| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285 |
- // Copyright 2012 The Go Authors. All rights reserved.
- // Use of this source code is governed by a BSD-style
- // license that can be found in the LICENSE file.
- #include "textflag.h"
- // func hasAsm() bool
- // returns whether AES-NI is supported
- TEXT ·hasAsm(SB),NOSPLIT,$0
- XORQ AX, AX
- INCL AX
- CPUID
- SHRQ $25, CX
- ANDQ $1, CX
- MOVB CX, ret+0(FP)
- RET
- // func encryptBlockAsm(nr int, xk *uint32, dst, src *byte)
- TEXT ·encryptBlockAsm(SB),NOSPLIT,$0
- MOVQ nr+0(FP), CX
- MOVQ xk+8(FP), AX
- MOVQ dst+16(FP), DX
- MOVQ src+24(FP), BX
- MOVUPS 0(AX), X1
- MOVUPS 0(BX), X0
- ADDQ $16, AX
- PXOR X1, X0
- SUBQ $12, CX
- JE Lenc196
- JB Lenc128
- Lenc256:
- MOVUPS 0(AX), X1
- AESENC X1, X0
- MOVUPS 16(AX), X1
- AESENC X1, X0
- ADDQ $32, AX
- Lenc196:
- MOVUPS 0(AX), X1
- AESENC X1, X0
- MOVUPS 16(AX), X1
- AESENC X1, X0
- ADDQ $32, AX
- Lenc128:
- MOVUPS 0(AX), X1
- AESENC X1, X0
- MOVUPS 16(AX), X1
- AESENC X1, X0
- MOVUPS 32(AX), X1
- AESENC X1, X0
- MOVUPS 48(AX), X1
- AESENC X1, X0
- MOVUPS 64(AX), X1
- AESENC X1, X0
- MOVUPS 80(AX), X1
- AESENC X1, X0
- MOVUPS 96(AX), X1
- AESENC X1, X0
- MOVUPS 112(AX), X1
- AESENC X1, X0
- MOVUPS 128(AX), X1
- AESENC X1, X0
- MOVUPS 144(AX), X1
- AESENCLAST X1, X0
- MOVUPS X0, 0(DX)
- RET
- // func decryptBlockAsm(nr int, xk *uint32, dst, src *byte)
- TEXT ·decryptBlockAsm(SB),NOSPLIT,$0
- MOVQ nr+0(FP), CX
- MOVQ xk+8(FP), AX
- MOVQ dst+16(FP), DX
- MOVQ src+24(FP), BX
- MOVUPS 0(AX), X1
- MOVUPS 0(BX), X0
- ADDQ $16, AX
- PXOR X1, X0
- SUBQ $12, CX
- JE Ldec196
- JB Ldec128
- Ldec256:
- MOVUPS 0(AX), X1
- AESDEC X1, X0
- MOVUPS 16(AX), X1
- AESDEC X1, X0
- ADDQ $32, AX
- Ldec196:
- MOVUPS 0(AX), X1
- AESDEC X1, X0
- MOVUPS 16(AX), X1
- AESDEC X1, X0
- ADDQ $32, AX
- Ldec128:
- MOVUPS 0(AX), X1
- AESDEC X1, X0
- MOVUPS 16(AX), X1
- AESDEC X1, X0
- MOVUPS 32(AX), X1
- AESDEC X1, X0
- MOVUPS 48(AX), X1
- AESDEC X1, X0
- MOVUPS 64(AX), X1
- AESDEC X1, X0
- MOVUPS 80(AX), X1
- AESDEC X1, X0
- MOVUPS 96(AX), X1
- AESDEC X1, X0
- MOVUPS 112(AX), X1
- AESDEC X1, X0
- MOVUPS 128(AX), X1
- AESDEC X1, X0
- MOVUPS 144(AX), X1
- AESDECLAST X1, X0
- MOVUPS X0, 0(DX)
- RET
- // func expandKeyAsm(nr int, key *byte, enc, dec *uint32) {
- // Note that round keys are stored in uint128 format, not uint32
- TEXT ·expandKeyAsm(SB),NOSPLIT,$0
- MOVQ nr+0(FP), CX
- MOVQ key+8(FP), AX
- MOVQ enc+16(FP), BX
- MOVQ dec+24(FP), DX
- MOVUPS (AX), X0
- // enc
- MOVUPS X0, (BX)
- ADDQ $16, BX
- PXOR X4, X4 // _expand_key_* expect X4 to be zero
- CMPL CX, $12
- JE Lexp_enc196
- JB Lexp_enc128
- Lexp_enc256:
- MOVUPS 16(AX), X2
- MOVUPS X2, (BX)
- ADDQ $16, BX
- AESKEYGENASSIST $0x01, X2, X1
- CALL _expand_key_256a<>(SB)
- AESKEYGENASSIST $0x01, X0, X1
- CALL _expand_key_256b<>(SB)
- AESKEYGENASSIST $0x02, X2, X1
- CALL _expand_key_256a<>(SB)
- AESKEYGENASSIST $0x02, X0, X1
- CALL _expand_key_256b<>(SB)
- AESKEYGENASSIST $0x04, X2, X1
- CALL _expand_key_256a<>(SB)
