2f50bcfeaf867804edd3fbc19ffe9104231464b3
[deliverable/binutils-gdb.git] / gas / config / tc-i386.c
1 /* tc-i386.c -- Assemble code for the Intel 80386
2 Copyright 1989, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
3 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007
4 Free Software Foundation, Inc.
5
6 This file is part of GAS, the GNU Assembler.
7
8 GAS is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 2, or (at your option)
11 any later version.
12
13 GAS is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
17
18 You should have received a copy of the GNU General Public License
19 along with GAS; see the file COPYING. If not, write to the Free
20 Software Foundation, 51 Franklin Street - Fifth Floor, Boston, MA
21 02110-1301, USA. */
22
23 /* Intel 80386 machine specific gas.
24 Written by Eliot Dresselhaus (eliot@mgm.mit.edu).
25 x86_64 support by Jan Hubicka (jh@suse.cz)
26 VIA PadLock support by Michal Ludvig (mludvig@suse.cz)
27 Bugs & suggestions are completely welcome. This is free software.
28 Please help us make it better. */
29
30 #include "as.h"
31 #include "safe-ctype.h"
32 #include "subsegs.h"
33 #include "dwarf2dbg.h"
34 #include "dw2gencfi.h"
35 #include "opcode/i386.h"
36 #include "elf/x86-64.h"
37
38 #ifndef REGISTER_WARNINGS
39 #define REGISTER_WARNINGS 1
40 #endif
41
42 #ifndef INFER_ADDR_PREFIX
43 #define INFER_ADDR_PREFIX 1
44 #endif
45
46 #ifndef SCALE1_WHEN_NO_INDEX
47 /* Specifying a scale factor besides 1 when there is no index is
48 futile. eg. `mov (%ebx,2),%al' does exactly the same as
49 `mov (%ebx),%al'. To slavishly follow what the programmer
50 specified, set SCALE1_WHEN_NO_INDEX to 0. */
51 #define SCALE1_WHEN_NO_INDEX 1
52 #endif
53
54 #ifndef DEFAULT_ARCH
55 #define DEFAULT_ARCH "i386"
56 #endif
57
58 #ifndef INLINE
59 #if __GNUC__ >= 2
60 #define INLINE __inline__
61 #else
62 #define INLINE
63 #endif
64 #endif
65
66 static void set_code_flag (int);
67 static void set_16bit_gcc_code_flag (int);
68 static void set_intel_syntax (int);
69 static void set_cpu_arch (int);
70 #ifdef TE_PE
71 static void pe_directive_secrel (int);
72 #endif
73 static void signed_cons (int);
74 static char *output_invalid (int c);
75 static int i386_operand (char *);
76 static int i386_intel_operand (char *, int);
77 static const reg_entry *parse_register (char *, char **);
78 static char *parse_insn (char *, char *);
79 static char *parse_operands (char *, const char *);
80 static void swap_operands (void);
81 static void swap_imm_operands (void);
82 static void optimize_imm (void);
83 static void optimize_disp (void);
84 static int match_template (void);
85 static int check_string (void);
86 static int process_suffix (void);
87 static int check_byte_reg (void);
88 static int check_long_reg (void);
89 static int check_qword_reg (void);
90 static int check_word_reg (void);
91 static int finalize_imm (void);
92 static int process_operands (void);
93 static const seg_entry *build_modrm_byte (void);
94 static void output_insn (void);
95 static void output_imm (fragS *, offsetT);
96 static void output_disp (fragS *, offsetT);
97 #ifndef I386COFF
98 static void s_bss (int);
99 #endif
100 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
101 static void handle_large_common (int small ATTRIBUTE_UNUSED);
102 #endif
103
104 static const char *default_arch = DEFAULT_ARCH;
105
106 /* 'md_assemble ()' gathers together information and puts it into a
107 i386_insn. */
108
109 union i386_op
110 {
111 expressionS *disps;
112 expressionS *imms;
113 const reg_entry *regs;
114 };
115
116 struct _i386_insn
117 {
118 /* TM holds the template for the insn were currently assembling. */
119 template tm;
120
121 /* SUFFIX holds the instruction mnemonic suffix if given.
122 (e.g. 'l' for 'movl') */
123 char suffix;
124
125 /* OPERANDS gives the number of given operands. */
126 unsigned int operands;
127
128 /* REG_OPERANDS, DISP_OPERANDS, MEM_OPERANDS, IMM_OPERANDS give the number
129 of given register, displacement, memory operands and immediate
130 operands. */
131 unsigned int reg_operands, disp_operands, mem_operands, imm_operands;
132
133 /* TYPES [i] is the type (see above #defines) which tells us how to
134 use OP[i] for the corresponding operand. */
135 unsigned int types[MAX_OPERANDS];
136
137 /* Displacement expression, immediate expression, or register for each
138 operand. */
139 union i386_op op[MAX_OPERANDS];
140
141 /* Flags for operands. */
142 unsigned int flags[MAX_OPERANDS];
143 #define Operand_PCrel 1
144
145 /* Relocation type for operand */
146 enum bfd_reloc_code_real reloc[MAX_OPERANDS];
147
148 /* BASE_REG, INDEX_REG, and LOG2_SCALE_FACTOR are used to encode
149 the base index byte below. */
150 const reg_entry *base_reg;
151 const reg_entry *index_reg;
152 unsigned int log2_scale_factor;
153
154 /* SEG gives the seg_entries of this insn. They are zero unless
155 explicit segment overrides are given. */
156 const seg_entry *seg[2];
157
158 /* PREFIX holds all the given prefix opcodes (usually null).
159 PREFIXES is the number of prefix opcodes. */
160 unsigned int prefixes;
161 unsigned char prefix[MAX_PREFIXES];
162
163 /* RM and SIB are the modrm byte and the sib byte where the
164 addressing modes of this insn are encoded. */
165
166 modrm_byte rm;
167 rex_byte rex;
168 sib_byte sib;
169 };
170
171 typedef struct _i386_insn i386_insn;
172
173 /* List of chars besides those in app.c:symbol_chars that can start an
174 operand. Used to prevent the scrubber eating vital white-space. */
175 const char extra_symbol_chars[] = "*%-(["
176 #ifdef LEX_AT
177 "@"
178 #endif
179 #ifdef LEX_QM
180 "?"
181 #endif
182 ;
183
184 #if (defined (TE_I386AIX) \
185 || ((defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)) \
186 && !defined (TE_GNU) \
187 && !defined (TE_LINUX) \
188 && !defined (TE_NETWARE) \
189 && !defined (TE_FreeBSD) \
190 && !defined (TE_NetBSD)))
191 /* This array holds the chars that always start a comment. If the
192 pre-processor is disabled, these aren't very useful. The option
193 --divide will remove '/' from this list. */
194 const char *i386_comment_chars = "#/";
195 #define SVR4_COMMENT_CHARS 1
196 #define PREFIX_SEPARATOR '\\'
197
198 #else
199 const char *i386_comment_chars = "#";
200 #define PREFIX_SEPARATOR '/'
201 #endif
202
203 /* This array holds the chars that only start a comment at the beginning of
204 a line. If the line seems to have the form '# 123 filename'
205 .line and .file directives will appear in the pre-processed output.
206 Note that input_file.c hand checks for '#' at the beginning of the
207 first line of the input file. This is because the compiler outputs
208 #NO_APP at the beginning of its output.
209 Also note that comments started like this one will always work if
210 '/' isn't otherwise defined. */
211 const char line_comment_chars[] = "#/";
212
213 const char line_separator_chars[] = ";";
214
215 /* Chars that can be used to separate mant from exp in floating point
216 nums. */
217 const char EXP_CHARS[] = "eE";
218
219 /* Chars that mean this number is a floating point constant
220 As in 0f12.456
221 or 0d1.2345e12. */
222 const char FLT_CHARS[] = "fFdDxX";
223
224 /* Tables for lexical analysis. */
225 static char mnemonic_chars[256];
226 static char register_chars[256];
227 static char operand_chars[256];
228 static char identifier_chars[256];
229 static char digit_chars[256];
230
231 /* Lexical macros. */
232 #define is_mnemonic_char(x) (mnemonic_chars[(unsigned char) x])
233 #define is_operand_char(x) (operand_chars[(unsigned char) x])
234 #define is_register_char(x) (register_chars[(unsigned char) x])
235 #define is_space_char(x) ((x) == ' ')
236 #define is_identifier_char(x) (identifier_chars[(unsigned char) x])
237 #define is_digit_char(x) (digit_chars[(unsigned char) x])
238
239 /* All non-digit non-letter characters that may occur in an operand. */
240 static char operand_special_chars[] = "%$-+(,)*._~/<>|&^!:[@]";
241
242 /* md_assemble() always leaves the strings it's passed unaltered. To
243 effect this we maintain a stack of saved characters that we've smashed
244 with '\0's (indicating end of strings for various sub-fields of the
245 assembler instruction). */
246 static char save_stack[32];
247 static char *save_stack_p;
248 #define END_STRING_AND_SAVE(s) \
249 do { *save_stack_p++ = *(s); *(s) = '\0'; } while (0)
250 #define RESTORE_END_STRING(s) \
251 do { *(s) = *--save_stack_p; } while (0)
252
253 /* The instruction we're assembling. */
254 static i386_insn i;
255
256 /* Possible templates for current insn. */
257 static const templates *current_templates;
258
259 /* Per instruction expressionS buffers: max displacements & immediates. */
260 static expressionS disp_expressions[MAX_MEMORY_OPERANDS];
261 static expressionS im_expressions[MAX_IMMEDIATE_OPERANDS];
262
263 /* Current operand we are working on. */
264 static int this_operand;
265
266 /* We support four different modes. FLAG_CODE variable is used to distinguish
267 these. */
268
269 enum flag_code {
270 CODE_32BIT,
271 CODE_16BIT,
272 CODE_64BIT };
273 #define NUM_FLAG_CODE ((int) CODE_64BIT + 1)
274
275 static enum flag_code flag_code;
276 static unsigned int object_64bit;
277 static int use_rela_relocations = 0;
278
279 /* The names used to print error messages. */
280 static const char *flag_code_names[] =
281 {
282 "32",
283 "16",
284 "64"
285 };
286
287 /* 1 for intel syntax,
288 0 if att syntax. */
289 static int intel_syntax = 0;
290
291 /* 1 if register prefix % not required. */
292 static int allow_naked_reg = 0;
293
294 /* Register prefix used for error message. */
295 static const char *register_prefix = "%";
296
297 /* Used in 16 bit gcc mode to add an l suffix to call, ret, enter,
298 leave, push, and pop instructions so that gcc has the same stack
299 frame as in 32 bit mode. */
300 static char stackop_size = '\0';
301
302 /* Non-zero to optimize code alignment. */
303 int optimize_align_code = 1;
304
305 /* Non-zero to quieten some warnings. */
306 static int quiet_warnings = 0;
307
308 /* CPU name. */
309 static const char *cpu_arch_name = NULL;
310 static const char *cpu_sub_arch_name = NULL;
311
312 /* CPU feature flags. */
313 static unsigned int cpu_arch_flags = CpuUnknownFlags | CpuNo64;
314
315 /* If we have selected a cpu we are generating instructions for. */
316 static int cpu_arch_tune_set = 0;
317
318 /* Cpu we are generating instructions for. */
319 static enum processor_type cpu_arch_tune = PROCESSOR_UNKNOWN;
320
321 /* CPU feature flags of cpu we are generating instructions for. */
322 static unsigned int cpu_arch_tune_flags = 0;
323
324 /* CPU instruction set architecture used. */
325 static enum processor_type cpu_arch_isa = PROCESSOR_UNKNOWN;
326
327 /* CPU feature flags of instruction set architecture used. */
328 static unsigned int cpu_arch_isa_flags = 0;
329
330 /* If set, conditional jumps are not automatically promoted to handle
331 larger than a byte offset. */
332 static unsigned int no_cond_jump_promotion = 0;
333
334 /* Pre-defined "_GLOBAL_OFFSET_TABLE_". */
335 static symbolS *GOT_symbol;
336
337 /* The dwarf2 return column, adjusted for 32 or 64 bit. */
338 unsigned int x86_dwarf2_return_column;
339
340 /* The dwarf2 data alignment, adjusted for 32 or 64 bit. */
341 int x86_cie_data_alignment;
342
343 /* Interface to relax_segment.
344 There are 3 major relax states for 386 jump insns because the
345 different types of jumps add different sizes to frags when we're
346 figuring out what sort of jump to choose to reach a given label. */
347
348 /* Types. */
349 #define UNCOND_JUMP 0
350 #define COND_JUMP 1
351 #define COND_JUMP86 2
352
353 /* Sizes. */
354 #define CODE16 1
355 #define SMALL 0
356 #define SMALL16 (SMALL | CODE16)
357 #define BIG 2
358 #define BIG16 (BIG | CODE16)
359
360 #ifndef INLINE
361 #ifdef __GNUC__
362 #define INLINE __inline__
363 #else
364 #define INLINE
365 #endif
366 #endif
367
368 #define ENCODE_RELAX_STATE(type, size) \
369 ((relax_substateT) (((type) << 2) | (size)))
370 #define TYPE_FROM_RELAX_STATE(s) \
371 ((s) >> 2)
372 #define DISP_SIZE_FROM_RELAX_STATE(s) \
373 ((((s) & 3) == BIG ? 4 : (((s) & 3) == BIG16 ? 2 : 1)))
374
375 /* This table is used by relax_frag to promote short jumps to long
376 ones where necessary. SMALL (short) jumps may be promoted to BIG
377 (32 bit long) ones, and SMALL16 jumps to BIG16 (16 bit long). We
378 don't allow a short jump in a 32 bit code segment to be promoted to
379 a 16 bit offset jump because it's slower (requires data size
380 prefix), and doesn't work, unless the destination is in the bottom
381 64k of the code segment (The top 16 bits of eip are zeroed). */
382
383 const relax_typeS md_relax_table[] =
384 {
385 /* The fields are:
386 1) most positive reach of this state,
387 2) most negative reach of this state,
388 3) how many bytes this mode will have in the variable part of the frag
389 4) which index into the table to try if we can't fit into this one. */
390
391 /* UNCOND_JUMP states. */
392 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (UNCOND_JUMP, BIG)},
393 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (UNCOND_JUMP, BIG16)},
394 /* dword jmp adds 4 bytes to frag:
395 0 extra opcode bytes, 4 displacement bytes. */
396 {0, 0, 4, 0},
397 /* word jmp adds 2 byte2 to frag:
398 0 extra opcode bytes, 2 displacement bytes. */
399 {0, 0, 2, 0},
400
401 /* COND_JUMP states. */
402 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP, BIG)},
403 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP, BIG16)},
404 /* dword conditionals adds 5 bytes to frag:
405 1 extra opcode byte, 4 displacement bytes. */
406 {0, 0, 5, 0},
407 /* word conditionals add 3 bytes to frag:
408 1 extra opcode byte, 2 displacement bytes. */
409 {0, 0, 3, 0},
410
411 /* COND_JUMP86 states. */
412 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP86, BIG)},
413 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP86, BIG16)},
414 /* dword conditionals adds 5 bytes to frag:
415 1 extra opcode byte, 4 displacement bytes. */
416 {0, 0, 5, 0},
417 /* word conditionals add 4 bytes to frag:
418 1 displacement byte and a 3 byte long branch insn. */
419 {0, 0, 4, 0}
420 };
421
422 static const arch_entry cpu_arch[] =
423 {
424 {"generic32", PROCESSOR_GENERIC32,
425 Cpu186|Cpu286|Cpu386},
426 {"generic64", PROCESSOR_GENERIC64,
427 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
428 |CpuMMX2|CpuSSE|CpuSSE2},
429 {"i8086", PROCESSOR_UNKNOWN,
430 0},
431 {"i186", PROCESSOR_UNKNOWN,
432 Cpu186},
433 {"i286", PROCESSOR_UNKNOWN,
434 Cpu186|Cpu286},
435 {"i386", PROCESSOR_GENERIC32,
436 Cpu186|Cpu286|Cpu386},
437 {"i486", PROCESSOR_I486,
438 Cpu186|Cpu286|Cpu386|Cpu486},
439 {"i586", PROCESSOR_PENTIUM,
440 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586},
441 {"i686", PROCESSOR_PENTIUMPRO,
442 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686},
443 {"pentium", PROCESSOR_PENTIUM,
444 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586},
445 {"pentiumpro",PROCESSOR_PENTIUMPRO,
446 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686},
447 {"pentiumii", PROCESSOR_PENTIUMPRO,
448 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuMMX},
449 {"pentiumiii",PROCESSOR_PENTIUMPRO,
450 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuMMX|CpuMMX2|CpuSSE},
451 {"pentium4", PROCESSOR_PENTIUM4,
452 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
453 |CpuMMX2|CpuSSE|CpuSSE2},
454 {"prescott", PROCESSOR_NOCONA,
455 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
456 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
457 {"nocona", PROCESSOR_NOCONA,
458 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
459 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
460 {"yonah", PROCESSOR_CORE,
461 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
462 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
463 {"core", PROCESSOR_CORE,
464 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
465 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
466 {"merom", PROCESSOR_CORE2,
467 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
468 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3},
469 {"core2", PROCESSOR_CORE2,
470 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
471 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3},
472 {"k6", PROCESSOR_K6,
473 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|CpuK6|CpuMMX},
474 {"k6_2", PROCESSOR_K6,
475 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|CpuK6|CpuMMX|Cpu3dnow},
476 {"athlon", PROCESSOR_ATHLON,
477 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
478 |CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA},
479 {"sledgehammer", PROCESSOR_K8,
480 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
481 |CpuSledgehammer|CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2},
482 {"opteron", PROCESSOR_K8,
483 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
484 |CpuSledgehammer|CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2},
485 {"k8", PROCESSOR_K8,
486 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
487 |CpuSledgehammer|CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2},
488 {"amdfam10", PROCESSOR_AMDFAM10,
489 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6|CpuSledgehammer
490 |CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2|CpuSSE3|CpuSSE4a
491 |CpuABM},
492 {".mmx", PROCESSOR_UNKNOWN,
493 CpuMMX},
494 {".sse", PROCESSOR_UNKNOWN,
495 CpuMMX|CpuMMX2|CpuSSE},
496 {".sse2", PROCESSOR_UNKNOWN,
497 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2},
498 {".sse3", PROCESSOR_UNKNOWN,
499 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
500 {".ssse3", PROCESSOR_UNKNOWN,
501 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3},
502 {".3dnow", PROCESSOR_UNKNOWN,
503 CpuMMX|Cpu3dnow},
504 {".3dnowa", PROCESSOR_UNKNOWN,
505 CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA},
506 {".padlock", PROCESSOR_UNKNOWN,
507 CpuPadLock},
508 {".pacifica", PROCESSOR_UNKNOWN,
509 CpuSVME},
510 {".svme", PROCESSOR_UNKNOWN,
511 CpuSVME},
512 {".sse4a", PROCESSOR_UNKNOWN,
513 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSE4a},
514 {".abm", PROCESSOR_UNKNOWN,
515 CpuABM}
516 };
517
518 const pseudo_typeS md_pseudo_table[] =
519 {
520 #if !defined(OBJ_AOUT) && !defined(USE_ALIGN_PTWO)
521 {"align", s_align_bytes, 0},
522 #else
523 {"align", s_align_ptwo, 0},
524 #endif
525 {"arch", set_cpu_arch, 0},
526 #ifndef I386COFF
527 {"bss", s_bss, 0},
528 #endif
529 {"ffloat", float_cons, 'f'},
530 {"dfloat", float_cons, 'd'},
531 {"tfloat", float_cons, 'x'},
532 {"value", cons, 2},
533 {"slong", signed_cons, 4},
534 {"noopt", s_ignore, 0},
535 {"optim", s_ignore, 0},
536 {"code16gcc", set_16bit_gcc_code_flag, CODE_16BIT},
537 {"code16", set_code_flag, CODE_16BIT},
538 {"code32", set_code_flag, CODE_32BIT},
539 {"code64", set_code_flag, CODE_64BIT},
540 {"intel_syntax", set_intel_syntax, 1},
541 {"att_syntax", set_intel_syntax, 0},
542 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
543 {"largecomm", handle_large_common, 0},
544 #else
545 {"file", (void (*) (int)) dwarf2_directive_file, 0},
546 {"loc", dwarf2_directive_loc, 0},
547 {"loc_mark_labels", dwarf2_directive_loc_mark_labels, 0},
548 #endif
549 #ifdef TE_PE
550 {"secrel32", pe_directive_secrel, 0},
551 #endif
552 {0, 0, 0}
553 };
554
555 /* For interface with expression (). */
556 extern char *input_line_pointer;
557
558 /* Hash table for instruction mnemonic lookup. */
559 static struct hash_control *op_hash;
560
561 /* Hash table for register lookup. */
562 static struct hash_control *reg_hash;
563 \f
564 void
565 i386_align_code (fragS *fragP, int count)
566 {
567 /* Various efficient no-op patterns for aligning code labels.
568 Note: Don't try to assemble the instructions in the comments.
569 0L and 0w are not legal. */
570 static const char f32_1[] =
571 {0x90}; /* nop */
572 static const char f32_2[] =
573 {0x66,0x90}; /* xchg %ax,%ax */
574 static const char f32_3[] =
575 {0x8d,0x76,0x00}; /* leal 0(%esi),%esi */
576 static const char f32_4[] =
577 {0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */
578 static const char f32_5[] =
579 {0x90, /* nop */
580 0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */
581 static const char f32_6[] =
582 {0x8d,0xb6,0x00,0x00,0x00,0x00}; /* leal 0L(%esi),%esi */
583 static const char f32_7[] =
584 {0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */
585 static const char f32_8[] =
586 {0x90, /* nop */
587 0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */
588 static const char f32_9[] =
589 {0x89,0xf6, /* movl %esi,%esi */
590 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
591 static const char f32_10[] =
592 {0x8d,0x76,0x00, /* leal 0(%esi),%esi */
593 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
594 static const char f32_11[] =
595 {0x8d,0x74,0x26,0x00, /* leal 0(%esi,1),%esi */
596 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
597 static const char f32_12[] =
598 {0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */
599 0x8d,0xbf,0x00,0x00,0x00,0x00}; /* leal 0L(%edi),%edi */
600 static const char f32_13[] =
601 {0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */
602 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
603 static const char f32_14[] =
604 {0x8d,0xb4,0x26,0x00,0x00,0x00,0x00, /* leal 0L(%esi,1),%esi */
605 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
606 static const char f32_15[] =
607 {0xeb,0x0d,0x90,0x90,0x90,0x90,0x90, /* jmp .+15; lotsa nops */
608 0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90};
609 static const char f16_3[] =
610 {0x8d,0x74,0x00}; /* lea 0(%esi),%esi */
611 static const char f16_4[] =
612 {0x8d,0xb4,0x00,0x00}; /* lea 0w(%si),%si */
613 static const char f16_5[] =
614 {0x90, /* nop */
615 0x8d,0xb4,0x00,0x00}; /* lea 0w(%si),%si */
616 static const char f16_6[] =
617 {0x89,0xf6, /* mov %si,%si */
618 0x8d,0xbd,0x00,0x00}; /* lea 0w(%di),%di */
619 static const char f16_7[] =
620 {0x8d,0x74,0x00, /* lea 0(%si),%si */
621 0x8d,0xbd,0x00,0x00}; /* lea 0w(%di),%di */
622 static const char f16_8[] =
623 {0x8d,0xb4,0x00,0x00, /* lea 0w(%si),%si */
624 0x8d,0xbd,0x00,0x00}; /* lea 0w(%di),%di */
625 static const char *const f32_patt[] = {
626 f32_1, f32_2, f32_3, f32_4, f32_5, f32_6, f32_7, f32_8,
627 f32_9, f32_10, f32_11, f32_12, f32_13, f32_14, f32_15
628 };
629 static const char *const f16_patt[] = {
630 f32_1, f32_2, f16_3, f16_4, f16_5, f16_6, f16_7, f16_8,
631 f32_15, f32_15, f32_15, f32_15, f32_15, f32_15, f32_15
632 };
633 /* nopl (%[re]ax) */
634 static const char alt_3[] =
635 {0x0f,0x1f,0x00};
636 /* nopl 0(%[re]ax) */
637 static const char alt_4[] =
638 {0x0f,0x1f,0x40,0x00};
639 /* nopl 0(%[re]ax,%[re]ax,1) */
640 static const char alt_5[] =
641 {0x0f,0x1f,0x44,0x00,0x00};
642 /* nopw 0(%[re]ax,%[re]ax,1) */
643 static const char alt_6[] =
644 {0x66,0x0f,0x1f,0x44,0x00,0x00};
645 /* nopl 0L(%[re]ax) */
646 static const char alt_7[] =
647 {0x0f,0x1f,0x80,0x00,0x00,0x00,0x00};
648 /* nopl 0L(%[re]ax,%[re]ax,1) */
649 static const char alt_8[] =
650 {0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
651 /* nopw 0L(%[re]ax,%[re]ax,1) */
652 static const char alt_9[] =
653 {0x66,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
654 /* nopw %cs:0L(%[re]ax,%[re]ax,1) */
655 static const char alt_10[] =
656 {0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
657 /* data16
658 nopw %cs:0L(%[re]ax,%[re]ax,1) */
659 static const char alt_long_11[] =
660 {0x66,
661 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
662 /* data16
663 data16
664 nopw %cs:0L(%[re]ax,%[re]ax,1) */
665 static const char alt_long_12[] =
666 {0x66,
667 0x66,
668 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
669 /* data16
670 data16
671 data16
672 nopw %cs:0L(%[re]ax,%[re]ax,1) */
673 static const char alt_long_13[] =
674 {0x66,
675 0x66,
676 0x66,
677 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
678 /* data16
679 data16
680 data16
681 data16
682 nopw %cs:0L(%[re]ax,%[re]ax,1) */
683 static const char alt_long_14[] =
684 {0x66,
685 0x66,
686 0x66,
687 0x66,
688 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
689 /* data16
690 data16
691 data16
692 data16
693 data16
694 nopw %cs:0L(%[re]ax,%[re]ax,1) */
695 static const char alt_long_15[] =
696 {0x66,
697 0x66,
698 0x66,
699 0x66,
700 0x66,
701 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
702 /* nopl 0(%[re]ax,%[re]ax,1)
703 nopw 0(%[re]ax,%[re]ax,1) */
704 static const char alt_short_11[] =
705 {0x0f,0x1f,0x44,0x00,0x00,
706 0x66,0x0f,0x1f,0x44,0x00,0x00};
707 /* nopw 0(%[re]ax,%[re]ax,1)
708 nopw 0(%[re]ax,%[re]ax,1) */
709 static const char alt_short_12[] =
710 {0x66,0x0f,0x1f,0x44,0x00,0x00,
711 0x66,0x0f,0x1f,0x44,0x00,0x00};
712 /* nopw 0(%[re]ax,%[re]ax,1)
713 nopl 0L(%[re]ax) */
714 static const char alt_short_13[] =
715 {0x66,0x0f,0x1f,0x44,0x00,0x00,
716 0x0f,0x1f,0x80,0x00,0x00,0x00,0x00};
717 /* nopl 0L(%[re]ax)
718 nopl 0L(%[re]ax) */
719 static const char alt_short_14[] =
720 {0x0f,0x1f,0x80,0x00,0x00,0x00,0x00,
721 0x0f,0x1f,0x80,0x00,0x00,0x00,0x00};
722 /* nopl 0L(%[re]ax)
723 nopl 0L(%[re]ax,%[re]ax,1) */
724 static const char alt_short_15[] =
725 {0x0f,0x1f,0x80,0x00,0x00,0x00,0x00,
726 0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
727 static const char *const alt_short_patt[] = {
728 f32_1, f32_2, alt_3, alt_4, alt_5, alt_6, alt_7, alt_8,
729 alt_9, alt_10, alt_short_11, alt_short_12, alt_short_13,
730 alt_short_14, alt_short_15
731 };
732 static const char *const alt_long_patt[] = {
733 f32_1, f32_2, alt_3, alt_4, alt_5, alt_6, alt_7, alt_8,
734 alt_9, alt_10, alt_long_11, alt_long_12, alt_long_13,
735 alt_long_14, alt_long_15
736 };
737
738 if (count <= 0 || count > 15)
739 return;
740
741 /* We need to decide which NOP sequence to use for 32bit and
742 64bit. When -mtune= is used:
743
744 1. For PROCESSOR_I486, PROCESSOR_PENTIUM and PROCESSOR_GENERIC32,
745 f32_patt will be used.
746 2. For PROCESSOR_K8 and PROCESSOR_AMDFAM10 in 64bit, NOPs with 0x66 prefix will be used.
747 3. For PROCESSOR_CORE2, alt_long_patt will be used.
748 4. For PROCESSOR_PENTIUMPRO, PROCESSOR_PENTIUM4, PROCESSOR_NOCONA,
749 PROCESSOR_CORE, PROCESSOR_CORE2, PROCESSOR_K6, PROCESSOR_ATHLON
750 and PROCESSOR_GENERIC64, alt_short_patt will be used.
751
752 When -mtune= isn't used, alt_short_patt will be used if
753 cpu_arch_isa_flags has Cpu686. Otherwise, f32_patt will be used.
