Skip to content

Commit

Permalink
Update assembly code for lsh256 and lsh512 implementations
Browse files Browse the repository at this point in the history
  • Loading branch information
@RyuaNerin
RyuaNerin committed Jun 19, 2024
1 parent 5adc6be commit 6b88f0d
Show file tree
Hide file tree
Showing 2 changed files with 14 additions and 45 deletions.
6 changes: 3 additions & 3 deletions lsh256/asm_arm64_neon.s
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@

#include "textflag.h"

TEXT ·__lsh256_neon_init(SB), NOSPLIT, $16
TEXT ·__lsh256_neon_init(SB), NOSPLIT, $0-16
MOVD ctx+0(FP), R0
MOVD algtype+8(FP), R1

Expand Down Expand Up @@ -58,7 +58,7 @@ DATA ·lCPI1_0<>+0x038(SB)/4, $0xffffffe8
DATA ·lCPI1_0<>+0x03C(SB)/4, $0xffffffe0
GLOBL ·lCPI1_0<>(SB), NOPTR|RODATA, $64

TEXT ·__lsh256_neon_update(SB), NOSPLIT, $32
TEXT ·__lsh256_neon_update(SB), NOSPLIT, $0-32
MOVD ctx+0(FP), R0
MOVD data_base+8(FP), R1
MOVD data_len+16(FP), R2
Expand Down Expand Up @@ -562,7 +562,7 @@ LBB1_56: // LBB1_56: ; =>This Inner Loop H
CBNZ R17, LBB1_13 // cbnz x17, LBB1_13
B LBB1_15 // b LBB1_15

TEXT ·__lsh256_neon_final(SB), NOSPLIT, $16
TEXT ·__lsh256_neon_final(SB), NOSPLIT, $0-16
MOVD ctx+0(FP), R0
MOVD hashval+8(FP), R1

Expand Down
53 changes: 11 additions & 42 deletions lsh512/asm_arm64_neon.s
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@

#include "textflag.h"

TEXT ·__lsh512_neon_init(SB), NOSPLIT, $16
TEXT ·__lsh512_neon_init(SB), NOSPLIT, $0-16
MOVD ctx+0(FP), R0
MOVD algtype+8(FP), R1

Expand Down Expand Up @@ -80,23 +80,14 @@ DATA ·lCPI1_0<>+0x070(SB)/8, $0xffffffffffffffe8
DATA ·lCPI1_0<>+0x078(SB)/8, $0xfffffffffffffff8
GLOBL ·lCPI1_0<>(SB), RODATA|NOPTR, $128

TEXT ·__lsh512_neon_update(SB), NOSPLIT, $144-32
// sub x9, sp, #176
TEXT ·__lsh512_neon_update(SB), NOSPLIT, $176-32
MOVD ctx+0(FP), R0
MOVD data_base+8(FP), R1
MOVD data_len+16(FP), R2
// data_len+24

