gnir 0.16.5

Automated mirror of ring - Safe, fast, small crypto using Rust.
Documentation
// This file is generated from a similarly-named Perl script in the BoringSSL
// source tree. Do not edit by hand.

#if defined(__has_feature)
#if __has_feature(memory_sanitizer) && !defined(OPENSSL_NO_ASM)
#define OPENSSL_NO_ASM
#endif
#endif

#if !defined(OPENSSL_NO_ASM)
.section	__TEXT,__const


.align	7	// totally strategic alignment
_vpaes_consts:
Lk_mc_forward:	//	mc_forward
.quad	0x0407060500030201, 0x0C0F0E0D080B0A09
.quad	0x080B0A0904070605, 0x000302010C0F0E0D
.quad	0x0C0F0E0D080B0A09, 0x0407060500030201
.quad	0x000302010C0F0E0D, 0x080B0A0904070605
Lk_mc_backward:	//	mc_backward
.quad	0x0605040702010003, 0x0E0D0C0F0A09080B
.quad	0x020100030E0D0C0F, 0x0A09080B06050407
.quad	0x0E0D0C0F0A09080B, 0x0605040702010003
.quad	0x0A09080B06050407, 0x020100030E0D0C0F
Lk_sr:	//	sr
.quad	0x0706050403020100, 0x0F0E0D0C0B0A0908
.quad	0x030E09040F0A0500, 0x0B06010C07020D08
.quad	0x0F060D040B020900, 0x070E050C030A0108
.quad	0x0B0E0104070A0D00, 0x0306090C0F020508

//
// "Hot" constants
//
Lk_inv:	//	inv, inva
.quad	0x0E05060F0D080180, 0x040703090A0B0C02
.quad	0x01040A060F0B0780, 0x030D0E0C02050809
Lk_ipt:	//	input transform (lo, hi)
.quad	0xC2B2E8985A2A7000, 0xCABAE09052227808
.quad	0x4C01307D317C4D00, 0xCD80B1FCB0FDCC81
Lk_sbo:	//	sbou, sbot
.quad	0xD0D26D176FBDC700, 0x15AABF7AC502A878
.quad	0xCFE474A55FBB6A00, 0x8E1E90D1412B35FA
Lk_sb1:	//	sb1u, sb1t
.quad	0x3618D415FAE22300, 0x3BF7CCC10D2ED9EF
.quad	0xB19BE18FCB503E00, 0xA5DF7A6E142AF544
Lk_sb2:	//	sb2u, sb2t
.quad	0x69EB88400AE12900, 0xC2A163C8AB82234A
.quad	0xE27A93C60B712400, 0x5EB7E955BC982FCD

//
//  Key schedule constants
//
Lk_dksd:	//	decryption key schedule: invskew x*D
.quad	0xFEB91A5DA3E44700, 0x0740E3A45A1DBEF9
.quad	0x41C277F4B5368300, 0x5FDC69EAAB289D1E
Lk_dksb:	//	decryption key schedule: invskew x*B
.quad	0x9A4FCA1F8550D500, 0x03D653861CC94C99
.quad	0x115BEDA7B6FC4A00, 0xD993256F7E3482C8
Lk_dkse:	//	decryption key schedule: invskew x*E + 0x63
.quad	0xD5031CCA1FC9D600, 0x53859A4C994F5086
.quad	0xA23196054FDC7BE8, 0xCD5EF96A20B31487
Lk_dks9:	//	decryption key schedule: invskew x*9
.quad	0xB6116FC87ED9A700, 0x4AED933482255BFC
.quad	0x4576516227143300, 0x8BB89FACE9DAFDCE

Lk_rcon:	//	rcon
.quad	0x1F8391B9AF9DEEB6, 0x702A98084D7C7D81

Lk_opt:	//	output transform
.quad	0xFF9F4929D6B66000, 0xF7974121DEBE6808
.quad	0x01EDBD5150BCEC00, 0xE10D5DB1B05C0CE0
Lk_deskew:	//	deskew tables: inverts the sbox's "skew"
.quad	0x07E4A34047A4E300, 0x1DFEB95A5DBEF91A
.quad	0x5F36B5DC83EA6900, 0x2841C2ABF49D1E77

.byte	86,101,99,116,111,114,32,80,101,114,109,117,116,97,116,105,111,110,32,65,69,83,32,102,111,114,32,65,82,77,118,56,44,32,77,105,107,101,32,72,97,109,98,117,114,103,32,40,83,116,97,110,102,111,114,100,32,85,110,105,118,101,114,115,105,116,121,41,0
.align	2

.align	6

.text
##
##  _aes_preheat
##
##  Fills register %r10 -> .aes_consts (so you can -fPIC)
##  and %xmm9-%xmm15 as specified below.
##

.align	4
_vpaes_encrypt_preheat:
	adrp	x10, Lk_inv@PAGE
	add	x10, x10, Lk_inv@PAGEOFF
	movi	v17.16b, #0x0f
	ld1	{v18.2d,v19.2d}, [x10],#32	// Lk_inv
	ld1	{v20.2d,v21.2d,v22.2d,v23.2d}, [x10],#64	// Lk_ipt, Lk_sbo
	ld1	{v24.2d,v25.2d,v26.2d,v27.2d}, [x10]		// Lk_sb1, Lk_sb2
	ret


##
##  _aes_encrypt_core
##
##  AES-encrypt %xmm0.
##
##  Inputs:
##     %xmm0 = input
##     %xmm9-%xmm15 as in _vpaes_preheat
##    (%rdx) = scheduled keys
##
##  Output in %xmm0
##  Clobbers  %xmm1-%xmm5, %r9, %r10, %r11, %rax
##  Preserves %xmm6 - %xmm8 so you get some local vectors
##
##

