X-Git-Url: https://git.gag.com/?a=blobdiff_plain;f=device%2Flib%2F_fsmul.c;h=95c12f110b3f8a20c4b1abd51fec47435e5ff443;hb=bf1925f901a9f45ad2529874f41e1e2b01f5bc9f;hp=09fd00a4fdba9092114e4d52538f139a366be3da;hpb=4e85b590d59de032a494ad439773f94d845db0f4;p=fw%2Fsdcc

diff --git a/device/lib/_fsmul.c b/device/lib/_fsmul.c
index 09fd00a4..95c12f11 100644
--- a/device/lib/_fsmul.c
+++ b/device/lib/_fsmul.c
@@ -1,3 +1,189 @@
+/* Floating point library in optimized assembly for 8051
+ * Copyright (c) 2004, Paul Stoffregen, paul@pjrc.com
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public License
+ * as published by the Free Software Foundation; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * This 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 General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
+ */
+
+
+#define SDCC_FLOAT_LIB
+#include <float.h>
+
+
+#ifdef FLOAT_ASM_MCS51
+
+// float __fsmul (float a, float b) reentrant
+static void dummy(void) _naked
+{
+	_asm
+	.globl	___fsmul
+___fsmul:
+	// extract the two inputs, placing them into:
+	//      sign     exponent   mantiassa
+	//      ----     --------   ---------
+	//  a:  sign_a   exp_a      r4/r3/r2
+	//  b:  sign_b   exp_b      r7/r6/r5
+
+	lcall	fsgetargs
+
+	// first check if either input is zero
+	cjne	r4, #0, 00002$
+00001$:
+	ljmp	fs_return_zero
+
+00002$:
+	mov	a, r7
+	jz	00001$
+
+	// compute final sign bit
+	jnb	sign_b, 00003$
+	cpl	sign_a
+00003$:
+
+	// add the exponents
+	mov	a, exp_a
+	add	a, exp_b
+	add	a, #130
+	mov	exp_a, a
+
+
+	// now we need to multipy r4/r3/r2 * r7/r6/r5
+	// ------------------------------------------
+	//	                        r2 * r5		<< 0
+	//	            r3 * r5  +  r2 * r6		<< 8
+	//	r4 * r5  +  r3 * r6  +  r2 * r7		<< 16
+	//	r4 * r6  +  r3 * r7			<< 24
+	//	r4 * r7					<< 32
+	//
+	// This adds quite a bit of code, but it is a LOT faster
+	// that three calls to __mululong...
+
+	// output goes into r4/r3/r2/r1/r0/xx
+
+	mov	a, r2
+	mov	b, r5
+	mul	ab			// r2 * r5
+	// discard lowest 8 bits
+	mov	r0, b
+	// range 0-FE
+
+	mov	a, r2
+	mov	b, r6
+	mul	ab			// r2 * r6
+	add	a, r0
+	mov	r0, a
+	clr	a
+	addc	a, b
+	mov	r1, a
+	// range 0-FEFF
+
+	mov	a, r3
+	mov	b, r5
+	mul	ab			// r3 * r5
+	add	a, r0
+	// discard lowest 8 bits
+	mov	a, r1
+	addc	a, b
+	mov	r1, a
+	clr	a
+	rlc	a
+	xch	a, r2
+	// range 0-1FD
+
+	mov	b, r7
+	mul	ab			// r2 * r7
+	add	a, r1
+	mov	r1, a
+	mov	a, r2
+	addc	a, b
+	mov	r2, a
+	// range 0-FFFE
+
+	mov	a, r3
+	mov	r0, a
+	mov	b, r6
+	mul	ab			// r3 * r6
+	add	a, r1
+	mov	r1, a
+	mov	a, r2
+	addc	a, b
+	mov	r2, a
+	clr	a
+	rlc	a
+	mov	r3, a
+	// range 0-1FDFF
+
+	mov	a, r4
+	mov	b, r5
+	mul	ab			// r4 * r5
+	add	a, r1
+	mov	r1, a
+	mov	a, r2
+	addc	a, b
+	mov	r2, a
+	clr	a
+	addc	a, r3
+	mov	r3, a
+	// range 0-2FC00
+
+	mov	a, r0 // r3
+	mov	b, r7
+	mul	ab			// r3 * r7
+	add	a, r2
+	mov	r2, a
+	mov	a, r3
+	addc	a, b
+	mov	r3, a
+	clr	a
+	rlc	a
+	xch	a, r4
+	// range 0-100FD00
+
+	mov	r5, a
+	mov	b, r6
+	mul	ab			// r4 * r6
+	add	a, r2
+	mov	r2, a
+	mov	a, r3
+	addc	a, b
+	mov	r3, a
+	clr	a
+	addc	a, r4
+	mov	r4, a
+	// range 0-1FEFE00
+
+	mov	a, r5 // r4
+	mov	b, r7
+	mul	ab			// r4 * r7
+	add	a, r3
+	mov	r3, a
+	mov	a, r4
+	addc	a, b
+	mov	r4, a
+	// range 40000000-FFFFFE00
+
+	jb	acc.7, 00010$
+	lcall	fs_normalize_a
+
+00010$:
+	ljmp	fs_round_and_return
+	_endasm;
+}
+
+#else
+
+
 /*
 ** libgcc support for software floating point.
 ** Copyright (C) 1991 by Pipeline Associates, Inc.  All rights reserved.
@@ -16,7 +202,6 @@
 
 /* (c)2000/2001: hacked a little by johan.knol@iduna.nl for sdcc */
 
-#include <float.h>
 
 union float_long
   {
@@ -71,6 +256,5 @@ float __fsmul (float a1, float a2) {
   return (fl1.f);
 }
 
-
-
+#endif