/ This Source Code Form is subject to the terms of the Mozilla Public
/ License, v. 2.0. If a copy of the MPL was not distributed with this
/ file, You can obtain one at http://mozilla.org/MPL/2.0/.

/ ** ARCFOUR implementation optimized for AMD64.
/ **
/ ** The throughput achieved by this code is about 320 MBytes/sec, on
/ ** a 1.8 GHz AMD Opteron (rev C0) processor.

.text
.align 16
.globl ARCFOUR
.type ARCFOUR,@function
ARCFOUR:
	pushq	%rbp
	pushq	%rbx
	movq	%rdi,		%rbp	/ key = ARG(key)
	movq	%rsi,		%rbx	/ rbx = ARG(len)
	movq	%rdx,		%rsi	/ in = ARG(in)
	movq	%rcx,		%rdi	/ out = ARG(out)
	movq	(%rbp),		%rcx	/ x = key->x
	movq	8(%rbp),	%rdx	/ y = key->y
	addq	$16,		%rbp	/ d = key->data
	incq	%rcx			/ x++
	andq	$255,		%rcx	/ x &= 0xff
	leaq	-8(%rbx,%rsi),	%rbx	/ rbx = in+len-8
	movq	%rbx,		%r9	/ tmp = in+len-8
	movq	0(%rbp,%rcx,8),	%rax	/ tx = d[x]
	cmpq	%rsi,		%rbx	/ cmp in with in+len-8
	jl	.Lend			/ jump if (in+len-8 < in)

.Lstart:
	addq	$8,		%rsi		/ increment in
	addq	$8,		%rdi		/ increment out

	/ generate the next 8 bytes of the rc4 stream into %r8
	movq	$8,		%r11		/ byte counter
1:	addb	%al,		%dl		/ y += tx
	movl	0(%rbp,%rdx,8),	%ebx		/ ty = d[y]
	movl	%ebx,		0(%rbp,%rcx,8)	/ d[x] = ty
	addb	%al,		%bl		/ val = ty + tx
	movl	%eax,		0(%rbp,%rdx,8)	/ d[y] = tx
	incb	%cl				/ x++		(NEXT ROUND)
	movl	0(%rbp,%rcx,8),	%eax		/ tx = d[x]	(NEXT ROUND)
	movb	0(%rbp,%rbx,8),	%r8b		/ val = d[val]
	decb	%r11b
	rorq	$8,		%r8		/ (ror does not change ZF)
	jnz 	1b

	/ xor 8 bytes
	xorq	-8(%rsi),	%r8
	cmpq	%r9,		%rsi		/ cmp in+len-8 with in
	movq	%r8,		-8(%rdi)
	jle	.Lstart				/ jump if (in <= in+len-8)

.Lend:
	addq	$8,		%r9		/ tmp = in+len

	/ handle the last bytes, one by one
1:	cmpq	%rsi,		%r9		/ cmp in with in+len
	jle	.Lfinished			/ jump if (in+len <= in)
	addb	%al,		%dl		/ y += tx
	movl	0(%rbp,%rdx,8),	%ebx		/ ty = d[y]
	movl	%ebx,		0(%rbp,%rcx,8)	/ d[x] = ty
	addb	%al,		%bl		/ val = ty + tx
	movl	%eax,		0(%rbp,%rdx,8)	/ d[y] = tx
	incb	%cl				/ x++		(NEXT ROUND)
	movl	0(%rbp,%rcx,8),	%eax		/ tx = d[x]	(NEXT ROUND)
	movb	0(%rbp,%rbx,8),	%r8b		/ val = d[val]
	xorb	(%rsi),		%r8b		/ xor 1 byte
	movb	%r8b,		(%rdi)
	incq	%rsi				/ in++
	incq	%rdi				/ out++
	jmp 1b

.Lfinished:
	decq	%rcx				/ x--
	movb	%dl,		-8(%rbp)	/ key->y = y
	movb	%cl,		-16(%rbp)	/ key->x = x
	popq	%rbx
	popq	%rbp
	ret
.L_ARCFOUR_end:
.size ARCFOUR,.L_ARCFOUR_end-ARCFOUR