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
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
|
/***************************************************************************
* __________ __ ___.
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
* \/ \/ \/ \/ \/
*
* Copyright (C) 2005 by Pedro Vasconcelos
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
****************************************************************************/
/* asm routines for wide math on the MCF5249 */
#if defined(CPU_COLDFIRE)
/* attribute for 16-byte alignment */
#define LINE_ATTR __attribute__ ((aligned (16)))
#ifndef _V_WIDE_MATH
#define _V_WIDE_MATH
static inline int32_t MULT32(int32_t x, int32_t y) {
asm volatile ("mac.l %[x], %[y], %%acc0;" /* multiply & shift */
"movclr.l %%acc0, %[x];" /* move & clear acc */
"asr.l #1, %[x];" /* no overflow test */
: [x] "+&d" (x)
: [y] "r" (y)
: "cc");
return x;
}
static inline int32_t MULT31(int32_t x, int32_t y) {
asm volatile ("mac.l %[x], %[y], %%acc0;" /* multiply */
"movclr.l %%acc0, %[x];" /* move and clear */
: [x] "+&r" (x)
: [y] "r" (y)
: "cc");
return x;
}
static inline int32_t MULT31_SHIFT15(int32_t x, int32_t y) {
int32_t r;
asm volatile ("mac.l %[x], %[y], %%acc0;" /* multiply */
"mulu.l %[y], %[x];" /* get lower half, avoid emac stall */
"movclr.l %%acc0, %[r];" /* get higher half */
"asl.l #8, %[r];" /* hi<<16, plus one free */
"asl.l #8, %[r];"
"lsr.l #8, %[x];" /* (unsigned)lo >> 15 */
"lsr.l #7, %[x];"
"or.l %[x], %[r];" /* logical-or results */
: [r] "=&d" (r), [x] "+d" (x)
: [y] "d" (y)
: "cc");
return r;
}
static inline
void XPROD31(int32_t a, int32_t b,
int32_t t, int32_t v,
int32_t *x, int32_t *y)
{
asm volatile ("mac.l %[a], %[t], %%acc0;"
"mac.l %[b], %[v], %%acc0;"
"mac.l %[b], %[t], %%acc1;"
"msac.l %[a], %[v], %%acc1;"
"movclr.l %%acc0, %[a];"
"move.l %[a], (%[x]);"
"movclr.l %%acc1, %[a];"
"move.l %[a], (%[y]);"
: [a] "+&r" (a)
: [x] "a" (x), [y] "a" (y),
[b] "r" (b), [t] "r" (t), [v] "r" (v)
: "cc", "memory");
}
static inline
void XNPROD31(int32_t a, int32_t b,
int32_t t, int32_t v,
int32_t *x, int32_t *y)
{
asm volatile ("mac.l %[a], %[t], %%acc0;"
"msac.l %[b], %[v], %%acc0;"
"mac.l %[b], %[t], %%acc1;"
"mac.l %[a], %[v], %%acc1;"
"movclr.l %%acc0, %[a];"
"move.l %[a], (%[x]);"
"movclr.l %%acc1, %[a];"
"move.l %[a], (%[y]);"
: [a] "+&r" (a)
: [x] "a" (x), [y] "a" (y),
[b] "r" (b), [t] "r" (t), [v] "r" (v)
: "cc", "memory");
}
#if 0 /* canonical Tremor definition */
#define XPROD32(_a, _b, _t, _v, _x, _y) \
{ (_x)=MULT32(_a,_t)+MULT32(_b,_v); \
(_y)=MULT32(_b,_t)-MULT32(_a,_v); }
#endif
/* this could lose the LSB by overflow, but i don't think it'll ever happen.
