1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
|
/* $OpenBSD: impyu.S,v 1.10 2003/04/10 17:27:58 mickey Exp $ */
/*
(c) Copyright 1986 HEWLETT-PACKARD COMPANY
To anyone who acknowledges that this file is provided "AS IS"
without any express or implied warranty:
permission to use, copy, modify, and distribute this file
for any purpose is hereby granted without fee, provided that
the above copyright notice and this notice appears in all
copies, and that the name of Hewlett-Packard Company not be
used in advertising or publicity pertaining to distribution
of the software without specific, written prior permission.
Hewlett-Packard Company makes no representations about the
suitability of this software for any purpose.
*/
/* @(#)impyu.s: Revision: 1.11.88.1 Date: 93/12/07 15:06:31 */
#include <machine/asm.h>
#define _LOCORE
#include <machine/frame.h>
;****************************************************************************
;
;Implement an integer multiply routine for 32-bit operands and 64-bit product
; with operand values of zero (multiplicand only) and 2**32reated specially.
; The algorithm uses the multiplier, four bits at a time, from right to left,
; to generate partial product. Execution speed is more important than program
; size in this implementation.
;
;******************************************************************************
;
; Definitions - General registers
;
gr0 .equ 0 ; General register zero
pu .equ 3 ; upper part of product
pl .equ 4 ; lower part of product
op2 .equ 4 ; multiplier
op1 .equ 5 ; multiplicand
cnt .equ 6 ; count in multiply
brindex .equ 7 ; index into the br. table
saveop2 .equ 8 ; save op2 if high bit of multiplicand
; is set
pc .equ 9 ; carry bit of product, = 00...01
pm .equ 10 ; value of -1 used in shifting
temp .equ 6
;****************************************************************************
.text
LEAF_ENTRY(u_xmpy)
stws,ma pu,4(sp) ; save registers on stack
stws,ma pl,4(sp) ; save registers on stack
stws,ma op1,4(sp) ; save registers on stack
stws,ma cnt,4(sp) ; save registers on stack
stws,ma brindex,4(sp) ; save registers on stack
stws,ma saveop2,4(sp) ; save registers on stack
stws,ma pc,4(sp) ; save registers on stack
stws,ma pm,4(sp) ; save registers on stack
;
; Start multiply process
;
ldws 0(arg0),op1 ; get multiplicand
ldws 0(arg1),op2 ; get multiplier
addib,= 0,op1,fini0 ; op1 = 0, product = 0
addi 0,gr0,pu ; clear product
bb,>= op1,0,mpy1 ; test msb of multiplicand
addi 0,gr0,saveop2 ; clear saveop2
;
; msb of multiplicand is set so will save multiplier for a final
; addition into the result
;
extru,= op1,31,31,op1 ; clear msb of multiplicand
b mpy1 ; if op1 < 2**32, start multiply
add op2,gr0,saveop2 ; save op2 in saveop2
shd gr0,op2,1,pu ; shift op2 left 31 for result
b fini ; go to finish
shd op2,gr0,1,pl
;
mpy1 addi -1,gr0,pm ; initialize pm to 111...1
addi 1,gr0,pc ; initialize pc to 00...01
movib,tr 8,cnt,mloop ; set count for mpy loop
extru op2,31,4,brindex ; 4 bits as index into table
;
.align 8
;
