ring/crypto/fipsmodule/ec/asm/p256-x86_64-asm.pl [3221:3272]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
	lea	$U2(%rsp), $r_ptr		# U1*H^2
	call	__ecp_nistz256_mul_mont$x	# p256_mul_mont(U2, U1, Hsqr);
___
{
#######################################################################
# operate in 4-5-0-1 "name space" that matches multiplication output
#
my ($acc0,$acc1,$acc2,$acc3,$t3,$t4)=($acc4,$acc5,$acc0,$acc1,$acc2,$acc3);
my ($poly1, $poly3)=($acc6,$acc7);

$code.=<<___;
	#lea	$U2(%rsp), $a_ptr
	#lea	$Hsqr(%rsp), $r_ptr	# 2*U1*H^2
	#call	__ecp_nistz256_mul_by_2	# ecp_nistz256_mul_by_2(Hsqr, U2);

	xor	$t4, $t4
	add	$acc0, $acc0		# a0:a3+a0:a3
	lea	$Rsqr(%rsp), $a_ptr
	adc	$acc1, $acc1
	 mov	$acc0, $t0
	adc	$acc2, $acc2
	adc	$acc3, $acc3
	 mov	$acc1, $t1
	adc	\$0, $t4

	sub	\$-1, $acc0
	 mov	$acc2, $t2
	sbb	$poly1, $acc1
	sbb	\$0, $acc2
	 mov	$acc3, $t3
	sbb	$poly3, $acc3
	sbb	\$0, $t4

	cmovc	$t0, $acc0
	mov	8*0($a_ptr), $t0
	cmovc	$t1, $acc1
	mov	8*1($a_ptr), $t1
	cmovc	$t2, $acc2
	mov	8*2($a_ptr), $t2
	cmovc	$t3, $acc3
	mov	8*3($a_ptr), $t3

	call	__ecp_nistz256_sub$x		# p256_sub(res_x, Rsqr, Hsqr);

	lea	$Hcub(%rsp), $b_ptr
	lea	$res_x(%rsp), $r_ptr
	call	__ecp_nistz256_sub_from$x	# p256_sub(res_x, res_x, Hcub);

	mov	$U2+8*0(%rsp), $t0
	mov	$U2+8*1(%rsp), $t1
	mov	$U2+8*2(%rsp), $t2
	mov	$U2+8*3(%rsp), $t3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



ring/crypto/fipsmodule/ec/asm/p256-x86_64-asm.pl [3547:3598]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
	lea	$U2(%rsp), $r_ptr		# U1*H^2
	call	__ecp_nistz256_mul_mont$x	# p256_mul_mont(U2, in1_x, Hsqr);
___
{
#######################################################################
# operate in 4-5-0-1 "name space" that matches multiplication output
#
my ($acc0,$acc1,$acc2,$acc3,$t3,$t4)=($acc4,$acc5,$acc0,$acc1,$acc2,$acc3);
my ($poly1, $poly3)=($acc6,$acc7);

$code.=<<___;
	#lea	$U2(%rsp), $a_ptr
	#lea	$Hsqr(%rsp), $r_ptr	# 2*U1*H^2
	#call	__ecp_nistz256_mul_by_2	# ecp_nistz256_mul_by_2(Hsqr, U2);

	xor	$t4, $t4
	add	$acc0, $acc0		# a0:a3+a0:a3
	lea	$Rsqr(%rsp), $a_ptr
	adc	$acc1, $acc1
	 mov	$acc0, $t0
	adc	$acc2, $acc2
	adc	$acc3, $acc3
	 mov	$acc1, $t1
	adc	\$0, $t4

	sub	\$-1, $acc0
	 mov	$acc2, $t2
	sbb	$poly1, $acc1
	sbb	\$0, $acc2
	 mov	$acc3, $t3
	sbb	$poly3, $acc3
	sbb	\$0, $t4

	cmovc	$t0, $acc0
	mov	8*0($a_ptr), $t0
	cmovc	$t1, $acc1
	mov	8*1($a_ptr), $t1
	cmovc	$t2, $acc2
	mov	8*2($a_ptr), $t2
	cmovc	$t3, $acc3
	mov	8*3($a_ptr), $t3

	call	__ecp_nistz256_sub$x		# p256_sub(res_x, Rsqr, Hsqr);

	lea	$Hcub(%rsp), $b_ptr
	lea	$res_x(%rsp), $r_ptr
	call	__ecp_nistz256_sub_from$x	# p256_sub(res_x, res_x, Hcub);

	mov	$U2+8*0(%rsp), $t0
	mov	$U2+8*1(%rsp), $t1
	mov	$U2+8*2(%rsp), $t2
	mov	$U2+8*3(%rsp), $t3
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -