Initial import

1 files changed, 678 insertions, 0 deletions

diff --git a/openssl/crypto/bn/asm/s390x.S b/openssl/crypto/bn/asm/s390x.S
new file mode 100755
index 00000000..43fcb79b
--- /dev/null
+++ b/openssl/crypto/bn/asm/s390x.S
@@ -0,0 +1,678 @@
+.ident "s390x.S, version 1.1"
+// ====================================================================
+// Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
+// project.
+//
+// Rights for redistribution and usage in source and binary forms are
+// granted according to the OpenSSL license. Warranty of any kind is
+// disclaimed.
+// ====================================================================
+
+.text
+
+#define zero	%r0
+
+// BN_ULONG bn_mul_add_words(BN_ULONG *r2,BN_ULONG *r3,int r4,BN_ULONG r5);
+.globl	bn_mul_add_words
+.type	bn_mul_add_words,@function
+.align	4
+bn_mul_add_words:
+	lghi	zero,0		// zero = 0
+	la	%r1,0(%r2)	// put rp aside
+	lghi	%r2,0		// i=0;
+	ltgfr	%r4,%r4
+	bler	%r14		// if (len<=0) return 0;
+
+	stmg	%r6,%r10,48(%r15)
+	lghi	%r10,3
+	lghi	%r8,0		// carry = 0
+	nr	%r10,%r4	// len%4
+	sra	%r4,2		// cnt=len/4
+	jz	.Loop1_madd	// carry is incidentally cleared if branch taken
+	algr	zero,zero	// clear carry
+
+.Loop4_madd:
+	lg	%r7,0(%r2,%r3)	// ap[i]
+	mlgr	%r6,%r5		// *=w
+	alcgr	%r7,%r8		// +=carry
+	alcgr	%r6,zero
+	alg	%r7,0(%r2,%r1)	// +=rp[i]
+	stg	%r7,0(%r2,%r1)	// rp[i]=
+
+	lg	%r9,8(%r2,%r3)
+	mlgr	%r8,%r5
+	alcgr	%r9,%r6
+	alcgr	%r8,zero
+	alg	%r9,8(%r2,%r1)
+	stg	%r9,8(%r2,%r1)
+
+	lg	%r7,16(%r2,%r3)
+	mlgr	%r6,%r5
+	alcgr	%r7,%r8
+	alcgr	%r6,zero
+	alg	%r7,16(%r2,%r1)
+	stg	%r7,16(%r2,%r1)
+
+	lg	%r9,24(%r2,%r3)
+	mlgr	%r8,%r5
+	alcgr	%r9,%r6
+	alcgr	%r8,zero
+	alg	%r9,24(%r2,%r1)
+	stg	%r9,24(%r2,%r1)
+
+	la	%r2,32(%r2)	// i+=4
+	brct	%r4,.Loop4_madd
+
+	la	%r10,1(%r10)		// see if len%4 is zero ...
+	brct	%r10,.Loop1_madd	// without touching condition code:-)
+
+.Lend_madd:
+	alcgr	%r8,zero	// collect carry bit
+	lgr	%r2,%r8
+	lmg	%r6,%r10,48(%r15)
+	br	%r14
+
+.Loop1_madd:
+	lg	%r7,0(%r2,%r3)	// ap[i]
+	mlgr	%r6,%r5		// *=w
+	alcgr	%r7,%r8		// +=carry
+	alcgr	%r6,zero
+	alg	%r7,0(%r2,%r1)	// +=rp[i]
+	stg	%r7,0(%r2,%r1)	// rp[i]=
+
+	lgr	%r8,%r6
+	la	%r2,8(%r2)	// i++
+	brct	%r10,.Loop1_madd
+
+	j	.Lend_madd
+.size	bn_mul_add_words,.-bn_mul_add_words
+
+// BN_ULONG bn_mul_words(BN_ULONG *r2,BN_ULONG *r3,int r4,BN_ULONG r5);
+.globl	bn_mul_words
+.type	bn_mul_words,@function
+.align	4
+bn_mul_words:
+	lghi	zero,0		// zero = 0
+	la	%r1,0(%r2)	// put rp aside
+	lghi	%r2,0		// i=0;
+	ltgfr	%r4,%r4
+	bler	%r14		// if (len<=0) return 0;
+
+	stmg	%r6,%r10,48(%r15)
+	lghi	%r10,3
+	lghi	%r8,0		// carry = 0
+	nr	%r10,%r4	// len%4
+	sra	%r4,2		// cnt=len/4
+	jz	.Loop1_mul	// carry is incidentally cleared if branch taken
+	algr	zero,zero	// clear carry
+
+.Loop4_mul:
+	lg	%r7,0(%r2,%r3)	// ap[i]
+	mlgr	%r6,%r5		// *=w
+	alcgr	%r7,%r8		// +=carry
+	stg	%r7,0(%r2,%r1)	// rp[i]=
+
+	lg	%r9,8(%r2,%r3)
+	mlgr	%r8,%r5
+	alcgr	%r9,%r6
+	stg	%r9,8(%r2,%r1)
+
+	lg	%r7,16(%r2,%r3)
+	mlgr	%r6,%r5
+	alcgr	%r7,%r8
+	stg	%r7,16(%r2,%r1)
+
+	lg	%r9,24(%r2,%r3)
+	mlgr	%r8,%r5
+	alcgr	%r9,%r6
+	stg	%r9,24(%r2,%r1)
+
+	la	%r2,32(%r2)	// i+=4
+	brct	%r4,.Loop4_mul
+
+	la	%r10,1(%r10)		// see if len%4 is zero ...
+	brct	%r10,.Loop1_mul		// without touching condition code:-)
+
+.Lend_mul:
+	alcgr	%r8,zero	// collect carry bit
+	lgr	%r2,%r8
+	lmg	%r6,%r10,48(%r15)
+	br	%r14
+
+.Loop1_mul:
+	lg	%r7,0(%r2,%r3)	// ap[i]
+	mlgr	%r6,%r5		// *=w
+	alcgr	%r7,%r8		// +=carry
+	stg	%r7,0(%r2,%r1)	// rp[i]=
+
+	lgr	%r8,%r6
+	la	%r2,8(%r2)	// i++
+	brct	%r10,.Loop1_mul
+
+	j	.Lend_mul
+.size	bn_mul_words,.-bn_mul_words
+
+// void bn_sqr_words(BN_ULONG *r2,BN_ULONG *r2,int r4)
+.globl	bn_sqr_words
+.type	bn_sqr_words,@function
+.align	4
+bn_sqr_words:
+	ltgfr	%r4,%r4
+	bler	%r14
+
+	stmg	%r6,%r7,48(%r15)
+	srag	%r1,%r4,2	// cnt=len/4
+	jz	.Loop1_sqr
+
+.Loop4_sqr:
+	lg	%r7,0(%r3)
+	mlgr	%r6,%r7
+	stg	%r7,0(%r2)
+	stg	%r6,8(%r2)
+
+	lg	%r7,8(%r3)
+	mlgr	%r6,%r7
+	stg	%r7,16(%r2)
+	stg	%r6,24(%r2)
+
+	lg	%r7,16(%r3)
+	mlgr	%r6,%r7
+	stg	%r7,32(%r2)
+	stg	%r6,40(%r2)
+
+	lg	%r7,24(%r3)
+	mlgr	%r6,%r7
+	stg	%r7,48(%r2)
+	stg	%r6,56(%r2)
+
+	la	%r3,32(%r3)
+	la	%r2,64(%r2)
+	brct	%r1,.Loop4_sqr
+
+	lghi	%r1,3
+	nr	%r4,%r1		// cnt=len%4
+	jz	.Lend_sqr
+
+.Loop1_sqr:
+	lg	%r7,0(%r3)
+	mlgr	%r6,%r7
+	stg	%r7,0(%r2)
+	stg	%r6,8(%r2)
+
+	la	%r3,8(%r3)
+	la	%r2,16(%r2)
+	brct	%r4,.Loop1_sqr
+
+.Lend_sqr:
+	lmg	%r6,%r7,48(%r15)
+	br	%r14
+.size	bn_sqr_words,.-bn_sqr_words
+
+// BN_ULONG bn_div_words(BN_ULONG h,BN_ULONG l,BN_ULONG d);