754
755 When -march= or .arch is used, we can't use anything beyond
756 cpu_arch_isa_flags. */
757
758 if (flag_code == CODE_16BIT)
759 {
760 memcpy (fragP->fr_literal + fragP->fr_fix,
761 f16_patt[count - 1], count);
762 if (count > 8)
763 /* Adjust jump offset. */
764 fragP->fr_literal[fragP->fr_fix + 1] = count - 2;
765 }
766 else if (flag_code == CODE_64BIT && cpu_arch_tune == PROCESSOR_K8)
767 {
768 int i;
769 int nnops = (count + 3) / 4;
770 int len = count / nnops;
771 int remains = count - nnops * len;
772 int pos = 0;
773
774 /* The recommended way to pad 64bit code is to use NOPs preceded
775 by maximally four 0x66 prefixes. Balance the size of nops. */
776 for (i = 0; i < remains; i++)
777 {
778 memset (fragP->fr_literal + fragP->fr_fix + pos, 0x66, len);
779 fragP->fr_literal[fragP->fr_fix + pos + len] = 0x90;
780 pos += len + 1;
781 }
782 for (; i < nnops; i++)
783 {
784 memset (fragP->fr_literal + fragP->fr_fix + pos, 0x66, len - 1);
785 fragP->fr_literal[fragP->fr_fix + pos + len - 1] = 0x90;
786 pos += len;
787 }
788 }
789 else
790 {
791 const char *const *patt = NULL;
792
793 if (cpu_arch_isa == PROCESSOR_UNKNOWN)
794 {
795 /* PROCESSOR_UNKNOWN means that all ISAs may be used. */
796 switch (cpu_arch_tune)
797 {
798 case PROCESSOR_UNKNOWN:
799 /* We use cpu_arch_isa_flags to check if we SHOULD
800 optimize for Cpu686. */
801 if ((cpu_arch_isa_flags & Cpu686) != 0)
802 patt = alt_short_patt;
803 else
804 patt = f32_patt;
805 break;
806 case PROCESSOR_CORE2:
807 patt = alt_long_patt;
808 break;
809 case PROCESSOR_PENTIUMPRO:
810 case PROCESSOR_PENTIUM4:
811 case PROCESSOR_NOCONA:
812 case PROCESSOR_CORE:
813 case PROCESSOR_K6:
814 case PROCESSOR_ATHLON:
815 case PROCESSOR_K8:
816 case PROCESSOR_GENERIC64:
817 case PROCESSOR_AMDFAM10:
818 patt = alt_short_patt;
819 break;
820 case PROCESSOR_I486:
821 case PROCESSOR_PENTIUM:
822 case PROCESSOR_GENERIC32:
823 patt = f32_patt;
824 break;
825 }
826 }
827 else
828 {
829 switch (cpu_arch_tune)
830 {
831 case PROCESSOR_UNKNOWN:
832 /* When cpu_arch_isa is net, cpu_arch_tune shouldn't be
833 PROCESSOR_UNKNOWN. */
834 abort ();
835 break;
836
837 case PROCESSOR_I486:
838 case PROCESSOR_PENTIUM:
839 case PROCESSOR_PENTIUMPRO:
840 case PROCESSOR_PENTIUM4:
841 case PROCESSOR_NOCONA:
842 case PROCESSOR_CORE:
843 case PROCESSOR_K6:
844 case PROCESSOR_ATHLON:
845 case PROCESSOR_K8:
846 case PROCESSOR_AMDFAM10:
847 case PROCESSOR_GENERIC32:
848 /* We use cpu_arch_isa_flags to check if we CAN optimize
849 for Cpu686. */
850 if ((cpu_arch_isa_flags & Cpu686) != 0)
851 patt = alt_short_patt;
852 else
853 patt = f32_patt;
854 break;
855 case PROCESSOR_CORE2:
856 if ((cpu_arch_isa_flags & Cpu686) != 0)
857 patt = alt_long_patt;
858 else
859 patt = f32_patt;
860 break;
861 case PROCESSOR_GENERIC64:
862 patt = alt_short_patt;
863 break;
864 }
865 }
866
867 memcpy (fragP->fr_literal + fragP->fr_fix,
868 patt[count - 1], count);
869 }
870 fragP->fr_var = count;
871 }
872
873 static INLINE unsigned int
874 mode_from_disp_size (unsigned int t)
875 {
876 return (t & Disp8) ? 1 : (t & (Disp16 | Disp32 | Disp32S)) ? 2 : 0;
877 }
878
879 static INLINE int
880 fits_in_signed_byte (offsetT num)
881 {
882 return (num >= -128) && (num <= 127);
883 }
884
885 static INLINE int
886 fits_in_unsigned_byte (offsetT num)
887 {
888 return (num & 0xff) == num;
889 }
890
891 static INLINE int
892 fits_in_unsigned_word (offsetT num)
893 {
894 return (num & 0xffff) == num;
895 }
896
897 static INLINE int
898 fits_in_signed_word (offsetT num)
899 {
900 return (-32768 <= num) && (num <= 32767);
901 }
902
903 static INLINE int
904 fits_in_signed_long (offsetT num ATTRIBUTE_UNUSED)
905 {
906 #ifndef BFD64
907 return 1;
908 #else
909 return (!(((offsetT) -1 << 31) & num)
910 || (((offsetT) -1 << 31) & num) == ((offsetT) -1 << 31));
911 #endif
912 } /* fits_in_signed_long() */
913
914 static INLINE int
915 fits_in_unsigned_long (offsetT num ATTRIBUTE_UNUSED)
916 {
917 #ifndef BFD64
918 return 1;
919 #else
920 return (num & (((offsetT) 2 << 31) - 1)) == num;
921 #endif
922 } /* fits_in_unsigned_long() */
923
924 static unsigned int
925 smallest_imm_type (offsetT num)
926 {
927 if (cpu_arch_flags != (Cpu186 | Cpu286 | Cpu386 | Cpu486 | CpuNo64))
928 {
929 /* This code is disabled on the 486 because all the Imm1 forms
930 in the opcode table are slower on the i486. They're the
931 versions with the implicitly specified single-position
932 displacement, which has another syntax if you really want to
933 use that form. */
934 if (num == 1)
935 return Imm1 | Imm8 | Imm8S | Imm16 | Imm32 | Imm32S | Imm64;
936 }
937 return (fits_in_signed_byte (num)
938 ? (Imm8S | Imm8 | Imm16 | Imm32 | Imm32S | Imm64)
939 : fits_in_unsigned_byte (num)
940 ? (Imm8 | Imm16 | Imm32 | Imm32S | Imm64)
941 : (fits_in_signed_word (num) || fits_in_unsigned_word (num))
942 ? (Imm16 | Imm32 | Imm32S | Imm64)
943 : fits_in_signed_long (num)
944 ? (Imm32 | Imm32S | Imm64)
945 : fits_in_unsigned_long (num)
946 ? (Imm32 | Imm64)
947 : Imm64);
948 }
949
950 static offsetT
951 offset_in_range (offsetT val, int size)
952 {
953 addressT mask;
954
955 switch (size)
956 {
957 case 1: mask = ((addressT) 1 << 8) - 1; break;
958 case 2: mask = ((addressT) 1 << 16) - 1; break;
959 case 4: mask = ((addressT) 2 << 31) - 1; break;
960 #ifdef BFD64
961 case 8: mask = ((addressT) 2 << 63) - 1; break;
962 #endif
963 default: abort ();
964 }
965
966 /* If BFD64, sign extend val. */
967 if (!use_rela_relocations)
968 if ((val & ~(((addressT) 2 << 31) - 1)) == 0)
969 val = (val ^ ((addressT) 1 << 31)) - ((addressT) 1 << 31);
970
971 if ((val & ~mask) != 0 && (val & ~mask) != ~mask)
972 {
973 char buf1[40], buf2[40];
974
975 sprint_value (buf1, val);
976 sprint_value (buf2, val & mask);
977 as_warn (_("%s shortened to %s"), buf1, buf2);
978 }
979 return val & mask;
980 }
981
982 /* Returns 0 if attempting to add a prefix where one from the same
983 class already exists, 1 if non rep/repne added, 2 if rep/repne
984 added. */
985 static int
986 add_prefix (unsigned int prefix)
987 {
988 int ret = 1;
989 unsigned int q;
990
991 if (prefix >= REX_OPCODE && prefix < REX_OPCODE + 16
992 && flag_code == CODE_64BIT)
993 {
994 if ((i.prefix[REX_PREFIX] & prefix & REX_MODE64)
995 || ((i.prefix[REX_PREFIX] & (REX_EXTX | REX_EXTY | REX_EXTZ))
996 && (prefix & (REX_EXTX | REX_EXTY | REX_EXTZ))))
997 ret = 0;
998 q = REX_PREFIX;
999 }
1000 else
1001 {
1002 switch (prefix)
1003 {
1004 default:
1005 abort ();
1006
1007 case CS_PREFIX_OPCODE:
1008 case DS_PREFIX_OPCODE:
1009 case ES_PREFIX_OPCODE:
1010 case FS_PREFIX_OPCODE:
1011 case GS_PREFIX_OPCODE:
1012 case SS_PREFIX_OPCODE:
1013 q = SEG_PREFIX;
1014 break;
1015
1016 case REPNE_PREFIX_OPCODE:
1017 case REPE_PREFIX_OPCODE:
1018 ret = 2;
1019 /* fall thru */
1020 case LOCK_PREFIX_OPCODE:
1021 q = LOCKREP_PREFIX;
1022 break;
1023
1024 case FWAIT_OPCODE:
1025 q = WAIT_PREFIX;
1026 break;
1027
1028 case ADDR_PREFIX_OPCODE:
1029 q = ADDR_PREFIX;
1030 break;
1031
1032 case DATA_PREFIX_OPCODE:
1033 q = DATA_PREFIX;
1034 break;
1035 }
1036 if (i.prefix[q] != 0)
1037 ret = 0;
1038 }
1039
1040 if (ret)
1041 {
1042 if (!i.prefix[q])
1043 ++i.prefixes;
1044 i.prefix[q] |= prefix;
1045 }
1046 else
1047 as_bad (_("same type of prefix used twice"));
1048
1049 return ret;
1050 }
1051
1052 static void
1053 set_code_flag (int value)
1054 {
1055 flag_code = value;
1056 cpu_arch_flags &= ~(Cpu64 | CpuNo64);
1057 cpu_arch_flags |= (flag_code == CODE_64BIT ? Cpu64 : CpuNo64);
1058 if (value == CODE_64BIT && !(cpu_arch_flags & CpuSledgehammer))
1059 {
1060 as_bad (_("64bit mode not supported on this CPU."));
1061 }
1062 if (value == CODE_32BIT && !(cpu_arch_flags & Cpu386))
1063 {
1064 as_bad (_("32bit mode not supported on this CPU."));
1065 }
1066 stackop_size = '\0';
1067 }
1068
1069 static void
1070 set_16bit_gcc_code_flag (int new_code_flag)
1071 {
1072 flag_code = new_code_flag;
1073 cpu_arch_flags &= ~(Cpu64 | CpuNo64);
1074 cpu_arch_flags |= (flag_code == CODE_64BIT ? Cpu64 : CpuNo64);
1075 stackop_size = LONG_MNEM_SUFFIX;
1076 }
1077
1078 static void
1079 set_intel_syntax (int syntax_flag)
1080 {
1081 /* Find out if register prefixing is specified. */
1082 int ask_naked_reg = 0;
1083
1084 SKIP_WHITESPACE ();
1085 if (!is_end_of_line[(unsigned char) *input_line_pointer])
1086 {
1087 char *string = input_line_pointer;
1088 int e = get_symbol_end ();
1089
1090 if (strcmp (string, "prefix") == 0)
1091 ask_naked_reg = 1;
1092 else if (strcmp (string, "noprefix") == 0)
1093 ask_naked_reg = -1;
1094 else
1095 as_bad (_("bad argument to syntax directive."));
1096 *input_line_pointer = e;
1097 }
1098 demand_empty_rest_of_line ();
1099
1100 intel_syntax = syntax_flag;
1101
1102 if (ask_naked_reg == 0)
1103 allow_naked_reg = (intel_syntax
1104 && (bfd_get_symbol_leading_char (stdoutput) != '\0'));
1105 else
1106 allow_naked_reg = (ask_naked_reg < 0);
1107
1108 identifier_chars['%'] = intel_syntax && allow_naked_reg ? '%' : 0;
1109 identifier_chars['$'] = intel_syntax ? '$' : 0;
1110 register_prefix = allow_naked_reg ? "" : "%";
1111 }
1112
1113 static void
1114 set_cpu_arch (int dummy ATTRIBUTE_UNUSED)
1115 {
1116 SKIP_WHITESPACE ();
1117
1118 if (!is_end_of_line[(unsigned char) *input_line_pointer])
1119 {
1120 char *string = input_line_pointer;
1121 int e = get_symbol_end ();
1122 unsigned int i;
1123
1124 for (i = 0; i < ARRAY_SIZE (cpu_arch); i++)
1125 {
1126 if (strcmp (string, cpu_arch[i].name) == 0)
1127 {
1128 if (*string != '.')
1129 {
1130 cpu_arch_name = cpu_arch[i].name;
1131 cpu_sub_arch_name = NULL;
1132 cpu_arch_flags = (cpu_arch[i].flags
1133 | (flag_code == CODE_64BIT ? Cpu64 : CpuNo64));
1134 cpu_arch_isa = cpu_arch[i].type;
1135 cpu_arch_isa_flags = cpu_arch[i].flags;
1136 if (!cpu_arch_tune_set)
1137 {
1138 cpu_arch_tune = cpu_arch_isa;
1139 cpu_arch_tune_flags = cpu_arch_isa_flags;
1140 }
1141 break;
1142 }
1143 if ((cpu_arch_flags | cpu_arch[i].flags) != cpu_arch_flags)
1144 {
1145 cpu_sub_arch_name = cpu_arch[i].name;
1146 cpu_arch_flags |= cpu_arch[i].flags;
1147 }
1148 *input_line_pointer = e;
1149 demand_empty_rest_of_line ();
1150 return;
1151 }
1152 }
1153 if (i >= ARRAY_SIZE (cpu_arch))
1154 as_bad (_("no such architecture: `%s'"), string);
1155
1156 *input_line_pointer = e;
1157 }
1158 else
1159 as_bad (_("missing cpu architecture"));
1160
1161 no_cond_jump_promotion = 0;
1162 if (*input_line_pointer == ','
1163 && !is_end_of_line[(unsigned char) input_line_pointer[1]])
1164 {
1165 char *string = ++input_line_pointer;
1166 int e = get_symbol_end ();
1167
1168 if (strcmp (string, "nojumps") == 0)
1169 no_cond_jump_promotion = 1;
1170 else if (strcmp (string, "jumps") == 0)
1171 ;
1172 else
1173 as_bad (_("no such architecture modifier: `%s'"), string);
1174
1175 *input_line_pointer = e;
1176 }
1177
1178 demand_empty_rest_of_line ();
1179 }
1180
1181 unsigned long
1182 i386_mach ()
1183 {
1184 if (!strcmp (default_arch, "x86_64"))
1185 return bfd_mach_x86_64;
1186 else if (!strcmp (default_arch, "i386"))
1187 return bfd_mach_i386_i386;
1188 else
1189 as_fatal (_("Unknown architecture"));
1190 }
1191 \f
1192 void
1193 md_begin ()
1194 {
1195 const char *hash_err;
1196
1197 /* Initialize op_hash hash table. */
1198 op_hash = hash_new ();
1199
1200 {
1201 const template *optab;
1202 templates *core_optab;
1203
1204 /* Setup for loop. */
1205 optab = i386_optab;
1206 core_optab = (templates *) xmalloc (sizeof (templates));
1207 core_optab->start = optab;
1208
1209 while (1)
1210 {
1211 ++optab;
1212 if (optab->name == NULL
1213 || strcmp (optab->name, (optab - 1)->name) != 0)
1214 {
1215 /* different name --> ship out current template list;
1216 add to hash table; & begin anew. */
1217 core_optab->end = optab;
1218 hash_err = hash_insert (op_hash,
1219 (optab - 1)->name,
1220 (PTR) core_optab);
1221 if (hash_err)
1222 {
1223 as_fatal (_("Internal Error: Can't hash %s: %s"),
1224 (optab - 1)->name,
1225 hash_err);
1226 }
1227 if (optab->name == NULL)
1228 break;
1229 core_optab = (templates *) xmalloc (sizeof (templates));
1230 core_optab->start = optab;
1231 }
1232 }
1233 }
1234
1235 /* Initialize reg_hash hash table. */
1236 reg_hash = hash_new ();
1237 {
1238 const reg_entry *regtab;
1239
1240 for (regtab = i386_regtab;
1241 regtab < i386_regtab + sizeof (i386_regtab) / sizeof (i386_regtab[0]);
1242 regtab++)
1243 {
1244 hash_err = hash_insert (reg_hash, regtab->reg_name, (PTR) regtab);
1245 if (hash_err)
1246 as_fatal (_("Internal Error: Can't hash %s: %s"),
1247 regtab->reg_name,
1248 hash_err);
1249 }
1250 }
1251
1252 /* Fill in lexical tables: mnemonic_chars, operand_chars. */
1253 {
1254 int c;
1255 char *p;
1256
1257 for (c = 0; c < 256; c++)
1258 {
1259 if (ISDIGIT (c))
1260 {
1261 digit_chars[c] = c;
1262 mnemonic_chars[c] = c;
1263 register_chars[c] = c;
1264 operand_chars[c] = c;
1265 }
1266 else if (ISLOWER (c))
1267 {
1268 mnemonic_chars[c] = c;
1269 register_chars[c] = c;
1270 operand_chars[c] = c;
1271 }
1272 else if (ISUPPER (c))
1273 {
1274 mnemonic_chars[c] = TOLOWER (c);
1275 register_chars[c] = mnemonic_chars[c];
1276 operand_chars[c] = c;
1277 }
1278
1279 if (ISALPHA (c) || ISDIGIT (c))
1280 identifier_chars[c] = c;
1281 else if (c >= 128)
1282 {
1283 identifier_chars[c] = c;
1284 operand_chars[c] = c;
1285 }
1286 }
1287
1288 #ifdef LEX_AT
1289 identifier_chars['@'] = '@';
1290 #endif
1291 #ifdef LEX_QM
1292 identifier_chars['?'] = '?';
1293 operand_chars['?'] = '?';
1294 #endif
1295 digit_chars['-'] = '-';
1296 mnemonic_chars['-'] = '-';
1297 identifier_chars['_'] = '_';
1298 identifier_chars['.'] = '.';
1299
1300 for (p = operand_special_chars; *p != '\0'; p++)
1301 operand_chars[(unsigned char) *p] = *p;
1302 }
1303
1304 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
1305 if (IS_ELF)
1306 {
1307 record_alignment (text_section, 2);
1308 record_alignment (data_section, 2);
1309 record_alignment (bss_section, 2);
1310 }
1311 #endif
1312
1313 if (flag_code == CODE_64BIT)
1314 {
1315 x86_dwarf2_return_column = 16;
1316 x86_cie_data_alignment = -8;
1317 }
1318 else
1319 {
1320 x86_dwarf2_return_column = 8;
1321 x86_cie_data_alignment = -4;
1322 }
1323 }
1324
1325 void
1326 i386_print_statistics (FILE *file)
1327 {
1328 hash_print_statistics (file, "i386 opcode", op_hash);
1329 hash_print_statistics (file, "i386 register", reg_hash);
1330 }
1331 \f
1332 #ifdef DEBUG386
1333
1334 /* Debugging routines for md_assemble. */
1335 static void pte (template *);
1336 static void pt (unsigned int);
1337 static void pe (expressionS *);
1338 static void ps (symbolS *);
1339
1340 static void
1341 pi (char *line, i386_insn *x)
1342 {
1343 unsigned int i;
1344
1345 fprintf (stdout, "%s: template ", line);
1346 pte (&x->tm);
1347 fprintf (stdout, " address: base %s index %s scale %x\n",
1348 x->base_reg ? x->base_reg->reg_name : "none",
1349 x->index_reg ? x->index_reg->reg_name : "none",
1350 x->log2_scale_factor);
1351 fprintf (stdout, " modrm: mode %x reg %x reg/mem %x\n",
1352 x->rm.mode, x->rm.reg, x->rm.regmem);
1353 fprintf (stdout, " sib: base %x index %x scale %x\n",
1354 x->sib.base, x->sib.index, x->sib.scale);
1355 fprintf (stdout, " rex: 64bit %x extX %x extY %x extZ %x\n",
1356 (x->rex & REX_MODE64) != 0,
1357 (x->rex & REX_EXTX) != 0,
1358 (x->rex & REX_EXTY) != 0,
1359 (x->rex & REX_EXTZ) != 0);
1360 for (i = 0; i < x->operands; i++)
1361 {
1362 fprintf (stdout, " #%d: ", i + 1);
1363 pt (x->types[i]);
1364 fprintf (stdout, "\n");
1365 if (x->types[i]
1366 & (Reg | SReg2 | SReg3 | Control | Debug | Test | RegMMX | RegXMM))
1367 fprintf (stdout, "%s\n", x->op[i].regs->reg_name);
1368 if (x->types[i] & Imm)
1369 pe (x->op[i].imms);
1370 if (x->types[i] & Disp)
1371 pe (x->op[i].disps);
1372 }
1373 }
1374
1375 static void
1376 pte (template *t)
1377 {
1378 unsigned int i;
1379 fprintf (stdout, " %d operands ", t->operands);
1380 fprintf (stdout, "opcode %x ", t->base_opcode);
1381 if (t->extension_opcode != None)
1382 fprintf (stdout, "ext %x ", t->extension_opcode);
1383 if (t->opcode_modifier & D)
1384 fprintf (stdout, "D");
1385 if (t->opcode_modifier & W)
1386 fprintf (stdout, "W");
1387 fprintf (stdout, "\n");
1388 for (i = 0; i < t->operands; i++)
1389 {
1390 fprintf (stdout, " #%d type ", i + 1);
1391 pt (t->operand_types[i]);
1392 fprintf (stdout, "\n");
1393 }
1394 }
1395
1396 static void
1397 pe (expressionS *e)
1398 {
1399 fprintf (stdout, " operation %d\n", e->X_op);
1400 fprintf (stdout, " add_number %ld (%lx)\n",
1401 (long) e->X_add_number, (long) e->X_add_number);
1402 if (e->X_add_symbol)
1403 {
1404 fprintf (stdout, " add_symbol ");
1405 ps (e->X_add_symbol);
1406 fprintf (stdout, "\n");
1407 }
1408 if (e->X_op_symbol)
1409 {
1410 fprintf (stdout, " op_symbol ");
1411 ps (e->X_op_symbol);
1412 fprintf (stdout, "\n");
1413 }
1414 }
1415
1416 static void
1417 ps (symbolS *s)
1418 {
1419 fprintf (stdout, "%s type %s%s",
1420 S_GET_NAME (s),
1421 S_IS_EXTERNAL (s) ? "EXTERNAL " : "",
1422 segment_name (S_GET_SEGMENT (s)));
1423 }
1424
1425 static struct type_name
1426 {
1427 unsigned int mask;
1428 char *tname;
1429 }
1430 const type_names[] =
1431 {
1432 { Reg8, "r8" },
1433 { Reg16, "r16" },
1434 { Reg32, "r32" },
1435 { Reg64, "r64" },
1436 { Imm8, "i8" },
1437 { Imm8S, "i8s" },
1438 { Imm16, "i16" },
1439 { Imm32, "i32" },
1440 { Imm32S, "i32s" },
1441 { Imm64, "i64" },
1442 { Imm1, "i1" },
1443 { BaseIndex, "BaseIndex" },
1444 { Disp8, "d8" },
1445 { Disp16, "d16" },
1446 { Disp32, "d32" },
1447 { Disp32S, "d32s" },
1448 { Disp64, "d64" },
1449 { InOutPortReg, "InOutPortReg" },
1450 { ShiftCount, "ShiftCount" },
1451 { Control, "control reg" },
1452 { Test, "test reg" },
1453 { Debug, "debug reg" },
1454 { FloatReg, "FReg" },
1455 { FloatAcc, "FAcc" },
1456 { SReg2, "SReg2" },
1457 { SReg3, "SReg3" },
1458 { Acc, "Acc" },
1459 { JumpAbsolute, "Jump Absolute" },
1460 { RegMMX, "rMMX" },
1461 { RegXMM, "rXMM" },
1462 { EsSeg, "es" },
1463 { 0, "" }
1464 };
1465
1466 static void
1467 pt (t)
1468 unsigned int t;
1469 {
1470 const struct type_name *ty;
1471
1472 for (ty = type_names; ty->mask; ty++)
1473 if (t & ty->mask)
1474 fprintf (stdout, "%s, ", ty->tname);
1475 fflush (stdout);
1476 }
1477
1478 #endif /* DEBUG386 */
1479 \f
1480 static bfd_reloc_code_real_type
1481 reloc (unsigned int size,
1482 int pcrel,
1483 int sign,
1484 bfd_reloc_code_real_type other)
1485 {
1486 if (other != NO_RELOC)
1487 {
1488 reloc_howto_type *reloc;
1489
1490 if (size == 8)
1491 switch (other)
1492 {
1493 case BFD_RELOC_X86_64_GOT32:
1494 return BFD_RELOC_X86_64_GOT64;
1495 break;
1496 case BFD_RELOC_X86_64_PLTOFF64:
1497 return BFD_RELOC_X86_64_PLTOFF64;
1498 break;
1499 case BFD_RELOC_X86_64_GOTPC32:
1500 other = BFD_RELOC_X86_64_GOTPC64;
1501 break;
1502 case BFD_RELOC_X86_64_GOTPCREL:
1503 other = BFD_RELOC_X86_64_GOTPCREL64;
1504 break;
1505 case BFD_RELOC_X86_64_TPOFF32:
1506 other = BFD_RELOC_X86_64_TPOFF64;
1507 break;
1508 case BFD_RELOC_X86_64_DTPOFF32:
1509 other = BFD_RELOC_X86_64_DTPOFF64;
1510 break;
1511 default:
1512 break;
1513 }
1514
1515 /* Sign-checking 4-byte relocations in 16-/32-bit code is pointless. */
1516 if (size == 4 && flag_code != CODE_64BIT)
1517 sign = -1;
1518
1519 reloc = bfd_reloc_type_lookup (stdoutput, other);
1520 if (!reloc)
1521 as_bad (_("unknown relocation (%u)"), other);
1522 else if (size != bfd_get_reloc_size (reloc))
1523 as_bad (_("%u-byte relocation cannot be applied to %u-byte field"),
1524 bfd_get_reloc_size (reloc),
1525 size);
1526 else if (pcrel && !reloc->pc_relative)
1527 as_bad (_("non-pc-relative relocation for pc-relative field"));
1528 else if ((reloc->complain_on_overflow == complain_overflow_signed
1529 && !sign)
1530 || (reloc->complain_on_overflow == complain_overflow_unsigned
1531 && sign > 0))
1532 as_bad (_("relocated field and relocation type differ in signedness"));
1533 else
1534 return other;
1535 return NO_RELOC;
1536 }
1537
1538 if (pcrel)
1539 {
1540 if (!sign)
1541 as_bad (_("there are no unsigned pc-relative relocations"));
1542 switch (size)
1543 {
1544 case 1: return BFD_RELOC_8_PCREL;
1545 case 2: return BFD_RELOC_16_PCREL;
1546 case 4: return BFD_RELOC_32_PCREL;
1547 case 8: return BFD_RELOC_64_PCREL;
1548 }
1549 as_bad (_("cannot do %u byte pc-relative relocation"), size);
1550 }
1551 else
1552 {
1553 if (sign > 0)
1554 switch (size)
1555 {
1556 case 4: return BFD_RELOC_X86_64_32S;
1557 }
1558 else
1559 switch (size)
1560 {
1561 case 1: return BFD_RELOC_8;
1562 case 2: return BFD_RELOC_16;
1563 case 4: return BFD_RELOC_32;
1564 case 8: return BFD_RELOC_64;
1565 }
1566 as_bad (_("cannot do %s %u byte relocation"),
1567 sign > 0 ? "signed" : "unsigned", size);
1568 }
1569
1570 abort ();
1571 return BFD_RELOC_NONE;
1572 }
1573
1574 /* Here we decide which fixups can be adjusted to make them relative to
1575 the beginning of the section instead of the symbol. Basically we need
1576 to make sure that the dynamic relocations are done correctly, so in
1577 some cases we force the original symbol to be used. */
1578
1579 int
1580 tc_i386_fix_adjustable (fixS *fixP ATTRIBUTE_UNUSED)
1581 {
1582 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
1583 if (!IS_ELF)
1584 return 1;
1585
1586 /* Don't adjust pc-relative references to merge sections in 64-bit
1587 mode. */
1588 if (use_rela_relocations
1589 && (S_GET_SEGMENT (fixP->fx_addsy)->flags & SEC_MERGE) != 0
1590 && fixP->fx_pcrel)
1591 return 0;
1592
1593 /* The x86_64 GOTPCREL are represented as 32bit PCrel relocations
1594 and changed later by validate_fix. */
1595 if (GOT_symbol && fixP->fx_subsy == GOT_symbol
1596 && fixP->fx_r_type == BFD_RELOC_32_PCREL)
1597 return 0;
1598
1599 /* adjust_reloc_syms doesn't know about the GOT. */
1600 if (fixP->fx_r_type == BFD_RELOC_386_GOTOFF
1601 || fixP->fx_r_type == BFD_RELOC_386_PLT32
1602 || fixP->fx_r_type == BFD_RELOC_386_GOT32
1603 || fixP->fx_r_type == BFD_RELOC_386_TLS_GD
1604 || fixP->fx_r_type == BFD_RELOC_386_TLS_LDM
1605 || fixP->fx_r_type == BFD_RELOC_386_TLS_LDO_32
1606 || fixP->fx_r_type == BFD_RELOC_386_TLS_IE_32
1607 || fixP->fx_r_type == BFD_RELOC_386_TLS_IE
1608 || fixP->fx_r_type == BFD_RELOC_386_TLS_GOTIE
1609 || fixP->fx_r_type == BFD_RELOC_386_TLS_LE_32
1610 || fixP->fx_r_type == BFD_RELOC_386_TLS_LE
1611 || fixP->fx_r_type == BFD_RELOC_386_TLS_GOTDESC
1612 || fixP->fx_r_type == BFD_RELOC_386_TLS_DESC_CALL
1613 || fixP->fx_r_type == BFD_RELOC_X86_64_PLT32
1614 || fixP->fx_r_type == BFD_RELOC_X86_64_GOT32
1615 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTPCREL
1616 || fixP->fx_r_type == BFD_RELOC_X86_64_TLSGD
1617 || fixP->fx_r_type == BFD_RELOC_X86_64_TLSLD
1618 || fixP->fx_r_type == BFD_RELOC_X86_64_DTPOFF32
1619 || fixP->fx_r_type == BFD_RELOC_X86_64_DTPOFF64
1620 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTTPOFF
1621 || fixP->fx_r_type == BFD_RELOC_X86_64_TPOFF32
1622 || fixP->fx_r_type == BFD_RELOC_X86_64_TPOFF64
1623 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTOFF64
1624 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTPC32_TLSDESC
1625 || fixP->fx_r_type == BFD_RELOC_X86_64_TLSDESC_CALL
1626 || fixP->fx_r_type == BFD_RELOC_VTABLE_INHERIT
1627 || fixP->fx_r_type == BFD_RELOC_VTABLE_ENTRY)
1628 return 0;
1629 #endif
1630 return 1;
1631 }
1632
1633 static int
1634 intel_float_operand (const char *mnemonic)
1635 {
1636 /* Note that the value returned is meaningful only for opcodes with (memory)
1637 operands, hence the code here is free to improperly handle opcodes that
1638 have no operands (for better performance and smaller code). */
1639
1640 if (mnemonic[0] != 'f')
1641 return 0; /* non-math */
1642
1643 switch (mnemonic[1])
1644 {
1645 /* fclex, fdecstp, fdisi, femms, feni, fincstp, finit, fsetpm, and
1646 the fs segment override prefix not currently handled because no
1647 call path can make opcodes without operands get here */
1648 case 'i':
1649 return 2 /* integer op */;
1650 case 'l':
1651 if (mnemonic[2] == 'd' && (mnemonic[3] == 'c' || mnemonic[3] == 'e'))
1652 return 3; /* fldcw/fldenv */
1653 break;
1654 case 'n':
1655 if (mnemonic[2] != 'o' /* fnop */)
1656 return 3; /* non-waiting control op */
1657 break;
1658 case 'r':
1659 if (mnemonic[2] == 's')
1660 return 3; /* frstor/frstpm */
1661 break;
1662 case 's':
1663 if (mnemonic[2] == 'a')
1664 return 3; /* fsave */
1665 if (mnemonic[2] == 't')
1666 {
1667 switch (mnemonic[3])
1668 {
1669 case 'c': /* fstcw */
1670 case 'd': /* fstdw */
1671 case 'e': /* fstenv */
1672 case 's': /* fsts[gw] */
1673 return 3;
1674 }
1675 }
1676 break;
1677 case 'x':
1678 if (mnemonic[2] == 'r' || mnemonic[2] == 's')
1679 return 0; /* fxsave/fxrstor are not really math ops */
1680 break;
1681 }
1682
1683 return 1;
1684 }
1685
1686 /* This is the guts of the machine-dependent assembler. LINE points to a
1687 machine dependent instruction. This function is supposed to emit
1688 the frags/bytes it assembles to. */
1689
1690 void
1691 md_assemble (line)
1692 char *line;
1693 {
1694 int j;
1695 char mnemonic[MAX_MNEM_SIZE];
1696
1697 /* Initialize globals. */
1698 memset (&i, '\0', sizeof (i));
1699 for (j = 0; j < MAX_OPERANDS; j++)
1700 i.reloc[j] = NO_RELOC;
1701 memset (disp_expressions, '\0', sizeof (disp_expressions));
1702 memset (im_expressions, '\0', sizeof (im_expressions));
1703 save_stack_p = save_stack;
1704
1705 /* First parse an instruction mnemonic & call i386_operand for the operands.
1706 We assume that the scrubber has arranged it so that line[0] is the valid
1707 start of a (possibly prefixed) mnemonic. */
1708
1709 line = parse_insn (line, mnemonic);
1710 if (line == NULL)
1711 return;
1712
1713 line = parse_operands (line, mnemonic);
1714 if (line == NULL)
1715 return;
1716
1717 /* The order of the immediates should be reversed
1718 for 2 immediates extrq and insertq instructions */
1719 if ((i.imm_operands == 2) &&
1720 ((strcmp (mnemonic, "extrq") == 0)
1721 || (strcmp (mnemonic, "insertq") == 0)))
1722 {
1723 swap_imm_operands ();
1724 /* "extrq" and insertq" are the only two instructions whose operands
1725 have to be reversed even though they have two immediate operands.
1726 */
1727 if (intel_syntax)
1728 swap_operands ();
1729 }
1730
1731 /* Now we've parsed the mnemonic into a set of templates, and have the
1732 operands at hand. */
1733
1734 /* All intel opcodes have reversed operands except for "bound" and
1735 "enter". We also don't reverse intersegment "jmp" and "call"
1736 instructions with 2 immediate operands so that the immediate segment
1737 precedes the offset, as it does when in AT&T mode. */
1738 if (intel_syntax && i.operands > 1
1739 && (strcmp (mnemonic, "bound") != 0)
1740 && (strcmp (mnemonic, "invlpga") != 0)
1741 && !((i.types[0] & Imm) && (i.types[1] & Imm)))
1742 swap_operands ();
1743
1744 if (i.imm_operands)
1745 optimize_imm ();
1746
1747 /* Don't optimize displacement for movabs since it only takes 64bit
1748 displacement. */
1749 if (i.disp_operands
1750 && (flag_code != CODE_64BIT
1751 || strcmp (mnemonic, "movabs") != 0))
1752 optimize_disp ();
1753
1754 /* Next, we find a template that matches the given insn,
1755 making sure the overlap of the given operands types is consistent
1756 with the template operand types. */
1757
1758 if (!match_template ())
1759 return;
1760
1761 if (intel_syntax)
1762 {
1763 /* Undo SYSV386_COMPAT brokenness when in Intel mode. See i386.h */
1764 if (SYSV386_COMPAT
1765 && (i.tm.base_opcode & 0xfffffde0) == 0xdce0)
1766 i.tm.base_opcode ^= FloatR;
1767
1768 /* Zap movzx and movsx suffix. The suffix may have been set from
1769 "word ptr" or "byte ptr" on the source operand, but we'll use
1770 the suffix later to choose the destination register. */
1771 if ((i.tm.base_opcode & ~9) == 0x0fb6)
1772 {
1773 if (i.reg_operands < 2
1774 && !i.suffix
1775 && (~i.tm.opcode_modifier
1776 & (No_bSuf
1777 | No_wSuf
1778 | No_lSuf
1779 | No_sSuf
1780 | No_xSuf
1781 | No_qSuf)))
1782 as_bad (_("ambiguous operand size for `%s'"), i.tm.name);
1783
1784 i.suffix = 0;
1785 }
1786 }
1787
1788 if (i.tm.opcode_modifier & FWait)
1789 if (!add_prefix (FWAIT_OPCODE))
1790 return;
1791
1792 /* Check string instruction segment overrides. */
1793 if ((i.tm.opcode_modifier & IsString) != 0 && i.mem_operands != 0)
1794 {
1795 if (!check_string ())
1796 return;
1797 }
1798
1799 if (!process_suffix ())
1800 return;
1801
1802 /* Make still unresolved immediate matches conform to size of immediate
1803 given in i.suffix. */
1804 if (!finalize_imm ())
1805 return;
1806
1807 if (i.types[0] & Imm1)
1808 i.imm_operands = 0; /* kludge for shift insns. */
1809 if (i.types[0] & ImplicitRegister)
1810 i.reg_operands--;
1811 if (i.types[1] & ImplicitRegister)
1812 i.reg_operands--;
1813 if (i.types[2] & ImplicitRegister)
1814 i.reg_operands--;
1815
1816 if (i.tm.opcode_modifier & ImmExt)
1817 {
1818 expressionS *exp;
1819
1820 if ((i.tm.cpu_flags & CpuSSE3) && i.operands > 0)
1821 {
1822 /* Streaming SIMD extensions 3 Instructions have the fixed
1823 operands with an opcode suffix which is coded in the same
1824 place as an 8-bit immediate field would be. Here we check
1825 those operands and remove them afterwards. */
1826 unsigned int x;
1827
1828 for (x = 0; x < i.operands; x++)
1829 if (i.op[x].regs->reg_num != x)
1830 as_bad (_("can't use register '%%%s' as operand %d in '%s'."),
1831 i.op[x].regs->reg_name, x + 1, i.tm.name);
1832 i.operands = 0;
1833 }
1834
1835 /* These AMD 3DNow! and Intel Katmai New Instructions have an
1836 opcode suffix which is coded in the same place as an 8-bit
1837 immediate field would be. Here we fake an 8-bit immediate
1838 operand from the opcode suffix stored in tm.extension_opcode. */
1839
1840 assert (i.imm_operands == 0 && i.operands <= 2 && 2 < MAX_OPERANDS);
1841
1842 exp = &im_expressions[i.imm_operands++];
1843 i.op[i.operands].imms = exp;
1844 i.types[i.operands++] = Imm8;
1845 exp->X_op = O_constant;
1846 exp->X_add_number = i.tm.extension_opcode;
1847 i.tm.extension_opcode = None;
1848 }
1849
1850 /* For insns with operands there are more diddles to do to the opcode. */
1851 if (i.operands)
1852 {
1853 if (!process_operands ())
1854 return;
1855 }
1856 else if (!quiet_warnings && (i.tm.opcode_modifier & Ugh) != 0)
1857 {
1858 /* UnixWare fsub no args is alias for fsubp, fadd -> faddp, etc. */
1859 as_warn (_("translating to `%sp'"), i.tm.name);
1860 }
1861
1862 /* Handle conversion of 'int $3' --> special int3 insn. */
1863 if (i.tm.base_opcode == INT_OPCODE && i.op[0].imms->X_add_number == 3)
1864 {
1865 i.tm.base_opcode = INT3_OPCODE;
1866 i.imm_operands = 0;
1867 }
1868
1869 if ((i.tm.opcode_modifier & (Jump | JumpByte | JumpDword))
1870 && i.op[0].disps->X_op == O_constant)
1871 {
1872 /* Convert "jmp constant" (and "call constant") to a jump (call) to
1873 the absolute address given by the constant. Since ix86 jumps and
1874 calls are pc relative, we need to generate a reloc. */
1875 i.op[0].disps->X_add_symbol = &abs_symbol;
1876 i.op[0].disps->X_op = O_symbol;
1877 }
1878
1879 if ((i.tm.opcode_modifier & Rex64) != 0)
1880 i.rex |= REX_MODE64;
1881
1882 /* For 8 bit registers we need an empty rex prefix. Also if the
1883 instruction already has a prefix, we need to convert old
1884 registers to new ones. */
1885
1886 if (((i.types[0] & Reg8) != 0
1887 && (i.op[0].regs->reg_flags & RegRex64) != 0)
1888 || ((i.types[1] & Reg8) != 0
1889 && (i.op[1].regs->reg_flags & RegRex64) != 0)
1890 || (((i.types[0] & Reg8) != 0 || (i.types[1] & Reg8) != 0)
1891 && i.rex != 0))
1892 {
1893 int x;
1894
1895 i.rex |= REX_OPCODE;
1896 for (x = 0; x < 2; x++)
1897 {
1898 /* Look for 8 bit operand that uses old registers. */
1899 if ((i.types[x] & Reg8) != 0
1900 && (i.op[x].regs->reg_flags & RegRex64) == 0)
1901 {
1902 /* In case it is "hi" register, give up. */
1903 if (i.op[x].regs->reg_num > 3)
1904 as_bad (_("can't encode register '%%%s' in an instruction requiring REX prefix."),
1905 i.op[x].regs->reg_name);
1906
1907 /* Otherwise it is equivalent to the extended register.
