aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/src
diff options
context:
space:
mode:
authorEgor Tensin <Egor.Tensin@gmail.com>2015-05-30 05:13:44 +0300
committerEgor Tensin <Egor.Tensin@gmail.com>2015-05-30 05:13:44 +0300
commit1109190cc2edd86f12cdd16847f3e80e48dccc50 (patch)
treea688b89f7e4ede0defaeb79ffb8e8a99ece5a395 /src
parentsupport AES-{128,192}-cbc (diff)
downloadaes-tools-1109190cc2edd86f12cdd16847f3e80e48dccc50.tar.gz
aes-tools-1109190cc2edd86f12cdd16847f3e80e48dccc50.zip
detach key expansion routines
Block encryption/decryption routines now don't expand key schedules on each call.
Diffstat (limited to 'src')
-rw-r--r--src/aes128.asm209
-rw-r--r--src/aes192.asm261
-rw-r--r--src/aes256.asm302
3 files changed, 366 insertions, 406 deletions
diff --git a/src/aes128.asm b/src/aes128.asm
index 5d62621..577e582 100644
--- a/src/aes128.asm
+++ b/src/aes128.asm
@@ -6,68 +6,55 @@
.xmm
.model flat
-.data
-
-align 10h
-key_schedule oword 11 dup(0)
-
-align 10h
-inverted_key_schedule oword 11 dup(0)
-
.code
-@raw_aes128ecb_encrypt@32 proc
- call expand_keys128
- pxor xmm0, [key_schedule]
- aesenc xmm0, [key_schedule + 10h]
- aesenc xmm0, [key_schedule + 20h]
- aesenc xmm0, [key_schedule + 30h]
- aesenc xmm0, [key_schedule + 40h]
- aesenc xmm0, [key_schedule + 50h]
- aesenc xmm0, [key_schedule + 60h]
- aesenc xmm0, [key_schedule + 70h]
- aesenc xmm0, [key_schedule + 80h]
- aesenc xmm0, [key_schedule + 90h]
- aesenclast xmm0, [key_schedule + 0A0h]
+@raw_aes128ecb_encrypt@20 proc
+ pxor xmm0, [ecx]
+ aesenc xmm0, [ecx + 10h]
+ aesenc xmm0, [ecx + 20h]
+ aesenc xmm0, [ecx + 30h]
+ aesenc xmm0, [ecx + 40h]
+ aesenc xmm0, [ecx + 50h]
+ aesenc xmm0, [ecx + 60h]
+ aesenc xmm0, [ecx + 70h]
+ aesenc xmm0, [ecx + 80h]
+ aesenc xmm0, [ecx + 90h]
+ aesenclast xmm0, [ecx + 0A0h]
ret
-@raw_aes128ecb_encrypt@32 endp
+@raw_aes128ecb_encrypt@20 endp
+
+@raw_aes128cbc_encrypt@24 proc
+ pxor xmm0, [edx]
+ jmp @raw_aes128ecb_encrypt@20
+@raw_aes128cbc_encrypt@24 endp
-@raw_aes128cbc_encrypt@36 proc
+@raw_aes128ecb_decrypt@20 proc
pxor xmm0, [ecx]
- jmp @raw_aes128ecb_encrypt@32
-@raw_aes128cbc_encrypt@36 endp
-
-@raw_aes128ecb_decrypt@32 proc
- call expand_keys128
- pxor xmm0, [inverted_key_schedule]
- aesdec xmm0, [inverted_key_schedule + 10h]
- aesdec xmm0, [inverted_key_schedule + 20h]
- aesdec xmm0, [inverted_key_schedule + 30h]
- aesdec xmm0, [inverted_key_schedule + 40h]
- aesdec xmm0, [inverted_key_schedule + 50h]
- aesdec xmm0, [inverted_key_schedule + 60h]
- aesdec xmm0, [inverted_key_schedule + 70h]
- aesdec xmm0, [inverted_key_schedule + 80h]
- aesdec xmm0, [inverted_key_schedule + 90h]
- aesdeclast xmm0, [inverted_key_schedule + 0A0h]
+ aesdec xmm0, [ecx + 10h]
+ aesdec xmm0, [ecx + 20h]
+ aesdec xmm0, [ecx + 30h]
+ aesdec xmm0, [ecx + 40h]
+ aesdec xmm0, [ecx + 50h]
+ aesdec xmm0, [ecx + 60h]
+ aesdec xmm0, [ecx + 70h]
+ aesdec xmm0, [ecx + 80h]
+ aesdec xmm0, [ecx + 90h]
+ aesdeclast xmm0, [ecx + 0A0h]
ret
-@raw_aes128ecb_decrypt@32 endp
+@raw_aes128ecb_decrypt@20 endp
-@raw_aes128cbc_decrypt@36 proc
- push ecx
- call @raw_aes128ecb_decrypt@32
- pop ecx
- pxor xmm0, [ecx]
+@raw_aes128cbc_decrypt@24 proc
+ call @raw_aes128ecb_decrypt@20
+ pxor xmm0, [edx]
ret
-@raw_aes128cbc_decrypt@36 endp
+@raw_aes128cbc_decrypt@24 endp
-expand_keys128 proc
+@raw_aes128_expand_key_schedule@20 proc
; A "word" (in terms of the FIPS 187 standard) is a 32-bit block.
; Words are denoted by `w[N]`.
;
; A key schedule is composed of 10 "regular" keys and a dumb key for
; the "whitening" step.
- ; It's stored in `key_schedule`.
;
; A key schedule is thus composed of 44 "words".
; The FIPS standard includes an algorithm to calculate these words via
@@ -104,52 +91,51 @@ expand_keys128 proc
; and RotWord using `aeskeygenassist`, which is used in this routine.
;
; Preconditions:
- ; * xmm1[127:96] == w[3],
- ; * xmm1[95:64] == w[2],
- ; * xmm1[63:32] == w[1],
- ; * xmm1[31:0] == w[0].
+ ; * xmm0[127:96] == w[3],
+ ; * xmm0[95:64] == w[2],
+ ; * xmm0[63:32] == w[1],
+ ; * xmm0[31:0] == w[0].
