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
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
|
/* mips16-opc.c. Mips16 opcode table.
Copyright (C) 1996-2016 Free Software Foundation, Inc.
Contributed by Ian Lance Taylor, Cygnus Support
This file is part of the GNU opcodes library.
This library 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 3, or (at your option)
any later version.
It 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 file; see the file COPYING. If not, write to the
Free Software Foundation, 51 Franklin Street - Fifth Floor, Boston,
MA 02110-1301, USA. */
#include "sysdep.h"
#include <stdio.h>
#include "opcode/mips.h"
#include "mips-formats.h"
static unsigned char reg_0_map[] = { 0 };
static unsigned char reg_29_map[] = { 29 };
static unsigned char reg_31_map[] = { 31 };
static unsigned char reg_m16_map[] = { 16, 17, 2, 3, 4, 5, 6, 7 };
static unsigned char reg32r_map[] = {
0, 8, 16, 24,
1, 9, 17, 25,
2, 10, 18, 26,
3, 11, 19, 27,
4, 12, 20, 28,
5, 13, 21, 29,
6, 14, 22, 30,
7, 15, 23, 31
};
/* Return the meaning of operand character TYPE, or null if it isn't
recognized. If the operand is affected by the EXTEND instruction,
EXTENDED_P selects between the extended and unextended forms.
The extended forms all have an lsb of 0. */
const struct mips_operand *
decode_mips16_operand (char type, bfd_boolean extended_p)
{
switch (type)
{
case '0': MAPPED_REG (0, 0, GP, reg_0_map);
case 'L': SPECIAL (6, 5, ENTRY_EXIT_LIST);
case 'M': SPECIAL (7, 0, SAVE_RESTORE_LIST);
case 'P': SPECIAL (0, 0, PC);
case 'R': MAPPED_REG (0, 0, GP, reg_31_map);
case 'S': MAPPED_REG (0, 0, GP, reg_29_map);
case 'X': REG (5, 0, GP);
case 'Y': MAPPED_REG (5, 3, GP, reg32r_map);
case 'Z': MAPPED_REG (3, 0, GP, reg_m16_map);
case 'a': JUMP (26, 0, 2);
case 'e': HINT (11, 0);
case 'i': JALX (26, 0, 2);
case 'l': SPECIAL (6, 5, ENTRY_EXIT_LIST);
case 'm': SPECIAL (7, 0, SAVE_RESTORE_LIST);
case 'v': OPTIONAL_MAPPED_REG (3, 8, GP, reg_m16_map);
case 'w': OPTIONAL_MAPPED_REG (3, 5, GP, reg_m16_map);
case 'x': MAPPED_REG (3, 8, GP, reg_m16_map);
case 'y': MAPPED_REG (3, 5, GP, reg_m16_map);
case 'z': MAPPED_REG (3, 2, GP, reg_m16_map);
}
if (extended_p)
switch (type)
{
case '<': UINT (5, 0);
case '[': UINT (6, 0);