1908 Since the encoding doesn't change this is merely
1909 cosmetic cleanup for debug output. */
1910
1911 i.op[x].regs = i.op[x].regs + 8;
1912 }
1913 }
1914 }
1915
1916 if (i.rex != 0)
1917 add_prefix (REX_OPCODE | i.rex);
1918
1919 /* We are ready to output the insn. */
1920 output_insn ();
1921 }
1922
1923 static char *
1924 parse_insn (char *line, char *mnemonic)
1925 {
1926 char *l = line;
1927 char *token_start = l;
1928 char *mnem_p;
1929 int supported;
1930 const template *t;
1931
1932 /* Non-zero if we found a prefix only acceptable with string insns. */
1933 const char *expecting_string_instruction = NULL;
1934
1935 while (1)
1936 {
1937 mnem_p = mnemonic;
1938 while ((*mnem_p = mnemonic_chars[(unsigned char) *l]) != 0)
1939 {
1940 mnem_p++;
1941 if (mnem_p >= mnemonic + MAX_MNEM_SIZE)
1942 {
1943 as_bad (_("no such instruction: `%s'"), token_start);
1944 return NULL;
1945 }
1946 l++;
1947 }
1948 if (!is_space_char (*l)
1949 && *l != END_OF_INSN
1950 && (intel_syntax
1951 || (*l != PREFIX_SEPARATOR
1952 && *l != ',')))
1953 {
1954 as_bad (_("invalid character %s in mnemonic"),
1955 output_invalid (*l));
1956 return NULL;
1957 }
1958 if (token_start == l)
1959 {
1960 if (!intel_syntax && *l == PREFIX_SEPARATOR)
1961 as_bad (_("expecting prefix; got nothing"));
1962 else
1963 as_bad (_("expecting mnemonic; got nothing"));
1964 return NULL;
1965 }
1966
1967 /* Look up instruction (or prefix) via hash table. */
1968 current_templates = hash_find (op_hash, mnemonic);
1969
1970 if (*l != END_OF_INSN
1971 && (!is_space_char (*l) || l[1] != END_OF_INSN)
1972 && current_templates
1973 && (current_templates->start->opcode_modifier & IsPrefix))
1974 {
1975 if (current_templates->start->cpu_flags
1976 & (flag_code != CODE_64BIT ? Cpu64 : CpuNo64))
1977 {
1978 as_bad ((flag_code != CODE_64BIT
1979 ? _("`%s' is only supported in 64-bit mode")
1980 : _("`%s' is not supported in 64-bit mode")),
1981 current_templates->start->name);
1982 return NULL;
1983 }
1984 /* If we are in 16-bit mode, do not allow addr16 or data16.
1985 Similarly, in 32-bit mode, do not allow addr32 or data32. */
1986 if ((current_templates->start->opcode_modifier & (Size16 | Size32))
1987 && flag_code != CODE_64BIT
1988 && (((current_templates->start->opcode_modifier & Size32) != 0)
1989 ^ (flag_code == CODE_16BIT)))
1990 {
1991 as_bad (_("redundant %s prefix"),
1992 current_templates->start->name);
1993 return NULL;
1994 }
1995 /* Add prefix, checking for repeated prefixes. */
1996 switch (add_prefix (current_templates->start->base_opcode))
1997 {
1998 case 0:
1999 return NULL;
2000 case 2:
2001 expecting_string_instruction = current_templates->start->name;
2002 break;
2003 }
2004 /* Skip past PREFIX_SEPARATOR and reset token_start. */
2005 token_start = ++l;
2006 }
2007 else
2008 break;
2009 }
2010
2011 if (!current_templates)
2012 {
2013 /* See if we can get a match by trimming off a suffix. */
2014 switch (mnem_p[-1])
2015 {
2016 case WORD_MNEM_SUFFIX:
2017 if (intel_syntax && (intel_float_operand (mnemonic) & 2))
2018 i.suffix = SHORT_MNEM_SUFFIX;
2019 else
2020 case BYTE_MNEM_SUFFIX:
2021 case QWORD_MNEM_SUFFIX:
2022 i.suffix = mnem_p[-1];
2023 mnem_p[-1] = '\0';
2024 current_templates = hash_find (op_hash, mnemonic);
2025 break;
2026 case SHORT_MNEM_SUFFIX:
2027 case LONG_MNEM_SUFFIX:
2028 if (!intel_syntax)
2029 {
2030 i.suffix = mnem_p[-1];
2031 mnem_p[-1] = '\0';
2032 current_templates = hash_find (op_hash, mnemonic);
2033 }
2034 break;
2035
2036 /* Intel Syntax. */
2037 case 'd':
2038 if (intel_syntax)
2039 {
2040 if (intel_float_operand (mnemonic) == 1)
2041 i.suffix = SHORT_MNEM_SUFFIX;
2042 else
2043 i.suffix = LONG_MNEM_SUFFIX;
2044 mnem_p[-1] = '\0';
2045 current_templates = hash_find (op_hash, mnemonic);
2046 }
2047 break;
2048 }
2049 if (!current_templates)
2050 {
2051 as_bad (_("no such instruction: `%s'"), token_start);
2052 return NULL;
2053 }
2054 }
2055
2056 if (current_templates->start->opcode_modifier & (Jump | JumpByte))
2057 {
2058 /* Check for a branch hint. We allow ",pt" and ",pn" for
2059 predict taken and predict not taken respectively.
2060 I'm not sure that branch hints actually do anything on loop
2061 and jcxz insns (JumpByte) for current Pentium4 chips. They
2062 may work in the future and it doesn't hurt to accept them
2063 now. */
2064 if (l[0] == ',' && l[1] == 'p')
2065 {
2066 if (l[2] == 't')
2067 {
2068 if (!add_prefix (DS_PREFIX_OPCODE))
2069 return NULL;
2070 l += 3;
2071 }
2072 else if (l[2] == 'n')
2073 {
2074 if (!add_prefix (CS_PREFIX_OPCODE))
2075 return NULL;
2076 l += 3;
2077 }
2078 }
2079 }
2080 /* Any other comma loses. */
2081 if (*l == ',')
2082 {
2083 as_bad (_("invalid character %s in mnemonic"),
2084 output_invalid (*l));
2085 return NULL;
2086 }
2087
2088 /* Check if instruction is supported on specified architecture. */
2089 supported = 0;
2090 for (t = current_templates->start; t < current_templates->end; ++t)
2091 {
2092 if (!((t->cpu_flags & ~(Cpu64 | CpuNo64))
2093 & ~(cpu_arch_flags & ~(Cpu64 | CpuNo64))))
2094 supported |= 1;
2095 if (!(t->cpu_flags & (flag_code == CODE_64BIT ? CpuNo64 : Cpu64)))
2096 supported |= 2;
2097 }
2098 if (!(supported & 2))
2099 {
2100 as_bad (flag_code == CODE_64BIT
2101 ? _("`%s' is not supported in 64-bit mode")
2102 : _("`%s' is only supported in 64-bit mode"),
2103 current_templates->start->name);
2104 return NULL;
2105 }
2106 if (!(supported & 1))
2107 {
2108 as_warn (_("`%s' is not supported on `%s%s'"),
2109 current_templates->start->name,
2110 cpu_arch_name,
2111 cpu_sub_arch_name ? cpu_sub_arch_name : "");
2112 }
2113 else if ((Cpu386 & ~cpu_arch_flags) && (flag_code != CODE_16BIT))
2114 {
2115 as_warn (_("use .code16 to ensure correct addressing mode"));
2116 }
2117
2118 /* Check for rep/repne without a string instruction. */
2119 if (expecting_string_instruction)
2120 {
2121 static templates override;
2122
2123 for (t = current_templates->start; t < current_templates->end; ++t)
2124 if (t->opcode_modifier & IsString)
2125 break;
2126 if (t >= current_templates->end)
2127 {
2128 as_bad (_("expecting string instruction after `%s'"),
2129 expecting_string_instruction);
2130 return NULL;
2131 }
2132 for (override.start = t; t < current_templates->end; ++t)
2133 if (!(t->opcode_modifier & IsString))
2134 break;
2135 override.end = t;
2136 current_templates = &override;
2137 }
2138
2139 return l;
2140 }
2141
2142 static char *
2143 parse_operands (char *l, const char *mnemonic)
2144 {
2145 char *token_start;
2146
2147 /* 1 if operand is pending after ','. */
2148 unsigned int expecting_operand = 0;
2149
2150 /* Non-zero if operand parens not balanced. */
2151 unsigned int paren_not_balanced;
2152
2153 while (*l != END_OF_INSN)
2154 {
2155 /* Skip optional white space before operand. */
2156 if (is_space_char (*l))
2157 ++l;
2158 if (!is_operand_char (*l) && *l != END_OF_INSN)
2159 {
2160 as_bad (_("invalid character %s before operand %d"),
2161 output_invalid (*l),
2162 i.operands + 1);
2163 return NULL;
2164 }
2165 token_start = l; /* after white space */
2166 paren_not_balanced = 0;
2167 while (paren_not_balanced || *l != ',')
2168 {
2169 if (*l == END_OF_INSN)
2170 {
2171 if (paren_not_balanced)
2172 {
2173 if (!intel_syntax)
2174 as_bad (_("unbalanced parenthesis in operand %d."),
2175 i.operands + 1);
2176 else
2177 as_bad (_("unbalanced brackets in operand %d."),
2178 i.operands + 1);
2179 return NULL;
2180 }
2181 else
2182 break; /* we are done */
2183 }
2184 else if (!is_operand_char (*l) && !is_space_char (*l))
2185 {
2186 as_bad (_("invalid character %s in operand %d"),
2187 output_invalid (*l),
2188 i.operands + 1);
2189 return NULL;
2190 }
2191 if (!intel_syntax)
2192 {
2193 if (*l == '(')
2194 ++paren_not_balanced;
2195 if (*l == ')')
2196 --paren_not_balanced;
2197 }
2198 else
2199 {
2200 if (*l == '[')
2201 ++paren_not_balanced;
2202 if (*l == ']')
2203 --paren_not_balanced;
2204 }
2205 l++;
2206 }
2207 if (l != token_start)
2208 { /* Yes, we've read in another operand. */
2209 unsigned int operand_ok;
2210 this_operand = i.operands++;
2211 if (i.operands > MAX_OPERANDS)
2212 {
2213 as_bad (_("spurious operands; (%d operands/instruction max)"),
2214 MAX_OPERANDS);
2215 return NULL;
2216 }
2217 /* Now parse operand adding info to 'i' as we go along. */
2218 END_STRING_AND_SAVE (l);
2219
2220 if (intel_syntax)
2221 operand_ok =
2222 i386_intel_operand (token_start,
2223 intel_float_operand (mnemonic));
2224 else
2225 operand_ok = i386_operand (token_start);
2226
2227 RESTORE_END_STRING (l);
2228 if (!operand_ok)
2229 return NULL;
2230 }
2231 else
2232 {
2233 if (expecting_operand)
2234 {
2235 expecting_operand_after_comma:
2236 as_bad (_("expecting operand after ','; got nothing"));
2237 return NULL;
2238 }
2239 if (*l == ',')
2240 {
2241 as_bad (_("expecting operand before ','; got nothing"));
2242 return NULL;
2243 }
2244 }
2245
2246 /* Now *l must be either ',' or END_OF_INSN. */
2247 if (*l == ',')
2248 {
2249 if (*++l == END_OF_INSN)
2250 {
2251 /* Just skip it, if it's \n complain. */
2252 goto expecting_operand_after_comma;
2253 }
2254 expecting_operand = 1;
2255 }
2256 }
2257 return l;
2258 }
2259
2260 static void
2261 swap_imm_operands (void)
2262 {
2263 union i386_op temp_op;
2264 unsigned int temp_type;
2265 enum bfd_reloc_code_real temp_reloc;
2266 int xchg1 = 0;
2267 int xchg2 = 1;
2268
2269 temp_type = i.types[xchg2];
2270 i.types[xchg2] = i.types[xchg1];
2271 i.types[xchg1] = temp_type;
2272 temp_op = i.op[xchg2];
2273 i.op[xchg2] = i.op[xchg1];
2274 i.op[xchg1] = temp_op;
2275 temp_reloc = i.reloc[xchg2];
2276 i.reloc[xchg2] = i.reloc[xchg1];
2277 i.reloc[xchg1] = temp_reloc;
2278 }
2279
2280
2281 static void
2282 swap_operands (void)
2283 {
2284 union i386_op temp_op;
2285 unsigned int temp_type;
2286 enum bfd_reloc_code_real temp_reloc;
2287 int xchg1, xchg2;
2288
2289 switch (i.operands)
2290 {
2291 case 4:
2292 xchg1 = 1;
2293 xchg2 = i.operands - 2;
2294 temp_type = i.types[xchg2];
2295 i.types[xchg2] = i.types[xchg1];
2296 i.types[xchg1] = temp_type;
2297 temp_op = i.op[xchg2];
2298 i.op[xchg2] = i.op[xchg1];
2299 i.op[xchg1] = temp_op;
2300 temp_reloc = i.reloc[xchg2];
2301 i.reloc[xchg2] = i.reloc[xchg1];
2302 i.reloc[xchg1] = temp_reloc;
2303 case 3:
2304 case 2:
2305 xchg1 = 0;
2306 xchg2 = i.operands - 1;
2307 temp_type = i.types[xchg2];
2308 i.types[xchg2] = i.types[xchg1];
2309 i.types[xchg1] = temp_type;
2310 temp_op = i.op[xchg2];
2311 i.op[xchg2] = i.op[xchg1];
2312 i.op[xchg1] = temp_op;
2313 temp_reloc = i.reloc[xchg2];
2314 i.reloc[xchg2] = i.reloc[xchg1];
2315 i.reloc[xchg1] = temp_reloc;
2316 break;
2317 default:
2318 abort ();
2319 }
2320
2321 if (i.mem_operands == 2)
2322 {
2323 const seg_entry *temp_seg;
2324 temp_seg = i.seg[0];
2325 i.seg[0] = i.seg[1];
2326 i.seg[1] = temp_seg;
2327 }
2328 }
2329
2330 /* Try to ensure constant immediates are represented in the smallest
2331 opcode possible. */
2332 static void
2333 optimize_imm (void)
2334 {
2335 char guess_suffix = 0;
2336 int op;
2337
2338 if (i.suffix)
2339 guess_suffix = i.suffix;
2340 else if (i.reg_operands)
2341 {
2342 /* Figure out a suffix from the last register operand specified.
2343 We can't do this properly yet, ie. excluding InOutPortReg,
2344 but the following works for instructions with immediates.
2345 In any case, we can't set i.suffix yet. */
2346 for (op = i.operands; --op >= 0;)
2347 if (i.types[op] & Reg)
2348 {
2349 if (i.types[op] & Reg8)
2350 guess_suffix = BYTE_MNEM_SUFFIX;
2351 else if (i.types[op] & Reg16)
2352 guess_suffix = WORD_MNEM_SUFFIX;
2353 else if (i.types[op] & Reg32)
2354 guess_suffix = LONG_MNEM_SUFFIX;
2355 else if (i.types[op] & Reg64)
2356 guess_suffix = QWORD_MNEM_SUFFIX;
2357 break;
2358 }
2359 }
2360 else if ((flag_code == CODE_16BIT) ^ (i.prefix[DATA_PREFIX] != 0))
2361 guess_suffix = WORD_MNEM_SUFFIX;
2362
2363 for (op = i.operands; --op >= 0;)
2364 if (i.types[op] & Imm)
2365 {
2366 switch (i.op[op].imms->X_op)
2367 {
2368 case O_constant:
2369 /* If a suffix is given, this operand may be shortened. */
2370 switch (guess_suffix)
2371 {
2372 case LONG_MNEM_SUFFIX:
2373 i.types[op] |= Imm32 | Imm64;
2374 break;
2375 case WORD_MNEM_SUFFIX:
2376 i.types[op] |= Imm16 | Imm32S | Imm32 | Imm64;
2377 break;
2378 case BYTE_MNEM_SUFFIX:
2379 i.types[op] |= Imm16 | Imm8 | Imm8S | Imm32S | Imm32 | Imm64;
2380 break;
2381 }
2382
2383 /* If this operand is at most 16 bits, convert it
2384 to a signed 16 bit number before trying to see
2385 whether it will fit in an even smaller size.
2386 This allows a 16-bit operand such as $0xffe0 to
2387 be recognised as within Imm8S range. */
2388 if ((i.types[op] & Imm16)
2389 && (i.op[op].imms->X_add_number & ~(offsetT) 0xffff) == 0)
2390 {
2391 i.op[op].imms->X_add_number =
2392 (((i.op[op].imms->X_add_number & 0xffff) ^ 0x8000) - 0x8000);
2393 }
2394 if ((i.types[op] & Imm32)
2395 && ((i.op[op].imms->X_add_number & ~(((offsetT) 2 << 31) - 1))
2396 == 0))
2397 {
2398 i.op[op].imms->X_add_number = ((i.op[op].imms->X_add_number
2399 ^ ((offsetT) 1 << 31))
2400 - ((offsetT) 1 << 31));
2401 }
2402 i.types[op] |= smallest_imm_type (i.op[op].imms->X_add_number);
2403
2404 /* We must avoid matching of Imm32 templates when 64bit
2405 only immediate is available. */
2406 if (guess_suffix == QWORD_MNEM_SUFFIX)
2407 i.types[op] &= ~Imm32;
2408 break;
2409
2410 case O_absent:
2411 case O_register:
2412 abort ();
2413
2414 /* Symbols and expressions. */
2415 default:
2416 /* Convert symbolic operand to proper sizes for matching, but don't
2417 prevent matching a set of insns that only supports sizes other
2418 than those matching the insn suffix. */
2419 {
2420 unsigned int mask, allowed = 0;
2421 const template *t;
2422
2423 for (t = current_templates->start; t < current_templates->end; ++t)
2424 allowed |= t->operand_types[op];
2425 switch (guess_suffix)
2426 {
2427 case QWORD_MNEM_SUFFIX:
2428 mask = Imm64 | Imm32S;
2429 break;
2430 case LONG_MNEM_SUFFIX:
2431 mask = Imm32;
2432 break;
2433 case WORD_MNEM_SUFFIX:
2434 mask = Imm16;
2435 break;
2436 case BYTE_MNEM_SUFFIX:
2437 mask = Imm8;
2438 break;
2439 default:
2440 mask = 0;
2441 break;
2442 }
2443 if (mask & allowed)
2444 i.types[op] &= mask;
2445 }
2446 break;
2447 }
2448 }
2449 }
2450
2451 /* Try to use the smallest displacement type too. */
2452 static void
2453 optimize_disp (void)
2454 {
2455 int op;
2456
2457 for (op = i.operands; --op >= 0;)
2458 if (i.types[op] & Disp)
2459 {
2460 if (i.op[op].disps->X_op == O_constant)
2461 {
2462 offsetT disp = i.op[op].disps->X_add_number;
2463
2464 if ((i.types[op] & Disp16)
2465 && (disp & ~(offsetT) 0xffff) == 0)
2466 {
2467 /* If this operand is at most 16 bits, convert
2468 to a signed 16 bit number and don't use 64bit
2469 displacement. */
2470 disp = (((disp & 0xffff) ^ 0x8000) - 0x8000);
2471 i.types[op] &= ~Disp64;
2472 }
2473 if ((i.types[op] & Disp32)
2474 && (disp & ~(((offsetT) 2 << 31) - 1)) == 0)
2475 {
2476 /* If this operand is at most 32 bits, convert
2477 to a signed 32 bit number and don't use 64bit
2478 displacement. */
2479 disp &= (((offsetT) 2 << 31) - 1);
2480 disp = (disp ^ ((offsetT) 1 << 31)) - ((addressT) 1 << 31);
2481 i.types[op] &= ~Disp64;
2482 }
2483 if (!disp && (i.types[op] & BaseIndex))
2484 {
2485 i.types[op] &= ~Disp;
2486 i.op[op].disps = 0;
2487 i.disp_operands--;
2488 }
2489 else if (flag_code == CODE_64BIT)
2490 {
2491 if (fits_in_signed_long (disp))
2492 {
2493 i.types[op] &= ~Disp64;
2494 i.types[op] |= Disp32S;
2495 }
2496 if (fits_in_unsigned_long (disp))
2497 i.types[op] |= Disp32;
2498 }
2499 if ((i.types[op] & (Disp32 | Disp32S | Disp16))
2500 && fits_in_signed_byte (disp))
2501 i.types[op] |= Disp8;
2502 }
2503 else if (i.reloc[op] == BFD_RELOC_386_TLS_DESC_CALL
2504 || i.reloc[op] == BFD_RELOC_X86_64_TLSDESC_CALL)
2505 {
2506 fix_new_exp (frag_now, frag_more (0) - frag_now->fr_literal, 0,
2507 i.op[op].disps, 0, i.reloc[op]);
2508 i.types[op] &= ~Disp;
2509 }
2510 else
2511 /* We only support 64bit displacement on constants. */
2512 i.types[op] &= ~Disp64;
2513 }
2514 }
2515
2516 static int
2517 match_template (void)
2518 {
2519 /* Points to template once we've found it. */
2520 const template *t;
2521 unsigned int overlap0, overlap1, overlap2, overlap3;
2522 unsigned int found_reverse_match;
2523 int suffix_check;
2524 unsigned int operand_types [MAX_OPERANDS];
2525 int addr_prefix_disp;
2526 unsigned int j;
2527
2528 #if MAX_OPERANDS != 4
2529 # error "MAX_OPERANDS must be 4."
2530 #endif
2531
2532 #define MATCH(overlap, given, template) \
2533 ((overlap & ~JumpAbsolute) \
2534 && (((given) & (BaseIndex | JumpAbsolute)) \
2535 == ((overlap) & (BaseIndex | JumpAbsolute))))
2536
2537 /* If given types r0 and r1 are registers they must be of the same type
2538 unless the expected operand type register overlap is null.
2539 Note that Acc in a template matches every size of reg. */
2540 #define CONSISTENT_REGISTER_MATCH(m0, g0, t0, m1, g1, t1) \
2541 (((g0) & Reg) == 0 || ((g1) & Reg) == 0 \
2542 || ((g0) & Reg) == ((g1) & Reg) \
2543 || ((((m0) & Acc) ? Reg : (t0)) & (((m1) & Acc) ? Reg : (t1)) & Reg) == 0 )
2544
2545 overlap0 = 0;
2546 overlap1 = 0;
2547 overlap2 = 0;
2548 overlap3 = 0;
2549 found_reverse_match = 0;
2550 for (j = 0; j < MAX_OPERANDS; j++)
2551 operand_types [j] = 0;
2552 addr_prefix_disp = -1;
2553 suffix_check = (i.suffix == BYTE_MNEM_SUFFIX
2554 ? No_bSuf
2555 : (i.suffix == WORD_MNEM_SUFFIX
2556 ? No_wSuf
2557 : (i.suffix == SHORT_MNEM_SUFFIX
2558 ? No_sSuf
2559 : (i.suffix == LONG_MNEM_SUFFIX
2560 ? No_lSuf
2561 : (i.suffix == QWORD_MNEM_SUFFIX
2562 ? No_qSuf
2563 : (i.suffix == LONG_DOUBLE_MNEM_SUFFIX
2564 ? No_xSuf : 0))))));
2565
2566 for (t = current_templates->start; t < current_templates->end; t++)
2567 {
2568 addr_prefix_disp = -1;
2569
2570 /* Must have right number of operands. */
2571 if (i.operands != t->operands)
2572 continue;
2573
2574 /* Check the suffix, except for some instructions in intel mode. */
2575 if ((t->opcode_modifier & suffix_check)
2576 && !(intel_syntax
2577 && (t->opcode_modifier & IgnoreSize)))
2578 continue;
2579
2580 for (j = 0; j < MAX_OPERANDS; j++)
2581 operand_types [j] = t->operand_types [j];
2582
2583 /* In general, don't allow 64-bit operands in 32-bit mode. */
2584 if (i.suffix == QWORD_MNEM_SUFFIX
2585 && flag_code != CODE_64BIT
2586 && (intel_syntax
2587 ? (!(t->opcode_modifier & IgnoreSize)
2588 && !intel_float_operand (t->name))
2589 : intel_float_operand (t->name) != 2)
2590 && (!(operand_types[0] & (RegMMX | RegXMM))
2591 || !(operand_types[t->operands > 1] & (RegMMX | RegXMM)))
2592 && (t->base_opcode != 0x0fc7
2593 || t->extension_opcode != 1 /* cmpxchg8b */))
2594 continue;
2595
2596 /* Do not verify operands when there are none. */
2597 else if (!t->operands)
2598 {
2599 if (t->cpu_flags & ~cpu_arch_flags)
2600 continue;
2601 /* We've found a match; break out of loop. */
2602 break;
2603 }
2604
2605 /* Address size prefix will turn Disp64/Disp32/Disp16 operand
2606 into Disp32/Disp16/Disp32 operand. */
2607 if (i.prefix[ADDR_PREFIX] != 0)
2608 {
2609 unsigned int DispOn = 0, DispOff = 0;
2610
2611 switch (flag_code)
2612 {
2613 case CODE_16BIT:
2614 DispOn = Disp32;
2615 DispOff = Disp16;
2616 break;
2617 case CODE_32BIT:
2618 DispOn = Disp16;
2619 DispOff = Disp32;
2620 break;
2621 case CODE_64BIT:
2622 DispOn = Disp32;
2623 DispOff = Disp64;
2624 break;
2625 }
2626
2627 for (j = 0; j < MAX_OPERANDS; j++)
2628 {
2629 /* There should be only one Disp operand. */
2630 if ((operand_types[j] & DispOff))
2631 {
2632 addr_prefix_disp = j;
2633 operand_types[j] |= DispOn;
2634 operand_types[j] &= ~DispOff;
2635 break;
2636 }
2637 }
2638 }
2639
2640 overlap0 = i.types[0] & operand_types[0];
2641 switch (t->operands)
2642 {
2643 case 1:
2644 if (!MATCH (overlap0, i.types[0], operand_types[0]))
2645 continue;
2646 break;
2647 case 2:
2648 case 3:
2649 case 4:
2650 overlap1 = i.types[1] & operand_types[1];
2651 if (!MATCH (overlap0, i.types[0], operand_types[0])
2652 || !MATCH (overlap1, i.types[1], operand_types[1])
2653 /* monitor in SSE3 is a very special case. The first
2654 register and the second register may have different
2655 sizes. */
2656 || !((t->base_opcode == 0x0f01
2657 && t->extension_opcode == 0xc8)
2658 || CONSISTENT_REGISTER_MATCH (overlap0, i.types[0],
2659 operand_types[0],
2660 overlap1, i.types[1],
2661 operand_types[1])))
2662 {
2663 /* Check if other direction is valid ... */
2664 if ((t->opcode_modifier & (D | FloatD)) == 0)
2665 continue;
2666
2667 /* Try reversing direction of operands. */
2668 overlap0 = i.types[0] & operand_types[1];
2669 overlap1 = i.types[1] & operand_types[0];
2670 if (!MATCH (overlap0, i.types[0], operand_types[1])
2671 || !MATCH (overlap1, i.types[1], operand_types[0])
2672 || !CONSISTENT_REGISTER_MATCH (overlap0, i.types[0],
2673 operand_types[1],
2674 overlap1, i.types[1],
2675 operand_types[0]))
2676 {
2677 /* Does not match either direction. */
2678 continue;
2679 }
2680 /* found_reverse_match holds which of D or FloatDR
2681 we've found. */
2682 found_reverse_match = t->opcode_modifier & (D | FloatDR);
2683 }
2684 else
2685 {
2686 /* Found a forward 2 operand match here. */
2687 switch (t->operands)
2688 {
2689 case 4:
2690 overlap3 = i.types[3] & operand_types[3];
2691 case 3:
2692 overlap2 = i.types[2] & operand_types[2];
2693 break;
2694 }
2695
2696 switch (t->operands)
2697 {
2698 case 4:
2699 if (!MATCH (overlap3, i.types[3], operand_types[3])
2700 || !CONSISTENT_REGISTER_MATCH (overlap2,
2701 i.types[2],
2702 operand_types[2],
2703 overlap3,
2704 i.types[3],
2705 operand_types[3]))
2706 continue;
2707 case 3:
2708 /* Here we make use of the fact that there are no
2709 reverse match 3 operand instructions, and all 3
2710 operand instructions only need to be checked for
2711 register consistency between operands 2 and 3. */
2712 if (!MATCH (overlap2, i.types[2], operand_types[2])
2713 || !CONSISTENT_REGISTER_MATCH (overlap1,
2714 i.types[1],
2715 operand_types[1],
2716 overlap2,
2717 i.types[2],
2718 operand_types[2]))
2719 continue;
2720 break;
2721 }
2722 }
2723 /* Found either forward/reverse 2, 3 or 4 operand match here:
2724 slip through to break. */
2725 }
2726 if (t->cpu_flags & ~cpu_arch_flags)
2727 {
2728 found_reverse_match = 0;
2729 continue;
2730 }
2731 /* We've found a match; break out of loop. */
2732 break;
2733 }
2734
2735 if (t == current_templates->end)
2736 {
2737 /* We found no match. */
2738 as_bad (_("suffix or operands invalid for `%s'"),
2739 current_templates->start->name);
2740 return 0;
2741 }
2742
2743 if (!quiet_warnings)
2744 {
2745 if (!intel_syntax
2746 && ((i.types[0] & JumpAbsolute)
2747 != (operand_types[0] & JumpAbsolute)))
2748 {
2749 as_warn (_("indirect %s without `*'"), t->name);
2750 }
2751
2752 if ((t->opcode_modifier & (IsPrefix | IgnoreSize))
2753 == (IsPrefix | IgnoreSize))
2754 {
2755 /* Warn them that a data or address size prefix doesn't
2756 affect assembly of the next line of code. */
2757 as_warn (_("stand-alone `%s' prefix"), t->name);
2758 }
2759 }
2760
2761 /* Copy the template we found. */
2762 i.tm = *t;
2763
2764 if (addr_prefix_disp != -1)
2765 i.tm.operand_types[addr_prefix_disp]
2766 = operand_types[addr_prefix_disp];
2767
2768 if (found_reverse_match)
2769 {
2770 /* If we found a reverse match we must alter the opcode
2771 direction bit. found_reverse_match holds bits to change
2772 (different for int & float insns). */
2773
2774 i.tm.base_opcode ^= found_reverse_match;
2775
2776 i.tm.operand_types[0] = operand_types[1];
2777 i.tm.operand_types[1] = operand_types[0];
2778 }
2779
2780 return 1;
2781 }
2782
2783 static int
2784 check_string (void)
2785 {
2786 int mem_op = (i.types[0] & AnyMem) ? 0 : 1;
2787 if ((i.tm.operand_types[mem_op] & EsSeg) != 0)
2788 {
2789 if (i.seg[0] != NULL && i.seg[0] != &es)
2790 {
2791 as_bad (_("`%s' operand %d must use `%%es' segment"),
2792 i.tm.name,
2793 mem_op + 1);
2794 return 0;
2795 }
2796 /* There's only ever one segment override allowed per instruction.
2797 This instruction possibly has a legal segment override on the
2798 second operand, so copy the segment to where non-string
2799 instructions store it, allowing common code. */
2800 i.seg[0] = i.seg[1];
2801 }
2802 else if ((i.tm.operand_types[mem_op + 1] & EsSeg) != 0)
2803 {
2804 if (i.seg[1] != NULL && i.seg[1] != &es)
2805 {
2806 as_bad (_("`%s' operand %d must use `%%es' segment"),
2807 i.tm.name,
2808 mem_op + 2);
2809 return 0;
2810 }
2811 }
2812 return 1;
2813 }
2814
2815 static int
2816 process_suffix (void)
2817 {
2818 /* If matched instruction specifies an explicit instruction mnemonic
2819 suffix, use it. */
2820 if (i.tm.opcode_modifier & (Size16 | Size32 | Size64))
2821 {
2822 if (i.tm.opcode_modifier & Size16)
2823 i.suffix = WORD_MNEM_SUFFIX;
2824 else if (i.tm.opcode_modifier & Size64)
2825 i.suffix = QWORD_MNEM_SUFFIX;
2826 else
2827 i.suffix = LONG_MNEM_SUFFIX;
2828 }
2829 else if (i.reg_operands)
2830 {
2831 /* If there's no instruction mnemonic suffix we try to invent one
2832 based on register operands. */
2833 if (!i.suffix)
2834 {
2835 /* We take i.suffix from the last register operand specified,
2836 Destination register type is more significant than source
2837 register type. */
2838 int op;
2839
2840 for (op = i.operands; --op >= 0;)
2841 if ((i.types[op] & Reg)
2842 && !(i.tm.operand_types[op] & InOutPortReg))
2843 {
2844 i.suffix = ((i.types[op] & Reg8) ? BYTE_MNEM_SUFFIX :
2845 (i.types[op] & Reg16) ? WORD_MNEM_SUFFIX :
2846 (i.types[op] & Reg64) ? QWORD_MNEM_SUFFIX :
2847 LONG_MNEM_SUFFIX);
2848 break;
2849 }
2850 }
2851 else if (i.suffix == BYTE_MNEM_SUFFIX)
2852 {
2853 if (!check_byte_reg ())
2854 return 0;
2855 }
2856 else if (i.suffix == LONG_MNEM_SUFFIX)
2857 {
2858 if (!check_long_reg ())
2859 return 0;
2860 }
2861 else if (i.suffix == QWORD_MNEM_SUFFIX)
2862 {
2863 if (!check_qword_reg ())
2864 return 0;
2865 }
2866 else if (i.suffix == WORD_MNEM_SUFFIX)
2867 {
2868 if (!check_word_reg ())
2869 return 0;
2870 }
2871 else if (intel_syntax && (i.tm.opcode_modifier & IgnoreSize))
2872 /* Do nothing if the instruction is going to ignore the prefix. */
2873 ;
2874 else
2875 abort ();
2876 }
2877 else if ((i.tm.opcode_modifier & DefaultSize)
2878 && !i.suffix
2879 /* exclude fldenv/frstor/fsave/fstenv */
2880 && (i.tm.opcode_modifier & No_sSuf))
2881 {
2882 i.suffix = stackop_size;
2883 }
2884 else if (intel_syntax
2885 && !i.suffix
2886 && ((i.tm.operand_types[0] & JumpAbsolute)
2887 || (i.tm.opcode_modifier & (JumpByte|JumpInterSegment))
2888 || (i.tm.base_opcode == 0x0f01 /* [ls][gi]dt */
2889 && i.tm.extension_opcode <= 3)))
2890 {
2891 switch (flag_code)
2892 {
2893 case CODE_64BIT:
2894 if (!(i.tm.opcode_modifier & No_qSuf))
2895 {
2896 i.suffix = QWORD_MNEM_SUFFIX;
2897 break;
2898 }
2899 case CODE_32BIT:
2900 if (!(i.tm.opcode_modifier & No_lSuf))
2901 i.suffix = LONG_MNEM_SUFFIX;
2902 break;
2903 case CODE_16BIT:
2904 if (!(i.tm.opcode_modifier & No_wSuf))
2905 i.suffix = WORD_MNEM_SUFFIX;
2906 break;
2907 }
2908 }
2909
2910 if (!i.suffix)
2911 {
2912 if (!intel_syntax)
2913 {
2914 if (i.tm.opcode_modifier & W)
2915 {
2916 as_bad (_("no instruction mnemonic suffix given and no register operands; can't size instruction"));
2917 return 0;
2918 }
2919 }
2920 else
2921 {
2922 unsigned int suffixes = (~i.tm.opcode_modifier
2923 & (No_bSuf
2924 | No_wSuf
2925 | No_lSuf
2926 | No_sSuf
2927 | No_xSuf
2928 | No_qSuf));
2929
2930 if ((i.tm.opcode_modifier & W)
2931 || ((suffixes & (suffixes - 1))
2932 && !(i.tm.opcode_modifier & (DefaultSize | IgnoreSize))))
2933 {
2934 as_bad (_("ambiguous operand size for `%s'"), i.tm.name);
2935 return 0;
2936 }
2937 }
2938 }
2939
2940 /* Change the opcode based on the operand size given by i.suffix;
2941 We don't need to change things for byte insns. */
2942
2943 if (i.suffix && i.suffix != BYTE_MNEM_SUFFIX)
2944 {
2945 /* It's not a byte, select word/dword operation. */
2946 if (i.tm.opcode_modifier & W)
2947 {
2948 if (i.tm.opcode_modifier & ShortForm)
2949 i.tm.base_opcode |= 8;
2950 else
2951 i.tm.base_opcode |= 1;
2952 }
2953
2954 /* Now select between word & dword operations via the operand
2955 size prefix, except for instructions that will ignore this
2956 prefix anyway. */
2957 if (i.tm.base_opcode == 0x0f01 && i.tm.extension_opcode == 0xc8)
2958 {
2959 /* monitor in SSE3 is a very special case. The default size
2960 of AX is the size of mode. The address size override
2961 prefix will change the size of AX. */
2962 if (i.op->regs[0].reg_type &
2963 (flag_code == CODE_32BIT ? Reg16 : Reg32))
2964 if (!add_prefix (ADDR_PREFIX_OPCODE))
2965 return 0;
2966 }
2967 else if (i.suffix != QWORD_MNEM_SUFFIX
2968 && i.suffix != LONG_DOUBLE_MNEM_SUFFIX
2969 && !(i.tm.opcode_modifier & (IgnoreSize | FloatMF))
2970 && ((i.suffix == LONG_MNEM_SUFFIX) == (flag_code == CODE_16BIT)
2971 || (flag_code == CODE_64BIT
2972 && (i.tm.opcode_modifier & JumpByte))))
2973 {
2974 unsigned int prefix = DATA_PREFIX_OPCODE;
2975
2976 if (i.tm.opcode_modifier & JumpByte) /* jcxz, loop */
2977 prefix = ADDR_PREFIX_OPCODE;
2978
2979 if (!add_prefix (prefix))
2980 return 0;
2981 }
2982
2983 /* Set mode64 for an operand. */
2984 if (i.suffix == QWORD_MNEM_SUFFIX
2985 && flag_code == CODE_64BIT
2986 && (i.tm.opcode_modifier & NoRex64) == 0)
2987 {
2988 /* Special case for xchg %rax,%rax. It is NOP and doesn't
2989 need rex64. */
2990 if (i.operands != 2
2991 || i.types [0] != (Acc | Reg64)
2992 || i.types [1] != (Acc | Reg64)
2993 || strcmp (i.tm.name, "xchg") != 0)
2994 i.rex |= REX_MODE64;
2995 }
2996
2997 /* Size floating point instruction. */
2998 if (i.suffix == LONG_MNEM_SUFFIX)
2999 if (i.tm.opcode_modifier & FloatMF)
3000 i.tm.base_opcode ^= 4;
3001 }
3002
3003 return 1;
3004 }
3005
3006 static int
3007 check_byte_reg (void)
3008 {
3009 int op;
3010
3011 for (op = i.operands; --op >= 0;)
3012 {
3013 /* If this is an eight bit register, it's OK. If it's the 16 or
3014 32 bit version of an eight bit register, we will just use the
3015 low portion, and that's OK too. */
3016 if (i.types[op] & Reg8)
3017 continue;
3018
3019 /* movzx and movsx should not generate this warning. */
3020 if (intel_syntax
3021 && (i.tm.base_opcode == 0xfb7
3022 || i.tm.base_opcode == 0xfb6
3023 || i.tm.base_opcode == 0x63
3024 || i.tm.base_opcode == 0xfbe
3025 || i.tm.base_opcode == 0xfbf))
3026 continue;
3027
3028 if ((i.types[op] & WordReg) && i.op[op].regs->reg_num < 4)
3029 {
3030 /* Prohibit these changes in the 64bit mode, since the
3031 lowering is more complicated. */
3032 if (flag_code == CODE_64BIT
3033 && (i.tm.operand_types[op] & InOutPortReg) == 0)
3034 {
3035 as_bad (_("Incorrect register `%s%s' used with `%c' suffix"),
3036 register_prefix, i.op[op].regs->reg_name,
3037 i.suffix);
3038 return 0;
3039 }
3040 #if REGISTER_WARNINGS
3041 if (!quiet_warnings
3042 && (i.tm.operand_types[op] & InOutPortReg) == 0)
3043 as_warn (_("using `%%%s' instead of `%%%s' due to `%c' suffix"),
3044 (i.op[op].regs + (i.types[op] & Reg16
3045 ? REGNAM_AL - REGNAM_AX
3046 : REGNAM_AL - REGNAM_EAX))->reg_name,
3047 i.op[op].regs->reg_name,
3048 i.suffix);
3049 #endif
3050 continue;
3051 }
3052 /* Any other register is bad. */
3053 if (i.types[op] & (Reg | RegMMX | RegXMM
3054 | SReg2 | SReg3
3055 | Control | Debug | Test
3056 | FloatReg | FloatAcc))
3057 {
3058 as_bad (_("`%%%s' not allowed with `%s%c'"),
3059 i.op[op].regs->reg_name,
3060 i.tm.name,
3061 i.suffix);
3062 return 0;
3063 }
3064 }
3065 return 1;
3066 }
3067
3068 static int
3069 check_long_reg (void)
3070 {
3071 int op;
3072
3073 for (op = i.operands; --op >= 0;)
3074 /* Reject eight bit registers, except where the template requires
3075 them. (eg. movzb) */
3076 if ((i.types[op] & Reg8) != 0
3077 && (i.tm.operand_types[op] & (Reg16 | Reg32 | Acc)) != 0)
3078 {
3079 as_bad (_("`%%%s' not allowed with `%s%c'"),
3080 i.op[op].regs->reg_name,
3081 i.tm.name,
3082 i.suffix);
3083 return 0;
3084 }
3085 /* Warn if the e prefix on a general reg is missing. */
3086 else if ((!quiet_warnings || flag_code == CODE_64BIT)
3087 && (i.types[op] & Reg16) != 0
3088 && (i.tm.operand_types[op] & (Reg32 | Acc)) != 0)
3089 {
3090 /* Prohibit these changes in the 64bit mode, since the
3091 lowering is more complicated. */
3092 if (flag_code == CODE_64BIT)
3093 {
3094 as_bad (_("Incorrect register `%s%s' used with `%c' suffix"),
3095 register_prefix, i.op[op].regs->reg_name,
3096 i.suffix);
3097 return 0;
3098 }
3099 #if REGISTER_WARNINGS
3100 else
3101 as_warn (_("using `%%%s' instead of `%%%s' due to `%c' suffix"),
3102 (i.op[op].regs + REGNAM_EAX - REGNAM_AX)->reg_name,
3103 i.op[op].regs->reg_name,
3104 i.suffix);
3105 #endif
3106 }
3107 /* Warn if the r prefix on a general reg is missing. */
3108 else if ((i.types[op] & Reg64) != 0
3109 && (i.tm.operand_types[op] & (Reg32 | Acc)) != 0)
3110 {
3111 as_bad (_("Incorrect register `%s%s' used with `%c' suffix"),
3112 register_prefix, i.op[op].regs->reg_name,
3113 i.suffix);
3114 return 0;
3115 }
3116 return 1;
3117 }
3118
3119 static int
3120 check_qword_reg (void)
3121 {
3122 int op;
3123
3124 for (op = i.operands; --op >= 0; )
3125 /* Reject eight bit registers, except where the template requires
3126 them. (eg. movzb) */
3127 if ((i.types[op] & Reg8) != 0
3128 && (i.tm.operand_types[op] & (Reg16 | Reg32 | Acc)) != 0)
3129 {
3130 as_bad (_("`%%%s' not allowed with `%s%c'"),
3131 i.op[op].regs->reg_name,
3132 i.tm.name,
3133 i.suffix);
3134 return 0;
3135 }
3136 /* Warn if the e prefix on a general reg is missing. */
3137 else if (((i.types[op] & Reg16) != 0
3138 || (i.types[op] & Reg32) != 0)
3139 && (i.tm.operand_types[op] & (Reg32 | Acc)) != 0)
3140 {
3141 /* Prohibit these changes in the 64bit mode, since the
3142 lowering is more complicated. */
3143 as_bad (_("Incorrect register `%s%s' used with `%c' suffix"),
3144 register_prefix, i.op[op].regs->reg_name,
3145 i.suffix);
3146 return 0;
3147 }
3148 return 1;
3149 }
3150
3151 static int
3152 check_word_reg (void)
3153 {
3154 int op;
3155 for (op = i.operands; --op >= 0;)
3156 /* Reject eight bit registers, except where the template requires
3157 them. (eg. movzb) */
3158 if ((i.types[op] & Reg8) != 0
3159 && (i.tm.operand_types[op] & (Reg16 | Reg32 | Acc)) != 0)
3160 {
3161 as_bad (_("`%%%s' not allowed with `%s%c'"),
3162 i.op[op].regs->reg_name,
3163 i.tm.name,
3164 i.suffix);
3165 return 0;
3166 }
3167 /* Warn if the e prefix on a general reg is present. */
3168 else if ((!quiet_warnings || flag_code == CODE_64BIT)
3169 && (i.types[op] & Reg32) != 0
3170 && (i.tm.operand_types[op] & (Reg16 | Acc)) != 0)
3171 {
3172 /* Prohibit these changes in the 64bit mode, since the
3173 lowering is more complicated. */
3174 if (flag_code == CODE_64BIT)
3175 {
3176 as_bad (_("Incorrect register `%s%s' used with `%c' suffix"),
3177 register_prefix, i.op[op].regs->reg_name,
3178 i.suffix);
3179 return 0;
3180 }
3181 else
3182 #if REGISTER_WARNINGS
3183 as_warn (_("using `%%%s' instead of `%%%s' due to `%c' suffix"),
3184 (i.op[op].regs + REGNAM_AX - REGNAM_EAX)->reg_name,
3185 i.op[op].regs->reg_name,
3186 i.suffix);
3187 #endif
3188 }
3189 return 1;
3190 }
3191
3192 static int
3193 finalize_imm (void)
3194 {
3195 unsigned int overlap0, overlap1, overlap2;
3196
3197 overlap0 = i.types[0] & i.tm.operand_types[0];
3198 if ((overlap0 & (Imm8 | Imm8S | Imm16 | Imm32 | Imm32S | Imm64))
3199 && overlap0 != Imm8 && overlap0 != Imm8S
3200 && overlap0 != Imm16 && overlap0 != Imm32S
3201 && overlap0 != Imm32 && overlap0 != Imm64)
3202 {
3203 if (i.suffix)
3204 {
3205 overlap0 &= (i.suffix == BYTE_MNEM_SUFFIX
3206 ? Imm8 | Imm8S
3207 : (i.suffix == WORD_MNEM_SUFFIX
3208 ? Imm16
3209 : (i.suffix == QWORD_MNEM_SUFFIX
3210 ? Imm64 | Imm32S
3211 : Imm32)));
3212 }
3213 else if (overlap0 == (Imm16 | Imm32S | Imm32)
3214 || overlap0 == (Imm16 | Imm32)
3215 || overlap0 == (Imm16 | Imm32S))
3216 {
3217 overlap0 = ((flag_code == CODE_16BIT) ^ (i.prefix[DATA_PREFIX] != 0)
3218 ? Imm16 : Imm32S);
3219 }
3220 if (overlap0 != Imm8 && overlap0 != Imm8S
3221 && overlap0 != Imm16 && overlap0 != Imm32S
3222 && overlap0 != Imm32 && overlap0 != Imm64)
3223 {
3224 as_bad (_("no instruction mnemonic suffix given; can't determine immediate size"));
3225 return 0;
3226 }
3227 }
3228 i.types[0] = overlap0;
3229
3230 overlap1 = i.types[1] & i.tm.operand_types[1];
3231 if ((overlap1 & (Imm8 | Imm8S | Imm16 | Imm32S | Imm32 | Imm64))
3232 && overlap1 != Imm8 && overlap1 != Imm8S
3233 && overlap1 != Imm16 && overlap1 != Imm32S
3234 && overlap1 != Imm32 && overlap1 != Imm64)
3235 {
3236 if (i.suffix)
3237 {
3238 overlap1 &= (i.suffix == BYTE_MNEM_SUFFIX
3239 ? Imm8 | Imm8S
3240 : (i.suffix == WORD_MNEM_SUFFIX
3241 ? Imm16
3242 : (i.suffix == QWORD_MNEM_SUFFIX
3243 ? Imm64 | Imm32S
3244 : Imm32)));
3245 }
3246 else if (overlap1 == (Imm16 | Imm32 | Imm32S)
3247 || overlap1 == (Imm16 | Imm32)
3248 || overlap1 == (Imm16 | Imm32S))
3249 {
3250 overlap1 = ((flag_code == CODE_16BIT) ^ (i.prefix[DATA_PREFIX] != 0)
3251 ? Imm16 : Imm32S);
3252 }
3253 if (overlap1 != Imm8 && overlap1 != Imm8S
3254 && overlap1 != Imm16 && overlap1 != Imm32S
3255 && overlap1 != Imm32 && overlap1 != Imm64)
3256 {
3257 as_bad (_("no instruction mnemonic suffix given; can't determine immediate size %x %c"),overlap1, i.suffix);
3258 return 0;
3259 }
3260 }
3261 i.types[1] = overlap1;
3262
3263 overlap2 = i.types[2] & i.tm.operand_types[2];
3264 assert ((overlap2 & Imm) == 0);
3265 i.types[2] = overlap2;
3266
3267 return 1;
3268 }
3269
3270 static int
3271 process_operands (void)
3272 {
3273 /* Default segment register this instruction will use for memory
3274 accesses. 0 means unknown. This is only for optimizing out
3275 unnecessary segment overrides. */
3276 const seg_entry *default_seg = 0;
3277
3278 /* The imul $imm, %reg instruction is converted into
3279 imul $imm, %reg, %reg, and the clr %reg instruction
3280 is converted into xor %reg, %reg. */
3281 if (i.tm.opcode_modifier & regKludge)
3282 {
3283 unsigned int first_reg_op = (i.types[0] & Reg) ? 0 : 1;
3284 /* Pretend we saw the extra register operand. */
3285 assert (i.reg_operands == 1
3286 && i.op[first_reg_op + 1].regs == 0);
3287 i.op[first_reg_op + 1].regs = i.op[first_reg_op].regs;
3288 i.types[first_reg_op + 1] = i.types[first_reg_op];
3289 i.operands++;
3290 i.reg_operands++;
3291 }
3292
3293 if (i.tm.opcode_modifier & ShortForm)
3294 {
3295 /* The register or float register operand is in operand 0 or 1. */
3296 unsigned int op = (i.types[0] & (Reg | FloatReg)) ? 0 : 1;
3297 /* Register goes in low 3 bits of opcode. */
3298 i.tm.base_opcode |= i.op[op].regs->reg_num;
3299 if ((i.op[op].regs->reg_flags & RegRex) != 0)
3300 i.rex |= REX_EXTZ;
3301 if (!quiet_warnings && (i.tm.opcode_modifier & Ugh) != 0)
3302 {
3303 /* Warn about some common errors, but press on regardless.
3304 The first case can be generated by gcc (<= 2.8.1). */
3305 if (i.operands == 2)
3306 {
3307 /* Reversed arguments on faddp, fsubp, etc. */
3308 as_warn (_("translating to `%s %%%s,%%%s'"), i.tm.name,
3309 i.op[1].regs->reg_name,
3310 i.op[0].regs->reg_name);
3311 }
3312 else
3313 {
3314 /* Extraneous `l' suffix on fp insn. */
3315 as_warn (_("translating to `%s %%%s'"), i.tm.name,
3316 i.op[0].regs->reg_name);
3317 }
3318 }
3319 }
3320 else if (i.tm.opcode_modifier & Modrm)
3321 {
3322 /* The opcode is completed (modulo i.tm.extension_opcode which
3323 must be put into the modrm byte). Now, we make the modrm and
3324 index base bytes based on all the info we've collected. */
3325
3326 default_seg = build_modrm_byte ();
3327 }
3328 else if (i.tm.opcode_modifier & (Seg2ShortForm | Seg3ShortForm))
3329 {
3330 if (i.tm.base_opcode == POP_SEG_SHORT
3331 && i.op[0].regs->reg_num == 1)
3332 {
3333 as_bad (_("you can't `pop %%cs'"));
3334 return 0;
3335 }
3336 i.tm.base_opcode |= (i.op[0].regs->reg_num << 3);
3337 if ((i.op[0].regs->reg_flags & RegRex) != 0)
3338 i.rex |= REX_EXTZ;
3339 }
3340 else if ((i.tm.base_opcode & ~(D | W)) == MOV_AX_DISP32)
3341 {
3342 default_seg = &ds;
3343 }
3344 else if ((i.tm.opcode_modifier & IsString) != 0)
3345 {
3346 /* For the string instructions that allow a segment override
3347 on one of their operands, the default segment is ds. */
3348 default_seg = &ds;
3349 }
3350
3351 if ((i.tm.base_opcode == 0x8d /* lea */
3352 || (i.tm.cpu_flags & CpuSVME))
3353 && i.seg[0] && !quiet_warnings)
3354 as_warn (_("segment override on `%s' is ineffectual"), i.tm.name);
3355
3356 /* If a segment was explicitly specified, and the specified segment
3357 is not the default, use an opcode prefix to select it. If we
3358 never figured out what the default segment is, then default_seg
3359 will be zero at this point, and the specified segment prefix will
3360 always be used. */
3361 if ((i.seg[0]) && (i.seg[0] != default_seg))
3362 {
3363 if (!add_prefix (i.seg[0]->seg_prefix))
3364 return 0;
3365 }
3366 return 1;
3367 }
3368
3369 static const seg_entry *
3370 build_modrm_byte (void)
3371 {
3372 const seg_entry *default_seg = 0;
3373
3374 /* i.reg_operands MUST be the number of real register operands;
3375 implicit registers do not count. */
3376 if (i.reg_operands == 2)
3377 {
3378 unsigned int source, dest;
3379
3380 switch (i.operands)
3381 {
3382 case 2:
3383 source = 0;
3384 break;
3385 case 3:
3386 /* When there are 3 operands, one of them may be immediate,
3387 which may be the first or the last operand. Otherwise,
3388 the first operand must be shift count register (cl). */
3389 assert (i.imm_operands == 1
3390 || (i.imm_operands == 0
3391 && (i.types[0] & ShiftCount)));
3392 source = (i.types[0] & (Imm | ShiftCount)) ? 1 : 0;
3393 break;
3394 case 4:
3395 /* When there are 4 operands, the first two must be immediate
3396 operands. The source operand will be the 3rd one. */
3397 assert (i.imm_operands == 2
3398 && (i.types[0] & Imm)
3399 && (i.types[1] & Imm));
3400 source = 2;
3401 break;
3402 default:
3403 abort ();
3404 }
3405
3406 dest = source + 1;
3407
3408 i.rm.mode = 3;
3409 /* One of the register operands will be encoded in the i.tm.reg
3410 field, the other in the combined i.tm.mode and i.tm.regmem
3411 fields. If no form of this instruction supports a memory
3412 destination operand, then we assume the source operand may
3413 sometimes be a memory operand and so we need to store the
3414 destination in the i.rm.reg field. */
3415 if ((i.tm.operand_types[dest] & AnyMem) == 0)
3416 {
3417 i.rm.reg = i.op[dest].regs->reg_num;
3418 i.rm.regmem = i.op[source].regs->reg_num;
3419 if ((i.op[dest].regs->reg_flags & RegRex) != 0)
3420 i.rex |= REX_EXTX;
3421 if ((i.op[source].regs->reg_flags & RegRex) != 0)
3422 i.rex |= REX_EXTZ;
3423 }
3424 else
3425 {
3426 i.rm.reg = i.op[source].regs->reg_num;
3427 i.rm.regmem = i.op[dest].regs->reg_num;
3428 if ((i.op[dest].regs->reg_flags & RegRex) != 0)
3429 i.rex |= REX_EXTZ;
3430 if ((i.op[source].regs->reg_flags & RegRex) != 0)
3431 i.rex |= REX_EXTX;
3432 }
3433 if (flag_code != CODE_64BIT && (i.rex & (REX_EXTX | REX_EXTZ)))
3434 {
3435 if (!((i.types[0] | i.types[1]) & Control))
3436 abort ();
3437 i.rex &= ~(REX_EXTX | REX_EXTZ);
3438 add_prefix (LOCK_PREFIX_OPCODE);
3439 }
3440 }
3441 else
3442 { /* If it's not 2 reg operands... */
3443 if (i.mem_operands)
3444 {
3445 unsigned int fake_zero_displacement = 0;
3446 unsigned int op;
3447
3448 for (op = 0; op < i.operands; op++)
3449 if ((i.types[op] & AnyMem))
3450 break;
3451 assert (op < i.operands);
3452
3453 default_seg = &ds;
3454
3455 if (i.base_reg == 0)
3456 {
3457 i.rm.mode = 0;
3458 if (!i.disp_operands)
3459 fake_zero_displacement = 1;
3460 if (i.index_reg == 0)
3461 {
3462 /* Operand is just <disp> */
3463 if (flag_code == CODE_64BIT)
3464 {
3465 /* 64bit mode overwrites the 32bit absolute
3466 addressing by RIP relative addressing and
3467 absolute addressing is encoded by one of the
3468 redundant SIB forms. */
3469 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3470 i.sib.base = NO_BASE_REGISTER;
3471 i.sib.index = NO_INDEX_REGISTER;
3472 i.types[op] = ((i.prefix[ADDR_PREFIX] == 0)
3473 ? Disp32S : Disp32);
3474 }
3475 else if ((flag_code == CODE_16BIT)
3476 ^ (i.prefix[ADDR_PREFIX] != 0))
3477 {
3478 i.rm.regmem = NO_BASE_REGISTER_16;
3479 i.types[op] = Disp16;
3480 }
3481 else
3482 {
3483 i.rm.regmem = NO_BASE_REGISTER;
3484 i.types[op] = Disp32;
3485 }
3486 }
3487 else /* !i.base_reg && i.index_reg */
3488 {
3489 i.sib.index = i.index_reg->reg_num;
3490 i.sib.base = NO_BASE_REGISTER;
3491 i.sib.scale = i.log2_scale_factor;
3492 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3493 i.types[op] &= ~Disp;
3494 if (flag_code != CODE_64BIT)
3495 i.types[op] |= Disp32; /* Must be 32 bit */
3496 else
3497 i.types[op] |= Disp32S;
3498 if ((i.index_reg->reg_flags & RegRex) != 0)
3499 i.rex |= REX_EXTY;
3500 }
3501 }
3502 /* RIP addressing for 64bit mode. */
3503 else if (i.base_reg->reg_type == BaseIndex)
3504 {
3505 i.rm.regmem = NO_BASE_REGISTER;
3506 i.types[op] &= ~ Disp;
3507 i.types[op] |= Disp32S;
3508 i.flags[op] |= Operand_PCrel;
3509 if (! i.disp_operands)
3510 fake_zero_displacement = 1;
3511 }
3512 else if (i.base_reg->reg_type & Reg16)
3513 {
3514 switch (i.base_reg->reg_num)
3515 {
3516 case 3: /* (%bx) */
3517 if (i.index_reg == 0)
3518 i.rm.regmem = 7;
3519 else /* (%bx,%si) -> 0, or (%bx,%di) -> 1 */
3520 i.rm.regmem = i.index_reg->reg_num - 6;
3521 break;
3522 case 5: /* (%bp) */
3523 default_seg = &ss;
3524 if (i.index_reg == 0)
3525 {
3526 i.rm.regmem = 6;
3527 if ((i.types[op] & Disp) == 0)
3528 {
3529 /* fake (%bp) into 0(%bp) */
3530 i.types[op] |= Disp8;
3531 fake_zero_displacement = 1;
3532 }
3533 }
3534 else /* (%bp,%si) -> 2, or (%bp,%di) -> 3 */
3535 i.rm.regmem = i.index_reg->reg_num - 6 + 2;
3536 break;
3537 default: /* (%si) -> 4 or (%di) -> 5 */
3538 i.rm.regmem = i.base_reg->reg_num - 6 + 4;
3539 }
3540 i.rm.mode = mode_from_disp_size (i.types[op]);
3541 }
3542 else /* i.base_reg and 32/64 bit mode */
3543 {
3544 if (flag_code == CODE_64BIT
3545 && (i.types[op] & Disp))
3546 i.types[op] = ((i.types[op] & Disp8)
3547 | (i.prefix[ADDR_PREFIX] == 0
3548 ? Disp32S : Disp32));
3549
3550 i.rm.regmem = i.base_reg->reg_num;
3551 if ((i.base_reg->reg_flags & RegRex) != 0)
3552 i.rex |= REX_EXTZ;
3553 i.sib.base = i.base_reg->reg_num;
3554 /* x86-64 ignores REX prefix bit here to avoid decoder
3555 complications. */
3556 if ((i.base_reg->reg_num & 7) == EBP_REG_NUM)
3557 {
3558 default_seg = &ss;
3559 if (i.disp_operands == 0)
3560 {
3561 fake_zero_displacement = 1;
3562 i.types[op] |= Disp8;
3563 }
3564 }
3565 else if (i.base_reg->reg_num == ESP_REG_NUM)
3566 {
3567 default_seg = &ss;
3568 }
3569 i.sib.scale = i.log2_scale_factor;
3570 if (i.index_reg == 0)
3571 {
3572 /* <disp>(%esp) becomes two byte modrm with no index
3573 register. We've already stored the code for esp
3574 in i.rm.regmem ie. ESCAPE_TO_TWO_BYTE_ADDRESSING.
3575 Any base register besides %esp will not use the
3576 extra modrm byte. */
3577 i.sib.index = NO_INDEX_REGISTER;
3578 #if !SCALE1_WHEN_NO_INDEX
3579 /* Another case where we force the second modrm byte. */
3580 if (i.log2_scale_factor)
3581 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3582 #endif
3583 }
3584 else
3585 {
3586 i.sib.index = i.index_reg->reg_num;
3587 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3588 if ((i.index_reg->reg_flags & RegRex) != 0)
3589 i.rex |= REX_EXTY;
3590 }
3591
3592 if (i.disp_operands
3593 && (i.reloc[op] == BFD_RELOC_386_TLS_DESC_CALL
3594 || i.reloc[op] == BFD_RELOC_X86_64_TLSDESC_CALL))
3595 i.rm.mode = 0;
3596 else
3597 i.rm.mode = mode_from_disp_size (i.types[op]);
3598 }
3599
3600 if (fake_zero_displacement)
3601 {
3602 /* Fakes a zero displacement assuming that i.types[op]
3603 holds the correct displacement size. */
3604 expressionS *exp;
3605
3606 assert (i.op[op].disps == 0);
3607 exp = &disp_expressions[i.disp_operands++];
3608 i.op[op].disps = exp;
3609 exp->X_op = O_constant;
3610 exp->X_add_number = 0;
3611 exp->X_add_symbol = (symbolS *) 0;
3612 exp->X_op_symbol = (symbolS *) 0;
3613 }
3614 }
3615
3616 /* Fill in i.rm.reg or i.rm.regmem field with register operand
3617 (if any) based on i.tm.extension_opcode. Again, we must be
3618 careful to make sure that segment/control/debug/test/MMX
3619 registers are coded into the i.rm.reg field. */
3620 if (i.reg_operands)
3621 {
3622 unsigned int op;
3623
3624 for (op = 0; op < i.operands; op++)
3625 if ((i.types[op] & (Reg | RegMMX | RegXMM
3626 | SReg2 | SReg3
3627 | Control | Debug | Test)))
3628 break;
3629 assert (op < i.operands);
3630
3631 /* If there is an extension opcode to put here, the register
3632 number must be put into the regmem field. */
3633 if (i.tm.extension_opcode != None)
3634 {
3635 i.rm.regmem = i.op[op].regs->reg_num;
3636 if ((i.op[op].regs->reg_flags & RegRex) != 0)
3637 i.rex |= REX_EXTZ;
3638 }
3639 else
3640 {
3641 i.rm.reg = i.op[op].regs->reg_num;
3642 if ((i.op[op].regs->reg_flags & RegRex) != 0)
3643 i.rex |= REX_EXTX;
3644 }
3645
3646 /* Now, if no memory operand has set i.rm.mode = 0, 1, 2 we
3647 must set it to 3 to indicate this is a register operand
3648 in the regmem field. */
3649 if (!i.mem_operands)
3650 i.rm.mode = 3;
3651 }
3652
3653 /* Fill in i.rm.reg field with extension opcode (if any). */
3654 if (i.tm.extension_opcode != None)
3655 i.rm.reg = i.tm.extension_opcode;
3656 }
3657 return default_seg;
3658 }
3659
3660 static void
3661 output_branch (void)
3662 {
3663 char *p;
3664 int code16;
3665 int prefix;
3666 relax_substateT subtype;
3667 symbolS *sym;
3668 offsetT off;
3669
3670 code16 = 0;
3671 if (flag_code == CODE_16BIT)
3672 code16 = CODE16;
3673
3674 prefix = 0;
3675 if (i.prefix[DATA_PREFIX] != 0)
3676 {
3677 prefix = 1;
3678 i.prefixes -= 1;
3679 code16 ^= CODE16;
3680 }
3681 /* Pentium4 branch hints. */
3682 if (i.prefix[SEG_PREFIX] == CS_PREFIX_OPCODE /* not taken */
3683 || i.prefix[SEG_PREFIX] == DS_PREFIX_OPCODE /* taken */)
3684 {
3685 prefix++;
3686 i.prefixes--;
3687 }
3688 if (i.prefix[REX_PREFIX] != 0)
3689 {
3690 prefix++;
3691 i.prefixes--;
3692 }
3693
3694 if (i.prefixes != 0 && !intel_syntax)
3695 as_warn (_("skipping prefixes on this instruction"));
3696
3697 /* It's always a symbol; End frag & setup for relax.
3698 Make sure there is enough room in this frag for the largest
3699 instruction we may generate in md_convert_frag. This is 2
3700 bytes for the opcode and room for the prefix and largest
3701 displacement. */
3702 frag_grow (prefix + 2 + 4);
3703 /* Prefix and 1 opcode byte go in fr_fix. */
3704 p = frag_more (prefix + 1);
3705 if (i.prefix[DATA_PREFIX] != 0)
3706 *p++ = DATA_PREFIX_OPCODE;
3707 if (i.prefix[SEG_PREFIX] == CS_PREFIX_OPCODE
3708 || i.prefix[SEG_PREFIX] == DS_PREFIX_OPCODE)
3709 *p++ = i.prefix[SEG_PREFIX];
3710 if (i.prefix[REX_PREFIX] != 0)
3711 *p++ = i.prefix[REX_PREFIX];
3712 *p = i.tm.base_opcode;
3713
3714 if ((unsigned char) *p == JUMP_PC_RELATIVE)
3715 subtype = ENCODE_RELAX_STATE (UNCOND_JUMP, SMALL);
3716 else if ((cpu_arch_flags & Cpu386) != 0)
3717 subtype = ENCODE_RELAX_STATE (COND_JUMP, SMALL);
3718 else
3719 subtype = ENCODE_RELAX_STATE (COND_JUMP86, SMALL);
3720 subtype |= code16;
3721
3722 sym = i.op[0].disps->X_add_symbol;
3723 off = i.op[0].disps->X_add_number;
3724
3725 if (i.op[0].disps->X_op != O_constant
3726 && i.op[0].disps->X_op != O_symbol)
3727 {
3728 /* Handle complex expressions. */
3729 sym = make_expr_symbol (i.op[0].disps);
3730 off = 0;
3731 }
3732
3733 /* 1 possible extra opcode + 4 byte displacement go in var part.
3734 Pass reloc in fr_var. */
3735 frag_var (rs_machine_dependent, 5, i.reloc[0], subtype, sym, off, p);
3736 }
3737
3738 static void
3739 output_jump (void)
3740 {
3741 char *p;
3742 int size;
3743 fixS *fixP;
3744
3745 if (i.tm.opcode_modifier & JumpByte)
3746 {
3747 /* This is a loop or jecxz type instruction. */
3748 size = 1;
3749 if (i.prefix[ADDR_PREFIX] != 0)
3750 {
3751 FRAG_APPEND_1_CHAR (ADDR_PREFIX_OPCODE);
3752 i.prefixes -= 1;
3753 }
3754 /* Pentium4 branch hints. */
3755 if (i.prefix[SEG_PREFIX] == CS_PREFIX_OPCODE /* not taken */
3756 || i.prefix[SEG_PREFIX] == DS_PREFIX_OPCODE /* taken */)
3757 {
3758 FRAG_APPEND_1_CHAR (i.prefix[SEG_PREFIX]);
3759 i.prefixes--;
3760 }
3761 }
3762 else
3763 {
3764 int code16;
3765
3766 code16 = 0;
3767 if (flag_code == CODE_16BIT)
3768 code16 = CODE16;
3769
3770 if (i.prefix[DATA_PREFIX] != 0)
3771 {
3772 FRAG_APPEND_1_CHAR (DATA_PREFIX_OPCODE);
3773 i.prefixes -= 1;
3774 code16 ^= CODE16;
3775 }
3776
3777 size = 4;
3778 if (code16)
3779 size = 2;
3780 }
3781
3782 if (i.prefix[REX_PREFIX] != 0)
3783 {
3784 FRAG_APPEND_1_CHAR (i.prefix[REX_PREFIX]);
3785 i.prefixes -= 1;
3786 }
3787
3788 if (i.prefixes != 0 && !intel_syntax)
3789 as_warn (_("skipping prefixes on this instruction"));
3790
3791 p = frag_more (1 + size);
3792 *p++ = i.tm.base_opcode;
3793
3794 fixP = fix_new_exp (frag_now, p - frag_now->fr_literal, size,
3795 i.op[0].disps, 1, reloc (size, 1, 1, i.reloc[0]));
3796
3797 /* All jumps handled here are signed, but don't use a signed limit
3798 check for 32 and 16 bit jumps as we want to allow wrap around at
3799 4G and 64k respectively. */
3800 if (size == 1)
3801 fixP->fx_signed = 1;
3802 }
3803
3804 static void
3805 output_interseg_jump (void)
3806 {
3807 char *p;
3808 int size;
3809 int prefix;
3810 int code16;
3811
3812 code16 = 0;
3813 if (flag_code == CODE_16BIT)
3814 code16 = CODE16;
3815
3816 prefix = 0;
3817 if (i.prefix[DATA_PREFIX] != 0)
3818 {
3819 prefix = 1;
3820 i.prefixes -= 1;
3821 code16 ^= CODE16;
3822 }
3823 if (i.prefix[REX_PREFIX] != 0)
3824 {
3825 prefix++;
3826 i.prefixes -= 1;
3827 }
3828
3829 size = 4;
3830 if (code16)
3831 size = 2;
3832
3833 if (i.prefixes != 0 && !intel_syntax)
3834 as_warn (_("skipping prefixes on this instruction"));
3835
3836 /* 1 opcode; 2 segment; offset */
3837 p = frag_more (prefix + 1 + 2 + size);
3838
3839 if (i.prefix[DATA_PREFIX] != 0)
3840 *p++ = DATA_PREFIX_OPCODE;
3841
3842 if (i.prefix[REX_PREFIX] != 0)
3843 *p++ = i.prefix[REX_PREFIX];
3844
3845 *p++ = i.tm.base_opcode;
3846 if (i.op[1].imms->X_op == O_constant)
3847 {
3848 offsetT n = i.op[1].imms->X_add_number;
3849
3850 if (size == 2
3851 && !fits_in_unsigned_word (n)
3852 && !fits_in_signed_word (n))
3853 {
3854 as_bad (_("16-bit jump out of range"));
3855 return;
3856 }
3857 md_number_to_chars (p, n, size);
3858 }
3859 else
3860 fix_new_exp (frag_now, p - frag_now->fr_literal, size,
3861 i.op[1].imms, 0, reloc (size, 0, 0, i.reloc[1]));
3862 if (i.op[0].imms->X_op != O_constant)
3863 as_bad (_("can't handle non absolute segment in `%s'"),
3864 i.tm.name);
3865 md_number_to_chars (p + size, (valueT) i.op[0].imms->X_add_number, 2);
3866 }
3867
3868 static void
3869 output_insn (void)
3870 {
3871 fragS *insn_start_frag;
3872 offsetT insn_start_off;
3873
3874 /* Tie dwarf2 debug info to the address at the start of the insn.
3875 We can't do this after the insn has been output as the current
3876 frag may have been closed off. eg. by frag_var. */
3877 dwarf2_emit_insn (0);
3878
3879 insn_start_frag = frag_now;
3880 insn_start_off = frag_now_fix ();
3881
3882 /* Output jumps. */
3883 if (i.tm.opcode_modifier & Jump)
3884 output_branch ();
3885 else if (i.tm.opcode_modifier & (JumpByte | JumpDword))
3886 output_jump ();
3887 else if (i.tm.opcode_modifier & JumpInterSegment)
3888 output_interseg_jump ();
3889 else
3890 {
3891 /* Output normal instructions here. */
3892 char *p;
3893 unsigned char *q;
3894 unsigned int prefix;
3895
3896 /* All opcodes on i386 have either 1 or 2 bytes. Supplemental
3897 Streaming SIMD extensions 3 Instructions have 3 bytes. We may
3898 use one more higher byte to specify a prefix the instruction
3899 requires. */
3900 if ((i.tm.cpu_flags & CpuSSSE3) != 0)
3901 {
3902 if (i.tm.base_opcode & 0xff000000)
3903 {
3904 prefix = (i.tm.base_opcode >> 24) & 0xff;
3905 goto check_prefix;
3906 }
3907 }
3908 else if ((i.tm.base_opcode & 0xff0000) != 0)
3909 {
3910 prefix = (i.tm.base_opcode >> 16) & 0xff;
3911 if ((i.tm.cpu_flags & CpuPadLock) != 0)
3912 {
3913 check_prefix:
3914 if (prefix != REPE_PREFIX_OPCODE
3915 || i.prefix[LOCKREP_PREFIX] != REPE_PREFIX_OPCODE)
3916 add_prefix (prefix);
3917 }
3918 else
3919 add_prefix (prefix);
3920 }
3921
3922 /* The prefix bytes. */
3923 for (q = i.prefix;
3924 q < i.prefix + sizeof (i.prefix) / sizeof (i.prefix[0]);
3925 q++)
3926 {
3927 if (*q)
3928 {
3929 p = frag_more (1);
3930 md_number_to_chars (p, (valueT) *q, 1);
3931 }
3932 }
3933
3934 /* Now the opcode; be careful about word order here! */
3935 if (fits_in_unsigned_byte (i.tm.base_opcode))
3936 {
3937 FRAG_APPEND_1_CHAR (i.tm.base_opcode);
3938 }
3939 else
3940 {
3941 if ((i.tm.cpu_flags & CpuSSSE3) != 0)
3942 {
3943 p = frag_more (3);
3944 *p++ = (i.tm.base_opcode >> 16) & 0xff;
3945 }
3946 else
3947 p = frag_more (2);
3948
3949 /* Put out high byte first: can't use md_number_to_chars! */
3950 *p++ = (i.tm.base_opcode >> 8) & 0xff;
3951 *p = i.tm.base_opcode & 0xff;
3952 }
3953
3954 /* Now the modrm byte and sib byte (if present). */
3955 if (i.tm.opcode_modifier & Modrm)
3956 {
3957 p = frag_more (1);
3958 md_number_to_chars (p,
3959 (valueT) (i.rm.regmem << 0
3960 | i.rm.reg << 3
3961 | i.rm.mode << 6),
3962 1);
3963 /* If i.rm.regmem == ESP (4)
3964 && i.rm.mode != (Register mode)
3965 && not 16 bit
3966 ==> need second modrm byte. */
3967 if (i.rm.regmem == ESCAPE_TO_TWO_BYTE_ADDRESSING
3968 && i.rm.mode != 3
3969 && !(i.base_reg && (i.base_reg->reg_type & Reg16) != 0))
3970 {
3971 p = frag_more (1);
3972 md_number_to_chars (p,
3973 (valueT) (i.sib.base << 0
3974 | i.sib.index << 3
3975 | i.sib.scale << 6),
3976 1);
3977 }
3978 }
3979
3980 if (i.disp_operands)
3981 output_disp (insn_start_frag, insn_start_off);
3982
3983 if (i.imm_operands)
3984 output_imm (insn_start_frag, insn_start_off);
3985 }
3986
3987 #ifdef DEBUG386
3988 if (flag_debug)
3989 {
3990 pi ("" /*line*/, &i);
3991 }
3992 #endif /* DEBUG386 */
3993 }
3994
3995 static void
3996 output_disp (fragS *insn_start_frag, offsetT insn_start_off)
3997 {
3998 char *p;
3999 unsigned int n;
4000
4001 for (n = 0; n < i.operands; n++)
4002 {
4003 if (i.types[n] & Disp)
4004 {
4005 if (i.op[n].disps->X_op == O_constant)
4006 {
4007 int size;
4008 offsetT val;
4009
4010 size = 4;
4011 if (i.types[n] & (Disp8 | Disp16 | Disp64))
4012 {
4013 size = 2;
4014 if (i.types[n] & Disp8)
4015 size = 1;
4016 if (i.types[n] & Disp64)
4017 size = 8;
4018 }
4019 val = offset_in_range (i.op[n].disps->X_add_number,
4020 size);
4021 p = frag_more (size);
4022 md_number_to_chars (p, val, size);
4023 }
4024 else
4025 {
4026 enum bfd_reloc_code_real reloc_type;
4027 int size = 4;
4028 int sign = 0;
4029 int pcrel = (i.flags[n] & Operand_PCrel) != 0;
4030
4031 /* The PC relative address is computed relative
4032 to the instruction boundary, so in case immediate
4033 fields follows, we need to adjust the value. */
4034 if (pcrel && i.imm_operands)
4035 {
4036 int imm_size = 4;
4037 unsigned int n1;
4038
4039 for (n1 = 0; n1 < i.operands; n1++)
4040 if (i.types[n1] & Imm)
4041 {
4042 if (i.types[n1] & (Imm8 | Imm8S | Imm16 | Imm64))
4043 {
4044 imm_size = 2;
4045 if (i.types[n1] & (Imm8 | Imm8S))
4046 imm_size = 1;
4047 if (i.types[n1] & Imm64)
4048 imm_size = 8;
4049 }
4050 break;
4051 }
4052 /* We should find the immediate. */
4053 if (n1 == i.operands)
4054 abort ();
4055 i.op[n].disps->X_add_number -= imm_size;
4056 }
4057
4058 if (i.types[n] & Disp32S)
4059 sign = 1;
4060
4061 if (i.types[n] & (Disp16 | Disp64))
4062 {
4063 size = 2;
4064 if (i.types[n] & Disp64)
4065 size = 8;
4066 }
4067
4068 p = frag_more (size);
4069 reloc_type = reloc (size, pcrel, sign, i.reloc[n]);
4070 if (GOT_symbol
4071 && GOT_symbol == i.op[n].disps->X_add_symbol
4072 && (((reloc_type == BFD_RELOC_32
4073 || reloc_type == BFD_RELOC_X86_64_32S
4074 || (reloc_type == BFD_RELOC_64
4075 && object_64bit))
4076 && (i.op[n].disps->X_op == O_symbol
4077 || (i.op[n].disps->X_op == O_add
4078 && ((symbol_get_value_expression
4079 (i.op[n].disps->X_op_symbol)->X_op)
4080 == O_subtract))))
4081 || reloc_type == BFD_RELOC_32_PCREL))
4082 {
4083 offsetT add;
4084
4085 if (insn_start_frag == frag_now)
4086 add = (p - frag_now->fr_literal) - insn_start_off;
4087 else
4088 {
4089 fragS *fr;
4090
4091 add = insn_start_frag->fr_fix - insn_start_off;
4092 for (fr = insn_start_frag->fr_next;
4093 fr && fr != frag_now; fr = fr->fr_next)
4094 add += fr->fr_fix;
4095 add += p - frag_now->fr_literal;
4096 }
4097
4098 if (!object_64bit)
4099 {
4100 reloc_type = BFD_RELOC_386_GOTPC;
4101 i.op[n].imms->X_add_number += add;
4102 }
4103 else if (reloc_type == BFD_RELOC_64)
4104 reloc_type = BFD_RELOC_X86_64_GOTPC64;
4105 else
4106 /* Don't do the adjustment for x86-64, as there
4107 the pcrel addressing is relative to the _next_
4108 insn, and that is taken care of in other code. */
4109 reloc_type = BFD_RELOC_X86_64_GOTPC32;
4110 }
4111 fix_new_exp (frag_now, p - frag_now->fr_literal, size,
4112 i.op[n].disps, pcrel, reloc_type);
4113 }
4114 }
4115 }
4116 }
4117
4118 static void
4119 output_imm (fragS *insn_start_frag, offsetT insn_start_off)
4120 {
4121 char *p;
4122 unsigned int n;
4123
4124 for (n = 0; n < i.operands; n++)
4125 {
4126 if (i.types[n] & Imm)
4127 {
4128 if (i.op[n].imms->X_op == O_constant)
4129 {
4130 int size;
4131 offsetT val;
4132
4133 size = 4;
4134 if (i.types[n] & (Imm8 | Imm8S | Imm16 | Imm64))
4135 {
4136 size = 2;
4137 if (i.types[n] & (Imm8 | Imm8S))
4138 size = 1;
4139 else if (i.types[n] & Imm64)
4140 size = 8;
4141 }
4142 val = offset_in_range (i.op[n].imms->X_add_number,
4143 size);
4144 p = frag_more (size);
4145 md_number_to_chars (p, val, size);
4146 }
4147 else
4148 {
4149 /* Not absolute_section.
4150 Need a 32-bit fixup (don't support 8bit
4151 non-absolute imms). Try to support other
4152 sizes ... */
4153 enum bfd_reloc_code_real reloc_type;
4154 int size = 4;
4155 int sign = 0;
4156
4157 if ((i.types[n] & (Imm32S))
4158 && (i.suffix == QWORD_MNEM_SUFFIX
4159 || (!i.suffix && (i.tm.opcode_modifier & No_lSuf))))
4160 sign = 1;
4161 if (i.types[n] & (Imm8 | Imm8S | Imm16 | Imm64))
4162 {
4163 size = 2;
4164 if (i.types[n] & (Imm8 | Imm8S))
4165 size = 1;
4166 if (i.types[n] & Imm64)
4167 size = 8;
4168 }
4169
4170 p = frag_more (size);
4171 reloc_type = reloc (size, 0, sign, i.reloc[n]);
4172
4173 /* This is tough to explain. We end up with this one if we
4174 * have operands that look like
4175 * "_GLOBAL_OFFSET_TABLE_+[.-.L284]". The goal here is to
4176 * obtain the absolute address of the GOT, and it is strongly
4177 * preferable from a performance point of view to avoid using
4178 * a runtime relocation for this. The actual sequence of
4179 * instructions often look something like:
4180 *
4181 * call .L66
4182 * .L66:
4183 * popl %ebx
4184 * addl $_GLOBAL_OFFSET_TABLE_+[.-.L66],%ebx
4185 *
4186 * The call and pop essentially return the absolute address
4187 * of the label .L66 and store it in %ebx. The linker itself
4188 * will ultimately change the first operand of the addl so
4189 * that %ebx points to the GOT, but to keep things simple, the
4190 * .o file must have this operand set so that it generates not
4191 * the absolute address of .L66, but the absolute address of
4192 * itself. This allows the linker itself simply treat a GOTPC
4193 * relocation as asking for a pcrel offset to the GOT to be
4194 * added in, and the addend of the relocation is stored in the
4195 * operand field for the instruction itself.
4196 *
4197 * Our job here is to fix the operand so that it would add
4198 * the correct offset so that %ebx would point to itself. The
4199 * thing that is tricky is that .-.L66 will point to the
4200 * beginning of the instruction, so we need to further modify
4201 * the operand so that it will point to itself. There are
4202 * other cases where you have something like:
4203 *
4204 * .long $_GLOBAL_OFFSET_TABLE_+[.-.L66]
4205 *
4206 * and here no correction would be required. Internally in
4207 * the assembler we treat operands of this form as not being
4208 * pcrel since the '.' is explicitly mentioned, and I wonder
4209 * whether it would simplify matters to do it this way. Who
4210 * knows. In earlier versions of the PIC patches, the
4211 * pcrel_adjust field was used to store the correction, but
4212 * since the expression is not pcrel, I felt it would be
4213 * confusing to do it this way. */
4214
4215 if ((reloc_type == BFD_RELOC_32
4216 || reloc_type == BFD_RELOC_X86_64_32S
4217 || reloc_type == BFD_RELOC_64)
4218 && GOT_symbol
4219 && GOT_symbol == i.op[n].imms->X_add_symbol
4220 && (i.op[n].imms->X_op == O_symbol
4221 || (i.op[n].imms->X_op == O_add
4222 && ((symbol_get_value_expression
4223 (i.op[n].imms->X_op_symbol)->X_op)
4224 == O_subtract))))
4225 {
4226 offsetT add;
4227
4228 if (insn_start_frag == frag_now)
4229 add = (p - frag_now->fr_literal) - insn_start_off;
4230 else
4231 {
4232 fragS *fr;
4233
4234 add = insn_start_frag->fr_fix - insn_start_off;
4235 for (fr = insn_start_frag->fr_next;
4236 fr && fr != frag_now; fr = fr->fr_next)
4237 add += fr->fr_fix;
4238 add += p - frag_now->fr_literal;
4239 }
4240
4241 if (!object_64bit)
4242 reloc_type = BFD_RELOC_386_GOTPC;
4243 else if (size == 4)
4244 reloc_type = BFD_RELOC_X86_64_GOTPC32;
4245 else if (size == 8)
4246 reloc_type = BFD_RELOC_X86_64_GOTPC64;
4247 i.op[n].imms->X_add_number += add;
4248 }
4249 fix_new_exp (frag_now, p - frag_now->fr_literal, size,
4250 i.op[n].imms, 0, reloc_type);
4251 }
4252 }
4253 }
4254 }
4255 \f
4256 /* x86_cons_fix_new is called via the expression parsing code when a
4257 reloc is needed. We use this hook to get the correct .got reloc. */
4258 static enum bfd_reloc_code_real got_reloc = NO_RELOC;
4259 static int cons_sign = -1;
4260
4261 void
4262 x86_cons_fix_new (fragS *frag, unsigned int off, unsigned int len,
4263 expressionS *exp)
4264 {
4265 enum bfd_reloc_code_real r = reloc (len, 0, cons_sign, got_reloc);
4266
4267 got_reloc = NO_RELOC;
4268
4269 #ifdef TE_PE
4270 if (exp->X_op == O_secrel)
4271 {
4272 exp->X_op = O_symbol;
4273 r = BFD_RELOC_32_SECREL;
4274 }
4275 #endif
4276
4277 fix_new_exp (frag, off, len, exp, 0, r);
4278 }
4279
4280 #if (!defined (OBJ_ELF) && !defined (OBJ_MAYBE_ELF)) || defined (LEX_AT)
4281 # define lex_got(reloc, adjust, types) NULL
4282 #else
4283 /* Parse operands of the form
4284 <symbol>@GOTOFF+<nnn>
4285 and similar .plt or .got references.
4286
4287 If we find one, set up the correct relocation in RELOC and copy the
4288 input string, minus the `@GOTOFF' into a malloc'd buffer for
4289 parsing by the calling routine. Return this buffer, and if ADJUST
4290 is non-null set it to the length of the string we removed from the
4291 input line. Otherwise return NULL. */
4292 static char *
4293 lex_got (enum bfd_reloc_code_real *reloc,
4294 int *adjust,
4295 unsigned int *types)
4296 {
4297 /* Some of the relocations depend on the size of what field is to
4298 be relocated. But in our callers i386_immediate and i386_displacement
4299 we don't yet know the operand size (this will be set by insn
4300 matching). Hence we record the word32 relocation here,
4301 and adjust the reloc according to the real size in reloc(). */
4302 static const struct {
4303 const char *str;
4304 const enum bfd_reloc_code_real rel[2];
4305 const unsigned int types64;
4306 } gotrel[] = {
4307 { "PLTOFF", { 0, BFD_RELOC_X86_64_PLTOFF64 }, Imm64 },
4308 { "PLT", { BFD_RELOC_386_PLT32, BFD_RELOC_X86_64_PLT32 }, Imm32|Imm32S|Disp32 },
4309 { "GOTPLT", { 0, BFD_RELOC_X86_64_GOTPLT64 }, Imm64|Disp64 },
4310 { "GOTOFF", { BFD_RELOC_386_GOTOFF, BFD_RELOC_X86_64_GOTOFF64 }, Imm64|Disp64 },
4311 { "GOTPCREL", { 0, BFD_RELOC_X86_64_GOTPCREL }, Imm32|Imm32S|Disp32 },
4312 { "TLSGD", { BFD_RELOC_386_TLS_GD, BFD_RELOC_X86_64_TLSGD }, Imm32|Imm32S|Disp32 },
4313 { "TLSLDM", { BFD_RELOC_386_TLS_LDM, 0 }, 0 },
4314 { "TLSLD", { 0, BFD_RELOC_X86_64_TLSLD }, Imm32|Imm32S|Disp32 },
4315 { "GOTTPOFF", { BFD_RELOC_386_TLS_IE_32, BFD_RELOC_X86_64_GOTTPOFF }, Imm32|Imm32S|Disp32 },
4316 { "TPOFF", { BFD_RELOC_386_TLS_LE_32, BFD_RELOC_X86_64_TPOFF32 }, Imm32|Imm32S|Imm64|Disp32|Disp64 },
4317 { "NTPOFF", { BFD_RELOC_386_TLS_LE, 0 }, 0 },
4318 { "DTPOFF", { BFD_RELOC_386_TLS_LDO_32, BFD_RELOC_X86_64_DTPOFF32 }, Imm32|Imm32S|Imm64|Disp32|Disp64 },
4319 { "GOTNTPOFF",{ BFD_RELOC_386_TLS_GOTIE, 0 }, 0 },
4320 { "INDNTPOFF",{ BFD_RELOC_386_TLS_IE, 0 }, 0 },
4321 { "GOT", { BFD_RELOC_386_GOT32, BFD_RELOC_X86_64_GOT32 }, Imm32|Imm32S|Disp32|Imm64 },
4322 { "TLSDESC", { BFD_RELOC_386_TLS_GOTDESC, BFD_RELOC_X86_64_GOTPC32_TLSDESC }, Imm32|Imm32S|Disp32 },
4323 { "TLSCALL", { BFD_RELOC_386_TLS_DESC_CALL, BFD_RELOC_X86_64_TLSDESC_CALL }, Imm32|Imm32S|Disp32 }
4324 };
4325 char *cp;
4326 unsigned int j;
4327
4328 if (!IS_ELF)
4329 return NULL;
4330
4331 for (cp = input_line_pointer; *cp != '@'; cp++)
4332 if (is_end_of_line[(unsigned char) *cp])
4333 return NULL;
4334
4335 for (j = 0; j < sizeof (gotrel) / sizeof (gotrel[0]); j++)
4336 {
4337 int len;
4338
4339 len = strlen (gotrel[j].str);
4340 if (strncasecmp (cp + 1, gotrel[j].str, len) == 0)
4341 {
4342 if (gotrel[j].rel[object_64bit] != 0)
4343 {
4344 int first, second;
4345 char *tmpbuf, *past_reloc;
4346
4347 *reloc = gotrel[j].rel[object_64bit];
4348 if (adjust)
4349 *adjust = len;
4350
4351 if (types)
4352 {
4353 if (flag_code != CODE_64BIT)
4354 *types = Imm32|Disp32;
4355 else
4356 *types = gotrel[j].types64;
4357 }
4358
4359 if (GOT_symbol == NULL)
4360 GOT_symbol = symbol_find_or_make (GLOBAL_OFFSET_TABLE_NAME);
4361
4362 /* Replace the relocation token with ' ', so that
4363 errors like foo@GOTOFF1 will be detected. */
4364
4365 /* The length of the first part of our input line. */
4366 first = cp - input_line_pointer;
4367
4368 /* The second part goes from after the reloc token until
4369 (and including) an end_of_line char. Don't use strlen
4370 here as the end_of_line char may not be a NUL. */
4371 past_reloc = cp + 1 + len;
4372 for (cp = past_reloc; !is_end_of_line[(unsigned char) *cp++]; )
4373 ;
4374 second = cp - past_reloc;
4375
4376 /* Allocate and copy string. The trailing NUL shouldn't
4377 be necessary, but be safe. */
4378 tmpbuf = xmalloc (first + second + 2);
4379 memcpy (tmpbuf, input_line_pointer, first);
4380 tmpbuf[first] = ' ';
4381 memcpy (tmpbuf + first + 1, past_reloc, second);
4382 tmpbuf[first + second + 1] = '\0';
4383 return tmpbuf;
4384 }
4385
4386 as_bad (_("@%s reloc is not supported with %d-bit output format"),
4387 gotrel[j].str, 1 << (5 + object_64bit));
4388 return NULL;
4389 }
4390 }
4391
4392 /* Might be a symbol version string. Don't as_bad here. */
4393 return NULL;
4394 }
4395
4396 void
4397 x86_cons (expressionS *exp, int size)
4398 {
4399 if (size == 4 || (object_64bit && size == 8))
4400 {
4401 /* Handle @GOTOFF and the like in an expression. */
4402 char *save;
4403 char *gotfree_input_line;
4404 int adjust;
4405
4406 save = input_line_pointer;
4407 gotfree_input_line = lex_got (&got_reloc, &adjust, NULL);
4408 if (gotfree_input_line)
4409 input_line_pointer = gotfree_input_line;
4410
4411 expression (exp);
4412
4413 if (gotfree_input_line)
4414 {
4415 /* expression () has merrily parsed up to the end of line,
4416 or a comma - in the wrong buffer. Transfer how far
4417 input_line_pointer has moved to the right buffer. */
4418 input_line_pointer = (save
4419 + (input_line_pointer - gotfree_input_line)
4420 + adjust);
4421 free (gotfree_input_line);
4422 }
4423 }
4424 else
4425 expression (exp);
4426 }
4427 #endif
4428
4429 static void signed_cons (int size)
4430 {
4431 if (flag_code == CODE_64BIT)
4432 cons_sign = 1;
4433 cons (size);
4434 cons_sign = -1;
4435 }
4436
4437 #ifdef TE_PE
4438 static void
4439 pe_directive_secrel (dummy)
4440 int dummy ATTRIBUTE_UNUSED;
4441 {
4442 expressionS exp;
4443
4444 do
4445 {
4446 expression (&exp);
4447 if (exp.X_op == O_symbol)
4448 exp.X_op = O_secrel;
4449
4450 emit_expr (&exp, 4);
4451 }
4452 while (*input_line_pointer++ == ',');
4453
4454 input_line_pointer--;
4455 demand_empty_rest_of_line ();
4456 }
4457 #endif
4458
4459 static int
4460 i386_immediate (char *imm_start)
4461 {
4462 char *save_input_line_pointer;
4463 char *gotfree_input_line;
4464 segT exp_seg = 0;
4465 expressionS *exp;
4466 unsigned int types = ~0U;
4467
4468 if (i.imm_operands == MAX_IMMEDIATE_OPERANDS)
4469 {
4470 as_bad (_("at most %d immediate operands are allowed"),
4471 MAX_IMMEDIATE_OPERANDS);
4472 return 0;
4473 }
4474
4475 exp = &im_expressions[i.imm_operands++];
4476 i.op[this_operand].imms = exp;
4477
4478 if (is_space_char (*imm_start))
4479 ++imm_start;
4480
4481 save_input_line_pointer = input_line_pointer;
4482 input_line_pointer = imm_start;
4483
4484 gotfree_input_line = lex_got (&i.reloc[this_operand], NULL, &types);
4485 if (gotfree_input_line)
4486 input_line_pointer = gotfree_input_line;
4487
4488 exp_seg = expression (exp);
4489
4490 SKIP_WHITESPACE ();
4491 if (*input_line_pointer)
4492 as_bad (_("junk `%s' after expression"), input_line_pointer);
4493
4494 input_line_pointer = save_input_line_pointer;
4495 if (gotfree_input_line)
4496 free (gotfree_input_line);
4497
4498 if (exp->X_op == O_absent || exp->X_op == O_big)
4499 {
4500 /* Missing or bad expr becomes absolute 0. */
4501 as_bad (_("missing or invalid immediate expression `%s' taken as 0"),
4502 imm_start);
4503 exp->X_op = O_constant;
4504 exp->X_add_number = 0;
4505 exp->X_add_symbol = (symbolS *) 0;
4506 exp->X_op_symbol = (symbolS *) 0;
4507 }
4508 else if (exp->X_op == O_constant)
4509 {
4510 /* Size it properly later. */
4511 i.types[this_operand] |= Imm64;
4512 /* If BFD64, sign extend val. */
4513 if (!use_rela_relocations)
4514 if ((exp->X_add_number & ~(((addressT) 2 << 31) - 1)) == 0)
4515 exp->X_add_number = (exp->X_add_number ^ ((addressT) 1 << 31)) - ((addressT) 1 << 31);
4516 }
4517 #if (defined (OBJ_AOUT) || defined (OBJ_MAYBE_AOUT))
4518 else if (OUTPUT_FLAVOR == bfd_target_aout_flavour
4519 && exp_seg != absolute_section
4520 && exp_seg != text_section
4521 && exp_seg != data_section
4522 && exp_seg != bss_section
4523 && exp_seg != undefined_section
4524 && !bfd_is_com_section (exp_seg))
4525 {
4526 as_bad (_("unimplemented segment %s in operand"), exp_seg->name);
4527 return 0;
4528 }
4529 #endif
4530 else if (!intel_syntax && exp->X_op == O_register)
4531 {
4532 as_bad (_("illegal immediate register operand %s"), imm_start);
4533 return 0;
4534 }
4535 else
4536 {
4537 /* This is an address. The size of the address will be
4538 determined later, depending on destination register,
4539 suffix, or the default for the section. */
4540 i.types[this_operand] |= Imm8 | Imm16 | Imm32 | Imm32S | Imm64;
4541 i.types[this_operand] &= types;
4542 }
4543
4544 return 1;
4545 }
4546
4547 static char *
4548 i386_scale (char *scale)
4549 {
4550 offsetT val;
4551 char *save = input_line_pointer;
4552
4553 input_line_pointer = scale;
4554 val = get_absolute_expression ();
4555
4556 switch (val)
4557 {
4558 case 1:
4559 i.log2_scale_factor = 0;
4560 break;
4561 case 2:
4562 i.log2_scale_factor = 1;
4563 break;
4564 case 4:
4565 i.log2_scale_factor = 2;
4566 break;
4567 case 8:
4568 i.log2_scale_factor = 3;
4569 break;
4570 default:
4571 {
4572 char sep = *input_line_pointer;
4573
4574 *input_line_pointer = '\0';
4575 as_bad (_("expecting scale factor of 1, 2, 4, or 8: got `%s'"),
4576 scale);
4577 *input_line_pointer = sep;
4578 input_line_pointer = save;
4579 return NULL;
4580 }
4581 }
4582 if (i.log2_scale_factor != 0 && i.index_reg == 0)
4583 {
4584 as_warn (_("scale factor of %d without an index register"),
4585 1 << i.log2_scale_factor);
4586 #if SCALE1_WHEN_NO_INDEX
4587 i.log2_scale_factor = 0;
4588 #endif
4589 }
4590 scale = input_line_pointer;
4591 input_line_pointer = save;
4592 return scale;
4593 }
4594
4595 static int
4596 i386_displacement (char *disp_start, char *disp_end)
4597 {
4598 expressionS *exp;
4599 segT exp_seg = 0;
4600 char *save_input_line_pointer;
4601 char *gotfree_input_line;
4602 int bigdisp, override;
4603 unsigned int types = Disp;
4604
4605 if (i.disp_operands == MAX_MEMORY_OPERANDS)
4606 {
4607 as_bad (_("at most %d displacement operands are allowed"),
4608 MAX_MEMORY_OPERANDS);
4609 return 0;
4610 }
4611
4612 if ((i.types[this_operand] & JumpAbsolute)
4613 || !(current_templates->start->opcode_modifier & (Jump | JumpDword)))
4614 {
4615 bigdisp = Disp32;
4616 override = (i.prefix[ADDR_PREFIX] != 0);
4617 }
4618 else
4619 {
4620 /* For PC-relative branches, the width of the displacement
4621 is dependent upon data size, not address size. */
4622 bigdisp = 0;
4623 override = (i.prefix[DATA_PREFIX] != 0);
4624 }
4625 if (flag_code == CODE_64BIT)
4626 {
4627 if (!bigdisp)
4628 bigdisp = ((override || i.suffix == WORD_MNEM_SUFFIX)
4629 ? Disp16
4630 : Disp32S | Disp32);
4631 else if (!override)
4632 bigdisp = Disp64 | Disp32S | Disp32;
4633 }
4634 else
4635 {
4636 if (!bigdisp)
4637 {
4638 if (!override)
4639 override = (i.suffix == (flag_code != CODE_16BIT
4640 ? WORD_MNEM_SUFFIX
4641 : LONG_MNEM_SUFFIX));
4642 bigdisp = Disp32;
4643 }
4644 if ((flag_code == CODE_16BIT) ^ override)
4645 bigdisp = Disp16;
4646 }
4647 i.types[this_operand] |= bigdisp;
4648
4649 exp = &disp_expressions[i.disp_operands];
4650 i.op[this_operand].disps = exp;
4651 i.disp_operands++;
4652 save_input_line_pointer = input_line_pointer;
4653 input_line_pointer = disp_start;
4654 END_STRING_AND_SAVE (disp_end);
4655
4656 #ifndef GCC_ASM_O_HACK
4657 #define GCC_ASM_O_HACK 0
4658 #endif
4659 #if GCC_ASM_O_HACK
4660 END_STRING_AND_SAVE (disp_end + 1);
4661 if ((i.types[this_operand] & BaseIndex) != 0
4662 && displacement_string_end[-1] == '+')
4663 {
4664 /* This hack is to avoid a warning when using the "o"
4665 constraint within gcc asm statements.