+.globl	bn_div_words
+.type	bn_div_words,@function
+.align	4
+bn_div_words:
+	dlgr	%r2,%r4
+	lgr	%r2,%r3
+	br	%r14
+.size	bn_div_words,.-bn_div_words
+
+// BN_ULONG bn_add_words(BN_ULONG *r2,BN_ULONG *r3,BN_ULONG *r4,int r5);
+.globl	bn_add_words
+.type	bn_add_words,@function
+.align	4
+bn_add_words:
+	la	%r1,0(%r2)	// put rp aside
+	lghi	%r2,0		// i=0
+	ltgfr	%r5,%r5
+	bler	%r14		// if (len<=0) return 0;
+
+	stg	%r6,48(%r15)
+	lghi	%r6,3
+	nr	%r6,%r5		// len%4
+	sra	%r5,2		// len/4, use sra because it sets condition code
+	jz	.Loop1_add	// carry is incidentally cleared if branch taken
+	algr	%r2,%r2		// clear carry
+
+.Loop4_add:
+	lg	%r0,0(%r2,%r3)
+	alcg	%r0,0(%r2,%r4)
+	stg	%r0,0(%r2,%r1)
+	lg	%r0,8(%r2,%r3)
+	alcg	%r0,8(%r2,%r4)
+	stg	%r0,8(%r2,%r1)
+	lg	%r0,16(%r2,%r3)
+	alcg	%r0,16(%r2,%r4)
+	stg	%r0,16(%r2,%r1)
+	lg	%r0,24(%r2,%r3)
+	alcg	%r0,24(%r2,%r4)
+	stg	%r0,24(%r2,%r1)
+
+	la	%r2,32(%r2)	// i+=4
+	brct	%r5,.Loop4_add
+
+	la	%r6,1(%r6)	// see if len%4 is zero ...
+	brct	%r6,.Loop1_add	// without touching condition code:-)
+
+.Lexit_add:
+	lghi	%r2,0
+	alcgr	%r2,%r2
+	lg	%r6,48(%r15)
+	br	%r14
+
+.Loop1_add:
+	lg	%r0,0(%r2,%r3)
+	alcg	%r0,0(%r2,%r4)
+	stg	%r0,0(%r2,%r1)
+
+	la	%r2,8(%r2)	// i++
+	brct	%r6,.Loop1_add
+
+	j	.Lexit_add
+.size	bn_add_words,.-bn_add_words
+
+// BN_ULONG bn_sub_words(BN_ULONG *r2,BN_ULONG *r3,BN_ULONG *r4,int r5);
+.globl	bn_sub_words
+.type	bn_sub_words,@function
+.align	4
+bn_sub_words:
+	la	%r1,0(%r2)	// put rp aside
+	lghi	%r2,0		// i=0
+	ltgfr	%r5,%r5
+	bler	%r14		// if (len<=0) return 0;
+
+	stg	%r6,48(%r15)
+	lghi	%r6,3
+	nr	%r6,%r5		// len%4
+	sra	%r5,2		// len/4, use sra because it sets condition code
+	jnz	.Loop4_sub	// borrow is incidentally cleared if branch taken
+	slgr	%r2,%r2		// clear borrow
+
+.Loop1_sub:
+	lg	%r0,0(%r2,%r3)
+	slbg	%r0,0(%r2,%r4)
+	stg	%r0,0(%r2,%r1)
+
+	la	%r2,8(%r2)	// i++
+	brct	%r6,.Loop1_sub
+	j	.Lexit_sub
+
+.Loop4_sub:
+	lg	%r0,0(%r2,%r3)
+	slbg	%r0,0(%r2,%r4)
+	stg	%r0,0(%r2,%r1)
+	lg	%r0,8(%r2,%r3)
+	slbg	%r0,8(%r2,%r4)
+	stg	%r0,8(%r2,%r1)
+	lg	%r0,16(%r2,%r3)
+	slbg	%r0,16(%r2,%r4)
+	stg	%r0,16(%r2,%r1)
+	lg	%r0,24(%r2,%r3)
+	slbg	%r0,24(%r2,%r4)
+	stg	%r0,24(%r2,%r1)
+
+	la	%r2,32(%r2)	// i+=4
+	brct	%r5,.Loop4_sub
+
+	la	%r6,1(%r6)	// see if len%4 is zero ...