- AESKEYGENASSIST $0x04, X0, X1
- CALL _expand_key_256b<>(SB)
- AESKEYGENASSIST $0x08, X2, X1
- CALL _expand_key_256a<>(SB)
- AESKEYGENASSIST $0x08, X0, X1
- CALL _expand_key_256b<>(SB)
- AESKEYGENASSIST $0x10, X2, X1
- CALL _expand_key_256a<>(SB)
- AESKEYGENASSIST $0x10, X0, X1
- CALL _expand_key_256b<>(SB)
- AESKEYGENASSIST $0x20, X2, X1
- CALL _expand_key_256a<>(SB)
- AESKEYGENASSIST $0x20, X0, X1
- CALL _expand_key_256b<>(SB)
- AESKEYGENASSIST $0x40, X2, X1
- CALL _expand_key_256a<>(SB)
- JMP Lexp_dec
- Lexp_enc196:
- MOVQ 16(AX), X2
- AESKEYGENASSIST $0x01, X2, X1
- CALL _expand_key_192a<>(SB)
- AESKEYGENASSIST $0x02, X2, X1
- CALL _expand_key_192b<>(SB)
- AESKEYGENASSIST $0x04, X2, X1
- CALL _expand_key_192a<>(SB)
- AESKEYGENASSIST $0x08, X2, X1
- CALL _expand_key_192b<>(SB)
- AESKEYGENASSIST $0x10, X2, X1
- CALL _expand_key_192a<>(SB)
- AESKEYGENASSIST $0x20, X2, X1
- CALL _expand_key_192b<>(SB)
- AESKEYGENASSIST $0x40, X2, X1
- CALL _expand_key_192a<>(SB)
- AESKEYGENASSIST $0x80, X2, X1
- CALL _expand_key_192b<>(SB)
- JMP Lexp_dec
- Lexp_enc128:
- AESKEYGENASSIST $0x01, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x02, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x04, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x08, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x10, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x20, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x40, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x80, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x1b, X0, X1
- CALL _expand_key_128<>(SB)
- AESKEYGENASSIST $0x36, X0, X1
- CALL _expand_key_128<>(SB)
- Lexp_dec:
- // dec
- SUBQ $16, BX
- MOVUPS (BX), X1
- MOVUPS X1, (DX)
- DECQ CX
- Lexp_dec_loop:
- MOVUPS -16(BX), X1
- AESIMC X1, X0
- MOVUPS X0, 16(DX)
- SUBQ $16, BX
- ADDQ $16, DX
- DECQ CX
- JNZ Lexp_dec_loop
- MOVUPS -16(BX), X0
- MOVUPS X0, 16(DX)
- RET
- TEXT _expand_key_128<>(SB),NOSPLIT,$0
- PSHUFD $0xff, X1, X1
- SHUFPS $0x10, X0, X4
- PXOR X4, X0
- SHUFPS $0x8c, X0, X4
- PXOR X4, X0
- PXOR X1, X0
- MOVUPS X0, (BX)
- ADDQ $16, BX
- RET
- TEXT _expand_key_192a<>(SB),NOSPLIT,$0
- PSHUFD $0x55, X1, X1
- SHUFPS $0x10, X0, X4
- PXOR X4, X0
- SHUFPS $0x8c, X0, X4
- PXOR X4, X0
- PXOR X1, X0
- MOVAPS X2, X5
- MOVAPS X2, X6
- PSLLDQ $0x4, X5
- PSHUFD $0xff, X0, X3
- PXOR X3, X2
- PXOR X5, X2
- MOVAPS X0, X1
- SHUFPS $0x44, X0, X6
- MOVUPS X6, (BX)
- SHUFPS $0x4e, X2, X1
- MOVUPS X1, 16(BX)
- ADDQ $32, BX
- RET
- TEXT _expand_key_192b<>(SB),NOSPLIT,$0
- PSHUFD $0x55, X1, X1
- SHUFPS $0x10, X0, X4
- PXOR X4, X0
- SHUFPS $0x8c, X0, X4
- PXOR X4, X0
- PXOR X1, X0
- MOVAPS X2, X5
- PSLLDQ $0x4, X5
- PSHUFD $0xff, X0, X3
- PXOR X3, X2
- PXOR X5, X2
- MOVUPS X0, (BX)
- ADDQ $16, BX
- RET
- TEXT _expand_key_256a<>(SB),NOSPLIT,$0
- JMP _expand_key_128<>(SB)
- TEXT _expand_key_256b<>(SB),NOSPLIT,$0
- PSHUFD $0xaa, X1, X1
- SHUFPS $0x10, X2, X4
- PXOR X4, X2
- SHUFPS $0x8c, X2, X4
- PXOR X4, X2
- PXOR X1, X2
- MOVUPS X2, (BX)
- ADDQ $16, BX
- RET
|