4666 For instance:
4667
4668 #define _set_tssldt_desc(n,addr,limit,type) \
4669 __asm__ __volatile__ ( \
4670 "movw %w2,%0\n\t" \
4671 "movw %w1,2+%0\n\t" \
4672 "rorl $16,%1\n\t" \
4673 "movb %b1,4+%0\n\t" \
4674 "movb %4,5+%0\n\t" \
4675 "movb $0,6+%0\n\t" \
4676 "movb %h1,7+%0\n\t" \
4677 "rorl $16,%1" \
4678 : "=o"(*(n)) : "q" (addr), "ri"(limit), "i"(type))
4679
4680 This works great except that the output assembler ends
4681 up looking a bit weird if it turns out that there is
4682 no offset. You end up producing code that looks like:
4683
4684 #APP
4685 movw $235,(%eax)
4686 movw %dx,2+(%eax)
4687 rorl $16,%edx
4688 movb %dl,4+(%eax)
4689 movb $137,5+(%eax)
4690 movb $0,6+(%eax)
4691 movb %dh,7+(%eax)
4692 rorl $16,%edx
4693 #NO_APP
4694
4695 So here we provide the missing zero. */
4696
4697 *displacement_string_end = '0';
4698 }
4699 #endif
4700 gotfree_input_line = lex_got (&i.reloc[this_operand], NULL, &types);
4701 if (gotfree_input_line)
4702 input_line_pointer = gotfree_input_line;
4703
4704 exp_seg = expression (exp);
4705
4706 SKIP_WHITESPACE ();
4707 if (*input_line_pointer)
4708 as_bad (_("junk `%s' after expression"), input_line_pointer);
4709 #if GCC_ASM_O_HACK
4710 RESTORE_END_STRING (disp_end + 1);
4711 #endif
4712 RESTORE_END_STRING (disp_end);
4713 input_line_pointer = save_input_line_pointer;
4714 if (gotfree_input_line)
4715 free (gotfree_input_line);
4716
4717 /* We do this to make sure that the section symbol is in
4718 the symbol table. We will ultimately change the relocation
4719 to be relative to the beginning of the section. */
4720 if (i.reloc[this_operand] == BFD_RELOC_386_GOTOFF
4721 || i.reloc[this_operand] == BFD_RELOC_X86_64_GOTPCREL
4722 || i.reloc[this_operand] == BFD_RELOC_X86_64_GOTOFF64)
4723 {
4724 if (exp->X_op != O_symbol)
4725 {
4726 as_bad (_("bad expression used with @%s"),
4727 (i.reloc[this_operand] == BFD_RELOC_X86_64_GOTPCREL
4728 ? "GOTPCREL"
4729 : "GOTOFF"));
4730 return 0;
4731 }
4732
4733 if (S_IS_LOCAL (exp->X_add_symbol)
4734 && S_GET_SEGMENT (exp->X_add_symbol) != undefined_section)
4735 section_symbol (S_GET_SEGMENT (exp->X_add_symbol));
4736 exp->X_op = O_subtract;
4737 exp->X_op_symbol = GOT_symbol;
4738 if (i.reloc[this_operand] == BFD_RELOC_X86_64_GOTPCREL)
4739 i.reloc[this_operand] = BFD_RELOC_32_PCREL;
4740 else if (i.reloc[this_operand] == BFD_RELOC_X86_64_GOTOFF64)
4741 i.reloc[this_operand] = BFD_RELOC_64;
4742 else
4743 i.reloc[this_operand] = BFD_RELOC_32;
4744 }
4745
4746 if (exp->X_op == O_absent || exp->X_op == O_big)
4747 {
4748 /* Missing or bad expr becomes absolute 0. */
4749 as_bad (_("missing or invalid displacement expression `%s' taken as 0"),
4750 disp_start);
4751 exp->X_op = O_constant;
4752 exp->X_add_number = 0;
4753 exp->X_add_symbol = (symbolS *) 0;
4754 exp->X_op_symbol = (symbolS *) 0;
4755 }
4756
4757 #if (defined (OBJ_AOUT) || defined (OBJ_MAYBE_AOUT))
4758 if (exp->X_op != O_constant
4759 && OUTPUT_FLAVOR == bfd_target_aout_flavour
4760 && exp_seg != absolute_section
4761 && exp_seg != text_section
4762 && exp_seg != data_section
4763 && exp_seg != bss_section
4764 && exp_seg != undefined_section
4765 && !bfd_is_com_section (exp_seg))
4766 {
4767 as_bad (_("unimplemented segment %s in operand"), exp_seg->name);
4768 return 0;
4769 }
4770 #endif
4771
4772 if (!(i.types[this_operand] & ~Disp))
4773 i.types[this_operand] &= types;
4774
4775 return 1;
4776 }
4777
4778 /* Make sure the memory operand we've been dealt is valid.
4779 Return 1 on success, 0 on a failure. */
4780
4781 static int
4782 i386_index_check (const char *operand_string)
4783 {
4784 int ok;
4785 #if INFER_ADDR_PREFIX
4786 int fudged = 0;
4787
4788 tryprefix:
4789 #endif
4790 ok = 1;
4791 if ((current_templates->start->cpu_flags & CpuSVME)
4792 && current_templates->end[-1].operand_types[0] == AnyMem)
4793 {
4794 /* Memory operands of SVME insns are special in that they only allow
4795 rAX as their memory address and ignore any segment override. */
4796 unsigned RegXX;
4797
4798 /* SKINIT is even more restrictive: it always requires EAX. */
4799 if (strcmp (current_templates->start->name, "skinit") == 0)
4800 RegXX = Reg32;
4801 else if (flag_code == CODE_64BIT)
4802 RegXX = i.prefix[ADDR_PREFIX] == 0 ? Reg64 : Reg32;
4803 else
4804 RegXX = ((flag_code == CODE_16BIT) ^ (i.prefix[ADDR_PREFIX] != 0)
4805 ? Reg16
4806 : Reg32);
4807 if (!i.base_reg
4808 || !(i.base_reg->reg_type & Acc)
4809 || !(i.base_reg->reg_type & RegXX)
4810 || i.index_reg
4811 || (i.types[0] & Disp))
4812 ok = 0;
4813 }
4814 else if (flag_code == CODE_64BIT)
4815 {
4816 unsigned RegXX = (i.prefix[ADDR_PREFIX] == 0 ? Reg64 : Reg32);
4817
4818 if ((i.base_reg
4819 && ((i.base_reg->reg_type & RegXX) == 0)
4820 && (i.base_reg->reg_type != BaseIndex
4821 || i.index_reg))
4822 || (i.index_reg
4823 && ((i.index_reg->reg_type & (RegXX | BaseIndex))
4824 != (RegXX | BaseIndex))))
4825 ok = 0;
4826 }
4827 else
4828 {
4829 if ((flag_code == CODE_16BIT) ^ (i.prefix[ADDR_PREFIX] != 0))
4830 {
4831 /* 16bit checks. */
4832 if ((i.base_reg
4833 && ((i.base_reg->reg_type & (Reg16 | BaseIndex | RegRex))
4834 != (Reg16 | BaseIndex)))
4835 || (i.index_reg
4836 && (((i.index_reg->reg_type & (Reg16 | BaseIndex))
4837 != (Reg16 | BaseIndex))
4838 || !(i.base_reg
4839 && i.base_reg->reg_num < 6
4840 && i.index_reg->reg_num >= 6
4841 && i.log2_scale_factor == 0))))
4842 ok = 0;
4843 }
4844 else
4845 {
4846 /* 32bit checks. */
4847 if ((i.base_reg
4848 && (i.base_reg->reg_type & (Reg32 | RegRex)) != Reg32)
4849 || (i.index_reg
4850 && ((i.index_reg->reg_type & (Reg32 | BaseIndex | RegRex))
4851 != (Reg32 | BaseIndex))))
4852 ok = 0;
4853 }
4854 }
4855 if (!ok)
4856 {
4857 #if INFER_ADDR_PREFIX
4858 if (i.prefix[ADDR_PREFIX] == 0)
4859 {
4860 i.prefix[ADDR_PREFIX] = ADDR_PREFIX_OPCODE;
4861 i.prefixes += 1;
4862 /* Change the size of any displacement too. At most one of
4863 Disp16 or Disp32 is set.
4864 FIXME. There doesn't seem to be any real need for separate
4865 Disp16 and Disp32 flags. The same goes for Imm16 and Imm32.
4866 Removing them would probably clean up the code quite a lot. */
4867 if (flag_code != CODE_64BIT && (i.types[this_operand] & (Disp16 | Disp32)))
4868 i.types[this_operand] ^= (Disp16 | Disp32);
4869 fudged = 1;
4870 goto tryprefix;
4871 }
4872 if (fudged)
4873 as_bad (_("`%s' is not a valid base/index expression"),
4874 operand_string);
4875 else
4876 #endif
4877 as_bad (_("`%s' is not a valid %s bit base/index expression"),
4878 operand_string,
4879 flag_code_names[flag_code]);
4880 }
4881 return ok;
4882 }
4883
4884 /* Parse OPERAND_STRING into the i386_insn structure I. Returns non-zero
4885 on error. */
4886
4887 static int
4888 i386_operand (char *operand_string)
4889 {
4890 const reg_entry *r;
4891 char *end_op;
4892 char *op_string = operand_string;
4893
4894 if (is_space_char (*op_string))
4895 ++op_string;
4896
4897 /* We check for an absolute prefix (differentiating,
4898 for example, 'jmp pc_relative_label' from 'jmp *absolute_label'. */
4899 if (*op_string == ABSOLUTE_PREFIX)
4900 {
4901 ++op_string;
4902 if (is_space_char (*op_string))
4903 ++op_string;
4904 i.types[this_operand] |= JumpAbsolute;
4905 }
4906
4907 /* Check if operand is a register. */
4908 if ((r = parse_register (op_string, &end_op)) != NULL)
4909 {
4910 /* Check for a segment override by searching for ':' after a
4911 segment register. */
4912 op_string = end_op;
4913 if (is_space_char (*op_string))
4914 ++op_string;
4915 if (*op_string == ':' && (r->reg_type & (SReg2 | SReg3)))
4916 {
4917 switch (r->reg_num)
4918 {
4919 case 0:
4920 i.seg[i.mem_operands] = &es;
4921 break;
4922 case 1:
4923 i.seg[i.mem_operands] = &cs;
4924 break;
4925 case 2:
4926 i.seg[i.mem_operands] = &ss;
4927 break;
4928 case 3:
4929 i.seg[i.mem_operands] = &ds;
4930 break;
4931 case 4:
4932 i.seg[i.mem_operands] = &fs;
4933 break;
4934 case 5:
4935 i.seg[i.mem_operands] = &gs;
4936 break;
4937 }
4938
4939 /* Skip the ':' and whitespace. */
4940 ++op_string;
4941 if (is_space_char (*op_string))
4942 ++op_string;
4943
4944 if (!is_digit_char (*op_string)
4945 && !is_identifier_char (*op_string)
4946 && *op_string != '('
4947 && *op_string != ABSOLUTE_PREFIX)
4948 {
4949 as_bad (_("bad memory operand `%s'"), op_string);
4950 return 0;
4951 }
4952 /* Handle case of %es:*foo. */
4953 if (*op_string == ABSOLUTE_PREFIX)
4954 {
4955 ++op_string;
4956 if (is_space_char (*op_string))
4957 ++op_string;
4958 i.types[this_operand] |= JumpAbsolute;
4959 }
4960 goto do_memory_reference;
4961 }
4962 if (*op_string)
4963 {
4964 as_bad (_("junk `%s' after register"), op_string);
4965 return 0;
4966 }
4967 i.types[this_operand] |= r->reg_type & ~BaseIndex;
4968 i.op[this_operand].regs = r;
4969 i.reg_operands++;
4970 }
4971 else if (*op_string == REGISTER_PREFIX)
4972 {
4973 as_bad (_("bad register name `%s'"), op_string);
4974 return 0;
4975 }
4976 else if (*op_string == IMMEDIATE_PREFIX)
4977 {
4978 ++op_string;
4979 if (i.types[this_operand] & JumpAbsolute)
4980 {
4981 as_bad (_("immediate operand illegal with absolute jump"));
4982 return 0;
4983 }
4984 if (!i386_immediate (op_string))
4985 return 0;
4986 }
4987 else if (is_digit_char (*op_string)
4988 || is_identifier_char (*op_string)
4989 || *op_string == '(')
4990 {
4991 /* This is a memory reference of some sort. */
4992 char *base_string;
4993
4994 /* Start and end of displacement string expression (if found). */
4995 char *displacement_string_start;
4996 char *displacement_string_end;
4997
4998 do_memory_reference:
4999 if ((i.mem_operands == 1
5000 && (current_templates->start->opcode_modifier & IsString) == 0)
5001 || i.mem_operands == 2)
5002 {
5003 as_bad (_("too many memory references for `%s'"),
5004 current_templates->start->name);
5005 return 0;
5006 }
5007
5008 /* Check for base index form. We detect the base index form by
5009 looking for an ')' at the end of the operand, searching
5010 for the '(' matching it, and finding a REGISTER_PREFIX or ','
5011 after the '('. */
5012 base_string = op_string + strlen (op_string);
5013
5014 --base_string;
5015 if (is_space_char (*base_string))
5016 --base_string;
5017
5018 /* If we only have a displacement, set-up for it to be parsed later. */
5019 displacement_string_start = op_string;
5020 displacement_string_end = base_string + 1;
5021
5022 if (*base_string == ')')
5023 {
5024 char *temp_string;
5025 unsigned int parens_balanced = 1;
5026 /* We've already checked that the number of left & right ()'s are
5027 equal, so this loop will not be infinite. */
5028 do
5029 {
5030 base_string--;
5031 if (*base_string == ')')
5032 parens_balanced++;
5033 if (*base_string == '(')
5034 parens_balanced--;
5035 }
5036 while (parens_balanced);
5037
5038 temp_string = base_string;
5039
5040 /* Skip past '(' and whitespace. */
5041 ++base_string;
5042 if (is_space_char (*base_string))
5043 ++base_string;
5044
5045 if (*base_string == ','
5046 || ((i.base_reg = parse_register (base_string, &end_op)) != NULL))
5047 {
5048 displacement_string_end = temp_string;
5049
5050 i.types[this_operand] |= BaseIndex;
5051
5052 if (i.base_reg)
5053 {
5054 base_string = end_op;
5055 if (is_space_char (*base_string))
5056 ++base_string;
5057 }
5058
5059 /* There may be an index reg or scale factor here. */
5060 if (*base_string == ',')
5061 {
5062 ++base_string;
5063 if (is_space_char (*base_string))
5064 ++base_string;
5065
5066 if ((i.index_reg = parse_register (base_string, &end_op)) != NULL)
5067 {
5068 base_string = end_op;
5069 if (is_space_char (*base_string))
5070 ++base_string;
5071 if (*base_string == ',')
5072 {
5073 ++base_string;
5074 if (is_space_char (*base_string))
5075 ++base_string;
5076 }
5077 else if (*base_string != ')')
5078 {
5079 as_bad (_("expecting `,' or `)' after index register in `%s'"),
5080 operand_string);
5081 return 0;
5082 }
5083 }
5084 else if (*base_string == REGISTER_PREFIX)
5085 {
5086 as_bad (_("bad register name `%s'"), base_string);
5087 return 0;
5088 }
5089
5090 /* Check for scale factor. */
5091 if (*base_string != ')')
5092 {
5093 char *end_scale = i386_scale (base_string);
5094
5095 if (!end_scale)
5096 return 0;
5097
5098 base_string = end_scale;
5099 if (is_space_char (*base_string))
5100 ++base_string;
5101 if (*base_string != ')')
5102 {
5103 as_bad (_("expecting `)' after scale factor in `%s'"),
5104 operand_string);
5105 return 0;
5106 }
5107 }
5108 else if (!i.index_reg)
5109 {
5110 as_bad (_("expecting index register or scale factor after `,'; got '%c'"),
5111 *base_string);
5112 return 0;
5113 }
5114 }
5115 else if (*base_string != ')')
5116 {
5117 as_bad (_("expecting `,' or `)' after base register in `%s'"),
5118 operand_string);
5119 return 0;
5120 }
5121 }
5122 else if (*base_string == REGISTER_PREFIX)
5123 {
5124 as_bad (_("bad register name `%s'"), base_string);
5125 return 0;
5126 }
5127 }
5128
5129 /* If there's an expression beginning the operand, parse it,
5130 assuming displacement_string_start and
5131 displacement_string_end are meaningful. */
5132 if (displacement_string_start != displacement_string_end)
5133 {
5134 if (!i386_displacement (displacement_string_start,
5135 displacement_string_end))
5136 return 0;
5137 }
5138
5139 /* Special case for (%dx) while doing input/output op. */
5140 if (i.base_reg
5141 && i.base_reg->reg_type == (Reg16 | InOutPortReg)
5142 && i.index_reg == 0
5143 && i.log2_scale_factor == 0
5144 && i.seg[i.mem_operands] == 0
5145 && (i.types[this_operand] & Disp) == 0)
5146 {
5147 i.types[this_operand] = InOutPortReg;
5148 return 1;
5149 }
5150
5151 if (i386_index_check (operand_string) == 0)
5152 return 0;
5153 i.mem_operands++;
5154 }
5155 else
5156 {
5157 /* It's not a memory operand; argh! */
5158 as_bad (_("invalid char %s beginning operand %d `%s'"),
5159 output_invalid (*op_string),
5160 this_operand + 1,
5161 op_string);
5162 return 0;
5163 }
5164 return 1; /* Normal return. */
5165 }
5166 \f
5167 /* md_estimate_size_before_relax()
5168
5169 Called just before relax() for rs_machine_dependent frags. The x86
5170 assembler uses these frags to handle variable size jump
5171 instructions.
5172
5173 Any symbol that is now undefined will not become defined.
5174 Return the correct fr_subtype in the frag.
5175 Return the initial "guess for variable size of frag" to caller.
5176 The guess is actually the growth beyond the fixed part. Whatever
5177 we do to grow the fixed or variable part contributes to our
5178 returned value. */
5179
5180 int
5181 md_estimate_size_before_relax (fragP, segment)
5182 fragS *fragP;
5183 segT segment;
5184 {
5185 /* We've already got fragP->fr_subtype right; all we have to do is
5186 check for un-relaxable symbols. On an ELF system, we can't relax
5187 an externally visible symbol, because it may be overridden by a
5188 shared library. */
5189 if (S_GET_SEGMENT (fragP->fr_symbol) != segment
5190 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5191 || (IS_ELF
5192 && (S_IS_EXTERNAL (fragP->fr_symbol)
5193 || S_IS_WEAK (fragP->fr_symbol)))
5194 #endif
5195 )
5196 {
5197 /* Symbol is undefined in this segment, or we need to keep a
5198 reloc so that weak symbols can be overridden. */
5199 int size = (fragP->fr_subtype & CODE16) ? 2 : 4;
5200 enum bfd_reloc_code_real reloc_type;
5201 unsigned char *opcode;
5202 int old_fr_fix;
5203
5204 if (fragP->fr_var != NO_RELOC)
5205 reloc_type = fragP->fr_var;
5206 else if (size == 2)
5207 reloc_type = BFD_RELOC_16_PCREL;
5208 else
5209 reloc_type = BFD_RELOC_32_PCREL;
5210
5211 old_fr_fix = fragP->fr_fix;
5212 opcode = (unsigned char *) fragP->fr_opcode;
5213
5214 switch (TYPE_FROM_RELAX_STATE (fragP->fr_subtype))
5215 {
5216 case UNCOND_JUMP:
5217 /* Make jmp (0xeb) a (d)word displacement jump. */
5218 opcode[0] = 0xe9;
5219 fragP->fr_fix += size;
5220 fix_new (fragP, old_fr_fix, size,
5221 fragP->fr_symbol,
5222 fragP->fr_offset, 1,
5223 reloc_type);
5224 break;
5225
5226 case COND_JUMP86:
5227 if (size == 2
5228 && (!no_cond_jump_promotion || fragP->fr_var != NO_RELOC))
5229 {
5230 /* Negate the condition, and branch past an
5231 unconditional jump. */
5232 opcode[0] ^= 1;
5233 opcode[1] = 3;
5234 /* Insert an unconditional jump. */
5235 opcode[2] = 0xe9;
5236 /* We added two extra opcode bytes, and have a two byte
5237 offset. */
5238 fragP->fr_fix += 2 + 2;
5239 fix_new (fragP, old_fr_fix + 2, 2,
5240 fragP->fr_symbol,
5241 fragP->fr_offset, 1,
5242 reloc_type);
5243 break;
5244 }
5245 /* Fall through. */
5246
5247 case COND_JUMP:
5248 if (no_cond_jump_promotion && fragP->fr_var == NO_RELOC)
5249 {
5250 fixS *fixP;
5251
5252 fragP->fr_fix += 1;
5253 fixP = fix_new (fragP, old_fr_fix, 1,
5254 fragP->fr_symbol,
5255 fragP->fr_offset, 1,
5256 BFD_RELOC_8_PCREL);
5257 fixP->fx_signed = 1;
5258 break;
5259 }
5260
5261 /* This changes the byte-displacement jump 0x7N
5262 to the (d)word-displacement jump 0x0f,0x8N. */
5263 opcode[1] = opcode[0] + 0x10;
5264 opcode[0] = TWO_BYTE_OPCODE_ESCAPE;
5265 /* We've added an opcode byte. */
5266 fragP->fr_fix += 1 + size;
5267 fix_new (fragP, old_fr_fix + 1, size,
5268 fragP->fr_symbol,
5269 fragP->fr_offset, 1,
5270 reloc_type);
5271 break;
5272
5273 default:
5274 BAD_CASE (fragP->fr_subtype);
5275 break;
5276 }
5277 frag_wane (fragP);
5278 return fragP->fr_fix - old_fr_fix;
5279 }
5280
5281 /* Guess size depending on current relax state. Initially the relax
5282 state will correspond to a short jump and we return 1, because
5283 the variable part of the frag (the branch offset) is one byte
5284 long. However, we can relax a section more than once and in that
5285 case we must either set fr_subtype back to the unrelaxed state,
5286 or return the value for the appropriate branch. */
5287 return md_relax_table[fragP->fr_subtype].rlx_length;
5288 }
5289
5290 /* Called after relax() is finished.
5291
5292 In: Address of frag.
5293 fr_type == rs_machine_dependent.
5294 fr_subtype is what the address relaxed to.
5295
5296 Out: Any fixSs and constants are set up.
5297 Caller will turn frag into a ".space 0". */
5298
5299 void
5300 md_convert_frag (abfd, sec, fragP)
5301 bfd *abfd ATTRIBUTE_UNUSED;
5302 segT sec ATTRIBUTE_UNUSED;
5303 fragS *fragP;
5304 {
5305 unsigned char *opcode;
5306 unsigned char *where_to_put_displacement = NULL;
5307 offsetT target_address;
5308 offsetT opcode_address;
5309 unsigned int extension = 0;
5310 offsetT displacement_from_opcode_start;
5311
5312 opcode = (unsigned char *) fragP->fr_opcode;
5313
5314 /* Address we want to reach in file space. */
5315 target_address = S_GET_VALUE (fragP->fr_symbol) + fragP->fr_offset;
5316
5317 /* Address opcode resides at in file space. */
5318 opcode_address = fragP->fr_address + fragP->fr_fix;
5319
5320 /* Displacement from opcode start to fill into instruction. */
5321 displacement_from_opcode_start = target_address - opcode_address;
5322
5323 if ((fragP->fr_subtype & BIG) == 0)
5324 {
5325 /* Don't have to change opcode. */
5326 extension = 1; /* 1 opcode + 1 displacement */
5327 where_to_put_displacement = &opcode[1];
5328 }
5329 else
5330 {
5331 if (no_cond_jump_promotion
5332 && TYPE_FROM_RELAX_STATE (fragP->fr_subtype) != UNCOND_JUMP)
5333 as_warn_where (fragP->fr_file, fragP->fr_line, _("long jump required"));
5334
5335 switch (fragP->fr_subtype)
5336 {
5337 case ENCODE_RELAX_STATE (UNCOND_JUMP, BIG):
5338 extension = 4; /* 1 opcode + 4 displacement */
5339 opcode[0] = 0xe9;
5340 where_to_put_displacement = &opcode[1];
5341 break;
5342
5343 case ENCODE_RELAX_STATE (UNCOND_JUMP, BIG16):
5344 extension = 2; /* 1 opcode + 2 displacement */
5345 opcode[0] = 0xe9;
5346 where_to_put_displacement = &opcode[1];
5347 break;
5348
5349 case ENCODE_RELAX_STATE (COND_JUMP, BIG):
5350 case ENCODE_RELAX_STATE (COND_JUMP86, BIG):
5351 extension = 5; /* 2 opcode + 4 displacement */
5352 opcode[1] = opcode[0] + 0x10;
5353 opcode[0] = TWO_BYTE_OPCODE_ESCAPE;
5354 where_to_put_displacement = &opcode[2];
5355 break;
5356
5357 case ENCODE_RELAX_STATE (COND_JUMP, BIG16):
5358 extension = 3; /* 2 opcode + 2 displacement */
5359 opcode[1] = opcode[0] + 0x10;
5360 opcode[0] = TWO_BYTE_OPCODE_ESCAPE;
5361 where_to_put_displacement = &opcode[2];
5362 break;
5363
5364 case ENCODE_RELAX_STATE (COND_JUMP86, BIG16):
5365 extension = 4;
5366 opcode[0] ^= 1;
5367 opcode[1] = 3;
5368 opcode[2] = 0xe9;
5369 where_to_put_displacement = &opcode[3];
5370 break;
5371
5372 default:
5373 BAD_CASE (fragP->fr_subtype);
5374 break;
5375 }
5376 }
5377
5378 /* If size if less then four we are sure that the operand fits,
5379 but if it's 4, then it could be that the displacement is larger
5380 then -/+ 2GB. */
5381 if (DISP_SIZE_FROM_RELAX_STATE (fragP->fr_subtype) == 4
5382 && object_64bit
5383 && ((addressT) (displacement_from_opcode_start - extension
5384 + ((addressT) 1 << 31))
5385 > (((addressT) 2 << 31) - 1)))
5386 {
5387 as_bad_where (fragP->fr_file, fragP->fr_line,
5388 _("jump target out of range"));
5389 /* Make us emit 0. */
5390 displacement_from_opcode_start = extension;
5391 }
5392 /* Now put displacement after opcode. */
5393 md_number_to_chars ((char *) where_to_put_displacement,
5394 (valueT) (displacement_from_opcode_start - extension),
5395 DISP_SIZE_FROM_RELAX_STATE (fragP->fr_subtype));
5396 fragP->fr_fix += extension;
5397 }
5398 \f
5399 /* Size of byte displacement jmp. */
5400 int md_short_jump_size = 2;
5401
5402 /* Size of dword displacement jmp. */
5403 int md_long_jump_size = 5;
5404
5405 void
5406 md_create_short_jump (ptr, from_addr, to_addr, frag, to_symbol)
5407 char *ptr;
5408 addressT from_addr, to_addr;
5409 fragS *frag ATTRIBUTE_UNUSED;
5410 symbolS *to_symbol ATTRIBUTE_UNUSED;
5411 {
5412 offsetT offset;
5413
5414 offset = to_addr - (from_addr + 2);
5415 /* Opcode for byte-disp jump. */
5416 md_number_to_chars (ptr, (valueT) 0xeb, 1);
5417 md_number_to_chars (ptr + 1, (valueT) offset, 1);
5418 }
5419
5420 void
5421 md_create_long_jump (ptr, from_addr, to_addr, frag, to_symbol)
5422 char *ptr;
5423 addressT from_addr, to_addr;
5424 fragS *frag ATTRIBUTE_UNUSED;
5425 symbolS *to_symbol ATTRIBUTE_UNUSED;
5426 {
5427 offsetT offset;
5428
5429 offset = to_addr - (from_addr + 5);
5430 md_number_to_chars (ptr, (valueT) 0xe9, 1);
5431 md_number_to_chars (ptr + 1, (valueT) offset, 4);
5432 }
5433 \f
5434 /* Apply a fixup (fixS) to segment data, once it has been determined
5435 by our caller that we have all the info we need to fix it up.
5436
5437 On the 386, immediates, displacements, and data pointers are all in
5438 the same (little-endian) format, so we don't need to care about which
5439 we are handling. */
5440
5441 void
5442 md_apply_fix (fixP, valP, seg)
5443 /* The fix we're to put in. */
5444 fixS *fixP;
5445 /* Pointer to the value of the bits. */
5446 valueT *valP;
5447 /* Segment fix is from. */
5448 segT seg ATTRIBUTE_UNUSED;
5449 {
5450 char *p = fixP->fx_where + fixP->fx_frag->fr_literal;
5451 valueT value = *valP;
5452
5453 #if !defined (TE_Mach)
5454 if (fixP->fx_pcrel)
5455 {
5456 switch (fixP->fx_r_type)
5457 {
5458 default:
5459 break;
5460
5461 case BFD_RELOC_64:
5462 fixP->fx_r_type = BFD_RELOC_64_PCREL;
5463 break;
5464 case BFD_RELOC_32:
5465 case BFD_RELOC_X86_64_32S:
5466 fixP->fx_r_type = BFD_RELOC_32_PCREL;
5467 break;
5468 case BFD_RELOC_16:
5469 fixP->fx_r_type = BFD_RELOC_16_PCREL;
5470 break;
5471 case BFD_RELOC_8:
5472 fixP->fx_r_type = BFD_RELOC_8_PCREL;
5473 break;
5474 }
5475 }
5476
5477 if (fixP->fx_addsy != NULL
5478 && (fixP->fx_r_type == BFD_RELOC_32_PCREL
5479 || fixP->fx_r_type == BFD_RELOC_64_PCREL
5480 || fixP->fx_r_type == BFD_RELOC_16_PCREL
5481 || fixP->fx_r_type == BFD_RELOC_8_PCREL)
5482 && !use_rela_relocations)
5483 {
5484 /* This is a hack. There should be a better way to handle this.