.align	4
_vpaes_encrypt_core:
	mov	x9, x2
	ldr	w8, [x2,#240]			// pull rounds
	adrp	x11, Lk_mc_forward@PAGE+16
	add	x11, x11, Lk_mc_forward@PAGEOFF+16
						// vmovdqa	.Lk_ipt(%rip),	%xmm2	# iptlo
	ld1	{v16.2d}, [x9], #16		// vmovdqu	(%r9),	%xmm5		# round0 key
	and	v1.16b, v7.16b, v17.16b		// vpand	%xmm9,	%xmm0,	%xmm1
	ushr	v0.16b, v7.16b, #4		// vpsrlb	$4,	%xmm0,	%xmm0
	tbl	v1.16b, {v20.16b}, v1.16b	// vpshufb	%xmm1,	%xmm2,	%xmm1
						// vmovdqa	.Lk_ipt+16(%rip), %xmm3	# ipthi
	tbl	v2.16b, {v21.16b}, v0.16b	// vpshufb	%xmm0,	%xmm3,	%xmm2
	eor	v0.16b, v1.16b, v16.16b		// vpxor	%xmm5,	%xmm1,	%xmm0
	eor	v0.16b, v0.16b, v2.16b		// vpxor	%xmm2,	%xmm0,	%xmm0
	b	Lenc_entry

.align	4
Lenc_loop:
	// middle of middle round
	add	x10, x11, #0x40
	tbl	v4.16b, {v25.16b}, v2.16b		// vpshufb	%xmm2,	%xmm13,	%xmm4	# 4 = sb1u
	ld1	{v1.2d}, [x11], #16		// vmovdqa	-0x40(%r11,%r10), %xmm1	# Lk_mc_forward[]
	tbl	v0.16b, {v24.16b}, v3.16b		// vpshufb	%xmm3,	%xmm12,	%xmm0	# 0 = sb1t
	eor	v4.16b, v4.16b, v16.16b		// vpxor	%xmm5,	%xmm4,	%xmm4	# 4 = sb1u + k
	tbl	v5.16b,	{v27.16b}, v2.16b		// vpshufb	%xmm2,	%xmm15,	%xmm5	# 4 = sb2u
	eor	v0.16b, v0.16b, v4.16b		// vpxor	%xmm4,	%xmm0,	%xmm0	# 0 = A
	tbl	v2.16b, {v26.16b}, v3.16b		// vpshufb	%xmm3,	%xmm14,	%xmm2	# 2 = sb2t
	ld1	{v4.2d}, [x10]			// vmovdqa	(%r11,%r10), %xmm4	# Lk_mc_backward[]
	tbl	v3.16b, {v0.16b}, v1.16b	// vpshufb	%xmm1,	%xmm0,	%xmm3	# 0 = B
	eor	v2.16b, v2.16b, v5.16b		// vpxor	%xmm5,	%xmm2,	%xmm2	# 2 = 2A
	tbl	v0.16b, {v0.16b}, v4.16b	// vpshufb	%xmm4,	%xmm0,	%xmm0	# 3 = D
	eor	v3.16b, v3.16b, v2.16b		// vpxor	%xmm2,	%xmm3,	%xmm3	# 0 = 2A+B
	tbl	v4.16b, {v3.16b}, v1.16b	// vpshufb	%xmm1,	%xmm3,	%xmm4	# 0 = 2B+C
	eor	v0.16b, v0.16b, v3.16b		// vpxor	%xmm3,	%xmm0,	%xmm0	# 3 = 2A+B+D
	and	x11, x11, #~(1<<6)		// and		$0x30,	%r11		# ... mod 4
	eor	v0.16b, v0.16b, v4.16b		// vpxor	%xmm4,	%xmm0, %xmm0	# 0 = 2A+3B+C+D
	sub	w8, w8, #1			// nr--

Lenc_entry:
	// top of round
	and	v1.16b, v0.16b, v17.16b		// vpand	%xmm0,	%xmm9,	%xmm1   # 0 = k
	ushr	v0.16b, v0.16b, #4		// vpsrlb	$4,	%xmm0,	%xmm0	# 1 = i
	tbl	v5.16b, {v19.16b}, v1.16b	// vpshufb	%xmm1,	%xmm11,	%xmm5	# 2 = a/k
	eor	v1.16b, v1.16b, v0.16b		// vpxor	%xmm0,	%xmm1,	%xmm1	# 0 = j
	tbl	v3.16b, {v18.16b}, v0.16b	// vpshufb	%xmm0, 	%xmm10,	%xmm3  	# 3 = 1/i
	tbl	v4.16b, {v18.16b}, v1.16b	// vpshufb	%xmm1, 	%xmm10,	%xmm4  	# 4 = 1/j
	eor	v3.16b, v3.16b, v5.16b		// vpxor	%xmm5,	%xmm3,	%xmm3	# 3 = iak = 1/i + a/k
	eor	v4.16b, v4.16b, v5.16b		// vpxor	%xmm5,	%xmm4,	%xmm4  	# 4 = jak = 1/j + a/k
	tbl	v2.16b, {v18.16b}, v3.16b	// vpshufb	%xmm3,	%xmm10,	%xmm2  	# 2 = 1/iak
	tbl	v3.16b, {v18.16b}, v4.16b	// vpshufb	%xmm4,	%xmm10,	%xmm3	# 3 = 1/jak
	eor	v2.16b, v2.16b, v1.16b		// vpxor	%xmm1,	%xmm2,	%xmm2  	# 2 = io
	eor	v3.16b, v3.16b, v0.16b		// vpxor	%xmm0,	%xmm3,	%xmm3	# 3 = jo
	ld1	{v16.2d}, [x9],#16		// vmovdqu	(%r9),	%xmm5
	cbnz	w8, Lenc_loop

	// middle of last round
	add	x10, x11, #0x80
						// vmovdqa	-0x60(%r10), %xmm4	# 3 : sbou	.Lk_sbo
						// vmovdqa	-0x50(%r10), %xmm0	# 0 : sbot	.Lk_sbo+16
	tbl	v4.16b, {v22.16b}, v2.16b		// vpshufb	%xmm2,	%xmm4,	%xmm4	# 4 = sbou
	ld1	{v1.2d}, [x10]			// vmovdqa	0x40(%r11,%r10), %xmm1	# Lk_sr[]
	tbl	v0.16b, {v23.16b}, v3.16b		// vpshufb	%xmm3,	%xmm0,	%xmm0	# 0 = sb1t
	eor	v4.16b, v4.16b, v16.16b		// vpxor	%xmm5,	%xmm4,	%xmm4	# 4 = sb1u + k
	eor	v0.16b, v0.16b, v4.16b		// vpxor	%xmm4,	%xmm0,	%xmm0	# 0 = A
	tbl	v0.16b, {v0.16b}, v1.16b	// vpshufb	%xmm1,	%xmm0,	%xmm0
	ret


.globl	_GFp_vpaes_encrypt
.private_extern	_GFp_vpaes_encrypt

.align	4
_GFp_vpaes_encrypt:
	stp	x29,x30,[sp,#-16]!
	add	x29,sp,#0

	ld1	{v7.16b}, [x0]
	bl	_vpaes_encrypt_preheat
	bl	_vpaes_encrypt_core
	st1	{v0.16b}, [x1]

	ldp	x29,x30,[sp],#16
	ret



.align	4
_vpaes_encrypt_2x:
	mov	x9, x2
	ldr	w8, [x2,#240]			// pull rounds
	adrp	x11, Lk_mc_forward@PAGE+16
	add	x11, x11, Lk_mc_forward@PAGEOFF+16
						// vmovdqa	.Lk_ipt(%rip),	%xmm2	# iptlo
	ld1	{v16.2d}, [x9], #16		// vmovdqu	(%r9),	%xmm5		# round0 key
	and	v1.16b,  v14.16b,  v17.16b	// vpand	%xmm9,	%xmm0,	%xmm1
	ushr	v0.16b,  v14.16b,  #4		// vpsrlb	$4,	%xmm0,	%xmm0
	and	v9.16b,  v15.16b,  v17.16b
	ushr	v8.16b,  v15.16b,  #4
	tbl	v1.16b,  {v20.16b}, v1.16b	// vpshufb	%xmm1,	%xmm2,	%xmm1
	tbl	v9.16b,  {v20.16b}, v9.16b
						// vmovdqa	.Lk_ipt+16(%rip), %xmm3	# ipthi
	tbl	v2.16b,  {v21.16b}, v0.16b	// vpshufb	%xmm0,	%xmm3,	%xmm2
	tbl	v10.16b, {v21.16b}, v8.16b
	eor	v0.16b,  v1.16b,   v16.16b	// vpxor	%xmm5,	%xmm1,	%xmm0
	eor	v8.16b,  v9.16b,   v16.16b
	eor	v0.16b,  v0.16b,   v2.16b	// vpxor	%xmm2,	%xmm0,	%xmm0
	eor	v8.16b,  v8.16b,   v10.16b
	b	Lenc_2x_entry

.align	4
Lenc_2x_loop:
	// middle of middle round
	add	x10, x11, #0x40
	tbl	v4.16b,  {v25.16b}, v2.16b	// vpshufb	%xmm2,	%xmm13,	%xmm4	# 4 = sb1u
	tbl	v12.16b, {v25.16b}, v10.16b
	ld1	{v1.2d}, [x11], #16		// vmovdqa	-0x40(%r11,%r10), %xmm1	# Lk_mc_forward[]
	tbl	v0.16b,  {v24.16b}, v3.16b	// vpshufb	%xmm3,	%xmm12,	%xmm0	# 0 = sb1t
	tbl	v8.16b,  {v24.16b}, v11.16b
	eor	v4.16b,  v4.16b,  v16.16b	// vpxor	%xmm5,	%xmm4,	%xmm4	# 4 = sb1u + k
	eor	v12.16b, v12.16b, v16.16b
	tbl	v5.16b,	 {v27.16b}, v2.16b	// vpshufb	%xmm2,	%xmm15,	%xmm5	# 4 = sb2u
	tbl	v13.16b, {v27.16b}, v10.16b
	eor	v0.16b,  v0.16b,  v4.16b	// vpxor	%xmm4,	%xmm0,	%xmm0	# 0 = A
	eor	v8.16b,  v8.16b,  v12.16b
	tbl	v2.16b,  {v26.16b}, v3.16b	// vpshufb	%xmm3,	%xmm14,	%xmm2	# 2 = sb2t
	tbl	v10.16b, {v26.16b}, v11.16b
	ld1	{v4.2d}, [x10]			// vmovdqa	(%r11,%r10), %xmm4	# Lk_mc_backward[]
	tbl	v3.16b,  {v0.16b}, v1.16b	// vpshufb	%xmm1,	%xmm0,	%xmm3	# 0 = B
	tbl	v11.16b, {v8.16b}, v1.16b
	eor	v2.16b,  v2.16b,  v5.16b	// vpxor	%xmm5,	%xmm2,	%xmm2	# 2 = 2A
	eor	v10.16b, v10.16b, v13.16b
	tbl	v0.16b,  {v0.16b}, v4.16b	// vpshufb	%xmm4,	%xmm0,	%xmm0	# 3 = D
	tbl	v8.16b,  {v8.16b}, v4.16b
	eor	v3.16b,  v3.16b,  v2.16b	// vpxor	%xmm2,	%xmm3,	%xmm3	# 0 = 2A+B
	eor	v11.16b, v11.16b, v10.16b
	tbl	v4.16b,  {v3.16b}, v1.16b	// vpshufb	%xmm1,	%xmm3,	%xmm4	# 0 = 2B+C
	tbl	v12.16b, {v11.16b},v1.16b
	eor	v0.16b,  v0.16b,  v3.16b	// vpxor	%xmm3,	%xmm0,	%xmm0	# 3 = 2A+B+D
	eor	v8.16b,  v8.16b,  v11.16b
	and	x11, x11, #~(1<<6)		// and		$0x30,	%r11		# ... mod 4
	eor	v0.16b,  v0.16b,  v4.16b	// vpxor	%xmm4,	%xmm0, %xmm0	# 0 = 2A+3B+C+D
	eor	v8.16b,  v8.16b,  v12.16b
	sub	w8, w8, #1			// nr--