b sh4c ; br. if sign overflow
sh4n shd pu,pl,4,pl ; shift product right 4 bits
addib,<= -1,cnt,mulend ; reduce count by 1, exit if
extru pu,27,28,pu ; <= zero
;
mloop blr brindex,gr0 ; br. into table
; entries of 2 words
extru op2,27,4,brindex ; next 4 bits into index
;
;
; branch table for the multiplication process with four multiplier bits
;
mtable ; two words per entry
;
; ---- bits = 0000 ---- shift product 4 bits -------------------------------
;
b sh4n+4 ; just shift partial
shd pu,pl,4,pl ; product right 4 bits
;
; ---- bits = 0001 ---- add op1, then shift 4 bits
;
addb,tr op1,pu,sh4n+4 ; add op1 to product, to shift
shd pu,pl,4,pl ; product right 4 bits
;
; ---- bits = 0010 ---- add op1, add op1, then shift 4 bits
;
addb,tr op1,pu,sh4n ; add 2*op1, to shift
addb,uv op1,pu,sh4c ; product right 4 bits
;
; ---- bits = 0011 ---- add op1, add 2*op1, shift 4 bits
;
addb,tr op1,pu,sh4n-4 ; add op1 & 2*op1, shift
sh1add,nuv op1,pu,pu ; product right 4 bits
;
; ---- bits = 0100 ---- shift 2, add op1, shift 2
;
b sh2sa
shd pu,pl,2,pl ; shift product 2 bits
;
; ---- bits = 0101 ---- add op1, shift 2, add op1, and shift 2 again
;
addb,tr op1,pu,sh2us ; add op1 to product
shd pu,pl,2,pl ; shift 2 bits
;
; ---- bits = 0110 ---- add op1, add op1, shift 2, add op1, and shift 2 again
;
addb,tr op1,pu,sh2c ; add 2*op1, to shift 2 bits
addb,nuv op1,pu,sh2us ; br. if not overflow
;
; ---- bits = 0111 ---- subtract op1, shift 3, add op1, and shift 1
;
b sh3s
sub pu,op1,pu ; subtract op1, br. to sh3s
;
; ---- bits = 1000 ---- shift 3, add op1, shift 1
;
b sh3sa
shd pu,pl,3,pl ; shift product right 3 bits
;
; ---- bits = 1001 ---- add op1, shift 3, add op1, shift 1
;
addb,tr op1,pu,sh3us ; add op1, to shift 3, add op1,
shd pu,pl,3,pl ; and shift 1
;
; ---- bits = 1010 ---- add op1, add op1, shift 3, add op1, shift 1
;
addb,tr op1,pu,sh3c ; add 2*op1, to shift 3 bits
addb,nuv op1,pu,sh3us ; br. if no overflow
;
; ---- bits = 1011 ---- add -op1, shift 2, add -op1, shift 2, inc. next index
;
addib,tr 1,brindex,sh2s ; add 1 to index, subtract op1,
sub pu,op1,pu ; shift 2 with minus sign
;
; ---- bits = 1100 ---- shift 2, subtract op1, shift 2, increment next index
;
addib,tr 1,brindex,sh2sb ; add 1 to index, to shift
shd pu,pl,2,pl ; shift right 2 bits signed
;
; ---- bits = 1101 ---- add op1, shift 2, add -op1, shift 2
;
addb,tr op1,pu,sh2ns ; add op1, to shift 2
shd pu,pl,2,pl ; right 2 unsigned, etc.
;
; ---- bits = 1110 ---- shift 1 signed, add -op1, shift 3 signed
;
addib,tr 1,brindex,sh1sa ; add 1 to index, to shift
shd pu,pl,1,pl ; shift 1 bit
;
; ---- bits = 1111 ---- add -op1, shift 4 signed
;
addib,tr 1,brindex,sh4s ; add 1 to index, subtract op1,
sub pu,op1,pu ; to shift 4 signed
;
; ---- bits = 10000 ---- shift 4 signed
;
addib,tr 1,brindex,sh4s+4 ; add 1 to index
shd pu,pl,4,pl ; shift 4 signed
;
; ---- end of table ---------------------------------------------------------
;
sh4s shd pu,pl,4,pl
addib,> -1,cnt,mloop ; decrement count, loop if > 0
shd pm,pu,4,pu ; shift 4, minus signed