if anyone think they can hear a bug caused by this, please try the above
version. */
#define XPROD32(_a, _b, _t, _v, _x, _y) \
asm volatile ("mac.l %[a], %[t], %%acc0;" \
"mac.l %[b], %[v], %%acc0;" \
"mac.l %[b], %[t], %%acc1;" \
"msac.l %[a], %[v], %%acc1;" \
"movclr.l %%acc0, %[x];" \
"asr.l #1, %[x];" \
"movclr.l %%acc1, %[y];" \
"asr.l #1, %[y];" \
: [x] "=&d" (_x), [y] "=&d" (_y) \
: [a] "r" (_a), [b] "r" (_b), \
[t] "r" (_t), [v] "r" (_v) \
: "cc");
#define XPROD31_R(_a, _b, _t, _v, _x, _y) \
asm volatile ("mac.l %[a], %[t], %%acc0;" \
"mac.l %[b], %[v], %%acc0;" \
"mac.l %[b], %[t], %%acc1;" \
"msac.l %[a], %[v], %%acc1;" \
"movclr.l %%acc0, %[x];" \
"movclr.l %%acc1, %[y];" \
: [x] "=&d" (_x), [y] "=&d" (_y) \
: [a] "r" (_a), [b] "r" (_b), \
[t] "r" (_t), [v] "r" (_v) \
: "cc");
#define XNPROD31_R(_a, _b, _t, _v, _x, _y) \
asm volatile ("mac.l %[a], %[t], %%acc0;" \
"msac.l %[b], %[v], %%acc0;" \
"mac.l %[b], %[t], %%acc1;" \
"mac.l %[a], %[v], %%acc1;" \
"movclr.l %%acc0, %[x];" \
"movclr.l %%acc1, %[y];" \
: [x] "=&d" (_x), [y] "=&d" (_y) \
: [a] "r" (_a), [b] "r" (_b), \
[t] "r" (_t), [v] "r" (_v) \
: "cc");
#ifndef _V_VECT_OPS
#define _V_VECT_OPS
/* asm versions of vector operations for block.c, window.c */
/* assumes MAC is initialized & accumulators cleared */
static inline
void vect_add(int32_t *x, const int32_t *y, int n)
{
/* align to 16 bytes */
while(n>0 && (int)x&15) {
*x++ += *y++;
n--;
}
asm volatile ("bra 1f;"
"0:" /* loop start */
"movem.l (%[x]), %%d0-%%d3;" /* fetch values */
"movem.l (%[y]), %%a0-%%a3;"
/* add */
"add.l %%a0, %%d0;"
"add.l %%a1, %%d1;"
"add.l %%a2, %%d2;"
"add.l %%a3, %%d3;"
/* store and advance */
"movem.l %%d0-%%d3, (%[x]);"
"lea.l (4*4, %[x]), %[x];"
"lea.l (4*4, %[y]), %[y];"
"subq.l #4, %[n];" /* done 4 elements */
"1: cmpi.l #4, %[n];"
"bge 0b;"
: [n] "+d" (n), [x] "+a" (x), [y] "+a" (y)
: : "%d0", "%d1", "%d2", "%d3", "%a0", "%a1", "%a2", "%a3",
"cc", "memory");
/* add final elements */
while (n>0) {
*x++ += *y++;
n--;
}
}
static inline
void vect_copy(int32_t *x, int32_t *y, int n)
{
/* align to 16 bytes */
while(n>0 && (int)x&15) {
*x++ = *y++;
n--;
}
asm volatile ("bra 1f;"
"0:" /* loop start */
"movem.l (%[y]), %%d0-%%d3;" /* fetch values */
"movem.l %%d0-%%d3, (%[x]);" /* store */
"lea.l (4*4, %[x]), %[x];" /* advance */
"lea.l (4*4, %[y]), %[y];"
"subq.l #4, %[n];" /* done 4 elements */
"1: cmpi.l #4, %[n];"
"bge 0b;"
: [n] "+d" (n), [x] "+a" (x), [y] "+a" (y)
: : "%d0", "%d1", "%d2", "%d3", "cc", "memory");
/* copy final elements */
while (n>0) {
*x++ = *y++;
n--;
}
}
static inline
void vect_mult_fw(int32_t *data, int32_t *window, int n)
{
/* ensure data is aligned to 16-bytes */
while(n>0 && (int)data&15) {
*data = MULT31(*data, *window);
data++;
window++;
n--;
}
asm volatile ("movem.l (%[d]), %%d0-%%d3;" /* loop start */
"movem.l (%[w]), %%a0-%%a3;" /* pre-fetch registers */
"lea.l (4*4, %[w]), %[w];"