case ']': UINT (6, 0);
case '4': SINT (15, 0);
case '5': SINT (16, 0);
case '6': SINT (16, 0);
case '8': SINT (16, 0);
case 'A': PCREL (16, 0, TRUE, 0, 2, FALSE, FALSE);
case 'B': PCREL (16, 0, TRUE, 0, 3, FALSE, FALSE);
case 'C': SINT (16, 0);
case 'D': SINT (16, 0);
case 'E': PCREL (16, 0, TRUE, 0, 2, FALSE, FALSE);
case 'H': SINT (16, 0);
case 'K': SINT (16, 0);
case 'U': UINT (16, 0);
case 'V': SINT (16, 0);
case 'W': SINT (16, 0);
case 'j': SINT (16, 0);
case 'k': SINT (16, 0);
case 'p': BRANCH (16, 0, 1);
case 'q': BRANCH (16, 0, 1);
}
else
switch (type)
{
case '<': INT_ADJ (3, 2, 8, 0, FALSE);
case '[': INT_ADJ (3, 2, 8, 0, FALSE);
case ']': INT_ADJ (3, 8, 8, 0, FALSE);
case '4': SINT (4, 0);
case '5': UINT (5, 0);
case '6': UINT (6, 5);
case '8': UINT (8, 0);
case 'A': PCREL (8, 0, FALSE, 2, 2, FALSE, FALSE);
case 'B': PCREL (5, 0, FALSE, 3, 3, FALSE, FALSE);
case 'C': INT_ADJ (8, 0, 255, 3, FALSE); /* (0 .. 255) << 3 */
case 'D': INT_ADJ (5, 0, 31, 3, FALSE); /* (0 .. 31) << 3 */
case 'E': PCREL (5, 0, FALSE, 2, 2, FALSE, FALSE);
case 'H': INT_ADJ (5, 0, 31, 1, FALSE); /* (0 .. 31) << 1 */
case 'K': INT_ADJ (8, 0, 127, 3, FALSE); /* (-128 .. 127) << 3 */
case 'U': UINT (8, 0);
case 'V': INT_ADJ (8, 0, 255, 2, FALSE); /* (0 .. 255) << 2 */
case 'W': INT_ADJ (5, 0, 31, 2, FALSE); /* (0 .. 31) << 2 */
case 'j': SINT (5, 0);
case 'k': SINT (8, 0);
case 'p': BRANCH (8, 0, 1);
case 'q': BRANCH (11, 0, 1);
}
return 0;
}
/* This is the opcodes table for the mips16 processor. The format of
this table is intentionally identical to the one in mips-opc.c.
However, the special letters that appear in the argument string are
different, and the table uses some different flags. */
/* Use some short hand macros to keep down the length of the lines in
the opcodes table. */
#define UBD INSN_UNCOND_BRANCH_DELAY
#define WR_1 INSN_WRITE_1
#define WR_2 INSN_WRITE_2
#define RD_1 INSN_READ_1
#define RD_2 INSN_READ_2
#define RD_3 INSN_READ_3
#define RD_4 INSN_READ_4
#define MOD_1 (WR_1|RD_1)
#define MOD_2 (WR_2|RD_2)
#define RD_T INSN_READ_GPR_24
#define WR_T INSN_WRITE_GPR_24
#define WR_31 INSN_WRITE_GPR_31
#define WR_HI INSN_WRITE_HI
#define WR_LO INSN_WRITE_LO
#define RD_HI INSN_READ_HI
#define RD_LO INSN_READ_LO
#define NODS INSN_NO_DELAY_SLOT