Lenc_2x_entry:
	// top of round
	and	v1.16b,  v0.16b, v17.16b	// vpand	%xmm0,	%xmm9,	%xmm1   # 0 = k
	ushr	v0.16b,  v0.16b, #4		// vpsrlb	$4,	%xmm0,	%xmm0	# 1 = i
	and	v9.16b,  v8.16b, v17.16b
	ushr	v8.16b,  v8.16b, #4
	tbl	v5.16b,  {v19.16b},v1.16b	// vpshufb	%xmm1,	%xmm11,	%xmm5	# 2 = a/k
	tbl	v13.16b, {v19.16b},v9.16b
	eor	v1.16b,  v1.16b,  v0.16b	// vpxor	%xmm0,	%xmm1,	%xmm1	# 0 = j
	eor	v9.16b,  v9.16b,  v8.16b
	tbl	v3.16b,  {v18.16b},v0.16b	// vpshufb	%xmm0, 	%xmm10,	%xmm3  	# 3 = 1/i
	tbl	v11.16b, {v18.16b},v8.16b
	tbl	v4.16b,  {v18.16b},v1.16b	// vpshufb	%xmm1, 	%xmm10,	%xmm4  	# 4 = 1/j
	tbl	v12.16b, {v18.16b},v9.16b
	eor	v3.16b,  v3.16b,  v5.16b	// vpxor	%xmm5,	%xmm3,	%xmm3	# 3 = iak = 1/i + a/k
	eor	v11.16b, v11.16b, v13.16b
	eor	v4.16b,  v4.16b,  v5.16b	// vpxor	%xmm5,	%xmm4,	%xmm4  	# 4 = jak = 1/j + a/k
	eor	v12.16b, v12.16b, v13.16b
	tbl	v2.16b,  {v18.16b},v3.16b	// vpshufb	%xmm3,	%xmm10,	%xmm2  	# 2 = 1/iak
	tbl	v10.16b, {v18.16b},v11.16b
	tbl	v3.16b,  {v18.16b},v4.16b	// vpshufb	%xmm4,	%xmm10,	%xmm3	# 3 = 1/jak
	tbl	v11.16b, {v18.16b},v12.16b
	eor	v2.16b,  v2.16b,  v1.16b	// vpxor	%xmm1,	%xmm2,	%xmm2  	# 2 = io
	eor	v10.16b, v10.16b, v9.16b
	eor	v3.16b,  v3.16b,  v0.16b	// vpxor	%xmm0,	%xmm3,	%xmm3	# 3 = jo
	eor	v11.16b, v11.16b, v8.16b
	ld1	{v16.2d}, [x9],#16		// vmovdqu	(%r9),	%xmm5
	cbnz	w8, Lenc_2x_loop

	// middle of last round
	add	x10, x11, #0x80
						// vmovdqa	-0x60(%r10), %xmm4	# 3 : sbou	.Lk_sbo
						// vmovdqa	-0x50(%r10), %xmm0	# 0 : sbot	.Lk_sbo+16
	tbl	v4.16b,  {v22.16b}, v2.16b	// vpshufb	%xmm2,	%xmm4,	%xmm4	# 4 = sbou
	tbl	v12.16b, {v22.16b}, v10.16b
	ld1	{v1.2d}, [x10]			// vmovdqa	0x40(%r11,%r10), %xmm1	# Lk_sr[]
	tbl	v0.16b,  {v23.16b}, v3.16b	// vpshufb	%xmm3,	%xmm0,	%xmm0	# 0 = sb1t
	tbl	v8.16b,  {v23.16b}, v11.16b
	eor	v4.16b,  v4.16b,  v16.16b	// vpxor	%xmm5,	%xmm4,	%xmm4	# 4 = sb1u + k
	eor	v12.16b, v12.16b, v16.16b
	eor	v0.16b,  v0.16b,  v4.16b	// vpxor	%xmm4,	%xmm0,	%xmm0	# 0 = A
	eor	v8.16b,  v8.16b,  v12.16b
	tbl	v0.16b,  {v0.16b},v1.16b	// vpshufb	%xmm1,	%xmm0,	%xmm0
	tbl	v1.16b,  {v8.16b},v1.16b
	ret