+	brct	%r6,.Loop1_sub	// without touching condition code:-)
+
+.Lexit_sub:
+	lghi	%r2,0
+	slbgr	%r2,%r2
+	lcgr	%r2,%r2
+	lg	%r6,48(%r15)
+	br	%r14
+.size	bn_sub_words,.-bn_sub_words
+
+#define c1	%r1
+#define c2	%r5
+#define c3	%r8
+
+#define mul_add_c(ai,bi,c1,c2,c3)	\
+	lg	%r7,ai*8(%r3);		\
+	mlg	%r6,bi*8(%r4);		\
+	algr	c1,%r7;			\
+	alcgr	c2,%r6;			\
+	alcgr	c3,zero
+
+// void bn_mul_comba8(BN_ULONG *r2,BN_ULONG *r3,BN_ULONG *r4);
+.globl	bn_mul_comba8
+.type	bn_mul_comba8,@function
+.align	4
+bn_mul_comba8:
+	stmg	%r6,%r8,48(%r15)
+
+	lghi	c1,0
+	lghi	c2,0
+	lghi	c3,0
+	lghi	zero,0
+
+	mul_add_c(0,0,c1,c2,c3);
+	stg	c1,0*8(%r2)
+	lghi	c1,0
+
+	mul_add_c(0,1,c2,c3,c1);
+	mul_add_c(1,0,c2,c3,c1);
+	stg	c2,1*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(2,0,c3,c1,c2);
+	mul_add_c(1,1,c3,c1,c2);
+	mul_add_c(0,2,c3,c1,c2);
+	stg	c3,2*8(%r2)
+	lghi	c3,0
+
+	mul_add_c(0,3,c1,c2,c3);
+	mul_add_c(1,2,c1,c2,c3);
+	mul_add_c(2,1,c1,c2,c3);
+	mul_add_c(3,0,c1,c2,c3);
+	stg	c1,3*8(%r2)
+	lghi	c1,0
+
+	mul_add_c(4,0,c2,c3,c1);
+	mul_add_c(3,1,c2,c3,c1);
+	mul_add_c(2,2,c2,c3,c1);
+	mul_add_c(1,3,c2,c3,c1);
+	mul_add_c(0,4,c2,c3,c1);
+	stg	c2,4*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(0,5,c3,c1,c2);
+	mul_add_c(1,4,c3,c1,c2);
+	mul_add_c(2,3,c3,c1,c2);
+	mul_add_c(3,2,c3,c1,c2);
+	mul_add_c(4,1,c3,c1,c2);
+	mul_add_c(5,0,c3,c1,c2);
+	stg	c3,5*8(%r2)
+	lghi	c3,0
+
+	mul_add_c(6,0,c1,c2,c3);
+	mul_add_c(5,1,c1,c2,c3);
+	mul_add_c(4,2,c1,c2,c3);
+	mul_add_c(3,3,c1,c2,c3);
+	mul_add_c(2,4,c1,c2,c3);
+	mul_add_c(1,5,c1,c2,c3);
+	mul_add_c(0,6,c1,c2,c3);
+	stg	c1,6*8(%r2)
+	lghi	c1,0
+
+	mul_add_c(0,7,c2,c3,c1);
+	mul_add_c(1,6,c2,c3,c1);
+	mul_add_c(2,5,c2,c3,c1);
+	mul_add_c(3,4,c2,c3,c1);
+	mul_add_c(4,3,c2,c3,c1);
+	mul_add_c(5,2,c2,c3,c1);
+	mul_add_c(6,1,c2,c3,c1);
+	mul_add_c(7,0,c2,c3,c1);
+	stg	c2,7*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(7,1,c3,c1,c2);
+	mul_add_c(6,2,c3,c1,c2);
+	mul_add_c(5,3,c3,c1,c2);
+	mul_add_c(4,4,c3,c1,c2);
+	mul_add_c(3,5,c3,c1,c2);
+	mul_add_c(2,6,c3,c1,c2);
+	mul_add_c(1,7,c3,c1,c2);
+	stg	c3,8*8(%r2)
+	lghi	c3,0
+
+	mul_add_c(2,7,c1,c2,c3);
+	mul_add_c(3,6,c1,c2,c3);
+	mul_add_c(4,5,c1,c2,c3);
+	mul_add_c(5,4,c1,c2,c3);
+	mul_add_c(6,3,c1,c2,c3);
+	mul_add_c(7,2,c1,c2,c3);
+	stg	c1,9*8(%r2)
+	lghi	c1,0
+
+	mul_add_c(7,3,c2,c3,c1);
+	mul_add_c(6,4,c2,c3,c1);