#define TRAP INSN_NO_DELAY_SLOT
#define RD_16 INSN2_READ_GPR_16
#define RD_SP INSN2_READ_SP
#define WR_SP INSN2_WRITE_SP
#define MOD_SP (RD_SP|WR_SP)
#define RD_31 INSN2_READ_GPR_31
#define RD_PC INSN2_READ_PC
#define UBR INSN2_UNCOND_BRANCH
#define CBR INSN2_COND_BRANCH
#define I1 INSN_ISA1
#define I3 INSN_ISA3
#define I32 INSN_ISA32
#define I64 INSN_ISA64
#define T3 INSN_3900
const struct mips_opcode mips16_opcodes[] =
{
/* name, args, match, mask, pinfo, pinfo2, membership, ase, exclusions */
{"nop", "", 0x6500, 0xffff, 0, RD_16, I1, 0, 0 }, /* move $0,$Z */
{"la", "x,A", 0x0800, 0xf800, WR_1, RD_PC, I1, 0, 0 },
{"abs", "x,w", 0, (int) M_ABS, INSN_MACRO, 0, I1, 0, 0 },
{"addiu", "y,x,4", 0x4000, 0xf810, WR_1|RD_2, 0, I1, 0, 0 },
{"addiu", "x,k", 0x4800, 0xf800, MOD_1, 0, I1, 0, 0 },
{"addiu", "S,K", 0x6300, 0xff00, 0, MOD_SP, I1, 0, 0 },
{"addiu", "S,S,K", 0x6300, 0xff00, 0, MOD_SP, I1, 0, 0 },
{"addiu", "x,P,V", 0x0800, 0xf800, WR_1, RD_PC, I1, 0, 0 },
{"addiu", "x,S,V", 0x0000, 0xf800, WR_1, RD_SP, I1, 0, 0 },
{"addu", "z,v,y", 0xe001, 0xf803, WR_1|RD_2|RD_3, 0, I1, 0, 0 },
{"addu", "y,x,4", 0x4000, 0xf810, WR_1|RD_2, 0, I1, 0, 0 },
{"addu", "x,k", 0x4800, 0xf800, MOD_1, 0, I1, 0, 0 },
{"addu", "S,K", 0x6300, 0xff00, 0, MOD_SP, I1, 0, 0 },
{"addu", "S,S,K", 0x6300, 0xff00, 0, MOD_SP, I1, 0, 0 },
{"addu", "x,P,V", 0x0800, 0xf800, WR_1, RD_PC, I1, 0, 0 },
{"addu", "x,S,V", 0x0000, 0xf800, WR_1, RD_SP, I1, 0, 0 },
{"and", "x,y", 0xe80c, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"b", "q", 0x1000, 0xf800, 0, UBR, I1, 0, 0 },
{"beq", "x,y,p", 0, (int) M_BEQ, INSN_MACRO, 0, I1, 0, 0 },
{"beq", "x,I,p", 0, (int) M_BEQ_I, INSN_MACRO, 0, I1, 0, 0 },
{"beqz", "x,p", 0x2000, 0xf800, RD_1, CBR, I1, 0, 0 },
{"bge", "x,y,p", 0, (int) M_BGE, INSN_MACRO, 0, I1, 0, 0 },
{"bge", "x,I,p", 0, (int) M_BGE_I, INSN_MACRO, 0, I1, 0, 0 },
{"bgeu", "x,y,p", 0, (int) M_BGEU, INSN_MACRO, 0, I1, 0, 0 },
{"bgeu", "x,I,p", 0, (int) M_BGEU_I, INSN_MACRO, 0, I1, 0, 0 },
{"bgt", "x,y,p", 0, (int) M_BGT, INSN_MACRO, 0, I1, 0, 0 },
{"bgt", "x,I,p", 0, (int) M_BGT_I, INSN_MACRO, 0, I1, 0, 0 },
{"bgtu", "x,y,p", 0, (int) M_BGTU, INSN_MACRO, 0, I1, 0, 0 },
{"bgtu", "x,I,p", 0, (int) M_BGTU_I, INSN_MACRO, 0, I1, 0, 0 },