- movdqa [key_schedule], xmm1 ; sets w[0], w[1], w[2], w[3]
+ movdqa [ecx], xmm0 ; sets w[0], w[1], w[2], w[3]
+ add ecx, 10h ; ecx = &w[4]
- lea ecx, [key_schedule + 10h] ; ecx = &w[4]
- aeskeygenassist xmm7, xmm1, 01h ; xmm7[127:96] = RotWord(SubWord(w[3]))^Rcon
+ aeskeygenassist xmm7, xmm0, 01h ; xmm7[127:96] = RotWord(SubWord(w[3]))^Rcon
call gen_round_key ; sets w[4], w[5], w[6], w[7]
- aeskeygenassist xmm7, xmm1, 02h ; xmm7[127:96] = RotWord(SubWord(w[7]))^Rcon
+ aeskeygenassist xmm7, xmm0, 02h ; xmm7[127:96] = RotWord(SubWord(w[7]))^Rcon
call gen_round_key ; sets w[8], w[9], w[10], w[11]
- aeskeygenassist xmm7, xmm1, 04h ; xmm7[127:96] = RotWord(SubWord(w[11]))^Rcon
+ aeskeygenassist xmm7, xmm0, 04h ; xmm7[127:96] = RotWord(SubWord(w[11]))^Rcon
call gen_round_key ; sets w[12], w[13], w[14], w[15]
- aeskeygenassist xmm7, xmm1, 08h ; xmm7[127:96] = RotWord(SubWord(w[15]))^Rcon
+ aeskeygenassist xmm7, xmm0, 08h ; xmm7[127:96] = RotWord(SubWord(w[15]))^Rcon
call gen_round_key ; sets w[16], w[17], w[18], w[19]
- aeskeygenassist xmm7, xmm1, 10h ; xmm7[127:96] = RotWord(SubWord(w[19]))^Rcon
+ aeskeygenassist xmm7, xmm0, 10h ; xmm7[127:96] = RotWord(SubWord(w[19]))^Rcon
call gen_round_key ; sets w[20], w[21], w[22], w[23]
- aeskeygenassist xmm7, xmm1, 20h ; xmm7[127:96] = RotWord(SubWord(w[23]))^Rcon
+ aeskeygenassist xmm7, xmm0, 20h ; xmm7[127:96] = RotWord(SubWord(w[23]))^Rcon
call gen_round_key ; sets w[24], w[25], w[26], w[27]
- aeskeygenassist xmm7, xmm1, 40h ; xmm7[127:96] = RotWord(SubWord(w[27]))^Rcon
+ aeskeygenassist xmm7, xmm0, 40h ; xmm7[127:96] = RotWord(SubWord(w[27]))^Rcon
call gen_round_key ; sets w[28], w[29], w[30], w[31]
- aeskeygenassist xmm7, xmm1, 80h ; xmm7[127:96] = RotWord(SubWord(w[31]))^Rcon
+ aeskeygenassist xmm7, xmm0, 80h ; xmm7[127:96] = RotWord(SubWord(w[31]))^Rcon
call gen_round_key ; sets w[32], w[33], w[34], w[35]
- aeskeygenassist xmm7, xmm1, 1Bh ; xmm7[127:96] = RotWord(SubWord(w[35]))^Rcon
+ aeskeygenassist xmm7, xmm0, 1Bh ; xmm7[127:96] = RotWord(SubWord(w[35]))^Rcon
call gen_round_key ; sets w[36], w[37], w[38], w[39]
- aeskeygenassist xmm7, xmm1, 36h ; xmm7[127:96] = RotWord(SubWord(w[39]))^Rcon
+ aeskeygenassist xmm7, xmm0, 36h ; xmm7[127:96] = RotWord(SubWord(w[39]))^Rcon
call gen_round_key ; sets w[40], w[41], w[42], w[43]
- call invert_key_schedule
ret
gen_round_key:
; Preconditions:
- ; * xmm1[127:96] == w[i+3],
- ; * xmm1[95:64] == w[i+2],
- ; * xmm1[63:32] == w[i+1],
- ; * xmm1[31:0] == w[i],
+ ; * xmm0[127:96] == w[i+3],
+ ; * xmm0[95:64] == w[i+2],
+ ; * xmm0[63:32] == w[i+1],
+ ; * xmm0[31:0] == w[i],
; * xmm7[127:96] == RotWord(SubWord(w[i+3]))^Rcon,
; * ecx == &w[i+4].
;
; Postconditions:
- ; * xmm1[127:96] == w[i+7] == RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i],
- ; * xmm1[95:64] == w[i+6] == RotWord(SubWord(w[i+3]))^Rcon^w[i+2]^w[i+1]^w[i],
- ; * xmm1[63:32] == w[i+5] == RotWord(SubWord(w[i+3]))^Rcon^w[i+1]^w[i],
- ; * xmm1[31:0] == w[i+4] == RotWord(SubWord(w[i+3]))^Rcon^w[i],
+ ; * xmm0[127:96] == w[i+7] == RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i],
+ ; * xmm0[95:64] == w[i+6] == RotWord(SubWord(w[i+3]))^Rcon^w[i+2]^w[i+1]^w[i],
+ ; * xmm0[63:32] == w[i+5] == RotWord(SubWord(w[i+3]))^Rcon^w[i+1]^w[i],
+ ; * xmm0[31:0] == w[i+4] == RotWord(SubWord(w[i+3]))^Rcon^w[i],
; * ecx == &w[i+8],
; * the value in xmm6 is also modified.
@@ -158,17 +144,17 @@ gen_round_key:
; w[i+2]^w[i+1]^w[i],
; w[i+1]^w[i] and
; w[i].
- movdqa xmm6, xmm1 ; xmm6 = xmm1
+ movdqa xmm6, xmm0 ; xmm6 = xmm0
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
- ; xmm1[127:96] == w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm1[95:64] == w[i+2]^w[i+1]^w[i]
- ; xmm1[63:32] == w[i+1]^w[i]
- ; xmm1[31:0] == w[i]
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
+ ; xmm0[127:96] == w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm0[95:64] == w[i+2]^w[i+1]^w[i]
+ ; xmm0[63:32] == w[i+1]^w[i]
+ ; xmm0[31:0] == w[i]
; Calculate
; w[i+7] == RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i],
@@ -176,51 +162,52 @@ gen_round_key:
; w[i+5] == RotWord(SubWord(w[i+3]))^Rcon^w[i+1]^w[i] and
; w[i+4] == RotWord(SubWord(w[i+3]))^Rcon^w[i].
pshufd xmm6, xmm7, 0FFh ; xmm6[127:96] = xmm6[95:64] = xmm6[63:32] = xmm6[31:0] = xmm7[127:96]
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
- ; xmm1[127:96] == w[i+7] == RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm1[95:64] == w[i+6] == RotWord(SubWord(w[i+3]))^Rcon^w[i+2]^w[i+1]^w[i]
- ; xmm1[63:32] == w[i+5] == RotWord(SubWord(w[i+3]))^Rcon^w[i+1]^w[i]
- ; xmm1[31:0] == w[i+4] == RotWord(SubWord(w[i+3]))^Rcon^w[i]
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
+ ; xmm0[127:96] == w[i+7] == RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm0[95:64] == w[i+6] == RotWord(SubWord(w[i+3]))^Rcon^w[i+2]^w[i+1]^w[i]
+ ; xmm0[63:32] == w[i+5] == RotWord(SubWord(w[i+3]))^Rcon^w[i+1]^w[i]
+ ; xmm0[31:0] == w[i+4] == RotWord(SubWord(w[i+3]))^Rcon^w[i]
; Set w[i+4], w[i+5], w[i+6] and w[i+7].
- movdqa [ecx], xmm1 ; w[i+7] = RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
+ movdqa [ecx], xmm0 ; w[i+7] = RotWord(SubWord(w[i+3]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
; w[i+6] = RotWord(SubWord(w[i+3]))^Rcon^w[i+2]^w[i+1]^w[i]
; w[i+5] = RotWord(SubWord(w[i+3]))^Rcon^w[i+1]^w[i]
; w[i+4] = RotWord(SubWord(w[i+3]))^Rcon^w[i]
add ecx, 10h ; ecx = &w[i+8]
ret
+@raw_aes128_expand_key_schedule@20 endp
-invert_key_schedule:
- movdqa xmm7, [key_schedule]
- movdqa xmm6, [key_schedule + 0A0h]
- movdqa [inverted_key_schedule], xmm6
- movdqa [inverted_key_schedule + 0A0h], xmm7
+@raw_aes128_invert_key_schedule@8 proc
+ movdqa xmm7, [ecx]
+ movdqa xmm6, [ecx + 0A0h]
+ movdqa [edx], xmm6
+ movdqa [edx + 0A0h], xmm7
- aesimc xmm7, [key_schedule + 10h]
- aesimc xmm6, [key_schedule + 90h]
- movdqa [inverted_key_schedule + 10h], xmm6
- movdqa [inverted_key_schedule + 90h], xmm7
+ aesimc xmm7, [ecx + 10h]
+ aesimc xmm6, [ecx + 90h]
+ movdqa [edx + 10h], xmm6
+ movdqa [edx + 90h], xmm7
- aesimc xmm7, [key_schedule + 20h]
- aesimc xmm6, [key_schedule + 80h]
- movdqa [inverted_key_schedule + 20h], xmm6
- movdqa [inverted_key_schedule + 80h], xmm7
+ aesimc xmm7, [ecx + 20h]
+ aesimc xmm6, [ecx + 80h]
+ movdqa [edx + 20h], xmm6
+ movdqa [edx + 80h], xmm7
- aesimc xmm7, [key_schedule + 30h]
- aesimc xmm6, [key_schedule + 70h]
- movdqa [inverted_key_schedule + 30h], xmm6
- movdqa [inverted_key_schedule + 70h], xmm7
+ aesimc xmm7, [ecx + 30h]
+ aesimc xmm6, [ecx + 70h]
+ movdqa [edx + 30h], xmm6
+ movdqa [edx + 70h], xmm7
- aesimc xmm7, [key_schedule + 40h]
- aesimc xmm6, [key_schedule + 60h]
- movdqa [inverted_key_schedule + 40h], xmm6
- movdqa [inverted_key_schedule + 60h], xmm7
+ aesimc xmm7, [ecx + 40h]
+ aesimc xmm6, [ecx + 60h]
+ movdqa [edx + 40h], xmm6
+ movdqa [edx + 60h], xmm7
- aesimc xmm7, [key_schedule + 50h]
- movdqa [inverted_key_schedule + 50h], xmm7
+ aesimc xmm7, [ecx + 50h]
+ movdqa [edx + 50h], xmm7
ret
-expand_keys128 endp
+@raw_aes128_invert_key_schedule@8 endp
end
diff --git a/src/aes192.asm b/src/aes192.asm
index 55fd540..21e0fa1 100644
--- a/src/aes192.asm
+++ b/src/aes192.asm
@@ -6,72 +6,59 @@
.xmm
.model flat
-.data
-
-align 10h
-key_schedule oword 13 dup(0)
-
-align 10h
-inverted_key_schedule oword 13 dup(0)
-
.code
-@raw_aes192ecb_encrypt@48 proc
- call expand_keys192
- pxor xmm0, [key_schedule]
- aesenc xmm0, [key_schedule + 10h]
- aesenc xmm0, [key_schedule + 20h]
- aesenc xmm0, [key_schedule + 30h]
- aesenc xmm0, [key_schedule + 40h]
- aesenc xmm0, [key_schedule + 50h]
- aesenc xmm0, [key_schedule + 60h]
- aesenc xmm0, [key_schedule + 70h]
- aesenc xmm0, [key_schedule + 80h]
- aesenc xmm0, [key_schedule + 90h]
- aesenc xmm0, [key_schedule + 0A0h]
- aesenc xmm0, [key_schedule + 0B0h]
- aesenclast xmm0, [key_schedule + 0C0h]
+@raw_aes192ecb_encrypt@20 proc
+ pxor xmm0, [ecx]
+ aesenc xmm0, [ecx + 10h]
+ aesenc xmm0, [ecx + 20h]
+ aesenc xmm0, [ecx + 30h]
+ aesenc xmm0, [ecx + 40h]
+ aesenc xmm0, [ecx + 50h]
+ aesenc xmm0, [ecx + 60h]
+ aesenc xmm0, [ecx + 70h]
+ aesenc xmm0, [ecx + 80h]
+ aesenc xmm0, [ecx + 90h]
+ aesenc xmm0, [ecx + 0A0h]
+ aesenc xmm0, [ecx + 0B0h]
+ aesenclast xmm0, [ecx + 0C0h]
ret
-@raw_aes192ecb_encrypt@48 endp
+@raw_aes192ecb_encrypt@20 endp
-@raw_aes192cbc_encrypt@52 proc
- pxor xmm0, [ecx]
- jmp @raw_aes192ecb_encrypt@48
-@raw_aes192cbc_encrypt@52 endp
+@raw_aes192cbc_encrypt@24 proc
+ pxor xmm0, [edx]
+ jmp @raw_aes192ecb_encrypt@20
+@raw_aes192cbc_encrypt@24 endp
-@raw_aes192ecb_decrypt@48 proc
- call expand_keys192
- pxor xmm0, [inverted_key_schedule]
- aesdec xmm0, [inverted_key_schedule + 10h]
- aesdec xmm0, [inverted_key_schedule + 20h]
- aesdec xmm0, [inverted_key_schedule + 30h]
- aesdec xmm0, [inverted_key_schedule + 40h]
- aesdec xmm0, [inverted_key_schedule + 50h]
- aesdec xmm0, [inverted_key_schedule + 60h]
- aesdec xmm0, [inverted_key_schedule + 70h]
- aesdec xmm0, [inverted_key_schedule + 80h]
- aesdec xmm0, [inverted_key_schedule + 90h]
- aesdec xmm0, [inverted_key_schedule + 0A0h]
- aesdec xmm0, [inverted_key_schedule + 0B0h]
- aesdeclast xmm0, [inverted_key_schedule + 0C0h]
+@raw_aes192ecb_decrypt@20 proc
+ pxor xmm0, [ecx]
+ aesdec xmm0, [ecx + 10h]
+ aesdec xmm0, [ecx + 20h]
+ aesdec xmm0, [ecx + 30h]
+ aesdec xmm0, [ecx + 40h]
+ aesdec xmm0, [ecx + 50h]
+ aesdec xmm0, [ecx + 60h]
+ aesdec xmm0, [ecx + 70h]
+ aesdec xmm0, [ecx + 80h]
+ aesdec xmm0, [ecx + 90h]
+ aesdec xmm0, [ecx + 0A0h]
+ aesdec xmm0, [ecx + 0B0h]
+ aesdeclast xmm0, [ecx + 0C0h]
ret
-@raw_aes192ecb_decrypt@48 endp
+@raw_aes192ecb_decrypt@20 endp
-@raw_aes192cbc_decrypt@52 proc
- push ecx
- call @raw_aes192ecb_decrypt@48
- pop ecx
- pxor xmm0, [ecx]
+@raw_aes192cbc_decrypt@24 proc
+ call @raw_aes192ecb_decrypt@20
+ pxor xmm0, [edx]
ret
-@raw_aes192cbc_decrypt@52 endp
+@raw_aes192cbc_decrypt@24 endp
-expand_keys192 proc
+@raw_aes192_expand_key_schedule@36 proc
; A "word" (in terms of the FIPS 187 standard) is a 32-bit block.
; Words are denoted by `w[N]`.
;
; A key schedule is composed of 12 "regular" keys and a dumb key for
; the "whitening" step.
- ; It's stored in `key_schedule`.
