), B
mov 8(HASH_PTR), C
mov 12(HASH_PTR), D
mov 16(HASH_PTR), E
mov %rsp, PRECALC_BUF
lea (2*4*80+32)(%rsp), WK_BUF
# Precalc WK for first 2 blocks
ADD_IF_GE BUFFER_PTR2, BLOCKS_CTR, 2, 64
.set i, 0
.rept 160
PRECALC i
.set i, i + 1
.endr
ADD_IF_GE BUFFER_PTR, BLOCKS_CTR, 3, 128
ADD_IF_GE BUFFER_PTR2, BLOCKS_CTR, 4, 128
xchg WK_BUF, PRECALC_BUF
.align 32
.L_loop:
test BLOCKS_CTR, BLOCKS_CTR
jnz .L_begin
.align 32
jmp .L_end
.align 32
.L_begin:
.set j, 0
.rept 5
RR j
.set j, j+2
.endr
.rept 25
RR j
.set j, j+2
.endr
sub $1, BLOCKS_CTR
ADD_IF_GE BUFFER_PTR, BLOCKS_CTR, 4, 128
.rept 10
RR j
.set j, j+2
.endr
UPDATE_HASH (HASH_PTR), A
UPDATE_HASH 4(HASH_PTR), TB
UPDATE_HASH 8(HASH_PTR), C
UPDATE_HASH 12(HASH_PTR), D
UPDATE_HASH 16(HASH_PTR), E
test BLOCKS_CTR, BLOCKS_CTR
jz .L_loop
mov TB, B
.set j, 0
.rept 10
RR j+80
.set j, j+2
.endr
.rept 10
RR j+80
.set j, j+2
.endr
.rept 10
RR j+80
.set j, j+2
.endr
sub $1, BLOCKS_CTR
ADD_IF_GE BUFFER_PTR2, BLOCKS_CTR, 4, 128
.rept 10
RR j+80
.set j, j+2
.endr
UPDATE_HASH (HASH_PTR), A
UPDATE_HASH 4(HASH_PTR), TB
UPDATE_HASH 8(HASH_PTR), C
UPDATE_HASH 12(HASH_PTR), D
UPDATE_HASH 16(HASH_PTR), E
mov A, TA
mov TB, A
mov C, TB
mov E, C
mov D, B
mov TA, D
REGALLOC
xchg WK_BUF, PRECALC_BUF
jmp .L_loop
.align 32
.L_end:
.endm
.macro SHA1_VECTOR_ASM name
SYM_FUNC_START(\name)
push %rbx
push %r12
push %r13
push %r14
push %r15
RESERVE_STACK = (W_SIZE*4 + 8+24)
push %rbp
mov %rsp, %rbp
and $~(0x20-1), %rsp
sub $RESERVE_STACK, %rsp
avx2_zeroupper
mov CTX, HASH_PTR
mov BUF, BUFFER_PTR
mov BUF, BUFFER_PTR2
mov CNT, BLOCKS_CTR
xmm_mov BSWAP_SHUFB_CTL(%rip), YMM_SHUFB_BSWAP
SHA1_PIPELINED_MAIN_BODY
avx2_zeroupper
mov %rbp, %rsp
pop %rbp
pop %r15
pop %r14
pop %r13
pop %r12
pop %rbx
RET
SYM_FUNC_END(\name)
.endm
.section .rodata
#define K1 0x5a827999
#define K2 0x6ed9eba1
#define K3 0x8f1bbcdc
#define K4 0xca62c1d6
.align 128
K_XMM_AR:
.long K1, K1, K1, K1
.long K1, K1, K1, K1
.long K2, K2, K2, K2
.long K2, K2, K2, K2
.long K3, K3, K3, K3
.long K3, K3, K3, K3
.long K4, K4, K4, K4
.long K4, K4, K4, K4
BSWAP_SHUFB_CTL:
.long 0x00010203
.long 0x04050607
.long 0x08090a0b
.long 0x0c0d0e0f
.long 0x00010203
.long 0x04050607
.long 0x08090a0b
.long 0x0c0d0e0f