########################################################
##                                                    ##
##                  AES key schedule                  ##
##                                                    ##
########################################################

.align	4
_vpaes_key_preheat:
	adrp	x10, Lk_inv@PAGE
	add	x10, x10, Lk_inv@PAGEOFF
	movi	v16.16b, #0x5b			// Lk_s63
	adrp	x11, Lk_sb1@PAGE
	add	x11, x11, Lk_sb1@PAGEOFF
	movi	v17.16b, #0x0f			// Lk_s0F
	ld1	{v18.2d,v19.2d,v20.2d,v21.2d}, [x10]		// Lk_inv, Lk_ipt
	adrp	x10, Lk_dksd@PAGE
	add	x10, x10, Lk_dksd@PAGEOFF
	ld1	{v22.2d,v23.2d}, [x11]		// Lk_sb1
	adrp	x11, Lk_mc_forward@PAGE
	add	x11, x11, Lk_mc_forward@PAGEOFF
	ld1	{v24.2d,v25.2d,v26.2d,v27.2d}, [x10],#64	// Lk_dksd, Lk_dksb
	ld1	{v28.2d,v29.2d,v30.2d,v31.2d}, [x10],#64	// Lk_dkse, Lk_dks9
	ld1	{v8.2d}, [x10]			// Lk_rcon
	ld1	{v9.2d}, [x11]			// Lk_mc_forward[0]
	ret



.align	4
_vpaes_schedule_core:
	stp	x29, x30, [sp,#-16]!
	add	x29,sp,#0

	bl	_vpaes_key_preheat		// load the tables

	ld1	{v0.16b}, [x0],#16		// vmovdqu	(%rdi),	%xmm0		# load key (unaligned)

	// input transform
	mov	v3.16b, v0.16b			// vmovdqa	%xmm0,	%xmm3
	bl	_vpaes_schedule_transform
	mov	v7.16b, v0.16b			// vmovdqa	%xmm0,	%xmm7

	adrp	x10, Lk_sr@PAGE		// lea	Lk_sr(%rip),%r10
	add	x10, x10, Lk_sr@PAGEOFF

	add	x8, x8, x10

	// encrypting, output zeroth round key after transform
	st1	{v0.2d}, [x2]			// vmovdqu	%xmm0,	(%rdx)

	cmp	w1, #192			// cmp	$192,	%esi
	b.hi	Lschedule_256
	b.eq	Lschedule_192
	// 128: fall though

##
##  .schedule_128
##
##  128-bit specific part of key schedule.
##
##  This schedule is really simple, because all its parts
##  are accomplished by the subroutines.
##
Lschedule_128:
	mov	x0, #10			// mov	$10, %esi

Loop_schedule_128:
	sub	x0, x0, #1			// dec	%esi
	bl	_vpaes_schedule_round
	cbz	x0, Lschedule_mangle_last
	bl	_vpaes_schedule_mangle		// write output
	b	Loop_schedule_128

##
##  .aes_schedule_192
##
##  192-bit specific part of key schedule.
##
##  The main body of this schedule is the same as the 128-bit
##  schedule, but with more smearing.  The long, high side is
##  stored in %xmm7 as before, and the short, low side is in
##  the high bits of %xmm6.
##
##  This schedule is somewhat nastier, however, because each
##  round produces 192 bits of key material, or 1.5 round keys.
##  Therefore, on each cycle we do 2 rounds and produce 3 round
##  keys.
##
.align	4
Lschedule_192:
	sub	x0, x0, #8
	ld1	{v0.16b}, [x0]		// vmovdqu	8(%rdi),%xmm0		# load key part 2 (very unaligned)
	bl	_vpaes_schedule_transform	// input transform
	mov	v6.16b, v0.16b			// vmovdqa	%xmm0,	%xmm6		# save short part
	eor	v4.16b, v4.16b, v4.16b		// vpxor	%xmm4,	%xmm4, %xmm4	# clear 4
	ins	v6.d[0], v4.d[0]		// vmovhlps	%xmm4,	%xmm6,	%xmm6		# clobber low side with zeros
	mov	x0, #4			// mov	$4,	%esi

Loop_schedule_192:
	sub	x0, x0, #1			// dec	%esi
	bl	_vpaes_schedule_round
	ext	v0.16b, v6.16b, v0.16b, #8	// vpalignr	$8,%xmm6,%xmm0,%xmm0
	bl	_vpaes_schedule_mangle		// save key n
	bl	_vpaes_schedule_192_smear
	bl	_vpaes_schedule_mangle		// save key n+1
	bl	_vpaes_schedule_round
	cbz	x0, Lschedule_mangle_last
	bl	_vpaes_schedule_mangle		// save key n+2
	bl	_vpaes_schedule_192_smear
	b	Loop_schedule_192

##
##  .aes_schedule_256
##
##  256-bit specific part of key schedule.
##
##  The structure here is very similar to the 128-bit
##  schedule, but with an additional "low side" in
##  %xmm6.  The low side's rounds are the same as the
##  high side's, except no rcon and no rotation.
##
.align	4
Lschedule_256:
	ld1	{v0.16b}, [x0]		// vmovdqu	16(%rdi),%xmm0		# load key part 2 (unaligned)
	bl	_vpaes_schedule_transform	// input transform
	mov	x0, #7			// mov	$7, %esi