5485 This covers for the fact that bfd_install_relocation will
5486 subtract the current location (for partial_inplace, PC relative
5487 relocations); see more below. */
5488 #ifndef OBJ_AOUT
5489 if (IS_ELF
5490 #ifdef TE_PE
5491 || OUTPUT_FLAVOR == bfd_target_coff_flavour
5492 #endif
5493 )
5494 value += fixP->fx_where + fixP->fx_frag->fr_address;
5495 #endif
5496 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5497 if (IS_ELF)
5498 {
5499 segT sym_seg = S_GET_SEGMENT (fixP->fx_addsy);
5500
5501 if ((sym_seg == seg
5502 || (symbol_section_p (fixP->fx_addsy)
5503 && sym_seg != absolute_section))
5504 && !generic_force_reloc (fixP))
5505 {
5506 /* Yes, we add the values in twice. This is because
5507 bfd_install_relocation subtracts them out again. I think
5508 bfd_install_relocation is broken, but I don't dare change
5509 it. FIXME. */
5510 value += fixP->fx_where + fixP->fx_frag->fr_address;
5511 }
5512 }
5513 #endif
5514 #if defined (OBJ_COFF) && defined (TE_PE)
5515 /* For some reason, the PE format does not store a
5516 section address offset for a PC relative symbol. */
5517 if (S_GET_SEGMENT (fixP->fx_addsy) != seg
5518 || S_IS_WEAK (fixP->fx_addsy))
5519 value += md_pcrel_from (fixP);
5520 #endif
5521 }
5522
5523 /* Fix a few things - the dynamic linker expects certain values here,
5524 and we must not disappoint it. */
5525 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5526 if (IS_ELF && fixP->fx_addsy)
5527 switch (fixP->fx_r_type)
5528 {
5529 case BFD_RELOC_386_PLT32:
5530 case BFD_RELOC_X86_64_PLT32:
5531 /* Make the jump instruction point to the address of the operand. At
5532 runtime we merely add the offset to the actual PLT entry. */
5533 value = -4;
5534 break;
5535
5536 case BFD_RELOC_386_TLS_GD:
5537 case BFD_RELOC_386_TLS_LDM:
5538 case BFD_RELOC_386_TLS_IE_32:
5539 case BFD_RELOC_386_TLS_IE:
5540 case BFD_RELOC_386_TLS_GOTIE:
5541 case BFD_RELOC_386_TLS_GOTDESC:
5542 case BFD_RELOC_X86_64_TLSGD:
5543 case BFD_RELOC_X86_64_TLSLD:
5544 case BFD_RELOC_X86_64_GOTTPOFF:
5545 case BFD_RELOC_X86_64_GOTPC32_TLSDESC:
5546 value = 0; /* Fully resolved at runtime. No addend. */
5547 /* Fallthrough */
5548 case BFD_RELOC_386_TLS_LE:
5549 case BFD_RELOC_386_TLS_LDO_32:
5550 case BFD_RELOC_386_TLS_LE_32:
5551 case BFD_RELOC_X86_64_DTPOFF32:
5552 case BFD_RELOC_X86_64_DTPOFF64:
5553 case BFD_RELOC_X86_64_TPOFF32:
5554 case BFD_RELOC_X86_64_TPOFF64:
5555 S_SET_THREAD_LOCAL (fixP->fx_addsy);
5556 break;
5557
5558 case BFD_RELOC_386_TLS_DESC_CALL:
5559 case BFD_RELOC_X86_64_TLSDESC_CALL:
5560 value = 0; /* Fully resolved at runtime. No addend. */
5561 S_SET_THREAD_LOCAL (fixP->fx_addsy);
5562 fixP->fx_done = 0;
5563 return;
5564
5565 case BFD_RELOC_386_GOT32:
5566 case BFD_RELOC_X86_64_GOT32:
5567 value = 0; /* Fully resolved at runtime. No addend. */
5568 break;
5569
5570 case BFD_RELOC_VTABLE_INHERIT:
5571 case BFD_RELOC_VTABLE_ENTRY:
5572 fixP->fx_done = 0;
5573 return;
5574
5575 default:
5576 break;
5577 }
5578 #endif /* defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) */
5579 *valP = value;
5580 #endif /* !defined (TE_Mach) */
5581
5582 /* Are we finished with this relocation now? */
5583 if (fixP->fx_addsy == NULL)
5584 fixP->fx_done = 1;
5585 else if (use_rela_relocations)
5586 {
5587 fixP->fx_no_overflow = 1;
5588 /* Remember value for tc_gen_reloc. */
5589 fixP->fx_addnumber = value;
5590 value = 0;
5591 }
5592
5593 md_number_to_chars (p, value, fixP->fx_size);
5594 }
5595 \f
5596 #define MAX_LITTLENUMS 6
5597
5598 /* Turn the string pointed to by litP into a floating point constant
5599 of type TYPE, and emit the appropriate bytes. The number of
5600 LITTLENUMS emitted is stored in *SIZEP. An error message is
5601 returned, or NULL on OK. */
5602
5603 char *
5604 md_atof (type, litP, sizeP)
5605 int type;
5606 char *litP;
5607 int *sizeP;
5608 {
5609 int prec;
5610 LITTLENUM_TYPE words[MAX_LITTLENUMS];
5611 LITTLENUM_TYPE *wordP;
5612 char *t;
5613
5614 switch (type)
5615 {
5616 case 'f':
5617 case 'F':
5618 prec = 2;
5619 break;
5620
5621 case 'd':
5622 case 'D':
5623 prec = 4;
5624 break;
5625
5626 case 'x':
5627 case 'X':
5628 prec = 5;
5629 break;
5630
5631 default:
5632 *sizeP = 0;
5633 return _("Bad call to md_atof ()");
5634 }
5635 t = atof_ieee (input_line_pointer, type, words);
5636 if (t)
5637 input_line_pointer = t;
5638
5639 *sizeP = prec * sizeof (LITTLENUM_TYPE);
5640 /* This loops outputs the LITTLENUMs in REVERSE order; in accord with
5641 the bigendian 386. */
5642 for (wordP = words + prec - 1; prec--;)
5643 {
5644 md_number_to_chars (litP, (valueT) (*wordP--), sizeof (LITTLENUM_TYPE));
5645 litP += sizeof (LITTLENUM_TYPE);
5646 }
5647 return 0;
5648 }
5649 \f
5650 static char output_invalid_buf[sizeof (unsigned char) * 2 + 6];
5651
5652 static char *
5653 output_invalid (int c)
5654 {
5655 if (ISPRINT (c))
5656 snprintf (output_invalid_buf, sizeof (output_invalid_buf),
5657 "'%c'", c);
5658 else
5659 snprintf (output_invalid_buf, sizeof (output_invalid_buf),
5660 "(0x%x)", (unsigned char) c);
5661 return output_invalid_buf;
5662 }
5663
5664 /* REG_STRING starts *before* REGISTER_PREFIX. */
5665
5666 static const reg_entry *
5667 parse_real_register (char *reg_string, char **end_op)
5668 {
5669 char *s = reg_string;
5670 char *p;
5671 char reg_name_given[MAX_REG_NAME_SIZE + 1];
5672 const reg_entry *r;
5673
5674 /* Skip possible REGISTER_PREFIX and possible whitespace. */
5675 if (*s == REGISTER_PREFIX)
5676 ++s;
5677
5678 if (is_space_char (*s))
5679 ++s;
5680
5681 p = reg_name_given;
5682 while ((*p++ = register_chars[(unsigned char) *s]) != '\0')
5683 {
5684 if (p >= reg_name_given + MAX_REG_NAME_SIZE)
5685 return (const reg_entry *) NULL;
5686 s++;
5687 }
5688
5689 /* For naked regs, make sure that we are not dealing with an identifier.
5690 This prevents confusing an identifier like `eax_var' with register
5691 `eax'. */
5692 if (allow_naked_reg && identifier_chars[(unsigned char) *s])
5693 return (const reg_entry *) NULL;
5694
5695 *end_op = s;
5696
5697 r = (const reg_entry *) hash_find (reg_hash, reg_name_given);
5698
5699 /* Handle floating point regs, allowing spaces in the (i) part. */
5700 if (r == i386_regtab /* %st is first entry of table */)
5701 {
5702 if (is_space_char (*s))
5703 ++s;
5704 if (*s == '(')
5705 {
5706 ++s;
5707 if (is_space_char (*s))
5708 ++s;
5709 if (*s >= '0' && *s <= '7')
5710 {
5711 r = &i386_float_regtab[*s - '0'];
5712 ++s;
5713 if (is_space_char (*s))
5714 ++s;
5715 if (*s == ')')
5716 {
5717 *end_op = s + 1;
5718 return r;
5719 }
5720 }
5721 /* We have "%st(" then garbage. */
5722 return (const reg_entry *) NULL;
5723 }
5724 }
5725
5726 if (r != NULL
5727 && ((r->reg_flags & (RegRex64 | RegRex)) | (r->reg_type & Reg64)) != 0
5728 && (r->reg_type != Control || !(cpu_arch_flags & CpuSledgehammer))
5729 && flag_code != CODE_64BIT)
5730 return (const reg_entry *) NULL;
5731
5732 return r;
5733 }
5734
5735 /* REG_STRING starts *before* REGISTER_PREFIX. */
5736
5737 static const reg_entry *
5738 parse_register (char *reg_string, char **end_op)
5739 {
5740 const reg_entry *r;
5741
5742 if (*reg_string == REGISTER_PREFIX || allow_naked_reg)
5743 r = parse_real_register (reg_string, end_op);
5744 else
5745 r = NULL;
5746 if (!r)
5747 {
5748 char *save = input_line_pointer;
5749 char c;
5750 symbolS *symbolP;
5751
5752 input_line_pointer = reg_string;
5753 c = get_symbol_end ();
5754 symbolP = symbol_find (reg_string);
5755 if (symbolP && S_GET_SEGMENT (symbolP) == reg_section)
5756 {
5757 const expressionS *e = symbol_get_value_expression (symbolP);
5758
5759 know (e->X_op == O_register);
5760 know (e->X_add_number >= 0 && (valueT) e->X_add_number < ARRAY_SIZE (i386_regtab));
5761 r = i386_regtab + e->X_add_number;
5762 *end_op = input_line_pointer;
5763 }
5764 *input_line_pointer = c;
5765 input_line_pointer = save;
5766 }
5767 return r;
5768 }
5769
5770 int
5771 i386_parse_name (char *name, expressionS *e, char *nextcharP)
5772 {
5773 const reg_entry *r;
5774 char *end = input_line_pointer;
5775
5776 *end = *nextcharP;
5777 r = parse_register (name, &input_line_pointer);
5778 if (r && end <= input_line_pointer)
5779 {
5780 *nextcharP = *input_line_pointer;
5781 *input_line_pointer = 0;
5782 e->X_op = O_register;
5783 e->X_add_number = r - i386_regtab;
5784 return 1;
5785 }
5786 input_line_pointer = end;
5787 *end = 0;
5788 return 0;
5789 }
5790
5791 void
5792 md_operand (expressionS *e)
5793 {
5794 if (*input_line_pointer == REGISTER_PREFIX)
5795 {
5796 char *end;
5797 const reg_entry *r = parse_real_register (input_line_pointer, &end);
5798
5799 if (r)
5800 {
5801 e->X_op = O_register;
5802 e->X_add_number = r - i386_regtab;
5803 input_line_pointer = end;
5804 }
5805 }
5806 }
5807
5808 \f
5809 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5810 const char *md_shortopts = "kVQ:sqn";
5811 #else
5812 const char *md_shortopts = "qn";
5813 #endif
5814
5815 #define OPTION_32 (OPTION_MD_BASE + 0)
5816 #define OPTION_64 (OPTION_MD_BASE + 1)
5817 #define OPTION_DIVIDE (OPTION_MD_BASE + 2)
5818 #define OPTION_MARCH (OPTION_MD_BASE + 3)
5819 #define OPTION_MTUNE (OPTION_MD_BASE + 4)
5820
5821 struct option md_longopts[] =
5822 {
5823 {"32", no_argument, NULL, OPTION_32},
5824 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) || defined(TE_PEP)
5825 {"64", no_argument, NULL, OPTION_64},
5826 #endif
5827 {"divide", no_argument, NULL, OPTION_DIVIDE},
5828 {"march", required_argument, NULL, OPTION_MARCH},
5829 {"mtune", required_argument, NULL, OPTION_MTUNE},
5830 {NULL, no_argument, NULL, 0}
5831 };
5832 size_t md_longopts_size = sizeof (md_longopts);
5833
5834 int
5835 md_parse_option (int c, char *arg)
5836 {
5837 unsigned int i;
5838
5839 switch (c)
5840 {
5841 case 'n':
5842 optimize_align_code = 0;
5843 break;
5844
5845 case 'q':
5846 quiet_warnings = 1;
5847 break;
5848
5849 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5850 /* -Qy, -Qn: SVR4 arguments controlling whether a .comment section
5851 should be emitted or not. FIXME: Not implemented. */
5852 case 'Q':
5853 break;
5854
5855 /* -V: SVR4 argument to print version ID. */
5856 case 'V':
5857 print_version_id ();
5858 break;
5859
5860 /* -k: Ignore for FreeBSD compatibility. */
5861 case 'k':
5862 break;
5863
5864 case 's':
5865 /* -s: On i386 Solaris, this tells the native assembler to use
5866 .stab instead of .stab.excl. We always use .stab anyhow. */
5867 break;
5868 #endif
5869 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) || defined(TE_PEP)
5870 case OPTION_64:
5871 {
5872 const char **list, **l;
5873
5874 list = bfd_target_list ();
5875 for (l = list; *l != NULL; l++)
5876 if (CONST_STRNEQ (*l, "elf64-x86-64")
5877 || strcmp (*l, "coff-x86-64") == 0
5878 || strcmp (*l, "pe-x86-64") == 0
5879 || strcmp (*l, "pei-x86-64") == 0)
5880 {
5881 default_arch = "x86_64";
5882 break;
5883 }
5884 if (*l == NULL)
5885 as_fatal (_("No compiled in support for x86_64"));
5886 free (list);
5887 }
5888 break;
5889 #endif
5890
5891 case OPTION_32:
5892 default_arch = "i386";
5893 break;
5894
5895 case OPTION_DIVIDE:
5896 #ifdef SVR4_COMMENT_CHARS
5897 {
5898 char *n, *t;
5899 const char *s;
5900
5901 n = (char *) xmalloc (strlen (i386_comment_chars) + 1);
5902 t = n;
5903 for (s = i386_comment_chars; *s != '\0'; s++)
5904 if (*s != '/')
5905 *t++ = *s;
5906 *t = '\0';
5907 i386_comment_chars = n;
5908 }
5909 #endif
5910 break;
5911
5912 case OPTION_MARCH:
5913 if (*arg == '.')
5914 as_fatal (_("Invalid -march= option: `%s'"), arg);
5915 for (i = 0; i < ARRAY_SIZE (cpu_arch); i++)
5916 {
5917 if (strcmp (arg, cpu_arch [i].name) == 0)
5918 {
5919 cpu_arch_isa = cpu_arch[i].type;
5920 cpu_arch_isa_flags = cpu_arch[i].flags;
5921 if (!cpu_arch_tune_set)
5922 {
5923 cpu_arch_tune = cpu_arch_isa;
5924 cpu_arch_tune_flags = cpu_arch_isa_flags;
5925 }
5926 break;
5927 }
5928 }
5929 if (i >= ARRAY_SIZE (cpu_arch))
5930 as_fatal (_("Invalid -march= option: `%s'"), arg);
5931 break;
5932
5933 case OPTION_MTUNE:
5934 if (*arg == '.')
5935 as_fatal (_("Invalid -mtune= option: `%s'"), arg);
5936 for (i = 0; i < ARRAY_SIZE (cpu_arch); i++)
5937 {
5938 if (strcmp (arg, cpu_arch [i].name) == 0)
5939 {
5940 cpu_arch_tune_set = 1;
5941 cpu_arch_tune = cpu_arch [i].type;
5942 cpu_arch_tune_flags = cpu_arch[i].flags;
5943 break;
5944 }
5945 }
5946 if (i >= ARRAY_SIZE (cpu_arch))
5947 as_fatal (_("Invalid -mtune= option: `%s'"), arg);
5948 break;
5949
5950 default:
5951 return 0;
5952 }
5953 return 1;
5954 }
5955
5956 void
5957 md_show_usage (stream)
5958 FILE *stream;
5959 {
5960 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5961 fprintf (stream, _("\
5962 -Q ignored\n\
5963 -V print assembler version number\n\
5964 -k ignored\n"));
5965 #endif
5966 fprintf (stream, _("\
5967 -n Do not optimize code alignment\n\
5968 -q quieten some warnings\n"));
5969 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5970 fprintf (stream, _("\
5971 -s ignored\n"));
5972 #endif
5973 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) || defined(TE_PEP)
5974 fprintf (stream, _("\
5975 --32/--64 generate 32bit/64bit code\n"));
5976 #endif
5977 #ifdef SVR4_COMMENT_CHARS
5978 fprintf (stream, _("\
5979 --divide do not treat `/' as a comment character\n"));
5980 #else
5981 fprintf (stream, _("\
5982 --divide ignored\n"));
5983 #endif
5984 fprintf (stream, _("\
5985 -march=CPU/-mtune=CPU generate code/optimize for CPU, where CPU is one of:\n\
5986 i386, i486, pentium, pentiumpro, pentium4, nocona,\n\
5987 core, core2, k6, athlon, k8, generic32, generic64\n"));
5988
5989 }
5990
5991 #if defined(TE_PEP)
5992 const char *
5993 x86_64_target_format (void)
5994 {
5995 if (strcmp (default_arch, "x86_64") == 0)
5996 {
5997 set_code_flag (CODE_64BIT);
5998 return COFF_TARGET_FORMAT;
5999 }
6000 else if (strcmp (default_arch, "i386") == 0)
6001 {
6002 set_code_flag (CODE_32BIT);
6003 return "coff-i386";
6004 }
6005
6006 as_fatal (_("Unknown architecture"));
6007 return NULL;
6008 }
6009 #endif
6010
6011 #if ((defined (OBJ_MAYBE_COFF) && defined (OBJ_MAYBE_AOUT)) \
6012 || defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF))
6013
6014 /* Pick the target format to use. */
6015
6016 const char *
6017 i386_target_format (void)
6018 {
6019 if (!strcmp (default_arch, "x86_64"))
6020 {
6021 set_code_flag (CODE_64BIT);
6022 if (cpu_arch_isa_flags == 0)
6023 cpu_arch_isa_flags = Cpu186|Cpu286|Cpu386|Cpu486
6024 |Cpu586|Cpu686|CpuP4|CpuMMX|CpuMMX2
6025 |CpuSSE|CpuSSE2;
6026 if (cpu_arch_tune_flags == 0)
6027 cpu_arch_tune_flags = Cpu186|Cpu286|Cpu386|Cpu486
6028 |Cpu586|Cpu686|CpuP4|CpuMMX|CpuMMX2
6029 |CpuSSE|CpuSSE2;
6030 }
6031 else if (!strcmp (default_arch, "i386"))
6032 {
6033 set_code_flag (CODE_32BIT);
6034 if (cpu_arch_isa_flags == 0)
6035 cpu_arch_isa_flags = Cpu186|Cpu286|Cpu386;
6036 if (cpu_arch_tune_flags == 0)
6037 cpu_arch_tune_flags = Cpu186|Cpu286|Cpu386;
6038 }
6039 else
6040 as_fatal (_("Unknown architecture"));
6041 switch (OUTPUT_FLAVOR)
6042 {
6043 #ifdef OBJ_MAYBE_AOUT
6044 case bfd_target_aout_flavour:
6045 return AOUT_TARGET_FORMAT;
6046 #endif
6047 #ifdef OBJ_MAYBE_COFF
6048 case bfd_target_coff_flavour:
6049 return "coff-i386";
6050 #endif
6051 #if defined (OBJ_MAYBE_ELF) || defined (OBJ_ELF)
6052 case bfd_target_elf_flavour:
6053 {
6054 if (flag_code == CODE_64BIT)
6055 {
6056 object_64bit = 1;
6057 use_rela_relocations = 1;
6058 }
6059 return flag_code == CODE_64BIT ? ELF_TARGET_FORMAT64 : ELF_TARGET_FORMAT;
6060 }
6061 #endif
6062 default:
6063 abort ();
6064 return NULL;
6065 }
6066 }
6067
6068 #endif /* OBJ_MAYBE_ more than one */
6069
6070 #if (defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF))
6071 void
6072 i386_elf_emit_arch_note (void)
6073 {
6074 if (IS_ELF && cpu_arch_name != NULL)
6075 {
6076 char *p;
6077 asection *seg = now_seg;
6078 subsegT subseg = now_subseg;
6079 Elf_Internal_Note i_note;
6080 Elf_External_Note e_note;
6081 asection *note_secp;
6082 int len;
6083
6084 /* Create the .note section. */
6085 note_secp = subseg_new (".note", 0);
6086 bfd_set_section_flags (stdoutput,
6087 note_secp,
6088 SEC_HAS_CONTENTS | SEC_READONLY);
6089
6090 /* Process the arch string. */
6091 len = strlen (cpu_arch_name);
6092
6093 i_note.namesz = len + 1;
6094 i_note.descsz = 0;
6095 i_note.type = NT_ARCH;
6096 p = frag_more (sizeof (e_note.namesz));
6097 md_number_to_chars (p, (valueT) i_note.namesz, sizeof (e_note.namesz));
6098 p = frag_more (sizeof (e_note.descsz));
6099 md_number_to_chars (p, (valueT) i_note.descsz, sizeof (e_note.descsz));
6100 p = frag_more (sizeof (e_note.type));
6101 md_number_to_chars (p, (valueT) i_note.type, sizeof (e_note.type));
6102 p = frag_more (len + 1);
6103 strcpy (p, cpu_arch_name);
6104
6105 frag_align (2, 0, 0);
6106
6107 subseg_set (seg, subseg);
6108 }
6109 }
6110 #endif
6111 \f
6112 symbolS *
6113 md_undefined_symbol (name)
6114 char *name;
6115 {
6116 if (name[0] == GLOBAL_OFFSET_TABLE_NAME[0]
6117 && name[1] == GLOBAL_OFFSET_TABLE_NAME[1]
6118 && name[2] == GLOBAL_OFFSET_TABLE_NAME[2]
6119 && strcmp (name, GLOBAL_OFFSET_TABLE_NAME) == 0)
6120 {
6121 if (!GOT_symbol)
6122 {
6123 if (symbol_find (name))
6124 as_bad (_("GOT already in symbol table"));
6125 GOT_symbol = symbol_new (name, undefined_section,
6126 (valueT) 0, &zero_address_frag);
6127 };
6128 return GOT_symbol;
6129 }
6130 return 0;
6131 }
6132
6133 /* Round up a section size to the appropriate boundary. */
6134
6135 valueT
6136 md_section_align (segment, size)
6137 segT segment ATTRIBUTE_UNUSED;
6138 valueT size;
6139 {
6140 #if (defined (OBJ_AOUT) || defined (OBJ_MAYBE_AOUT))
6141 if (OUTPUT_FLAVOR == bfd_target_aout_flavour)
6142 {
6143 /* For a.out, force the section size to be aligned. If we don't do
6144 this, BFD will align it for us, but it will not write out the
6145 final bytes of the section. This may be a bug in BFD, but it is
6146 easier to fix it here since that is how the other a.out targets
6147 work. */
6148 int align;
6149
6150 align = bfd_get_section_alignment (stdoutput, segment);
6151 size = ((size + (1 << align) - 1) & ((valueT) -1 << align));
6152 }
6153 #endif
6154
6155 return size;
6156 }
6157
6158 /* On the i386, PC-relative offsets are relative to the start of the
6159 next instruction. That is, the address of the offset, plus its
6160 size, since the offset is always the last part of the insn. */
6161
6162 long
6163 md_pcrel_from (fixS *fixP)
6164 {
6165 return fixP->fx_size + fixP->fx_where + fixP->fx_frag->fr_address;
6166 }
6167
6168 #ifndef I386COFF
6169
6170 static void
6171 s_bss (int ignore ATTRIBUTE_UNUSED)
6172 {
6173 int temp;
6174
6175 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
6176 if (IS_ELF)
6177 obj_elf_section_change_hook ();
6178 #endif
6179 temp = get_absolute_expression ();
6180 subseg_set (bss_section, (subsegT) temp);
6181 demand_empty_rest_of_line ();
6182 }
6183
6184 #endif
6185
6186 void
6187 i386_validate_fix (fixS *fixp)
6188 {
6189 if (fixp->fx_subsy && fixp->fx_subsy == GOT_symbol)
6190 {
6191 if (fixp->fx_r_type == BFD_RELOC_32_PCREL)
6192 {
6193 if (!object_64bit)
6194 abort ();
6195 fixp->fx_r_type = BFD_RELOC_X86_64_GOTPCREL;
6196 }
6197 else
6198 {
6199 if (!object_64bit)
6200 fixp->fx_r_type = BFD_RELOC_386_GOTOFF;
6201 else
6202 fixp->fx_r_type = BFD_RELOC_X86_64_GOTOFF64;
6203 }
6204 fixp->fx_subsy = 0;
6205 }
6206 }
6207
6208 arelent *
6209 tc_gen_reloc (section, fixp)
6210 asection *section ATTRIBUTE_UNUSED;
6211 fixS *fixp;
6212 {
6213 arelent *rel;
6214 bfd_reloc_code_real_type code;
6215
6216 switch (fixp->fx_r_type)
6217 {
6218 case BFD_RELOC_X86_64_PLT32:
6219 case BFD_RELOC_X86_64_GOT32:
6220 case BFD_RELOC_X86_64_GOTPCREL:
6221 case BFD_RELOC_386_PLT32:
6222 case BFD_RELOC_386_GOT32:
6223 case BFD_RELOC_386_GOTOFF:
6224 case BFD_RELOC_386_GOTPC:
6225 case BFD_RELOC_386_TLS_GD:
6226 case BFD_RELOC_386_TLS_LDM:
6227 case BFD_RELOC_386_TLS_LDO_32:
6228 case BFD_RELOC_386_TLS_IE_32:
6229 case BFD_RELOC_386_TLS_IE:
6230 case BFD_RELOC_386_TLS_GOTIE:
6231 case BFD_RELOC_386_TLS_LE_32:
6232 case BFD_RELOC_386_TLS_LE:
6233 case BFD_RELOC_386_TLS_GOTDESC:
6234 case BFD_RELOC_386_TLS_DESC_CALL:
6235 case BFD_RELOC_X86_64_TLSGD:
6236 case BFD_RELOC_X86_64_TLSLD:
6237 case BFD_RELOC_X86_64_DTPOFF32:
6238 case BFD_RELOC_X86_64_DTPOFF64:
6239 case BFD_RELOC_X86_64_GOTTPOFF:
6240 case BFD_RELOC_X86_64_TPOFF32:
6241 case BFD_RELOC_X86_64_TPOFF64:
6242 case BFD_RELOC_X86_64_GOTOFF64:
6243 case BFD_RELOC_X86_64_GOTPC32:
6244 case BFD_RELOC_X86_64_GOT64:
6245 case BFD_RELOC_X86_64_GOTPCREL64:
6246 case BFD_RELOC_X86_64_GOTPC64:
6247 case BFD_RELOC_X86_64_GOTPLT64:
6248 case BFD_RELOC_X86_64_PLTOFF64:
6249 case BFD_RELOC_X86_64_GOTPC32_TLSDESC:
6250 case BFD_RELOC_X86_64_TLSDESC_CALL:
6251 case BFD_RELOC_RVA:
6252 case BFD_RELOC_VTABLE_ENTRY:
6253 case BFD_RELOC_VTABLE_INHERIT:
6254 #ifdef TE_PE
6255 case BFD_RELOC_32_SECREL:
6256 #endif
6257 code = fixp->fx_r_type;
6258 break;
6259 case BFD_RELOC_X86_64_32S:
6260 if (!fixp->fx_pcrel)
6261 {
6262 /* Don't turn BFD_RELOC_X86_64_32S into BFD_RELOC_32. */
6263 code = fixp->fx_r_type;
6264 break;
6265 }
6266 default:
6267 if (fixp->fx_pcrel)
6268 {
6269 switch (fixp->fx_size)
6270 {
6271 default:
6272 as_bad_where (fixp->fx_file, fixp->fx_line,
6273 _("can not do %d byte pc-relative relocation"),
6274 fixp->fx_size);
6275 code = BFD_RELOC_32_PCREL;
6276 break;
6277 case 1: code = BFD_RELOC_8_PCREL; break;
6278 case 2: code = BFD_RELOC_16_PCREL; break;
6279 case 4: code = BFD_RELOC_32_PCREL; break;
6280 #ifdef BFD64
6281 case 8: code = BFD_RELOC_64_PCREL; break;
6282 #endif
6283 }
6284 }
6285 else
6286 {
6287 switch (fixp->fx_size)
6288 {
6289 default:
6290 as_bad_where (fixp->fx_file, fixp->fx_line,
6291 _("can not do %d byte relocation"),
6292 fixp->fx_size);
6293 code = BFD_RELOC_32;
6294 break;
6295 case 1: code = BFD_RELOC_8; break;
6296 case 2: code = BFD_RELOC_16; break;
6297 case 4: code = BFD_RELOC_32; break;
6298 #ifdef BFD64
6299 case 8: code = BFD_RELOC_64; break;
6300 #endif
6301 }
6302 }
6303 break;
6304 }
6305
6306 if ((code == BFD_RELOC_32
6307 || code == BFD_RELOC_32_PCREL
6308 || code == BFD_RELOC_X86_64_32S)
6309 && GOT_symbol
6310 && fixp->fx_addsy == GOT_symbol)
6311 {
6312 if (!object_64bit)
6313 code = BFD_RELOC_386_GOTPC;
6314 else
6315 code = BFD_RELOC_X86_64_GOTPC32;
6316 }
6317 if ((code == BFD_RELOC_64 || code == BFD_RELOC_64_PCREL)
6318 && GOT_symbol
6319 && fixp->fx_addsy == GOT_symbol)
6320 {
6321 code = BFD_RELOC_X86_64_GOTPC64;
6322 }
6323
6324 rel = (arelent *) xmalloc (sizeof (arelent));
6325 rel->sym_ptr_ptr = (asymbol **) xmalloc (sizeof (asymbol *));
6326 *rel->sym_ptr_ptr = symbol_get_bfdsym (fixp->fx_addsy);
6327
6328 rel->address = fixp->fx_frag->fr_address + fixp->fx_where;
6329
6330 if (!use_rela_relocations)
6331 {
6332 /* HACK: Since i386 ELF uses Rel instead of Rela, encode the
6333 vtable entry to be used in the relocation's section offset. */
6334 if (fixp->fx_r_type == BFD_RELOC_VTABLE_ENTRY)
6335 rel->address = fixp->fx_offset;
6336
6337 rel->addend = 0;
6338 }
6339 /* Use the rela in 64bit mode. */
6340 else
6341 {
6342 if (!fixp->fx_pcrel)
6343 rel->addend = fixp->fx_offset;
6344 else
6345 switch (code)
6346 {
6347 case BFD_RELOC_X86_64_PLT32:
6348 case BFD_RELOC_X86_64_GOT32:
6349 case BFD_RELOC_X86_64_GOTPCREL:
6350 case BFD_RELOC_X86_64_TLSGD:
6351 case BFD_RELOC_X86_64_TLSLD:
6352 case BFD_RELOC_X86_64_GOTTPOFF:
6353 case BFD_RELOC_X86_64_GOTPC32_TLSDESC:
6354 case BFD_RELOC_X86_64_TLSDESC_CALL:
6355 rel->addend = fixp->fx_offset - fixp->fx_size;
6356 break;
6357 default:
6358 rel->addend = (section->vma
6359 - fixp->fx_size
6360 + fixp->fx_addnumber
6361 + md_pcrel_from (fixp));
6362 break;
6363 }
6364 }
6365
6366 rel->howto = bfd_reloc_type_lookup (stdoutput, code);
6367 if (rel->howto == NULL)
6368 {
6369 as_bad_where (fixp->fx_file, fixp->fx_line,
6370 _("cannot represent relocation type %s"),
6371 bfd_get_reloc_code_name (code));
6372 /* Set howto to a garbage value so that we can keep going. */
6373 rel->howto = bfd_reloc_type_lookup (stdoutput, BFD_RELOC_32);
6374 assert (rel->howto != NULL);
6375 }
6376
6377 return rel;
6378 }
6379
6380 \f
6381 /* Parse operands using Intel syntax. This implements a recursive descent
6382 parser based on the BNF grammar published in Appendix B of the MASM 6.1
6383 Programmer's Guide.
6384
6385 FIXME: We do not recognize the full operand grammar defined in the MASM
6386 documentation. In particular, all the structure/union and
6387 high-level macro operands are missing.
6388
6389 Uppercase words are terminals, lower case words are non-terminals.
6390 Objects surrounded by double brackets '[[' ']]' are optional. Vertical
6391 bars '|' denote choices. Most grammar productions are implemented in
6392 functions called 'intel_<production>'.
6393
6394 Initial production is 'expr'.
6395
6396 addOp + | -
6397
6398 alpha [a-zA-Z]
6399
6400 binOp & | AND | \| | OR | ^ | XOR
6401
6402 byteRegister AL | AH | BL | BH | CL | CH | DL | DH
6403
6404 constant digits [[ radixOverride ]]
6405
6406 dataType BYTE | WORD | DWORD | FWORD | QWORD | TBYTE | OWORD | XMMWORD
6407
6408 digits decdigit
6409 | digits decdigit
6410 | digits hexdigit
6411
6412 decdigit [0-9]
6413
6414 e04 e04 addOp e05
6415 | e05
6416
6417 e05 e05 binOp e06
6418 | e06
6419
6420 e06 e06 mulOp e09
6421 | e09
6422
6423 e09 OFFSET e10
6424 | SHORT e10
6425 | + e10
6426 | - e10
6427 | ~ e10
6428 | NOT e10
6429 | e09 PTR e10
6430 | e09 : e10
6431 | e10
6432
6433 e10 e10 [ expr ]
6434 | e11
6435
6436 e11 ( expr )
6437 | [ expr ]
6438 | constant
6439 | dataType
6440 | id
6441 | $
6442 | register
6443
6444 => expr expr cmpOp e04
6445 | e04
6446
6447 gpRegister AX | EAX | BX | EBX | CX | ECX | DX | EDX
6448 | BP | EBP | SP | ESP | DI | EDI | SI | ESI
6449
6450 hexdigit a | b | c | d | e | f
6451 | A | B | C | D | E | F
6452
6453 id alpha
6454 | id alpha
6455 | id decdigit
6456
6457 mulOp * | / | % | MOD | << | SHL | >> | SHR
6458
6459 quote " | '
6460
6461 register specialRegister
6462 | gpRegister
6463 | byteRegister
6464
6465 segmentRegister CS | DS | ES | FS | GS | SS
6466
6467 specialRegister CR0 | CR2 | CR3 | CR4
6468 | DR0 | DR1 | DR2 | DR3 | DR6 | DR7
6469 | TR3 | TR4 | TR5 | TR6 | TR7
6470
6471 We simplify the grammar in obvious places (e.g., register parsing is
6472 done by calling parse_register) and eliminate immediate left recursion
6473 to implement a recursive-descent parser.
6474
6475 expr e04 expr'
6476
6477 expr' cmpOp e04 expr'
6478 | Empty
6479
6480 e04 e05 e04'
6481
6482 e04' addOp e05 e04'
6483 | Empty
6484
6485 e05 e06 e05'
6486
6487 e05' binOp e06 e05'
6488 | Empty
6489
6490 e06 e09 e06'
6491
6492 e06' mulOp e09 e06'
6493 | Empty
6494
6495 e09 OFFSET e10 e09'
6496 | SHORT e10'
6497 | + e10'
6498 | - e10'
6499 | ~ e10'
6500 | NOT e10'
6501 | e10 e09'
6502
6503 e09' PTR e10 e09'
6504 | : e10 e09'
6505 | Empty
6506
6507 e10 e11 e10'
6508
6509 e10' [ expr ] e10'
6510 | Empty
6511
6512 e11 ( expr )
6513 | [ expr ]
6514 | BYTE
6515 | WORD
6516 | DWORD
6517 | FWORD
6518 | QWORD
6519 | TBYTE
6520 | OWORD
6521 | XMMWORD
6522 | .