;
; A key schedule is thus composed of 52 "words".
; The FIPS standard includes an algorithm to calculate these words via
@@ -116,59 +103,58 @@ expand_keys192 proc
; and RotWord using `aeskeygenassist`, which is used in this routine.
;
; Preconditions:
- ; * xmm2[63:32] == w[5],
- ; * xmm2[31:0] == w[4],
- ; * xmm1[127:96] == w[3],
- ; * xmm1[95:64] == w[2],
- ; * xmm1[63:32] == w[1],
- ; * xmm1[31:0] == w[0].
+ ; * xmm1[63:32] == w[5],
+ ; * xmm1[31:0] == w[4],
+ ; * xmm0[127:96] == w[3],
+ ; * xmm0[95:64] == w[2],
+ ; * xmm0[63:32] == w[1],
+ ; * xmm0[31:0] == w[0].
- movdqa [key_schedule], xmm1 ; sets w[0], w[1], w[2], w[3]
- movdqa [key_schedule + 10h], xmm2 ; sets w[4], w[5]
+ movdqa [ecx], xmm0 ; sets w[0], w[1], w[2], w[3]
+ movdqa [ecx + 10h], xmm1 ; sets w[4], w[5]
+ add ecx, 18h ; ecx = &w[6]
- lea ecx, [key_schedule + 18h] ; ecx = &w[6]
- aeskeygenassist xmm7, xmm2, 1 ; xmm7[63:32] = RotWord(SubWord(w[5]))^Rcon,
+ aeskeygenassist xmm7, xmm1, 1 ; xmm7[63:32] = RotWord(SubWord(w[5]))^Rcon,
call gen_round_key ; sets w[6], w[7], w[8], w[9], w[10], w[11]
- aeskeygenassist xmm7, xmm2, 2 ; xmm7[63:32] = RotWord(SubWord(w[11]))^Rcon
+ aeskeygenassist xmm7, xmm1, 2 ; xmm7[63:32] = RotWord(SubWord(w[11]))^Rcon
call gen_round_key ; sets w[12], w[13], w[14], w[15], w[16], w[17]
- aeskeygenassist xmm7, xmm2, 4 ; xmm7[63:32] = RotWord(SubWord(w[17]))^Rcon
+ aeskeygenassist xmm7, xmm1, 4 ; xmm7[63:32] = RotWord(SubWord(w[17]))^Rcon
call gen_round_key ; sets w[18], w[19], w[20], w[21], w[22], w[23]
- aeskeygenassist xmm7, xmm2, 8 ; xmm7[63:32] = RotWord(SubWord(w[23]))^Rcon
+ aeskeygenassist xmm7, xmm1, 8 ; xmm7[63:32] = RotWord(SubWord(w[23]))^Rcon
call gen_round_key ; sets w[24], w[25], w[26], w[27], w[28], w[29]
- aeskeygenassist xmm7, xmm2, 10h ; xmm7[63:32] = RotWord(SubWord(w[29]))^Rcon
+ aeskeygenassist xmm7, xmm1, 10h ; xmm7[63:32] = RotWord(SubWord(w[29]))^Rcon
call gen_round_key ; sets w[30], w[31], w[32], w[33], w[34], w[35]
- aeskeygenassist xmm7, xmm2, 20h ; xmm7[63:32] = RotWord(SubWord(w[35]))^Rcon
+ aeskeygenassist xmm7, xmm1, 20h ; xmm7[63:32] = RotWord(SubWord(w[35]))^Rcon
call gen_round_key ; sets w[36], w[37], w[38], w[39], w[40], w[41]
- aeskeygenassist xmm7, xmm2, 40h ; xmm7[63:32] = RotWord(SubWord(w[41]))^Rcon
+ aeskeygenassist xmm7, xmm1, 40h ; xmm7[63:32] = RotWord(SubWord(w[41]))^Rcon
call gen_round_key ; sets w[42], w[43], w[44], w[45], w[46], w[47]
- aeskeygenassist xmm7, xmm2, 80h ; xmm7[63:32] = RotWord(SubWord(w[49]))^Rcon
+ aeskeygenassist xmm7, xmm1, 80h ; xmm7[63:32] = RotWord(SubWord(w[49]))^Rcon
call gen_round_key ; sets w[48], w[49], w[50], w[51], w[52], w[53] // FIXME
- call invert_key_schedule
ret
gen_round_key:
; Preconditions:
- ; * xmm2[127:96] == 0,
- ; * xmm2[95:64] == 0,
- ; * xmm2[63:32] == w[i+5],
- ; * xmm2[31:0] == w[i+4],
- ; * xmm1[127:96] == w[i+3],
- ; * xmm1[95:64] == w[i+2],
- ; * xmm1[63:32] == w[i+1],
- ; * xmm1[31:0] == w[i],
+ ; * xmm1[127:96] == 0,
+ ; * xmm1[95:64] == 0,
+ ; * xmm1[63:32] == w[i+5],
+ ; * xmm1[31:0] == w[i+4],
+ ; * xmm0[127:96] == w[i+3],
+ ; * xmm0[95:64] == w[i+2],
+ ; * xmm0[63:32] == w[i+1],
+ ; * xmm0[31:0] == w[i],
; * xmm7[63:32] == RotWord(SubWord(w[i+5]))^Rcon,
; * ecx == &w[i+6].
;
; Postconditions:
- ; * xmm2[127:96] == 0,
- ; * xmm2[95:64] == 0,
- ; * xmm2[63:32] == w[i+11] == RotWord(SubWord(w[i+5]))^Rcon^w[i+5]^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i],
- ; * xmm2[31:0] == w[i+10] == RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i],
- ; * xmm1[127:96] == w[i+9] == RotWord(SubWord(w[i+5]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i],
- ; * xmm1[95:64] == w[i+8] == RotWord(SubWord(w[i+5]))^Rcon^w[i+2]^w[i+1]^w[i],
- ; * xmm1[63:32] == w[i+7] == RotWord(SubWord(w[i+5]))^Rcon^w[i+1]^w[i],
- ; * xmm1[31:0] == w[i+6] == RotWord(SubWord(w[i+5]))^Rcon^w[i],
+ ; * xmm1[127:96] == 0,
+ ; * xmm1[95:64] == 0,
+ ; * xmm1[63:32] == w[i+11] == RotWord(SubWord(w[i+5]))^Rcon^w[i+5]^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i],
+ ; * xmm1[31:0] == w[i+10] == RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i],
+ ; * xmm0[127:96] == w[i+9] == RotWord(SubWord(w[i+5]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i],
+ ; * xmm0[95:64] == w[i+8] == RotWord(SubWord(w[i+5]))^Rcon^w[i+2]^w[i+1]^w[i],
+ ; * xmm0[63:32] == w[i+7] == RotWord(SubWord(w[i+5]))^Rcon^w[i+1]^w[i],
+ ; * xmm0[31:0] == w[i+6] == RotWord(SubWord(w[i+5]))^Rcon^w[i],
; * ecx == &w[i+12],
; * the value in xmm6 is also modified.
@@ -177,17 +163,17 @@ gen_round_key:
; w[i+2]^w[i+1]^w[i],
; w[i+1]^w[i] and
; w[i].
- movdqa xmm6, xmm1 ; xmm6 = xmm1
+ movdqa xmm6, xmm0 ; xmm6 = xmm0
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
- ; xmm1[127:96] == w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm1[95:64] == w[i+2]^w[i+1]^w[i]
- ; xmm1[63:32] == w[i+1]^w[i]
- ; xmm1[31:0] == w[i]
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
+ ; xmm0[127:96] == w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm0[95:64] == w[i+2]^w[i+1]^w[i]
+ ; xmm0[63:32] == w[i+1]^w[i]
+ ; xmm0[31:0] == w[i]
; Calculate
; w[i+9] == RotWord(SubWord(w[i+5]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i],
@@ -195,14 +181,14 @@ gen_round_key:
; w[i+7] == RotWord(SubWord(w[i+5]))^Rcon^w[i+1]^w[i] and
; w[i+6] == RotWord(SubWord(w[i+5]))^Rcon^w[i].
pshufd xmm6, xmm7, 55h ; xmm6[127:96] = xmm6[95:64] = xmm6[63:32] = xmm6[31:0] = xmm7[63:32]
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
- ; xmm1[127:96] == w[i+9] == RotWord(SubWord(w[i+5]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm1[95:64] == w[i+8] == RotWord(SubWord(w[i+5]))^Rcon^w[i+2]^w[i+1]^w[i]
- ; xmm1[63:32] == w[i+7] == RotWord(SubWord(w[i+5]))^Rcon^w[i+1]^w[i]
- ; xmm1[31:0] == w[i+6] == RotWord(SubWord(w[i+5]))^Rcon^w[i]
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
+ ; xmm0[127:96] == w[i+9] == RotWord(SubWord(w[i+5]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm0[95:64] == w[i+8] == RotWord(SubWord(w[i+5]))^Rcon^w[i+2]^w[i+1]^w[i]
+ ; xmm0[63:32] == w[i+7] == RotWord(SubWord(w[i+5]))^Rcon^w[i+1]^w[i]
+ ; xmm0[31:0] == w[i+6] == RotWord(SubWord(w[i+5]))^Rcon^w[i]
; Set w[i+6], w[i+7], w[i+8] and w[i+9].
- movdqu [ecx], xmm1 ; w[i+6] = RotWord(SubWord(w[i+5]))^Rcon^w[i]
+ movdqu [ecx], xmm0 ; w[i+6] = RotWord(SubWord(w[i+5]))^Rcon^w[i]
; w[i+7] = RotWord(SubWord(w[i+5]))^Rcon^w[i+1]^w[i]
; w[i+8] = RotWord(SubWord(w[i+5]))^Rcon^w[i+2]^w[i+1]^w[i]
; w[i+9] = RotWord(SubWord(w[i+5]))^Rcon^w[i+3]^w[i+2]^w[i+1]^w[i]
@@ -211,62 +197,63 @@ gen_round_key:
; Calculate
; w[i+5]^w[i+4],
; w[i+4].
- pshufd xmm6, xmm2, 0F3h ; xmm6 = xmm2[31:0] << 32
- pxor xmm2, xmm6 ; xmm2 ^= xmm7
- ; xmm2[63:32] == w[i+5]^w[i+4]
- ; xmm2[31:0] == w[i+4]
+ pshufd xmm6, xmm1, 0F3h ; xmm6 = xmm1[31:0] << 32
+ pxor xmm1, xmm6 ; xmm1 ^= xmm7
+ ; xmm1[63:32] == w[i+5]^w[i+4]
+ ; xmm1[31:0] == w[i+4]
; Calculate
; w[i+10] == RotWord(SubWord(w[i+5]))^Rcon^w[i+5]^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i],
; w[i+11] == RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i].
- pshufd xmm6, xmm1, 0FFh ; xmm6[127:96] = xmm6[95:64] = xmm6[63:32] = xmm6[31:0] = xmm1[127:96]
+ pshufd xmm6, xmm0, 0FFh ; xmm6[127:96] = xmm6[95:64] = xmm6[63:32] = xmm6[31:0] = xmm0[127:96]
psrldq xmm6, 8 ; xmm6 >>= 64
- pxor xmm2, xmm6 ; xmm2 ^= xmm6
- ; xmm2[63:32] == w[i+11] == RotWord(SubWord(w[i+5]))^Rcon^w[i+5]^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm2[31:0] == w[i+10] == RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
+ pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ ; xmm1[63:32] == w[i+11] == RotWord(SubWord(w[i+5]))^Rcon^w[i+5]^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm1[31:0] == w[i+10] == RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
; Set w[i+10] and w[i+11].
- movq qword ptr [ecx], xmm2 ; w[i+10] = RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
+ movq qword ptr [ecx], xmm1 ; w[i+10] = RotWord(SubWord(w[i+5]))^Rcon^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
; w[i+11] = RotWord(SubWord(w[i+5]))^Rcon^w[i+5]^w[i+4]^w[i+3]^w[i+2]^w[i+1]^w[i]
add ecx, 8 ; ecx = &w[i+12]
ret
+@raw_aes192_expand_key_schedule@36 endp
-invert_key_schedule:
- movdqa xmm7, [key_schedule]
- movdqa xmm6, [key_schedule + 0C0h]
- movdqa [inverted_key_schedule], xmm6
- movdqa [inverted_key_schedule + 0C0h], xmm7
+@raw_aes192_invert_key_schedule@8 proc
+ movdqa xmm7, [ecx]
+ movdqa xmm6, [ecx + 0C0h]
+ movdqa [edx], xmm6
+ movdqa [edx + 0C0h], xmm7
- aesimc xmm7, [key_schedule + 10h]
- aesimc xmm6, [key_schedule + 0B0h]
- movdqa [inverted_key_schedule + 10h], xmm6
- movdqa [inverted_key_schedule + 0B0h], xmm7
+ aesimc xmm7, [ecx + 10h]
+ aesimc xmm6, [ecx + 0B0h]
+ movdqa [edx + 10h], xmm6
+ movdqa [edx + 0B0h], xmm7
- aesimc xmm7, [key_schedule + 20h]
- aesimc xmm6, [key_schedule + 0A0h]
- movdqa [inverted_key_schedule + 20h], xmm6
- movdqa [inverted_key_schedule + 0A0h], xmm7
+ aesimc xmm7, [ecx + 20h]
+ aesimc xmm6, [ecx + 0A0h]
+ movdqa [edx + 20h], xmm6
+ movdqa [edx + 0A0h], xmm7
- aesimc xmm7, [key_schedule + 30h]
- aesimc xmm6, [key_schedule + 90h]
- movdqa [inverted_key_schedule + 30h], xmm6
- movdqa [inverted_key_schedule + 90h], xmm7
+ aesimc xmm7, [ecx + 30h]
+ aesimc xmm6, [ecx + 90h]
+ movdqa [edx + 30h], xmm6
+ movdqa [edx + 90h], xmm7
- aesimc xmm7, [key_schedule + 40h]
- aesimc xmm6, [key_schedule + 80h]
- movdqa [inverted_key_schedule + 40h], xmm6
- movdqa [inverted_key_schedule + 80h], xmm7