{"ble", "x,y,p", 0, (int) M_BLE, INSN_MACRO, 0, I1, 0, 0 },
{"ble", "x,I,p", 0, (int) M_BLE_I, INSN_MACRO, 0, I1, 0, 0 },
{"bleu", "x,y,p", 0, (int) M_BLEU, INSN_MACRO, 0, I1, 0, 0 },
{"bleu", "x,I,p", 0, (int) M_BLEU_I, INSN_MACRO, 0, I1, 0, 0 },
{"blt", "x,y,p", 0, (int) M_BLT, INSN_MACRO, 0, I1, 0, 0 },
{"blt", "x,I,p", 0, (int) M_BLT_I, INSN_MACRO, 0, I1, 0, 0 },
{"bltu", "x,y,p", 0, (int) M_BLTU, INSN_MACRO, 0, I1, 0, 0 },
{"bltu", "x,I,p", 0, (int) M_BLTU_I, INSN_MACRO, 0, I1, 0, 0 },
{"bne", "x,y,p", 0, (int) M_BNE, INSN_MACRO, 0, I1, 0, 0 },
{"bne", "x,I,p", 0, (int) M_BNE_I, INSN_MACRO, 0, I1, 0, 0 },
{"bnez", "x,p", 0x2800, 0xf800, RD_1, CBR, I1, 0, 0 },
{"break", "6", 0xe805, 0xf81f, TRAP, 0, I1, 0, 0 },
{"bteqz", "p", 0x6000, 0xff00, RD_T, CBR, I1, 0, 0 },
{"btnez", "p", 0x6100, 0xff00, RD_T, CBR, I1, 0, 0 },
{"cmpi", "x,U", 0x7000, 0xf800, RD_1|WR_T, 0, I1, 0, 0 },
{"cmp", "x,y", 0xe80a, 0xf81f, RD_1|RD_2|WR_T, 0, I1, 0, 0 },
{"cmp", "x,U", 0x7000, 0xf800, RD_1|WR_T, 0, I1, 0, 0 },
{"dla", "y,E", 0xfe00, 0xff00, WR_1, RD_PC, I3, 0, 0 },
{"daddiu", "y,x,4", 0x4010, 0xf810, WR_1|RD_2, 0, I3, 0, 0 },
{"daddiu", "y,j", 0xfd00, 0xff00, MOD_1, 0, I3, 0, 0 },
{"daddiu", "S,K", 0xfb00, 0xff00, 0, MOD_SP, I3, 0, 0 },
{"daddiu", "S,S,K", 0xfb00, 0xff00, 0, MOD_SP, I3, 0, 0 },
{"daddiu", "y,P,W", 0xfe00, 0xff00, WR_1, RD_PC, I3, 0, 0 },
{"daddiu", "y,S,W", 0xff00, 0xff00, WR_1, RD_SP, I3, 0, 0 },
{"daddu", "z,v,y", 0xe000, 0xf803, WR_1|RD_2|RD_3, 0, I3, 0, 0 },
{"daddu", "y,x,4", 0x4010, 0xf810, WR_1|RD_2, 0, I3, 0, 0 },
{"daddu", "y,j", 0xfd00, 0xff00, MOD_1, 0, I3, 0, 0 },
{"daddu", "S,K", 0xfb00, 0xff00, 0, MOD_SP, I3, 0, 0 },
{"daddu", "S,S,K", 0xfb00, 0xff00, 0, MOD_SP, I3, 0, 0 },
{"daddu", "y,P,W", 0xfe00, 0xff00, WR_1, RD_PC, I3, 0, 0 },
{"daddu", "y,S,W", 0xff00, 0xff00, WR_1, RD_SP, I3, 0, 0 },
{"ddiv", "0,x,y", 0xe81e, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I3, 0, 0 },
{"ddiv", "z,v,y", 0, (int) M_DDIV_3, INSN_MACRO, 0, I3, 0, 0 },
{"ddivu", "0,x,y", 0xe81f, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I3, 0, 0 },
{"ddivu", "z,v,y", 0, (int) M_DDIVU_3, INSN_MACRO, 0, I3, 0, 0 },
{"div", "0,x,y", 0xe81a, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I1, 0, 0 },
{"div", "z,v,y", 0, (int) M_DIV_3, INSN_MACRO, 0, I1, 0, 0 },
{"divu", "0,x,y", 0xe81b, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I1, 0, 0 },
{"divu", "z,v,y", 0, (int) M_DIVU_3, INSN_MACRO, 0, I1, 0, 0 },
{"dmul", "z,v,y", 0, (int) M_DMUL, INSN_MACRO, 0, I3, 0, 0 },
{"dmult", "x,y", 0xe81c, 0xf81f, RD_1|RD_2|WR_HI|WR_LO, 0, I3, 0, 0 },
{"dmultu", "x,y", 0xe81d, 0xf81f, RD_1|RD_2|WR_HI|WR_LO, 0, I3, 0, 0 },
{"drem", "0,x,y", 0xe81e, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I3, 0, 0 },
{"drem", "z,v,y", 0, (int) M_DREM_3, INSN_MACRO, 0, I3, 0, 0 },
{"dremu", "0,x,y", 0xe81f, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I3, 0, 0 },
{"dremu", "z,v,y", 0, (int) M_DREMU_3, INSN_MACRO, 0, I3, 0, 0 },
{"dsllv", "y,x", 0xe814, 0xf81f, MOD_1|RD_2, 0, I3, 0, 0 },
{"dsll", "x,w,[", 0x3001, 0xf803, WR_1|RD_2, 0, I3, 0, 0 },
{"dsll", "y,x", 0xe814, 0xf81f, MOD_1|RD_2, 0, I3, 0, 0 },
{"dsrav", "y,x", 0xe817, 0xf81f, MOD_1|RD_2, 0, I3, 0, 0 },
{"dsra", "y,]", 0xe813, 0xf81f, MOD_1, 0, I3, 0, 0 },
{"dsra", "y,x", 0xe817, 0xf81f, MOD_1|RD_2, 0, I3, 0, 0 },
{"dsrlv", "y,x", 0xe816, 0xf81f, MOD_1|RD_2, 0, I3, 0, 0 },
{"dsrl", "y,]", 0xe808, 0xf81f, MOD_1, 0, I3, 0, 0 },
{"dsrl", "y,x", 0xe816, 0xf81f, MOD_1|RD_2, 0, I3, 0, 0 },
{"dsubu", "z,v,y", 0xe002, 0xf803, WR_1|RD_2|RD_3, 0, I3, 0, 0 },
{"dsubu", "y,x,I", 0, (int) M_DSUBU_I, INSN_MACRO, 0, I3, 0, 0 },
{"dsubu", "y,I", 0, (int) M_DSUBU_I_2, INSN_MACRO, 0, I3, 0, 0 },
{"exit", "L", 0xed09, 0xff1f, TRAP, 0, I1, 0, 0 },
{"exit", "L", 0xee09, 0xff1f, TRAP, 0, I1, 0, 0 },
{"exit", "", 0xef09, 0xffff, TRAP, 0, I1, 0, 0 },
{"exit", "L", 0xef09, 0xff1f, TRAP, 0, I1, 0, 0 },
{"entry", "", 0xe809, 0xffff, TRAP, 0, I1, 0, 0 },
{"entry", "l", 0xe809, 0xf81f, TRAP, 0, I1, 0, 0 },
{"extend", "e", 0xf000, 0xf800, 0, 0, I1, 0, 0 },
{"jalr", "x", 0xe840, 0xf8ff, RD_1|WR_31|UBD, 0, I1, 0, 0 },
{"jalr", "R,x", 0xe840, 0xf8ff, RD_2|WR_31|UBD, 0, I1, 0, 0 },
{"jal", "x", 0xe840, 0xf8ff, RD_1|WR_31|UBD, 0, I1, 0, 0 },
{"jal", "R,x", 0xe840, 0xf8ff, RD_2|WR_31|UBD, 0, I1, 0, 0 },
{"jal", "a", 0x1800, 0xfc00, WR_31|UBD, 0, I1, 0, 0 },
{"jalx", "i", 0x1c00, 0xfc00, WR_31|UBD, 0, I1, 0, 0 },
{"jr", "x", 0xe800, 0xf8ff, RD_1|UBD, 0, I1, 0, 0 },
{"jr", "R", 0xe820, 0xffff, UBD, RD_31, I1, 0, 0 },
{"j", "x", 0xe800, 0xf8ff, RD_1|UBD, 0, I1, 0, 0 },
{"j", "R", 0xe820, 0xffff, UBD, RD_31, I1, 0, 0 },
/* MIPS16e compact jumps. We keep them near the ordinary jumps
so that we easily find them when converting a normal jump
to a compact one. */
{"jalrc", "x", 0xe8c0, 0xf8ff, RD_1|WR_31|NODS, UBR, I32, 0, 0 },
{"jalrc", "R,x", 0xe8c0, 0xf8ff, RD_2|WR_31|NODS, UBR, I32, 0, 0 },
{"jrc", "x", 0xe880, 0xf8ff, RD_1|NODS, UBR, I32, 0, 0 },
{"jrc", "R", 0xe8a0, 0xffff, NODS, RD_31|UBR, I32, 0, 0 },
{"lb", "y,5(x)", 0x8000, 0xf800, WR_1|RD_3, 0, I1, 0, 0 },
{"lbu", "y,5(x)", 0xa000, 0xf800, WR_1|RD_3, 0, I1, 0, 0 },
{"ld", "y,D(x)", 0x3800, 0xf800, WR_1|RD_3, 0, I3, 0, 0 },
{"ld", "y,B", 0xfc00, 0xff00, WR_1, RD_PC, I3, 0, 0 },
{"ld", "y,D(P)", 0xfc00, 0xff00, WR_1, RD_PC, I3, 0, 0 },
{"ld", "y,D(S)", 0xf800, 0xff00, WR_1, RD_SP, I3, 0, 0 },
{"lh", "y,H(x)", 0x8800, 0xf800, WR_1|RD_3, 0, I1, 0, 0 },
{"lhu", "y,H(x)", 0xa800, 0xf800, WR_1|RD_3, 0, I1, 0, 0 },
{"li", "x,U", 0x6800, 0xf800, WR_1, 0, I1, 0, 0 },
{"lw", "y,W(x)", 0x9800, 0xf800, WR_1|RD_3, 0, I1, 0, 0 },
{"lw", "x,A", 0xb000, 0xf800, WR_1, RD_PC, I1, 0, 0 },
{"lw", "x,V(P)", 0xb000, 0xf800, WR_1, RD_PC, I1, 0, 0 },
{"lw", "x,V(S)", 0x9000, 0xf800, WR_1, RD_SP, I1, 0, 0 },
{"lwu", "y,W(x)", 0xb800, 0xf800, WR_1|RD_3, 0, I3, 0, 0 },
{"mfhi", "x", 0xe810, 0xf8ff, WR_1|RD_HI, 0, I1, 0, 0 },
{"mflo", "x", 0xe812, 0xf8ff, WR_1|RD_LO, 0, I1, 0, 0 },
{"move", "y,X", 0x6700, 0xff00, WR_1|RD_2, 0, I1, 0, 0 },
{"move", "Y,Z", 0x6500, 0xff00, WR_1|RD_2, 0, I1, 0, 0 },
{"mul", "z,v,y", 0, (int) M_MUL, INSN_MACRO, 0, I1, 0, 0 },
{"mult", "x,y", 0xe818, 0xf81f, RD_1|RD_2|WR_HI|WR_LO, 0, I1, 0, 0 },
{"multu", "x,y", 0xe819, 0xf81f, RD_1|RD_2|WR_HI|WR_LO, 0, I1, 0, 0 },
{"neg", "x,w", 0xe80b, 0xf81f, WR_1|RD_2, 0, I1, 0, 0 },
{"not", "x,w", 0xe80f, 0xf81f, WR_1|RD_2, 0, I1, 0, 0 },
{"or", "x,y", 0xe80d, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"rem", "0,x,y", 0xe81a, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I1, 0, 0 },
{"rem", "z,v,y", 0, (int) M_REM_3, INSN_MACRO, 0, I1, 0, 0 },
{"remu", "0,x,y", 0xe81b, 0xf81f, RD_2|RD_3|WR_HI|WR_LO, 0, I1, 0, 0 },
{"remu", "z,v,y", 0, (int) M_REMU_3, INSN_MACRO, 0, I1, 0, 0 },
{"sb", "y,5(x)", 0xc000, 0xf800, RD_1|RD_3, 0, I1, 0, 0 },
{"sd", "y,D(x)", 0x7800, 0xf800, RD_1|RD_3, 0, I3, 0, 0 },