Loop_schedule_256:
	sub	x0, x0, #1			// dec	%esi
	bl	_vpaes_schedule_mangle		// output low result
	mov	v6.16b, v0.16b			// vmovdqa	%xmm0,	%xmm6		# save cur_lo in xmm6

	// high round
	bl	_vpaes_schedule_round
	cbz	x0, Lschedule_mangle_last
	bl	_vpaes_schedule_mangle

	// low round. swap xmm7 and xmm6
	dup	v0.4s, v0.s[3]			// vpshufd	$0xFF,	%xmm0,	%xmm0
	movi	v4.16b, #0
	mov	v5.16b, v7.16b			// vmovdqa	%xmm7,	%xmm5
	mov	v7.16b, v6.16b			// vmovdqa	%xmm6,	%xmm7
	bl	_vpaes_schedule_low_round
	mov	v7.16b, v5.16b			// vmovdqa	%xmm5,	%xmm7

	b	Loop_schedule_256

##
##  .aes_schedule_mangle_last
##
##  Mangler for last round of key schedule
##  Mangles %xmm0
##    when encrypting, outputs out(%xmm0) ^ 63
##    when decrypting, outputs unskew(%xmm0)
##
##  Always called right before return... jumps to cleanup and exits
##
.align	4
Lschedule_mangle_last:
	// schedule last round key from xmm0
	adrp	x11, Lk_deskew@PAGE	// lea	Lk_deskew(%rip),%r11	# prepare to deskew
	add	x11, x11, Lk_deskew@PAGEOFF

	cbnz	w3, Lschedule_mangle_last_dec

	// encrypting
	ld1	{v1.2d}, [x8]			// vmovdqa	(%r8,%r10),%xmm1
	adrp	x11, Lk_opt@PAGE		// lea	Lk_opt(%rip),	%r11		# prepare to output transform
	add	x11, x11, Lk_opt@PAGEOFF
	add	x2, x2, #32			// add	$32,	%rdx
	tbl	v0.16b, {v0.16b}, v1.16b	// vpshufb	%xmm1,	%xmm0,	%xmm0		# output permute

Lschedule_mangle_last_dec:
	ld1	{v20.2d,v21.2d}, [x11]		// reload constants
	sub	x2, x2, #16			// add	$-16,	%rdx
	eor	v0.16b, v0.16b, v16.16b		// vpxor	Lk_s63(%rip),	%xmm0,	%xmm0
	bl	_vpaes_schedule_transform	// output transform
	st1	{v0.2d}, [x2]			// vmovdqu	%xmm0,	(%rdx)		# save last key

	// cleanup
	eor	v0.16b, v0.16b, v0.16b		// vpxor	%xmm0,	%xmm0,	%xmm0
	eor	v1.16b, v1.16b, v1.16b		// vpxor	%xmm1,	%xmm1,	%xmm1
	eor	v2.16b, v2.16b, v2.16b		// vpxor	%xmm2,	%xmm2,	%xmm2
	eor	v3.16b, v3.16b, v3.16b		// vpxor	%xmm3,	%xmm3,	%xmm3
	eor	v4.16b, v4.16b, v4.16b		// vpxor	%xmm4,	%xmm4,	%xmm4
	eor	v5.16b, v5.16b, v5.16b		// vpxor	%xmm5,	%xmm5,	%xmm5
	eor	v6.16b, v6.16b, v6.16b		// vpxor	%xmm6,	%xmm6,	%xmm6
	eor	v7.16b, v7.16b, v7.16b		// vpxor	%xmm7,	%xmm7,	%xmm7
	ldp	x29, x30, [sp],#16
	ret


##
##  .aes_schedule_192_smear
##
##  Smear the short, low side in the 192-bit key schedule.
##
##  Inputs:
##    %xmm7: high side, b  a  x  y
##    %xmm6:  low side, d  c  0  0
##    %xmm13: 0
##
##  Outputs:
##    %xmm6: b+c+d  b+c  0  0
##    %xmm0: b+c+d  b+c  b  a
##

.align	4
_vpaes_schedule_192_smear:
	movi	v1.16b, #0
	dup	v0.4s, v7.s[3]
	ins	v1.s[3], v6.s[2]	// vpshufd	$0x80,	%xmm6,	%xmm1	# d c 0 0 -> c 0 0 0
	ins	v0.s[0], v7.s[2]	// vpshufd	$0xFE,	%xmm7,	%xmm0	# b a _ _ -> b b b a
	eor	v6.16b, v6.16b, v1.16b	// vpxor	%xmm1,	%xmm6,	%xmm6	# -> c+d c 0 0
	eor	v1.16b, v1.16b, v1.16b	// vpxor	%xmm1,	%xmm1,	%xmm1
	eor	v6.16b, v6.16b, v0.16b	// vpxor	%xmm0,	%xmm6,	%xmm6	# -> b+c+d b+c b a
	mov	v0.16b, v6.16b		// vmovdqa	%xmm6,	%xmm0
	ins	v6.d[0], v1.d[0]	// vmovhlps	%xmm1,	%xmm6,	%xmm6	# clobber low side with zeros
	ret


##
##  .aes_schedule_round
##
##  Runs one main round of the key schedule on %xmm0, %xmm7
##
##  Specifically, runs subbytes on the high dword of %xmm0
##  then rotates it by one byte and xors into the low dword of
##  %xmm7.
##
##  Adds rcon from low byte of %xmm8, then rotates %xmm8 for
##  next rcon.
##
##  Smears the dwords of %xmm7 by xoring the low into the
##  second low, result into third, result into highest.
##
##  Returns results in %xmm7 = %xmm0.
##  Clobbers %xmm1-%xmm4, %r11.
##