"bra 1f;" /* jump to loop condition */
"0:" /* loop body */
/* multiply and load next window values */
"mac.l %%d0, %%a0, (%[w])+, %%a0, %%acc0;"
"mac.l %%d1, %%a1, (%[w])+, %%a1, %%acc1;"
"mac.l %%d2, %%a2, (%[w])+, %%a2, %%acc2;"
"mac.l %%d3, %%a3, (%[w])+, %%a3, %%acc3;"
"movclr.l %%acc0, %%d0;" /* get the products */
"movclr.l %%acc1, %%d1;"
"movclr.l %%acc2, %%d2;"
"movclr.l %%acc3, %%d3;"
/* store and advance */
"movem.l %%d0-%%d3, (%[d]);"
"lea.l (4*4, %[d]), %[d];"
"movem.l (%[d]), %%d0-%%d3;"
"subq.l #4, %[n];" /* done 4 elements */
"1: cmpi.l #4, %[n];"
"bge 0b;"
/* multiply final elements */
"tst.l %[n];"
"beq 1f;" /* n=0 */
"mac.l %%d0, %%a0, %%acc0;"
"movclr.l %%acc0, %%d0;"
"move.l %%d0, (%[d])+;"
"subq.l #1, %[n];"
"beq 1f;" /* n=1 */
"mac.l %%d1, %%a1, %%acc0;"
"movclr.l %%acc0, %%d1;"
"move.l %%d1, (%[d])+;"
"subq.l #1, %[n];"
"beq 1f;" /* n=2 */
/* otherwise n = 3 */
"mac.l %%d2, %%a2, %%acc0;"
"movclr.l %%acc0, %%d2;"
"move.l %%d2, (%[d])+;"
"1:"
: [n] "+d" (n), [d] "+a" (data), [w] "+a" (window)
: : "%d0", "%d1", "%d2", "%d3", "%a0", "%a1", "%a2", "%a3",
"cc", "memory");
}
static inline
void vect_mult_bw(int32_t *data, int32_t *window, int n)
{
/* ensure at least data is aligned to 16-bytes */
while(n>0 && (int)data&15) {
*data = MULT31(*data, *window);
data++;
window--;
n--;
}
asm volatile ("lea.l (-3*4, %[w]), %[w];" /* loop start */
"movem.l (%[d]), %%d0-%%d3;" /* pre-fetch registers */
"movem.l (%[w]), %%a0-%%a3;"
"bra 1f;" /* jump to loop condition */
"0:" /* loop body */
/* multiply and load next window value */
"mac.l %%d0, %%a3, -(%[w]), %%a3, %%acc0;"
"mac.l %%d1, %%a2, -(%[w]), %%a2, %%acc1;"
"mac.l %%d2, %%a1, -(%[w]), %%a1, %%acc2;"
"mac.l %%d3, %%a0, -(%[w]), %%a0, %%acc3;"
"movclr.l %%acc0, %%d0;" /* get the products */
"movclr.l %%acc1, %%d1;"
"movclr.l %%acc2, %%d2;"
"movclr.l %%acc3, %%d3;"
/* store and advance */
"movem.l %%d0-%%d3, (%[d]);"
"lea.l (4*4, %[d]), %[d];"
"movem.l (%[d]), %%d0-%%d3;"
"subq.l #4, %[n];" /* done 4 elements */
"1: cmpi.l #4, %[n];"
"bge 0b;"
/* multiply final elements */
"tst.l %[n];"
"beq 1f;" /* n=0 */
"mac.l %%d0, %%a3, %%acc0;"
"movclr.l %%acc0, %%d0;"
"move.l %%d0, (%[d])+;"
"subq.l #1, %[n];"
"beq 1f;" /* n=1 */
"mac.l %%d1, %%a2, %%acc0;"
"movclr.l %%acc0, %%d1;"
"move.l %%d1, (%[d])+;"
"subq.l #1, %[n];"
"beq 1f;" /* n=2 */
/* otherwise n = 3 */
"mac.l %%d2, %%a1, %%acc0;"
"movclr.l %%acc0, %%d2;"
"move.l %%d2, (%[d])+;"
"1:"
: [n] "+d" (n), [d] "+a" (data), [w] "+a" (window)
: : "%d0", "%d1", "%d2", "%d3", "%a0", "%a1", "%a2", "%a3",
"cc", "memory");
}
#endif
#endif
/* not used anymore */
/*
#ifndef _V_CLIP_MATH
#define _V_CLIP_MATH
* this is portable C and simple; why not use this as default?
static inline int32_t CLIP_TO_15(register int32_t x) {
register int32_t hi=32767, lo=-32768;
return (x>=hi ? hi : (x<=lo ? lo : x));
}
#endif
*/
#else
#define LINE_ATTR
#endif
|