+ aesimc xmm7, [ecx + 40h]
+ aesimc xmm6, [ecx + 80h]
+ movdqa [edx + 40h], xmm6
+ movdqa [edx + 80h], xmm7
- aesimc xmm7, [key_schedule + 50h]
- aesimc xmm6, [key_schedule + 70h]
- movdqa [inverted_key_schedule + 50h], xmm6
- movdqa [inverted_key_schedule + 70h], xmm7
+ aesimc xmm7, [ecx + 50h]
+ aesimc xmm6, [ecx + 70h]
+ movdqa [edx + 50h], xmm6
+ movdqa [edx + 70h], xmm7
- aesimc xmm7, [key_schedule + 60h]
- movdqa [inverted_key_schedule + 60h], xmm7
+ aesimc xmm7, [ecx + 60h]
+ movdqa [edx + 60h], xmm7
ret
-expand_keys192 endp
+@raw_aes192_invert_key_schedule@8 endp
end
diff --git a/src/aes256.asm b/src/aes256.asm
index acea8e3..e59a253 100644
--- a/src/aes256.asm
+++ b/src/aes256.asm
@@ -6,76 +6,63 @@
.xmm
.model flat
-.data
-
-align 10h
-key_schedule oword 15 dup(0)
-
-align 10h
-inverse_key_schedule oword 15 dup(0)
-
.code
-@raw_aes256ecb_encrypt@48 proc
- call expand_keys256
- pxor xmm0, [key_schedule]
- aesenc xmm0, [key_schedule + 10h]
- aesenc xmm0, [key_schedule + 20h]
- aesenc xmm0, [key_schedule + 30h]
- aesenc xmm0, [key_schedule + 40h]
- aesenc xmm0, [key_schedule + 50h]
- aesenc xmm0, [key_schedule + 60h]
- aesenc xmm0, [key_schedule + 70h]
- aesenc xmm0, [key_schedule + 80h]
- aesenc xmm0, [key_schedule + 90h]
- aesenc xmm0, [key_schedule + 0A0h]
- aesenc xmm0, [key_schedule + 0B0h]
- aesenc xmm0, [key_schedule + 0C0h]
- aesenc xmm0, [key_schedule + 0D0h]
- aesenclast xmm0, [key_schedule + 0E0h]
+@raw_aes256ecb_encrypt@20 proc
+ pxor xmm0, [ecx]
+ aesenc xmm0, [ecx + 10h]
+ aesenc xmm0, [ecx + 20h]
+ aesenc xmm0, [ecx + 30h]
+ aesenc xmm0, [ecx + 40h]
+ aesenc xmm0, [ecx + 50h]
+ aesenc xmm0, [ecx + 60h]
+ aesenc xmm0, [ecx + 70h]
+ aesenc xmm0, [ecx + 80h]
+ aesenc xmm0, [ecx + 90h]
+ aesenc xmm0, [ecx + 0A0h]
+ aesenc xmm0, [ecx + 0B0h]
+ aesenc xmm0, [ecx + 0C0h]
+ aesenc xmm0, [ecx + 0D0h]
+ aesenclast xmm0, [ecx + 0E0h]
ret
-@raw_aes256ecb_encrypt@48 endp
+@raw_aes256ecb_encrypt@20 endp
+
+@raw_aes256cbc_encrypt@24 proc
+ pxor xmm0, [edx]
+ jmp @raw_aes256ecb_encrypt@20
+@raw_aes256cbc_encrypt@24 endp
-@raw_aes256cbc_encrypt@52 proc
+@raw_aes256ecb_decrypt@20 proc
pxor xmm0, [ecx]
- jmp @raw_aes256ecb_encrypt@48
-@raw_aes256cbc_encrypt@52 endp
-
-@raw_aes256ecb_decrypt@48 proc
- call expand_keys256
- pxor xmm0, [inverse_key_schedule]
- aesdec xmm0, [inverse_key_schedule + 10h]
- aesdec xmm0, [inverse_key_schedule + 20h]
- aesdec xmm0, [inverse_key_schedule + 30h]
- aesdec xmm0, [inverse_key_schedule + 40h]
- aesdec xmm0, [inverse_key_schedule + 50h]
- aesdec xmm0, [inverse_key_schedule + 60h]
- aesdec xmm0, [inverse_key_schedule + 70h]
- aesdec xmm0, [inverse_key_schedule + 80h]
- aesdec xmm0, [inverse_key_schedule + 90h]
- aesdec xmm0, [inverse_key_schedule + 0A0h]
- aesdec xmm0, [inverse_key_schedule + 0B0h]
- aesdec xmm0, [inverse_key_schedule + 0C0h]
- aesdec xmm0, [inverse_key_schedule + 0D0h]
- aesdeclast xmm0, [inverse_key_schedule + 0E0h]
+ aesdec xmm0, [ecx + 10h]
+ aesdec xmm0, [ecx + 20h]
+ aesdec xmm0, [ecx + 30h]
+ aesdec xmm0, [ecx + 40h]
+ aesdec xmm0, [ecx + 50h]
+ aesdec xmm0, [ecx + 60h]
+ aesdec xmm0, [ecx + 70h]
+ aesdec xmm0, [ecx + 80h]
+ aesdec xmm0, [ecx + 90h]
+ aesdec xmm0, [ecx + 0A0h]
+ aesdec xmm0, [ecx + 0B0h]
+ aesdec xmm0, [ecx + 0C0h]
+ aesdec xmm0, [ecx + 0D0h]
+ aesdeclast xmm0, [ecx + 0E0h]
ret
-@raw_aes256ecb_decrypt@48 endp
+@raw_aes256ecb_decrypt@20 endp
-@raw_aes256cbc_decrypt@52 proc
- push ecx
- call @raw_aes256ecb_decrypt@48
- pop ecx
- pxor xmm0, [ecx]
+@raw_aes256cbc_decrypt@24 proc
+ call @raw_aes256ecb_decrypt@20
+ pxor xmm0, [edx]
ret
-@raw_aes256cbc_decrypt@52 endp
+@raw_aes256cbc_decrypt@24 endp
-expand_keys256 proc
+@raw_aes256_expand_key_schedule@36 proc
; A "word" (in terms of the FIPS 187 standard) is a 32-bit block.
; Words are denoted by `w[N]`.
;
; A key schedule is composed of 14 "regular" keys and a dumb key for
; the "whitening" step.
- ; It's stored in `key_schedule`.
;
; A key schedule is thus composed of 60 "words".
; The FIPS standard includes an algorithm to calculate these words via
@@ -128,99 +115,97 @@ expand_keys256 proc
; and RotWord using `aeskeygenassist`, which is used in this routine.
;
; Preconditions:
- ; * xmm2[127:96] == w[7],
- ; * xmm2[95:64] == w[6],
- ; * xmm2[63:32] == w[5],
- ; * xmm2[31:0] == w[4],
- ; * xmm1[127:96] == w[3],
- ; * xmm1[95:64] == w[2],
- ; * xmm1[63:32] == w[1],
- ; * xmm1[31:0] == w[0].
-
- movdqa [key_schedule], xmm1 ; sets w[0], w[1], w[2], w[3]
- movdqa [key_schedule + 10h], xmm2 ; sets w[4], w[5], w[6], w[7]
-
- lea ecx, [key_schedule + 20h] ; ecx = &w[8]
-
- aeskeygenassist xmm7, xmm2, 1h ; xmm7[127:96] = RotWord(SubWord(w[7]))^Rcon
+ ; * xmm1[127:96] == w[7],
+ ; * xmm1[95:64] == w[6],
+ ; * xmm1[63:32] == w[5],
+ ; * xmm1[31:0] == w[4],
+ ; * xmm0[127:96] == w[3],
+ ; * xmm0[95:64] == w[2],
+ ; * xmm0[63:32] == w[1],
+ ; * xmm0[31:0] == w[0].
+
+ movdqa [ecx], xmm0 ; sets w[0], w[1], w[2], w[3]
+ movdqa [ecx + 10h], xmm1 ; sets w[4], w[5], w[6], w[7]
+ lea ecx, [ecx + 20h] ; ecx = &w[8]
+
+ aeskeygenassist xmm7, xmm1, 1h ; xmm7[127:96] = RotWord(SubWord(w[7]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[8], w[9], w[10], w[11]
- aeskeygenassist xmm7, xmm2, 0 ; xmm7[95:64] = SubWord(w[11])
+ aeskeygenassist xmm7, xmm1, 0 ; xmm7[95:64] = SubWord(w[11])
pshufd xmm7, xmm7, 0AAh ; xmm7[127:96] = xmm7[63:32] = xmm7[31:0] = xmm7[95:64]
call gen_round_key ; sets w[12], w[13], w[14], w[15]
- aeskeygenassist xmm7, xmm2, 2h ; xmm7[127:96] = RotWord(SubWord(w[15]))^Rcon
+ aeskeygenassist xmm7, xmm1, 2h ; xmm7[127:96] = RotWord(SubWord(w[15]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[16], w[17], w[18], w[19]
- aeskeygenassist xmm7, xmm2, 0 ; xmm7[95:64] = SubWord(w[19])
+ aeskeygenassist xmm7, xmm1, 0 ; xmm7[95:64] = SubWord(w[19])
pshufd xmm7, xmm7, 0AAh ; xmm7[127:96] = xmm7[63:32] = xmm7[31:0] = xmm7[95:64]
call gen_round_key ; sets w[20], w[21], w[22], w[23]
- aeskeygenassist xmm7, xmm2, 4h ; xmm7[127:96] = RotWord(SubWord(w[23]))^Rcon
+ aeskeygenassist xmm7, xmm1, 4h ; xmm7[127:96] = RotWord(SubWord(w[23]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[24], w[25], w[26], w[27]
- aeskeygenassist xmm7, xmm2, 0 ; xmm7[95:64] = SubWord(w[27])
+ aeskeygenassist xmm7, xmm1, 0 ; xmm7[95:64] = SubWord(w[27])
pshufd xmm7, xmm7, 0AAh ; xmm7[127:96] = xmm7[63:32] = xmm7[31:0] = xmm7[95:64]
call gen_round_key ; sets w[28], w[29], w[30], w[31]
- aeskeygenassist xmm7, xmm2, 8h ; xmm7[127:96] = RotWord(SubWord(w[31]))^Rcon
+ aeskeygenassist xmm7, xmm1, 8h ; xmm7[127:96] = RotWord(SubWord(w[31]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[32], w[33], w[34], w[35]
- aeskeygenassist xmm7, xmm2, 0 ; xmm7[95:64] = SubWord(w[35])
+ aeskeygenassist xmm7, xmm1, 0 ; xmm7[95:64] = SubWord(w[35])
pshufd xmm7, xmm7, 0AAh ; xmm7[127:96] = xmm7[63:32] = xmm7[31:0] = xmm7[95:64]
call gen_round_key ; sets w[36], w[37], w[38], w[39]
- aeskeygenassist xmm7, xmm2, 10h ; xmm7[127:96] = RotWord(SubWord(w[39]))^Rcon
+ aeskeygenassist xmm7, xmm1, 10h ; xmm7[127:96] = RotWord(SubWord(w[39]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[40], w[41], w[42], w[43]
- aeskeygenassist xmm7, xmm2, 0 ; xmm7[95:64] = SubWord(w[43])
+ aeskeygenassist xmm7, xmm1, 0 ; xmm7[95:64] = SubWord(w[43])
pshufd xmm7, xmm7, 0AAh ; xmm7[127:96] = xmm7[63:32] = xmm7[31:0] = xmm7[95:64]
call gen_round_key ; sets w[44], w[45], w[46], w[47]
- aeskeygenassist xmm7, xmm2, 20h ; xmm7[127:96] = RotWord(SubWord(w[47]))^Rcon
+ aeskeygenassist xmm7, xmm1, 20h ; xmm7[127:96] = RotWord(SubWord(w[47]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[48], w[49], w[50], w[51]
- aeskeygenassist xmm7, xmm2, 0 ; xmm7[95:64] = SubWord(w[51])
+ aeskeygenassist xmm7, xmm1, 0 ; xmm7[95:64] = SubWord(w[51])
pshufd xmm7, xmm7, 0AAh ; xmm7[127:96] = xmm7[63:32] = xmm7[31:0] = xmm7[95:64]
call gen_round_key ; sets w[52], w[53], w[54], w[55]
- aeskeygenassist xmm7, xmm2, 40h ; xmm7[127:96] = RotWord(SubWord(w[55]))^Rcon
+ aeskeygenassist xmm7, xmm1, 40h ; xmm7[127:96] = RotWord(SubWord(w[55]))^Rcon
pshufd xmm7, xmm7, 0FFh ; xmm7[95:64] = xmm7[63:32] = xmm7[31:0] = xmm7[127:96]
call gen_round_key ; sets w[56], w[57], w[58], w[59]
- call invert_key_schedule
ret
gen_round_key:
; Preconditions:
- ; * xmm2[127:96] == w[i+7],
- ; * xmm2[95:64] == w[i+6],
- ; * xmm2[63:32] == w[i+5],
- ; * xmm2[31:0] == w[i+4],
- ; * xmm1[127:96] == w[i+3],
- ; * xmm1[95:64] == w[i+2],
- ; * xmm1[63:32] == w[i+1],
- ; * xmm1[31:0] == w[i],
+ ; * xmm1[127:96] == w[i+7],
+ ; * xmm1[95:64] == w[i+6],
+ ; * xmm1[63:32] == w[i+5],
+ ; * xmm1[31:0] == w[i+4],
+ ; * xmm0[127:96] == w[i+3],
+ ; * xmm0[95:64] == w[i+2],
+ ; * xmm0[63:32] == w[i+1],
+ ; * xmm0[31:0] == w[i],
; * xmm7[127:96] == xmm7[95:64] == xmm7[63:32] == xmm7[31:0] == HWGEN,
; where HWGEN is either RotWord(SubWord(w[i+7]))^Rcon or SubWord(w[i+7]),
; depending on the number of the round being processed,
; * ecx == &w[i+8].
;
; Postconditions:
- ; * xmm2[127:96] == w[i+11] == HWGEN^w[i+3]^w[i+2]^w[i+1]^w[i],
- ; * xmm2[95:64] == w[i+10] == HWGEN^w[i+2]^w[i+1]^w[i],
- ; * xmm2[63:32] == w[i+9] == HWGEN^w[i+1]^w[i],
- ; * xmm2[31:0] == w[i+8] == HWGEN^w[i],
- ; * xmm1[127:96] == w[i+7],
- ; * xmm1[95:64] == w[i+6],
- ; * xmm1[63:32] == w[i+5],
- ; * xmm1[31:0] == w[i+4],
+ ; * xmm1[127:96] == w[i+11] == HWGEN^w[i+3]^w[i+2]^w[i+1]^w[i],
+ ; * xmm1[95:64] == w[i+10] == HWGEN^w[i+2]^w[i+1]^w[i],
+ ; * xmm1[63:32] == w[i+9] == HWGEN^w[i+1]^w[i],
+ ; * xmm1[31:0] == w[i+8] == HWGEN^w[i],
+ ; * xmm0[127:96] == w[i+7],
+ ; * xmm0[95:64] == w[i+6],
+ ; * xmm0[63:32] == w[i+5],
+ ; * xmm0[31:0] == w[i+4],
; * ecx == &w[i+12],
; * the value in xmm6 is also modified.
@@ -229,83 +214,84 @@ gen_round_key:
; w[i+2]^w[i+1]^w[i],
; w[i+1]^w[i] and
; w[i].
- movdqa xmm6, xmm1 ; xmm6 = xmm1
+ movdqa xmm6, xmm0 ; xmm6 = xmm0
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
pslldq xmm6, 4 ; xmm6 <<= 32
- pxor xmm1, xmm6 ; xmm1 ^= xmm6
- ; xmm1[127:96] == w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm1[95:64] == w[i+2]^w[i+1]^w[i]
- ; xmm1[63:32] == w[i+1]^w[i]
- ; xmm1[31:0] == w[i]
+ pxor xmm0, xmm6 ; xmm0 ^= xmm6
+ ; xmm0[127:96] == w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm0[95:64] == w[i+2]^w[i+1]^w[i]
+ ; xmm0[63:32] == w[i+1]^w[i]
+ ; xmm0[31:0] == w[i]
; Calculate
; HWGEN^w[i+3]^w[i+2]^w[i+1]^w[i],
; HWGEN^w[i+2]^w[i+1]^w[i],
; HWGEN^w[i+1]^w[i] and
; HWGEN^w[i].
- pxor xmm1, xmm7 ; xmm1 ^= xmm7
- ; xmm1[127:96] == w[i+11] == HWGEN^w[i+3]^w[i+2]^w[i+1]^w[i]
- ; xmm1[95:64] == w[i+10] == HWGEN^w[i+2]^w[i+1]^w[i]
- ; xmm1[63:32] == w[i+9] == HWGEN^w[i+1]^w[i]
- ; xmm1[31:0] == w[i+8] == HWGEN^w[i]
+ pxor xmm0, xmm7 ; xmm0 ^= xmm7
+ ; xmm0[127:96] == w[i+11] == HWGEN^w[i+3]^w[i+2]^w[i+1]^w[i]
+ ; xmm0[95:64] == w[i+10] == HWGEN^w[i+2]^w[i+1]^w[i]
+ ; xmm0[63:32] == w[i+9] == HWGEN^w[i+1]^w[i]
+ ; xmm0[31:0] == w[i+8] == HWGEN^w[i]
; Set w[i+8], w[i+9], w[i+10] and w[i+11].
- movdqa [ecx], xmm1 ; w[i+8] = HWGEN^w[i]
+ movdqa [ecx], xmm0 ; w[i+8] = HWGEN^w[i]
; w[i+9] = HWGEN^w[i+1]^w[i]
; w[i+10] = HWGEN^w[i+2]^w[i+1]^w[i]
; w[i+11] = HWGEN^w[i+3]^w[i+2]^w[i+1]^w[i]
add ecx, 10h ; ecx = &w[i+12]
- ; Swap the values in xmm1 and xmm2.
- pxor xmm1, xmm2
- pxor xmm2, xmm1
- pxor xmm1, xmm2
+ ; Swap the values in xmm0 and xmm1.
+ pxor xmm0, xmm1
+ pxor xmm1, xmm0
+ pxor xmm0, xmm1
ret
-
-invert_key_schedule:
- movdqa xmm7, [key_schedule ]
- movdqa xmm6, [key_schedule + 0E0h]
- movdqa [inverse_key_schedule ], xmm6
- movdqa [inverse_key_schedule + 0E0h], xmm7
-
- aesimc xmm7, [key_schedule + 10h]
- aesimc xmm6, [key_schedule + 0D0h]
- movdqa [inverse_key_schedule + 10h], xmm6
- movdqa [inverse_key_schedule + 0D0h], xmm7
-
- aesimc xmm7, [key_schedule + 20h]
- aesimc xmm6, [key_schedule + 0C0h]
- movdqa [inverse_key_schedule + 20h], xmm6
- movdqa [inverse_key_schedule + 0C0h], xmm7
-
- aesimc xmm7, [key_schedule + 30h]
- aesimc xmm6, [key_schedule + 0B0h]
- movdqa [inverse_key_schedule + 30h], xmm6
- movdqa [inverse_key_schedule + 0B0h], xmm7
-
- aesimc xmm7, [key_schedule + 40h]
- aesimc xmm6, [key_schedule + 0A0h]
- movdqa [inverse_key_schedule + 40h], xmm6
- movdqa [inverse_key_schedule + 0A0h], xmm7
-
- aesimc xmm7, [key_schedule + 50h]
- aesimc xmm6, [key_schedule + 90h]
- movdqa [inverse_key_schedule + 50h], xmm6
- movdqa [inverse_key_schedule + 90h], xmm7
-
- aesimc xmm7, [key_schedule + 60h]
- aesimc xmm6, [key_schedule + 80h]
- movdqa [inverse_key_schedule + 60h], xmm6
- movdqa [inverse_key_schedule + 80h], xmm7
-
- aesimc xmm7, [key_schedule + 70h]
- movdqa [inverse_key_schedule + 70h], xmm7
+@raw_aes256_expand_key_schedule@36 endp
+
+@raw_aes256_invert_key_schedule@8 proc
+ movdqa xmm7, [ecx]
+ movdqa xmm6, [ecx + 0E0h]
+ movdqa [edx], xmm6
+ movdqa [edx + 0E0h], xmm7
+
+ aesimc xmm7, [ecx + 10h]
+ aesimc xmm6, [ecx + 0D0h]
+ movdqa [edx + 10h], xmm6
+ movdqa [edx + 0D0h], xmm7
+
+ aesimc xmm7, [ecx + 20h]
+ aesimc xmm6, [ecx + 0C0h]
+ movdqa [edx + 20h], xmm6
+ movdqa [edx + 0C0h], xmm7
+
+ aesimc xmm7, [ecx + 30h]
+ aesimc xmm6, [ecx + 0B0h]
+ movdqa [edx + 30h], xmm6
+ movdqa [edx + 0B0h], xmm7
+
+ aesimc xmm7, [ecx + 40h]
+ aesimc xmm6, [ecx + 0A0h]
+ movdqa [edx + 40h], xmm6
+ movdqa [edx + 0A0h], xmm7
+
+ aesimc xmm7, [ecx + 50h]
+ aesimc xmm6, [ecx + 90h]
+ movdqa [edx + 50h], xmm6
+ movdqa [edx + 90h], xmm7
+
+ aesimc xmm7, [ecx + 60h]
+ aesimc xmm6, [ecx + 80h]
+ movdqa [edx + 60h], xmm6
+ movdqa [edx + 80h], xmm7
+
+ aesimc xmm7, [ecx + 70h]
+ movdqa [edx + 70h], xmm7
ret
-expand_keys256 endp
+@raw_aes256_invert_key_schedule@8 endp
end