+	mul_add_c(5,5,c2,c3,c1);
+	mul_add_c(4,6,c2,c3,c1);
+	mul_add_c(3,7,c2,c3,c1);
+	stg	c2,10*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(4,7,c3,c1,c2);
+	mul_add_c(5,6,c3,c1,c2);
+	mul_add_c(6,5,c3,c1,c2);
+	mul_add_c(7,4,c3,c1,c2);
+	stg	c3,11*8(%r2)
+	lghi	c3,0
+
+	mul_add_c(7,5,c1,c2,c3);
+	mul_add_c(6,6,c1,c2,c3);
+	mul_add_c(5,7,c1,c2,c3);
+	stg	c1,12*8(%r2)
+	lghi	c1,0
+
+
+	mul_add_c(6,7,c2,c3,c1);
+	mul_add_c(7,6,c2,c3,c1);
+	stg	c2,13*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(7,7,c3,c1,c2);
+	stg	c3,14*8(%r2)
+	stg	c1,15*8(%r2)
+
+	lmg	%r6,%r8,48(%r15)
+	br	%r14
+.size	bn_mul_comba8,.-bn_mul_comba8
+
+// void bn_mul_comba4(BN_ULONG *r2,BN_ULONG *r3,BN_ULONG *r4);
+.globl	bn_mul_comba4
+.type	bn_mul_comba4,@function
+.align	4
+bn_mul_comba4:
+	stmg	%r6,%r8,48(%r15)
+
+	lghi	c1,0
+	lghi	c2,0
+	lghi	c3,0
+	lghi	zero,0
+
+	mul_add_c(0,0,c1,c2,c3);
+	stg	c1,0*8(%r3)
+	lghi	c1,0
+
+	mul_add_c(0,1,c2,c3,c1);
+	mul_add_c(1,0,c2,c3,c1);
+	stg	c2,1*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(2,0,c3,c1,c2);
+	mul_add_c(1,1,c3,c1,c2);
+	mul_add_c(0,2,c3,c1,c2);
+	stg	c3,2*8(%r2)
+	lghi	c3,0
+
+	mul_add_c(0,3,c1,c2,c3);
+	mul_add_c(1,2,c1,c2,c3);
+	mul_add_c(2,1,c1,c2,c3);
+	mul_add_c(3,0,c1,c2,c3);
+	stg	c1,3*8(%r2)
+	lghi	c1,0
+
+	mul_add_c(3,1,c2,c3,c1);
+	mul_add_c(2,2,c2,c3,c1);
+	mul_add_c(1,3,c2,c3,c1);
+	stg	c2,4*8(%r2)
+	lghi	c2,0
+
+	mul_add_c(2,3,c3,c1,c2);
+	mul_add_c(3,2,c3,c1,c2);
+	stg	c3,5*8(%r2)
+	lghi	c3,0
+
+	mul_add_c(3,3,c1,c2,c3);
+	stg	c1,6*8(%r2)
+	stg	c2,7*8(%r2)
+
+	stmg	%r6,%r8,48(%r15)
+	br	%r14
+.size	bn_mul_comba4,.-bn_mul_comba4
+
+#define sqr_add_c(ai,c1,c2,c3)		\
+	lg	%r7,ai*8(%r3);		\
+	mlgr	%r6,%r7;		\
+	algr	c1,%r7;			\
+	alcgr	c2,%r6;			\
+	alcgr	c3,zero
+
+#define sqr_add_c2(ai,aj,c1,c2,c3)	\
+	lg	%r7,ai*8(%r3);		\
+	mlg	%r6,aj*8(%r3);		\
+	algr	c1,%r7;			\
+	alcgr	c2,%r6;			\
+	alcgr	c3,zero;		\
+	algr	c1,%r7;			\
+	alcgr	c2,%r6;			\
+	alcgr	c3,zero
+
+// void bn_sqr_comba8(BN_ULONG *r2,BN_ULONG *r3);
+.globl	bn_sqr_comba8
+.type	bn_sqr_comba8,@function
+.align	4
+bn_sqr_comba8:
+	stmg	%r6,%r8,48(%r15)
+
+	lghi	c1,0
+	lghi	c2,0
+	lghi	c3,0
+	lghi	zero,0
+
+	sqr_add_c(0,c1,c2,c3);
+	stg	c1,0*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c2(1,0,c2,c3,c1);
+	stg	c2,1*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c(1,c3,c1,c2);
+	sqr_add_c2(2,0,c3,c1,c2);