addb,tr op1,pu,lastadd ; do one more add, then finish
addb,=,n saveop2,gr0,fini ; check saveop2
;
sh4c addib,> -1,cnt,mloop ; decrement count, loop if > 0
shd pc,pu,4,pu ; shift 4 with overflow
b lastadd ; end of multiply
addb,=,n saveop2,gr0,fini ; check saveop2
;
sh3c shd pu,pl,3,pl ; shift product 3 bits
shd pc,pu,3,pu ; shift 3 signed
addb,tr op1,pu,sh1 ; add op1, to shift 1 bit
shd pu,pl,1,pl
;
sh3us extru pu,28,29,pu ; shift 3 unsigned
addb,tr op1,pu,sh1 ; add op1, to shift 1 bit
shd pu,pl,1,pl
;
sh3sa extrs pu,28,29,pu ; shift 3 signed
addb,tr op1,pu,sh1 ; add op1, to shift 1 bit
shd pu,pl,1,pl
;
sh3s shd pu,pl,3,pl ; shift 3 minus signed
shd pm,pu,3,pu
addb,tr op1,pu,sh1 ; add op1, to shift 1 bit
shd pu,pl,1,pl
;
sh1 addib,> -1,cnt,mloop ; loop if count > 0
extru pu,30,31,pu
b lastadd ; end of multiply
addb,=,n saveop2,gr0,fini ; check saveop2
;
sh2ns addib,tr 1,brindex,sh2sb+4 ; increment index
extru pu,29,30,pu ; shift unsigned
;
sh2s shd pu,pl,2,pl ; shift with minus sign
shd pm,pu,2,pu ;
sub pu,op1,pu ; subtract op1
shd pu,pl,2,pl ; shift with minus sign
addib,> -1,cnt,mloop ; decrement count, loop if > 0
shd pm,pu,2,pu ; shift with minus sign
addb,tr op1,pu,lastadd ; do one more add, then finish
addb,=,n saveop2,gr0,fini ; check saveop2
;
sh2sb extrs pu,29,30,pu ; shift 2 signed
sub pu,op1,pu ; subtract op1 from product
shd pu,pl,2,pl ; shift with minus sign
addib,> -1,cnt,mloop ; decrement count, loop if > 0
shd pm,pu,2,pu ; shift with minus sign
addb,tr op1,pu,lastadd ; do one more add, then finish
addb,=,n saveop2,gr0,fini ; check saveop2
;
sh1sa extrs pu,30,31,pu ; signed
sub pu,op1,pu ; subtract op1 from product
shd pu,pl,3,pl ; shift 3 with minus sign
addib,> -1,cnt,mloop ; decrement count, loop if >0
shd pm,pu,3,pu
addb,tr op1,pu,lastadd ; do one more add, then finish
addb,=,n saveop2,gr0,fini ; check saveop2
;
fini0 movib,tr 0,pl,fini ; product = 0 as op1 = 0
stws pu,0(arg2) ; save high part of result
;
sh2us extru pu,29,30,pu ; shift 2 unsigned
addb,tr op1,pu,sh2a ; add op1
shd pu,pl,2,pl ; shift 2 bits
;
sh2c shd pu,pl,2,pl
shd pc,pu,2,pu ; shift with carry
addb,tr op1,pu,sh2a ; add op1 to product
shd pu,pl,2,pl ; br. to sh2 to shift pu
;
sh2sa extrs pu,29,30,pu ; shift with sign
addb,tr op1,pu,sh2a ; add op1 to product
shd pu,pl,2,pl ; br. to sh2 to shift pu
;
sh2a addib,> -1,cnt,mloop ; loop if count > 0
extru pu,29,30,pu
;
mulend addb,=,n saveop2,gr0,fini ; check saveop2
lastadd shd saveop2,gr0,1,temp ; if saveop2 <> 0, shift it
shd gr0,saveop2,1,saveop2 ; left 31 and add to result
add pl,temp,pl
addc pu,saveop2,pu
;
; finish
;
fini stws pu,0(arg2) ; save high part of result
stws pl,4(arg2) ; save low part of result
ldws,mb -4(sp),pm ; restore registers
ldws,mb -4(sp),pc ; restore registers
ldws,mb -4(sp),saveop2 ; restore registers
ldws,mb -4(sp),brindex ; restore registers
ldws,mb -4(sp),cnt ; restore registers
ldws,mb -4(sp),op1 ; restore registers
ldws,mb -4(sp),pl ; restore registers
bv 0(rp) ; return
ldws,mb -4(sp),pu ; restore registers
EXIT(u_xmpy)
.end
|
