arm: Add optimized submul_1

Written from scratch rather than copied from GMP, due to LGPL 2.1 vs
GPL 3, but tested with the GMP testsuite.

This is 50% faster than the generic code as measured on Cortex-A15.
It is 25% slower than the current GMP routine on the same core.

1 files changed, 68 insertions, 0 deletions

diff --git a/ports/sysdeps/arm/submul_1.S b/ports/sysdeps/arm/submul_1.S
new file mode 100644
index 0000000000..3805eca9be
--- /dev/null
+++ b/ports/sysdeps/arm/submul_1.S
@@ -0,0 +1,68 @@
+/* mpn_submul_1 -- multiply and subtract bignums.
+   Copyright (C) 2013 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library.  If not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include <sysdep.h>
+
+	.syntax unified
+	.text
+
+@		cycles/limb
+@ StrongArm	   ?
+@ Cortex-A8	   ?
+@ Cortex-A9	   ?
+@ Cortex-A15	   4
+
+/* mp_limb_t mpn_submul_1(res_ptr, src1_ptr, size, s2_limb) */
+
+ENTRY (__mpn_submul_1)
+	push	{ r4, r5, r6, r7 }
+	cfi_adjust_cfa_offset (16)
+	cfi_rel_offset (r4, 0)
+	cfi_rel_offset (r5, 4)
+	cfi_rel_offset (r6, 8)
+	cfi_rel_offset (r7, 12)
+
+	ldr	r6, [r1], #4
+	ldr	r7, [r0]
+	mov	r4, #0			/* init carry in */
+	b	1f
+0:
+	ldr	r6, [r1], #4		/* load next ul */
+	adds	r5, r5, r4		/* (lpl, c) = lpl + cl */
+	adc	r4, ip, #0		/* cl = hpl + c */
+	subs	r5, r7, r5		/* (lpl, !c) = rl - lpl */
+	ldr	r7, [r0, #4]		/* load next rl */
+	it	cc
+	addcc	r4, r4, #1		/* cl += !c */
+	str	r5, [r0], #4
+1:
+	umull	r5, ip, r6, r3		/* (hpl, lpl) = ul * vl */
+	subs	r2, r2, #1
+	bne	0b
+
+	adds	r5, r5, r4		/* (lpl, c) = lpl + cl */
+	adc	r4, ip, #0		/* cl = hpl + c */
+	subs	r5, r7, r5		/* (lpl, !c) = rl - lpl */
+	str	r5, [r0], #4
+	it	cc
+	addcc	r4, r4, #1		/* cl += !c */
+	mov	r0, r4			/* return carry */
+
+	pop	{ r4, r5, r6, r7 }
+	DO_RET	(lr)
+END (__mpn_submul_1)