+	stg	c3,2*8(%r2)
+	lghi	c3,0
+
+	sqr_add_c2(3,0,c1,c2,c3);
+	sqr_add_c2(2,1,c1,c2,c3);
+	stg	c1,3*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c(2,c2,c3,c1);
+	sqr_add_c2(3,1,c2,c3,c1);
+	sqr_add_c2(4,0,c2,c3,c1);
+	stg	c2,4*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c2(5,0,c3,c1,c2);
+	sqr_add_c2(4,1,c3,c1,c2);
+	sqr_add_c2(3,2,c3,c1,c2);
+	stg	c3,5*8(%r2)
+	lghi	c3,0
+
+	sqr_add_c(3,c1,c2,c3);
+	sqr_add_c2(4,2,c1,c2,c3);
+	sqr_add_c2(5,1,c1,c2,c3);
+	sqr_add_c2(6,0,c1,c2,c3);
+	stg	c1,6*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c2(7,0,c2,c3,c1);
+	sqr_add_c2(6,1,c2,c3,c1);
+	sqr_add_c2(5,2,c2,c3,c1);
+	sqr_add_c2(4,3,c2,c3,c1);
+	stg	c2,7*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c(4,c3,c1,c2);
+	sqr_add_c2(5,3,c3,c1,c2);
+	sqr_add_c2(6,2,c3,c1,c2);
+	sqr_add_c2(7,1,c3,c1,c2);
+	stg	c3,8*8(%r2)
+	lghi	c3,0
+
+	sqr_add_c2(7,2,c1,c2,c3);
+	sqr_add_c2(6,3,c1,c2,c3);
+	sqr_add_c2(5,4,c1,c2,c3);
+	stg	c1,9*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c(5,c2,c3,c1);
+	sqr_add_c2(6,4,c2,c3,c1);
+	sqr_add_c2(7,3,c2,c3,c1);
+	stg	c2,10*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c2(7,4,c3,c1,c2);
+	sqr_add_c2(6,5,c3,c1,c2);
+	stg	c3,11*8(%r2)
+	lghi	c3,0
+
+	sqr_add_c(6,c1,c2,c3);
+	sqr_add_c2(7,5,c1,c2,c3);
+	stg	c1,12*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c2(7,6,c2,c3,c1);
+	stg	c2,13*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c(7,c3,c1,c2);
+	stg	c3,14*8(%r2)
+	stg	c1,15*8(%r2)
+
+	lmg	%r6,%r8,48(%r15)
+	br	%r14
+.size	bn_sqr_comba8,.-bn_sqr_comba8
+
+// void bn_sqr_comba4(BN_ULONG *r2,BN_ULONG *r3);
+.globl bn_sqr_comba4
+.type	bn_sqr_comba4,@function
+.align	4
+bn_sqr_comba4:
+	stmg	%r6,%r8,48(%r15)
+
+	lghi	c1,0
+	lghi	c2,0
+	lghi	c3,0
+	lghi	zero,0
+
+	sqr_add_c(0,c1,c2,c3);
+	stg	c1,0*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c2(1,0,c2,c3,c1);
+	stg	c2,1*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c(1,c3,c1,c2);
+	sqr_add_c2(2,0,c3,c1,c2);
+	stg	c3,2*8(%r2)
+	lghi	c3,0
+
+	sqr_add_c2(3,0,c1,c2,c3);
+	sqr_add_c2(2,1,c1,c2,c3);
+	stg	c1,3*8(%r2)
+	lghi	c1,0
+
+	sqr_add_c(2,c2,c3,c1);
+	sqr_add_c2(3,1,c2,c3,c1);
+	stg	c2,4*8(%r2)
+	lghi	c2,0
+
+	sqr_add_c2(3,2,c3,c1,c2);
+	stg	c3,5*8(%r2)
+	lghi	c3,0
+
+	sqr_add_c(3,c1,c2,c3);
+	stg	c1,6*8(%r2)
+	stg	c2,7*8(%r2)
+
+	lmg	%r6,%r8,48(%r15)
+	br	%r14
+.size	bn_sqr_comba4,.-bn_sqr_comba4