.align	4
_vpaes_schedule_round:
	// extract rcon from xmm8
	movi	v4.16b, #0			// vpxor	%xmm4,	%xmm4,	%xmm4
	ext	v1.16b, v8.16b, v4.16b, #15	// vpalignr	$15,	%xmm8,	%xmm4,	%xmm1
	ext	v8.16b, v8.16b, v8.16b, #15	// vpalignr	$15,	%xmm8,	%xmm8,	%xmm8
	eor	v7.16b, v7.16b, v1.16b		// vpxor	%xmm1,	%xmm7,	%xmm7

	// rotate
	dup	v0.4s, v0.s[3]			// vpshufd	$0xFF,	%xmm0,	%xmm0
	ext	v0.16b, v0.16b, v0.16b, #1	// vpalignr	$1,	%xmm0,	%xmm0,	%xmm0

	// fall through...

	// low round: same as high round, but no rotation and no rcon.
_vpaes_schedule_low_round:
	// smear xmm7
	ext	v1.16b, v4.16b, v7.16b, #12	// vpslldq	$4,	%xmm7,	%xmm1
	eor	v7.16b, v7.16b, v1.16b		// vpxor	%xmm1,	%xmm7,	%xmm7
	ext	v4.16b, v4.16b, v7.16b, #8	// vpslldq	$8,	%xmm7,	%xmm4

	// subbytes
	and	v1.16b, v0.16b, v17.16b		// vpand	%xmm9,	%xmm0,	%xmm1		# 0 = k
	ushr	v0.16b, v0.16b, #4		// vpsrlb	$4,	%xmm0,	%xmm0		# 1 = i
	eor	v7.16b, v7.16b, v4.16b		// vpxor	%xmm4,	%xmm7,	%xmm7
	tbl	v2.16b, {v19.16b}, v1.16b	// vpshufb	%xmm1,	%xmm11,	%xmm2		# 2 = a/k
	eor	v1.16b, v1.16b, v0.16b		// vpxor	%xmm0,	%xmm1,	%xmm1		# 0 = j
	tbl	v3.16b, {v18.16b}, v0.16b	// vpshufb	%xmm0, 	%xmm10,	%xmm3		# 3 = 1/i
	eor	v3.16b, v3.16b, v2.16b		// vpxor	%xmm2,	%xmm3,	%xmm3		# 3 = iak = 1/i + a/k
	tbl	v4.16b, {v18.16b}, v1.16b	// vpshufb	%xmm1,	%xmm10,	%xmm4		# 4 = 1/j
	eor	v7.16b, v7.16b, v16.16b		// vpxor	Lk_s63(%rip),	%xmm7,	%xmm7
	tbl	v3.16b, {v18.16b}, v3.16b	// vpshufb	%xmm3,	%xmm10,	%xmm3		# 2 = 1/iak
	eor	v4.16b, v4.16b, v2.16b		// vpxor	%xmm2,	%xmm4,	%xmm4		# 4 = jak = 1/j + a/k
	tbl	v2.16b, {v18.16b}, v4.16b	// vpshufb	%xmm4,	%xmm10,	%xmm2		# 3 = 1/jak
	eor	v3.16b, v3.16b, v1.16b		// vpxor	%xmm1,	%xmm3,	%xmm3		# 2 = io
	eor	v2.16b, v2.16b, v0.16b		// vpxor	%xmm0,	%xmm2,	%xmm2		# 3 = jo
	tbl	v4.16b, {v23.16b}, v3.16b	// vpshufb	%xmm3,	%xmm13,	%xmm4		# 4 = sbou
	tbl	v1.16b, {v22.16b}, v2.16b	// vpshufb	%xmm2,	%xmm12,	%xmm1		# 0 = sb1t
	eor	v1.16b, v1.16b, v4.16b		// vpxor	%xmm4,	%xmm1,	%xmm1		# 0 = sbox output

	// add in smeared stuff
	eor	v0.16b, v1.16b, v7.16b		// vpxor	%xmm7,	%xmm1,	%xmm0
	eor	v7.16b, v1.16b, v7.16b		// vmovdqa	%xmm0,	%xmm7
	ret


##
##  .aes_schedule_transform
##
##  Linear-transform %xmm0 according to tables at (%r11)
##
##  Requires that %xmm9 = 0x0F0F... as in preheat
##  Output in %xmm0
##  Clobbers %xmm1, %xmm2
##

.align	4
_vpaes_schedule_transform:
	and	v1.16b, v0.16b, v17.16b		// vpand	%xmm9,	%xmm0,	%xmm1
	ushr	v0.16b, v0.16b, #4		// vpsrlb	$4,	%xmm0,	%xmm0
						// vmovdqa	(%r11),	%xmm2 	# lo
	tbl	v2.16b, {v20.16b}, v1.16b	// vpshufb	%xmm1,	%xmm2,	%xmm2
						// vmovdqa	16(%r11),	%xmm1 # hi
	tbl	v0.16b, {v21.16b}, v0.16b	// vpshufb	%xmm0,	%xmm1,	%xmm0
	eor	v0.16b, v0.16b, v2.16b		// vpxor	%xmm2,	%xmm0,	%xmm0
	ret


##
##  .aes_schedule_mangle
##
##  Mangle xmm0 from (basis-transformed) standard version
##  to our version.
##
##  On encrypt,
##    xor with 0x63
##    multiply by circulant 0,1,1,1
##    apply shiftrows transform
##
##  On decrypt,
##    xor with 0x63
##    multiply by "inverse mixcolumns" circulant E,B,D,9
##    deskew
##    apply shiftrows transform
##
##
##  Writes out to (%rdx), and increments or decrements it
##  Keeps track of round number mod 4 in %r8
##  Preserves xmm0
##  Clobbers xmm1-xmm5
##

.align	4
_vpaes_schedule_mangle:
	mov	v4.16b, v0.16b			// vmovdqa	%xmm0,	%xmm4	# save xmm0 for later
						// vmovdqa	.Lk_mc_forward(%rip),%xmm5

	// encrypting
	eor	v4.16b, v0.16b, v16.16b		// vpxor	Lk_s63(%rip),	%xmm0,	%xmm4
	add	x2, x2, #16			// add	$16,	%rdx
	tbl	v4.16b, {v4.16b}, v9.16b	// vpshufb	%xmm5,	%xmm4,	%xmm4
	tbl	v1.16b, {v4.16b}, v9.16b	// vpshufb	%xmm5,	%xmm4,	%xmm1
	tbl	v3.16b, {v1.16b}, v9.16b	// vpshufb	%xmm5,	%xmm1,	%xmm3
	eor	v4.16b, v4.16b, v1.16b		// vpxor	%xmm1,	%xmm4,	%xmm4
	ld1	{v1.2d}, [x8]			// vmovdqa	(%r8,%r10),	%xmm1
	eor	v3.16b, v3.16b, v4.16b		// vpxor	%xmm4,	%xmm3,	%xmm3

Lschedule_mangle_both:
	tbl	v3.16b, {v3.16b}, v1.16b	// vpshufb	%xmm1,	%xmm3,	%xmm3
	add	x8, x8, #64-16			// add	$-16,	%r8
	and	x8, x8, #~(1<<6)		// and	$0x30,	%r8
	st1	{v3.2d}, [x2]			// vmovdqu	%xmm3,	(%rdx)
	ret


.globl	_GFp_vpaes_set_encrypt_key
.private_extern	_GFp_vpaes_set_encrypt_key

.align	4
_GFp_vpaes_set_encrypt_key:
	stp	x29,x30,[sp,#-16]!
	add	x29,sp,#0
	stp	d8,d9,[sp,#-16]!	// ABI spec says so

	lsr	w9, w1, #5		// shr	$5,%eax
	add	w9, w9, #5		// $5,%eax
	str	w9, [x2,#240]		// mov	%eax,240(%rdx)	# AES_KEY->rounds = nbits/32+5;

	mov	w3, #0		// mov	$0,%ecx
	mov	x8, #0x30		// mov	$0x30,%r8d
	bl	_vpaes_schedule_core
	eor	x0, x0, x0

	ldp	d8,d9,[sp],#16
	ldp	x29,x30,[sp],#16
	ret

.globl	_GFp_vpaes_ctr32_encrypt_blocks
.private_extern	_GFp_vpaes_ctr32_encrypt_blocks

.align	4
_GFp_vpaes_ctr32_encrypt_blocks:
	stp	x29,x30,[sp,#-16]!
	add	x29,sp,#0
	stp	d8,d9,[sp,#-16]!	// ABI spec says so
	stp	d10,d11,[sp,#-16]!
	stp	d12,d13,[sp,#-16]!
	stp	d14,d15,[sp,#-16]!

	cbz	x2, Lctr32_done

	// Note, unlike the other functions, x2 here is measured in blocks,
	// not bytes.
	mov	x17, x2
	mov	x2,  x3

	// Load the IV and counter portion.
	ldr	w6, [x4, #12]
	ld1	{v7.16b}, [x4]

	bl	_vpaes_encrypt_preheat
	tst	x17, #1
	rev	w6, w6		// The counter is big-endian.
	b.eq	Lctr32_prep_loop

	// Handle one block so the remaining block count is even for
	// _vpaes_encrypt_2x.
	ld1	{v6.16b}, [x0], #16	// Load input ahead of time
	bl	_vpaes_encrypt_core
	eor	v0.16b, v0.16b, v6.16b	// XOR input and result
	st1	{v0.16b}, [x1], #16
	subs	x17, x17, #1
	// Update the counter.
	add	w6, w6, #1
	rev	w7, w6
	mov	v7.s[3], w7
	b.ls	Lctr32_done

Lctr32_prep_loop:
	// _vpaes_encrypt_core takes its input from v7, while _vpaes_encrypt_2x
	// uses v14 and v15.
	mov	v15.16b, v7.16b
	mov	v14.16b, v7.16b
	add	w6, w6, #1
	rev	w7, w6
	mov	v15.s[3], w7

Lctr32_loop:
	ld1	{v6.16b,v7.16b}, [x0], #32	// Load input ahead of time
	bl	_vpaes_encrypt_2x
	eor	v0.16b, v0.16b, v6.16b		// XOR input and result
	eor	v1.16b, v1.16b, v7.16b		// XOR input and result (#2)
	st1	{v0.16b,v1.16b}, [x1], #32
	subs	x17, x17, #2
	// Update the counter.
	add	w7, w6, #1
	add	w6, w6, #2
	rev	w7, w7
	mov	v14.s[3], w7
	rev	w7, w6
	mov	v15.s[3], w7
	b.hi	Lctr32_loop

Lctr32_done:
	ldp	d14,d15,[sp],#16
	ldp	d12,d13,[sp],#16
	ldp	d10,d11,[sp],#16
	ldp	d8,d9,[sp],#16
	ldp	x29,x30,[sp],#16
	ret

#endif  // !OPENSSL_NO_ASM