// ; %bb.0:
WORD $0x6db93bef // stp d15, d14, [sp, #-112]! ; 16-byte Folded Spill
WORD $0x6d0133ed // stp d13, d12, [sp, #16] ; 16-byte Folded Spill
WORD $0x6d022beb // stp d11, d10, [sp, #32] ; 16-byte Folded Spill
WORD $0x6d0323e9 // stp d9, d8, [sp, #48] ; 16-byte Folded Spill
WORD $0xa9046ffc // stp x28, x27, [sp, #64] ; 16-byte Folded Spill
WORD $0xa9054ff4 // stp x20, x19, [sp, #80] ; 16-byte Folded Spill
WORD $0xa9067bfd // stp x29, x30, [sp, #96] ; 16-byte Folded Spill
WORD $0x910183fd // add x29, sp, #96
WORD $0xd102c3e9 // sub x9, sp, #176
WORD $0x927ced3f // and sp, x9, #0xfffffffffffffff0
WORD $0xb9401003 // ldr w3, [x0, #16]
WORD $0x8b020068 // add x8, x3, x2
WORD $0xf103fd1f // cmp x8, #255
Expand All @@ -114,23 +105,23 @@ TEXT ·__lsh512_neon_update(SB), NOSPLIT, $144-32
WORD $0x91028129 // add x9, x9, #160
WORD $0xf101013f // cmp x9, #64
BHS LBB1_31 // b.hs LBB1_31
LBB1_4: // LBB1_4:
LBB1_4: // LBB1_4:
WORD $0xd2800009 // mov x9, #0 ; =0x0
LBB1_5: // LBB1_5:
LBB1_5: // LBB1_5:
WORD $0xcb090108 // sub x8, x8, x9
WORD $0x8b03012a // add x10, x9, x3
WORD $0x8b00014a // add x10, x10, x0
WORD $0x9102814a // add x10, x10, #160
WORD $0x8b090029 // add x9, x1, x9
LBB1_6: // LBB1_6: ; =>This Inner Loop Header: Depth=1
LBB1_6: // LBB1_6: ; =>This Inner Loop Header: Depth=1
WORD $0x3840152b // ldrb w11, [x9], #1
WORD $0x3800154b // strb w11, [x10], #1
WORD $0xf1000508 // subs x8, x8, #1
BNE LBB1_6 // b.ne LBB1_6
LBB1_7: // LBB1_7:
LBB1_7: // LBB1_7:
WORD $0x0b020062 // add w2, w3, w2
B LBB1_29 // b LBB1_29
LBB1_8: // LBB1_8:
LBB1_8: // LBB1_8:
MOVD $·lCPI1_0<>(SB), R9 // adrp x9, lCPI1_0@PAGE
// adrp x10, lCPI1_1@PAGE
// adrp x11, lCPI1_2@PAGE
Expand Down Expand Up @@ -672,14 +663,6 @@ LBB1_28: // LBB1_28: ; =>This Inner Loop H
LBB1_29: // LBB1_29:
WORD $0xb9001002 // str w2, [x0, #16]
LBB1_30: // LBB1_30:
WORD $0xd10183bf // sub sp, x29, #96
WORD $0xa9467bfd // ldp x29, x30, [sp, #96] ; 16-byte Folded Reload
WORD $0xa9454ff4 // ldp x20, x19, [sp, #80] ; 16-byte Folded Reload
WORD $0xa9446ffc // ldp x28, x27, [sp, #64] ; 16-byte Folded Reload
WORD $0x6d4323e9 // ldp d9, d8, [sp, #48] ; 16-byte Folded Reload
WORD $0x6d422beb // ldp d11, d10, [sp, #32] ; 16-byte Folded Reload
WORD $0x6d4133ed // ldp d13, d12, [sp, #16] ; 16-byte Folded Reload
WORD $0x6cc73bef // ldp d15, d14, [sp], #112 ; 16-byte Folded Reload
RET // ret
LBB1_31: // LBB1_31:
WORD $0xf101011f // cmp x8, #64
Expand Down Expand Up @@ -811,19 +794,12 @@ LBB1_56: // LBB1_56: ; =>This Inner Loop H
CBNZ R6, LBB1_13 // cbnz x6, LBB1_13
B LBB1_15 // b LBB1_15

TEXT ·__lsh512_neon_final(SB), NOSPLIT, $96-16
// sub x9, sp, #176
TEXT ·__lsh512_neon_final(SB), NOSPLIT, $176-16
MOVD ctx+0(FP), R0
MOVD hashval+8(FP), R1

// ; %bb.0:
WORD $0x6dbb3bef // stp d15, d14, [sp, #-80]! ; 16-byte Folded Spill
WORD $0x6d0133ed // stp d13, d12, [sp, #16] ; 16-byte Folded Spill
WORD $0x6d022beb // stp d11, d10, [sp, #32] ; 16-byte Folded Spill
WORD $0x6d0323e9 // stp d9, d8, [sp, #48] ; 16-byte Folded Spill
WORD $0xa9047bfd // stp x29, x30, [sp, #64] ; 16-byte Folded Spill
WORD $0x910103fd // add x29, sp, #64
WORD $0xd102c3e9 // sub x9, sp, #176
WORD $0x927ced3f // and sp, x9, #0xfffffffffffffff0
WORD $0xb9401009 // ldr w9, [x0, #16]
WORD $0x8b090008 // add x8, x0, x9
WORD $0x5280100a // mov w10, #128 ; =0x80
Expand Down Expand Up @@ -1129,11 +1105,4 @@ LBB2_15: // LBB2_15: ; =>This Inner Loop H
WORD $0x3d800820 // str q0, [x1, #32]
WORD $0x3dc01400 // ldr q0, [x0, #80]
WORD $0x3d800c20 // str q0, [x1, #48]
WORD $0xd10103bf // sub sp, x29, #64
WORD $0xa9447bfd // ldp x29, x30, [sp, #64] ; 16-byte Folded Reload
WORD $0x6d4323e9 // ldp d9, d8, [sp, #48] ; 16-byte Folded Reload
WORD $0x6d422beb // ldp d11, d10, [sp, #32] ; 16-byte Folded Reload
WORD $0x6d4133ed // ldp d13, d12, [sp, #16] ; 16-byte Folded Reload
WORD $0x6cc53bef // ldp d15, d14, [sp], #80 ; 16-byte Folded Reload
RET // ret

RET // ret

0 comments on commit 6b88f0d

Please sign in to comment.