{"sd", "y,D(S)", 0xf900, 0xff00, RD_1, RD_SP, I3, 0, 0 },
{"sd", "R,C(S)", 0xfa00, 0xff00, 0, RD_31|RD_SP, I3, 0, 0 },
{"sh", "y,H(x)", 0xc800, 0xf800, RD_1|RD_3, 0, I1, 0, 0 },
{"sllv", "y,x", 0xe804, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"sll", "x,w,<", 0x3000, 0xf803, WR_1|RD_2, 0, I1, 0, 0 },
{"sll", "y,x", 0xe804, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"slti", "x,8", 0x5000, 0xf800, RD_1|WR_T, 0, I1, 0, 0 },
{"slt", "x,y", 0xe802, 0xf81f, RD_1|RD_2|WR_T, 0, I1, 0, 0 },
{"slt", "x,8", 0x5000, 0xf800, RD_1|WR_T, 0, I1, 0, 0 },
{"sltiu", "x,8", 0x5800, 0xf800, RD_1|WR_T, 0, I1, 0, 0 },
{"sltu", "x,y", 0xe803, 0xf81f, RD_1|RD_2|WR_T, 0, I1, 0, 0 },
{"sltu", "x,8", 0x5800, 0xf800, RD_1|WR_T, 0, I1, 0, 0 },
{"srav", "y,x", 0xe807, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"sra", "x,w,<", 0x3003, 0xf803, WR_1|RD_2, 0, I1, 0, 0 },
{"sra", "y,x", 0xe807, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"srlv", "y,x", 0xe806, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"srl", "x,w,<", 0x3002, 0xf803, WR_1|RD_2, 0, I1, 0, 0 },
{"srl", "y,x", 0xe806, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
{"subu", "z,v,y", 0xe003, 0xf803, WR_1|RD_2|RD_3, 0, I1, 0, 0 },
{"subu", "y,x,I", 0, (int) M_SUBU_I, INSN_MACRO, 0, I1, 0, 0 },
{"subu", "x,I", 0, (int) M_SUBU_I_2, INSN_MACRO, 0, I1, 0, 0 },
{"sw", "y,W(x)", 0xd800, 0xf800, RD_1|RD_3, 0, I1, 0, 0 },
{"sw", "x,V(S)", 0xd000, 0xf800, RD_1, RD_SP, I1, 0, 0 },
{"sw", "R,V(S)", 0x6200, 0xff00, 0, RD_31|RD_SP, I1, 0, 0 },
{"xor", "x,y", 0xe80e, 0xf81f, MOD_1|RD_2, 0, I1, 0, 0 },
/* MIPS16e additions; see above for compact jumps. */
{"restore", "M", 0x6400, 0xff80, WR_31|NODS, MOD_SP, I32, 0, 0 },
{"save", "m", 0x6480, 0xff80, NODS, RD_31|MOD_SP, I32, 0, 0 },
{"sdbbp", "6", 0xe801, 0xf81f, TRAP, 0, I32, 0, 0 },
{"seb", "x", 0xe891, 0xf8ff, MOD_1, 0, I32, 0, 0 },
{"seh", "x", 0xe8b1, 0xf8ff, MOD_1, 0, I32, 0, 0 },
{"sew", "x", 0xe8d1, 0xf8ff, MOD_1, 0, I64, 0, 0 },
{"zeb", "x", 0xe811, 0xf8ff, MOD_1, 0, I32, 0, 0 },
{"zeh", "x", 0xe831, 0xf8ff, MOD_1, 0, I32, 0, 0 },
{"zew", "x", 0xe851, 0xf8ff, MOD_1, 0, I64, 0, 0 },
};
const int bfd_mips16_num_opcodes =
((sizeof mips16_opcodes) / (sizeof (mips16_opcodes[0])));
|