6523 | $
6524 | register
6525 | id
6526 | constant */
6527
6528 /* Parsing structure for the intel syntax parser. Used to implement the
6529 semantic actions for the operand grammar. */
6530 struct intel_parser_s
6531 {
6532 char *op_string; /* The string being parsed. */
6533 int got_a_float; /* Whether the operand is a float. */
6534 int op_modifier; /* Operand modifier. */
6535 int is_mem; /* 1 if operand is memory reference. */
6536 int in_offset; /* >=1 if parsing operand of offset. */
6537 int in_bracket; /* >=1 if parsing operand in brackets. */
6538 const reg_entry *reg; /* Last register reference found. */
6539 char *disp; /* Displacement string being built. */
6540 char *next_operand; /* Resume point when splitting operands. */
6541 };
6542
6543 static struct intel_parser_s intel_parser;
6544
6545 /* Token structure for parsing intel syntax. */
6546 struct intel_token
6547 {
6548 int code; /* Token code. */
6549 const reg_entry *reg; /* Register entry for register tokens. */
6550 char *str; /* String representation. */
6551 };
6552
6553 static struct intel_token cur_token, prev_token;
6554
6555 /* Token codes for the intel parser. Since T_SHORT is already used
6556 by COFF, undefine it first to prevent a warning. */
6557 #define T_NIL -1
6558 #define T_CONST 1
6559 #define T_REG 2
6560 #define T_BYTE 3
6561 #define T_WORD 4
6562 #define T_DWORD 5
6563 #define T_FWORD 6
6564 #define T_QWORD 7
6565 #define T_TBYTE 8
6566 #define T_XMMWORD 9
6567 #undef T_SHORT
6568 #define T_SHORT 10
6569 #define T_OFFSET 11
6570 #define T_PTR 12
6571 #define T_ID 13
6572 #define T_SHL 14
6573 #define T_SHR 15
6574
6575 /* Prototypes for intel parser functions. */
6576 static int intel_match_token (int);
6577 static void intel_putback_token (void);
6578 static void intel_get_token (void);
6579 static int intel_expr (void);
6580 static int intel_e04 (void);
6581 static int intel_e05 (void);
6582 static int intel_e06 (void);
6583 static int intel_e09 (void);
6584 static int intel_e10 (void);
6585 static int intel_e11 (void);
6586
6587 static int
6588 i386_intel_operand (char *operand_string, int got_a_float)
6589 {
6590 int ret;
6591 char *p;
6592
6593 p = intel_parser.op_string = xstrdup (operand_string);
6594 intel_parser.disp = (char *) xmalloc (strlen (operand_string) + 1);
6595
6596 for (;;)
6597 {
6598 /* Initialize token holders. */
6599 cur_token.code = prev_token.code = T_NIL;
6600 cur_token.reg = prev_token.reg = NULL;
6601 cur_token.str = prev_token.str = NULL;
6602
6603 /* Initialize parser structure. */
6604 intel_parser.got_a_float = got_a_float;
6605 intel_parser.op_modifier = 0;
6606 intel_parser.is_mem = 0;
6607 intel_parser.in_offset = 0;
6608 intel_parser.in_bracket = 0;
6609 intel_parser.reg = NULL;
6610 intel_parser.disp[0] = '\0';
6611 intel_parser.next_operand = NULL;
6612
6613 /* Read the first token and start the parser. */
6614 intel_get_token ();
6615 ret = intel_expr ();
6616
6617 if (!ret)
6618 break;
6619
6620 if (cur_token.code != T_NIL)
6621 {
6622 as_bad (_("invalid operand for '%s' ('%s' unexpected)"),
6623 current_templates->start->name, cur_token.str);
6624 ret = 0;
6625 }
6626 /* If we found a memory reference, hand it over to i386_displacement
6627 to fill in the rest of the operand fields. */
6628 else if (intel_parser.is_mem)
6629 {
6630 if ((i.mem_operands == 1
6631 && (current_templates->start->opcode_modifier & IsString) == 0)
6632 || i.mem_operands == 2)
6633 {
6634 as_bad (_("too many memory references for '%s'"),
6635 current_templates->start->name);
6636 ret = 0;
6637 }
6638 else
6639 {
6640 char *s = intel_parser.disp;
6641 i.mem_operands++;
6642
6643 if (!quiet_warnings && intel_parser.is_mem < 0)
6644 /* See the comments in intel_bracket_expr. */
6645 as_warn (_("Treating `%s' as memory reference"), operand_string);
6646
6647 /* Add the displacement expression. */
6648 if (*s != '\0')
6649 ret = i386_displacement (s, s + strlen (s));
6650 if (ret)
6651 {
6652 /* Swap base and index in 16-bit memory operands like
6653 [si+bx]. Since i386_index_check is also used in AT&T
6654 mode we have to do that here. */
6655 if (i.base_reg
6656 && i.index_reg
6657 && (i.base_reg->reg_type & Reg16)
6658 && (i.index_reg->reg_type & Reg16)
6659 && i.base_reg->reg_num >= 6
6660 && i.index_reg->reg_num < 6)
6661 {
6662 const reg_entry *base = i.index_reg;
6663
6664 i.index_reg = i.base_reg;
6665 i.base_reg = base;
6666 }
6667 ret = i386_index_check (operand_string);
6668 }
6669 }
6670 }
6671
6672 /* Constant and OFFSET expressions are handled by i386_immediate. */
6673 else if ((intel_parser.op_modifier & (1 << T_OFFSET))
6674 || intel_parser.reg == NULL)
6675 ret = i386_immediate (intel_parser.disp);
6676
6677 if (intel_parser.next_operand && this_operand >= MAX_OPERANDS - 1)
6678 ret = 0;
6679 if (!ret || !intel_parser.next_operand)
6680 break;
6681 intel_parser.op_string = intel_parser.next_operand;
6682 this_operand = i.operands++;
6683 }
6684
6685 free (p);
6686 free (intel_parser.disp);
6687
6688 return ret;
6689 }
6690
6691 #define NUM_ADDRESS_REGS (!!i.base_reg + !!i.index_reg)
6692
6693 /* expr e04 expr'
6694
6695 expr' cmpOp e04 expr'
6696 | Empty */
6697 static int
6698 intel_expr (void)
6699 {
6700 /* XXX Implement the comparison operators. */
6701 return intel_e04 ();
6702 }
6703
6704 /* e04 e05 e04'
6705
6706 e04' addOp e05 e04'
6707 | Empty */
6708 static int
6709 intel_e04 (void)
6710 {
6711 int nregs = -1;
6712
6713 for (;;)
6714 {
6715 if (!intel_e05())
6716 return 0;
6717
6718 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
6719 i.base_reg = i386_regtab + REGNAM_AL; /* al is invalid as base */
6720
6721 if (cur_token.code == '+')
6722 nregs = -1;
6723 else if (cur_token.code == '-')
6724 nregs = NUM_ADDRESS_REGS;
6725 else
6726 return 1;
6727
6728 strcat (intel_parser.disp, cur_token.str);
6729 intel_match_token (cur_token.code);
6730 }
6731 }
6732
6733 /* e05 e06 e05'
6734
6735 e05' binOp e06 e05'
6736 | Empty */
6737 static int
6738 intel_e05 (void)
6739 {
6740 int nregs = ~NUM_ADDRESS_REGS;
6741
6742 for (;;)
6743 {
6744 if (!intel_e06())
6745 return 0;
6746
6747 if (cur_token.code == '&' || cur_token.code == '|' || cur_token.code == '^')
6748 {
6749 char str[2];
6750
6751 str[0] = cur_token.code;
6752 str[1] = 0;
6753 strcat (intel_parser.disp, str);
6754 }
6755 else
6756 break;
6757
6758 intel_match_token (cur_token.code);
6759
6760 if (nregs < 0)
6761 nregs = ~nregs;
6762 }
6763 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
6764 i.base_reg = i386_regtab + REGNAM_AL + 1; /* cl is invalid as base */
6765 return 1;
6766 }
6767
6768 /* e06 e09 e06'
6769
6770 e06' mulOp e09 e06'
6771 | Empty */
6772 static int
6773 intel_e06 (void)
6774 {
6775 int nregs = ~NUM_ADDRESS_REGS;
6776
6777 for (;;)
6778 {
6779 if (!intel_e09())
6780 return 0;
6781
6782 if (cur_token.code == '*' || cur_token.code == '/' || cur_token.code == '%')
6783 {
6784 char str[2];
6785
6786 str[0] = cur_token.code;
6787 str[1] = 0;
6788 strcat (intel_parser.disp, str);
6789 }
6790 else if (cur_token.code == T_SHL)
6791 strcat (intel_parser.disp, "<<");
6792 else if (cur_token.code == T_SHR)
6793 strcat (intel_parser.disp, ">>");
6794 else
6795 break;
6796
6797 intel_match_token (cur_token.code);
6798
6799 if (nregs < 0)
6800 nregs = ~nregs;
6801 }
6802 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
6803 i.base_reg = i386_regtab + REGNAM_AL + 2; /* dl is invalid as base */
6804 return 1;
6805 }
6806
6807 /* e09 OFFSET e09
6808 | SHORT e09
6809 | + e09
6810 | - e09
6811 | ~ e09
6812 | NOT e09
6813 | e10 e09'
6814
6815 e09' PTR e10 e09'
6816 | : e10 e09'
6817 | Empty */
6818 static int
6819 intel_e09 (void)
6820 {
6821 int nregs = ~NUM_ADDRESS_REGS;
6822 int in_offset = 0;
6823
6824 for (;;)
6825 {
6826 /* Don't consume constants here. */
6827 if (cur_token.code == '+' || cur_token.code == '-')
6828 {
6829 /* Need to look one token ahead - if the next token
6830 is a constant, the current token is its sign. */
6831 int next_code;
6832
6833 intel_match_token (cur_token.code);
6834 next_code = cur_token.code;
6835 intel_putback_token ();
6836 if (next_code == T_CONST)
6837 break;
6838 }
6839
6840 /* e09 OFFSET e09 */
6841 if (cur_token.code == T_OFFSET)
6842 {
6843 if (!in_offset++)
6844 ++intel_parser.in_offset;
6845 }
6846
6847 /* e09 SHORT e09 */
6848 else if (cur_token.code == T_SHORT)
6849 intel_parser.op_modifier |= 1 << T_SHORT;
6850
6851 /* e09 + e09 */
6852 else if (cur_token.code == '+')
6853 strcat (intel_parser.disp, "+");
6854
6855 /* e09 - e09
6856 | ~ e09
6857 | NOT e09 */
6858 else if (cur_token.code == '-' || cur_token.code == '~')
6859 {
6860 char str[2];
6861
6862 if (nregs < 0)
6863 nregs = ~nregs;
6864 str[0] = cur_token.code;
6865 str[1] = 0;
6866 strcat (intel_parser.disp, str);
6867 }
6868
6869 /* e09 e10 e09' */
6870 else
6871 break;
6872
6873 intel_match_token (cur_token.code);
6874 }
6875
6876 for (;;)
6877 {
6878 if (!intel_e10 ())
6879 return 0;
6880
6881 /* e09' PTR e10 e09' */
6882 if (cur_token.code == T_PTR)
6883 {
6884 char suffix;
6885
6886 if (prev_token.code == T_BYTE)
6887 suffix = BYTE_MNEM_SUFFIX;
6888
6889 else if (prev_token.code == T_WORD)
6890 {
6891 if (current_templates->start->name[0] == 'l'
6892 && current_templates->start->name[2] == 's'
6893 && current_templates->start->name[3] == 0)
6894 suffix = BYTE_MNEM_SUFFIX; /* so it will cause an error */
6895 else if (intel_parser.got_a_float == 2) /* "fi..." */
6896 suffix = SHORT_MNEM_SUFFIX;
6897 else
6898 suffix = WORD_MNEM_SUFFIX;
6899 }
6900
6901 else if (prev_token.code == T_DWORD)
6902 {
6903 if (current_templates->start->name[0] == 'l'
6904 && current_templates->start->name[2] == 's'
6905 && current_templates->start->name[3] == 0)
6906 suffix = WORD_MNEM_SUFFIX;
6907 else if (flag_code == CODE_16BIT
6908 && (current_templates->start->opcode_modifier
6909 & (Jump | JumpDword)))
6910 suffix = LONG_DOUBLE_MNEM_SUFFIX;
6911 else if (intel_parser.got_a_float == 1) /* "f..." */
6912 suffix = SHORT_MNEM_SUFFIX;
6913 else
6914 suffix = LONG_MNEM_SUFFIX;
6915 }
6916
6917 else if (prev_token.code == T_FWORD)
6918 {
6919 if (current_templates->start->name[0] == 'l'
6920 && current_templates->start->name[2] == 's'
6921 && current_templates->start->name[3] == 0)
6922 suffix = LONG_MNEM_SUFFIX;
6923 else if (!intel_parser.got_a_float)
6924 {
6925 if (flag_code == CODE_16BIT)
6926 add_prefix (DATA_PREFIX_OPCODE);
6927 suffix = LONG_DOUBLE_MNEM_SUFFIX;
6928 }
6929 else
6930 suffix = BYTE_MNEM_SUFFIX; /* so it will cause an error */
6931 }
6932
6933 else if (prev_token.code == T_QWORD)
6934 {
6935 if (intel_parser.got_a_float == 1) /* "f..." */
6936 suffix = LONG_MNEM_SUFFIX;
6937 else
6938 suffix = QWORD_MNEM_SUFFIX;
6939 }
6940
6941 else if (prev_token.code == T_TBYTE)
6942 {
6943 if (intel_parser.got_a_float == 1)
6944 suffix = LONG_DOUBLE_MNEM_SUFFIX;
6945 else
6946 suffix = BYTE_MNEM_SUFFIX; /* so it will cause an error */
6947 }
6948
6949 else if (prev_token.code == T_XMMWORD)
6950 {
6951 /* XXX ignored for now, but accepted since gcc uses it */
6952 suffix = 0;
6953 }
6954
6955 else
6956 {
6957 as_bad (_("Unknown operand modifier `%s'"), prev_token.str);
6958 return 0;
6959 }
6960
6961 /* Operands for jump/call using 'ptr' notation denote absolute
6962 addresses. */
6963 if (current_templates->start->opcode_modifier & (Jump | JumpDword))
6964 i.types[this_operand] |= JumpAbsolute;
6965
6966 if (current_templates->start->base_opcode == 0x8d /* lea */)
6967 ;
6968 else if (!i.suffix)
6969 i.suffix = suffix;
6970 else if (i.suffix != suffix)
6971 {
6972 as_bad (_("Conflicting operand modifiers"));
6973 return 0;
6974 }
6975
6976 }
6977
6978 /* e09' : e10 e09' */
6979 else if (cur_token.code == ':')
6980 {
6981 if (prev_token.code != T_REG)
6982 {
6983 /* While {call,jmp} SSSS:OOOO is MASM syntax only when SSSS is a
6984 segment/group identifier (which we don't have), using comma
6985 as the operand separator there is even less consistent, since
6986 there all branches only have a single operand. */
6987 if (this_operand != 0
6988 || intel_parser.in_offset
6989 || intel_parser.in_bracket
6990 || (!(current_templates->start->opcode_modifier
6991 & (Jump|JumpDword|JumpInterSegment))
6992 && !(current_templates->start->operand_types[0]
6993 & JumpAbsolute)))
6994 return intel_match_token (T_NIL);
6995 /* Remember the start of the 2nd operand and terminate 1st
6996 operand here.
6997 XXX This isn't right, yet (when SSSS:OOOO is right operand of
6998 another expression), but it gets at least the simplest case
6999 (a plain number or symbol on the left side) right. */
7000 intel_parser.next_operand = intel_parser.op_string;
7001 *--intel_parser.op_string = '\0';
7002 return intel_match_token (':');
7003 }
7004 }
7005
7006 /* e09' Empty */
7007 else
7008 break;
7009
7010 intel_match_token (cur_token.code);
7011
7012 }
7013
7014 if (in_offset)
7015 {
7016 --intel_parser.in_offset;
7017 if (nregs < 0)
7018 nregs = ~nregs;
7019 if (NUM_ADDRESS_REGS > nregs)
7020 {
7021 as_bad (_("Invalid operand to `OFFSET'"));
7022 return 0;
7023 }
7024 intel_parser.op_modifier |= 1 << T_OFFSET;
7025 }
7026
7027 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
7028 i.base_reg = i386_regtab + REGNAM_AL + 3; /* bl is invalid as base */
7029 return 1;
7030 }
7031
7032 static int
7033 intel_bracket_expr (void)
7034 {
7035 int was_offset = intel_parser.op_modifier & (1 << T_OFFSET);
7036 const char *start = intel_parser.op_string;
7037 int len;
7038
7039 if (i.op[this_operand].regs)
7040 return intel_match_token (T_NIL);
7041
7042 intel_match_token ('[');
7043
7044 /* Mark as a memory operand only if it's not already known to be an
7045 offset expression. If it's an offset expression, we need to keep
7046 the brace in. */
7047 if (!intel_parser.in_offset)
7048 {
7049 ++intel_parser.in_bracket;
7050
7051 /* Operands for jump/call inside brackets denote absolute addresses. */
7052 if (current_templates->start->opcode_modifier & (Jump | JumpDword))
7053 i.types[this_operand] |= JumpAbsolute;
7054
7055 /* Unfortunately gas always diverged from MASM in a respect that can't
7056 be easily fixed without risking to break code sequences likely to be
7057 encountered (the testsuite even check for this): MASM doesn't consider
7058 an expression inside brackets unconditionally as a memory reference.
7059 When that is e.g. a constant, an offset expression, or the sum of the
7060 two, this is still taken as a constant load. gas, however, always
7061 treated these as memory references. As a compromise, we'll try to make
7062 offset expressions inside brackets work the MASM way (since that's
7063 less likely to be found in real world code), but make constants alone
7064 continue to work the traditional gas way. In either case, issue a
7065 warning. */
7066 intel_parser.op_modifier &= ~was_offset;
7067 }
7068 else
7069 strcat (intel_parser.disp, "[");
7070
7071 /* Add a '+' to the displacement string if necessary. */
7072 if (*intel_parser.disp != '\0'
7073 && *(intel_parser.disp + strlen (intel_parser.disp) - 1) != '+')
7074 strcat (intel_parser.disp, "+");
7075
7076 if (intel_expr ()
7077 && (len = intel_parser.op_string - start - 1,
7078 intel_match_token (']')))
7079 {
7080 /* Preserve brackets when the operand is an offset expression. */
7081 if (intel_parser.in_offset)
7082 strcat (intel_parser.disp, "]");
7083 else
7084 {
7085 --intel_parser.in_bracket;
7086 if (i.base_reg || i.index_reg)
7087 intel_parser.is_mem = 1;
7088 if (!intel_parser.is_mem)
7089 {
7090 if (!(intel_parser.op_modifier & (1 << T_OFFSET)))
7091 /* Defer the warning until all of the operand was parsed. */
7092 intel_parser.is_mem = -1;
7093 else if (!quiet_warnings)
7094 as_warn (_("`[%.*s]' taken to mean just `%.*s'"), len, start, len, start);
7095 }
7096 }
7097 intel_parser.op_modifier |= was_offset;
7098
7099 return 1;
7100 }
7101 return 0;
7102 }
7103
7104 /* e10 e11 e10'
7105
7106 e10' [ expr ] e10'
7107 | Empty */
7108 static int
7109 intel_e10 (void)
7110 {
7111 if (!intel_e11 ())
7112 return 0;
7113
7114 while (cur_token.code == '[')
7115 {
7116 if (!intel_bracket_expr ())
7117 return 0;
7118 }
7119
7120 return 1;
7121 }
7122
7123 /* e11 ( expr )
7124 | [ expr ]
7125 | BYTE
7126 | WORD
7127 | DWORD
7128 | FWORD
7129 | QWORD
7130 | TBYTE
7131 | OWORD
7132 | XMMWORD
7133 | $
7134 | .
7135 | register
7136 | id
7137 | constant */
7138 static int
7139 intel_e11 (void)
7140 {
7141 switch (cur_token.code)
7142 {
7143 /* e11 ( expr ) */
7144 case '(':
7145 intel_match_token ('(');
7146 strcat (intel_parser.disp, "(");
7147
7148 if (intel_expr () && intel_match_token (')'))
7149 {
7150 strcat (intel_parser.disp, ")");
7151 return 1;
7152 }
7153 return 0;
7154
7155 /* e11 [ expr ] */
7156 case '[':
7157 return intel_bracket_expr ();
7158
7159 /* e11 $
7160 | . */
7161 case '.':
7162 strcat (intel_parser.disp, cur_token.str);
7163 intel_match_token (cur_token.code);
7164
7165 /* Mark as a memory operand only if it's not already known to be an
7166 offset expression. */
7167 if (!intel_parser.in_offset)
7168 intel_parser.is_mem = 1;
7169
7170 return 1;
7171
7172 /* e11 register */
7173 case T_REG:
7174 {
7175 const reg_entry *reg = intel_parser.reg = cur_token.reg;
7176
7177 intel_match_token (T_REG);
7178
7179 /* Check for segment change. */
7180 if (cur_token.code == ':')
7181 {
7182 if (!(reg->reg_type & (SReg2 | SReg3)))
7183 {
7184 as_bad (_("`%s' is not a valid segment register"), reg->reg_name);
7185 return 0;
7186 }
7187 else if (i.seg[i.mem_operands])
7188 as_warn (_("Extra segment override ignored"));
7189 else
7190 {
7191 if (!intel_parser.in_offset)
7192 intel_parser.is_mem = 1;
7193 switch (reg->reg_num)
7194 {
7195 case 0:
7196 i.seg[i.mem_operands] = &es;
7197 break;
7198 case 1:
7199 i.seg[i.mem_operands] = &cs;
7200 break;
7201 case 2:
7202 i.seg[i.mem_operands] = &ss;
7203 break;
7204 case 3:
7205 i.seg[i.mem_operands] = &ds;
7206 break;
7207 case 4:
7208 i.seg[i.mem_operands] = &fs;
7209 break;
7210 case 5:
7211 i.seg[i.mem_operands] = &gs;
7212 break;
7213 }
7214 }
7215 }
7216
7217 /* Not a segment register. Check for register scaling. */
7218 else if (cur_token.code == '*')
7219 {
7220 if (!intel_parser.in_bracket)
7221 {
7222 as_bad (_("Register scaling only allowed in memory operands"));
7223 return 0;
7224 }
7225
7226 if (reg->reg_type & Reg16) /* Disallow things like [si*1]. */
7227 reg = i386_regtab + REGNAM_AX + 4; /* sp is invalid as index */
7228 else if (i.index_reg)
7229 reg = i386_regtab + REGNAM_EAX + 4; /* esp is invalid as index */
7230
7231 /* What follows must be a valid scale. */
7232 intel_match_token ('*');
7233 i.index_reg = reg;
7234 i.types[this_operand] |= BaseIndex;
7235
7236 /* Set the scale after setting the register (otherwise,
7237 i386_scale will complain) */
7238 if (cur_token.code == '+' || cur_token.code == '-')
7239 {
7240 char *str, sign = cur_token.code;
7241 intel_match_token (cur_token.code);
7242 if (cur_token.code != T_CONST)
7243 {
7244 as_bad (_("Syntax error: Expecting a constant, got `%s'"),
7245 cur_token.str);
7246 return 0;
7247 }
7248 str = (char *) xmalloc (strlen (cur_token.str) + 2);
7249 strcpy (str + 1, cur_token.str);
7250 *str = sign;
7251 if (!i386_scale (str))
7252 return 0;
7253 free (str);
7254 }
7255 else if (!i386_scale (cur_token.str))
7256 return 0;
7257 intel_match_token (cur_token.code);
7258 }
7259
7260 /* No scaling. If this is a memory operand, the register is either a
7261 base register (first occurrence) or an index register (second
7262 occurrence). */
7263 else if (intel_parser.in_bracket)
7264 {
7265
7266 if (!i.base_reg)
7267 i.base_reg = reg;
7268 else if (!i.index_reg)
7269 i.index_reg = reg;
7270 else
7271 {
7272 as_bad (_("Too many register references in memory operand"));
7273 return 0;
7274 }
7275
7276 i.types[this_operand] |= BaseIndex;
7277 }
7278
7279 /* It's neither base nor index. */
7280 else if (!intel_parser.in_offset && !intel_parser.is_mem)
7281 {
7282 i.types[this_operand] |= reg->reg_type & ~BaseIndex;
7283 i.op[this_operand].regs = reg;
7284 i.reg_operands++;
7285 }
7286 else
7287 {
7288 as_bad (_("Invalid use of register"));
7289 return 0;
7290 }
7291
7292 /* Since registers are not part of the displacement string (except
7293 when we're parsing offset operands), we may need to remove any
7294 preceding '+' from the displacement string. */
7295 if (*intel_parser.disp != '\0'
7296 && !intel_parser.in_offset)
7297 {
7298 char *s = intel_parser.disp;
7299 s += strlen (s) - 1;
7300 if (*s == '+')
7301 *s = '\0';
7302 }
7303
7304 return 1;
7305 }
7306
7307 /* e11 BYTE
7308 | WORD
7309 | DWORD
7310 | FWORD
7311 | QWORD
7312 | TBYTE
7313 | OWORD
7314 | XMMWORD */
7315 case T_BYTE:
7316 case T_WORD:
7317 case T_DWORD:
7318 case T_FWORD:
7319 case T_QWORD:
7320 case T_TBYTE:
7321 case T_XMMWORD:
7322 intel_match_token (cur_token.code);
7323
7324 if (cur_token.code == T_PTR)
7325 return 1;
7326
7327 /* It must have been an identifier. */
7328 intel_putback_token ();
7329 cur_token.code = T_ID;
7330 /* FALLTHRU */
7331
7332 /* e11 id
7333 | constant */
7334 case T_ID:
7335 if (!intel_parser.in_offset && intel_parser.is_mem <= 0)
7336 {
7337 symbolS *symbolP;
7338
7339 /* The identifier represents a memory reference only if it's not
7340 preceded by an offset modifier and if it's not an equate. */
7341 symbolP = symbol_find(cur_token.str);
7342 if (!symbolP || S_GET_SEGMENT(symbolP) != absolute_section)
7343 intel_parser.is_mem = 1;
7344 }
7345 /* FALLTHRU */
7346
7347 case T_CONST:
7348 case '-':
7349 case '+':
7350 {
7351 char *save_str, sign = 0;
7352
7353 /* Allow constants that start with `+' or `-'. */
7354 if (cur_token.code == '-' || cur_token.code == '+')
7355 {
7356 sign = cur_token.code;
7357 intel_match_token (cur_token.code);
7358 if (cur_token.code != T_CONST)
7359 {
7360 as_bad (_("Syntax error: Expecting a constant, got `%s'"),
7361 cur_token.str);
7362 return 0;
7363 }
7364 }
7365
7366 save_str = (char *) xmalloc (strlen (cur_token.str) + 2);
7367 strcpy (save_str + !!sign, cur_token.str);
7368 if (sign)
7369 *save_str = sign;
7370
7371 /* Get the next token to check for register scaling. */
7372 intel_match_token (cur_token.code);
7373
7374 /* Check if this constant is a scaling factor for an index register. */
7375 if (cur_token.code == '*')
7376 {
7377 if (intel_match_token ('*') && cur_token.code == T_REG)
7378 {
7379 const reg_entry *reg = cur_token.reg;
7380
7381 if (!intel_parser.in_bracket)
7382 {
7383 as_bad (_("Register scaling only allowed in memory operands"));
7384 return 0;
7385 }
7386
7387 if (reg->reg_type & Reg16) /* Disallow things like [1*si]. */
7388 reg = i386_regtab + REGNAM_AX + 4; /* sp is invalid as index */
7389 else if (i.index_reg)
7390 reg = i386_regtab + REGNAM_EAX + 4; /* esp is invalid as index */
7391
7392 /* The constant is followed by `* reg', so it must be
7393 a valid scale. */
7394 i.index_reg = reg;
7395 i.types[this_operand] |= BaseIndex;
7396
7397 /* Set the scale after setting the register (otherwise,
7398 i386_scale will complain) */
7399 if (!i386_scale (save_str))
7400 return 0;
7401 intel_match_token (T_REG);
7402
7403 /* Since registers are not part of the displacement
7404 string, we may need to remove any preceding '+' from
7405 the displacement string. */
7406 if (*intel_parser.disp != '\0')
7407 {
7408 char *s = intel_parser.disp;
7409 s += strlen (s) - 1;
7410 if (*s == '+')
7411 *s = '\0';
7412 }
7413
7414 free (save_str);
7415
7416 return 1;
7417 }
7418
7419 /* The constant was not used for register scaling. Since we have
7420 already consumed the token following `*' we now need to put it
7421 back in the stream. */
7422 intel_putback_token ();
7423 }
7424
7425 /* Add the constant to the displacement string. */
7426 strcat (intel_parser.disp, save_str);
7427 free (save_str);
7428
7429 return 1;
7430 }
7431 }
7432
7433 as_bad (_("Unrecognized token '%s'"), cur_token.str);
7434 return 0;
7435 }
7436
7437 /* Match the given token against cur_token. If they match, read the next
7438 token from the operand string. */
7439 static int
7440 intel_match_token (int code)
7441 {
7442 if (cur_token.code == code)
7443 {
7444 intel_get_token ();
7445 return 1;
7446 }
7447 else
7448 {
7449 as_bad (_("Unexpected token `%s'"), cur_token.str);
7450 return 0;
7451 }
7452 }
7453
7454 /* Read a new token from intel_parser.op_string and store it in cur_token. */
7455 static void
7456 intel_get_token (void)
7457 {
7458 char *end_op;
7459 const reg_entry *reg;
7460 struct intel_token new_token;
7461
7462 new_token.code = T_NIL;
7463 new_token.reg = NULL;
7464 new_token.str = NULL;
7465
7466 /* Free the memory allocated to the previous token and move
7467 cur_token to prev_token. */
7468 if (prev_token.str)
7469 free (prev_token.str);
7470
7471 prev_token = cur_token;
7472
7473 /* Skip whitespace. */
7474 while (is_space_char (*intel_parser.op_string))
7475 intel_parser.op_string++;
7476
7477 /* Return an empty token if we find nothing else on the line. */
7478 if (*intel_parser.op_string == '\0')
7479 {
7480 cur_token = new_token;
7481 return;
7482 }
7483
7484 /* The new token cannot be larger than the remainder of the operand
7485 string. */
7486 new_token.str = (char *) xmalloc (strlen (intel_parser.op_string) + 1);
7487 new_token.str[0] = '\0';
7488
7489 if (strchr ("0123456789", *intel_parser.op_string))
7490 {
7491 char *p = new_token.str;
7492 char *q = intel_parser.op_string;
7493 new_token.code = T_CONST;
7494
7495 /* Allow any kind of identifier char to encompass floating point and
7496 hexadecimal numbers. */
7497 while (is_identifier_char (*q))
7498 *p++ = *q++;
7499 *p = '\0';
7500
7501 /* Recognize special symbol names [0-9][bf]. */
7502 if (strlen (intel_parser.op_string) == 2
7503 && (intel_parser.op_string[1] == 'b'
7504 || intel_parser.op_string[1] == 'f'))
7505 new_token.code = T_ID;
7506 }
7507
7508 else if ((reg = parse_register (intel_parser.op_string, &end_op)) != NULL)
7509 {
7510 size_t len = end_op - intel_parser.op_string;
7511
7512 new_token.code = T_REG;
7513 new_token.reg = reg;
7514
7515 memcpy (new_token.str, intel_parser.op_string, len);
7516 new_token.str[len] = '\0';
7517 }
7518
7519 else if (is_identifier_char (*intel_parser.op_string))
7520 {
7521 char *p = new_token.str;
7522 char *q = intel_parser.op_string;
7523
7524 /* A '.' or '$' followed by an identifier char is an identifier.
7525 Otherwise, it's operator '.' followed by an expression. */
7526 if ((*q == '.' || *q == '$') && !is_identifier_char (*(q + 1)))
7527 {
7528 new_token.code = '.';
7529 new_token.str[0] = '.';
7530 new_token.str[1] = '\0';
7531 }
7532 else
7533 {
7534 while (is_identifier_char (*q) || *q == '@')
7535 *p++ = *q++;
7536 *p = '\0';
7537
7538 if (strcasecmp (new_token.str, "NOT") == 0)
7539 new_token.code = '~';
7540
7541 else if (strcasecmp (new_token.str, "MOD") == 0)
7542 new_token.code = '%';
7543
7544 else if (strcasecmp (new_token.str, "AND") == 0)
7545 new_token.code = '&';
7546
7547 else if (strcasecmp (new_token.str, "OR") == 0)
7548 new_token.code = '|';
7549
7550 else if (strcasecmp (new_token.str, "XOR") == 0)
7551 new_token.code = '^';
7552
7553 else if (strcasecmp (new_token.str, "SHL") == 0)
7554 new_token.code = T_SHL;
7555
7556 else if (strcasecmp (new_token.str, "SHR") == 0)
7557 new_token.code = T_SHR;
7558
7559 else if (strcasecmp (new_token.str, "BYTE") == 0)
7560 new_token.code = T_BYTE;
7561
7562 else if (strcasecmp (new_token.str, "WORD") == 0)
7563 new_token.code = T_WORD;
7564
7565 else if (strcasecmp (new_token.str, "DWORD") == 0)
7566 new_token.code = T_DWORD;
7567
7568 else if (strcasecmp (new_token.str, "FWORD") == 0)
7569 new_token.code = T_FWORD;
7570
7571 else if (strcasecmp (new_token.str, "QWORD") == 0)
7572 new_token.code = T_QWORD;
7573
7574 else if (strcasecmp (new_token.str, "TBYTE") == 0
7575 /* XXX remove (gcc still uses it) */
7576 || strcasecmp (new_token.str, "XWORD") == 0)
7577 new_token.code = T_TBYTE;
7578
7579 else if (strcasecmp (new_token.str, "XMMWORD") == 0
7580 || strcasecmp (new_token.str, "OWORD") == 0)
7581 new_token.code = T_XMMWORD;
7582
7583 else if (strcasecmp (new_token.str, "PTR") == 0)
7584 new_token.code = T_PTR;
7585
7586 else if (strcasecmp (new_token.str, "SHORT") == 0)
7587 new_token.code = T_SHORT;
7588
7589 else if (strcasecmp (new_token.str, "OFFSET") == 0)
7590 {
7591 new_token.code = T_OFFSET;
7592
7593 /* ??? This is not mentioned in the MASM grammar but gcc
7594 makes use of it with -mintel-syntax. OFFSET may be
7595 followed by FLAT: */
7596 if (strncasecmp (q, " FLAT:", 6) == 0)
7597 strcat (new_token.str, " FLAT:");
7598 }
7599
7600 /* ??? This is not mentioned in the MASM grammar. */
7601 else if (strcasecmp (new_token.str, "FLAT") == 0)
7602 {
7603 new_token.code = T_OFFSET;
7604 if (*q == ':')
7605 strcat (new_token.str, ":");
7606 else
7607 as_bad (_("`:' expected"));
7608 }
7609
7610 else
7611 new_token.code = T_ID;
7612 }
7613 }
7614
7615 else if (strchr ("+-/*%|&^:[]()~", *intel_parser.op_string))
7616 {
7617 new_token.code = *intel_parser.op_string;
7618 new_token.str[0] = *intel_parser.op_string;
7619 new_token.str[1] = '\0';
7620 }
7621
7622 else if (strchr ("<>", *intel_parser.op_string)
7623 && *intel_parser.op_string == *(intel_parser.op_string + 1))
7624 {
7625 new_token.code = *intel_parser.op_string == '<' ? T_SHL : T_SHR;
7626 new_token.str[0] = *intel_parser.op_string;
7627 new_token.str[1] = *intel_parser.op_string;
7628 new_token.str[2] = '\0';
7629 }
7630
7631 else
7632 as_bad (_("Unrecognized token `%s'"), intel_parser.op_string);
7633
7634 intel_parser.op_string += strlen (new_token.str);
7635 cur_token = new_token;
7636 }
7637
7638 /* Put cur_token back into the token stream and make cur_token point to
7639 prev_token. */
7640 static void
7641 intel_putback_token (void)
7642 {
7643 if (cur_token.code != T_NIL)
7644 {
7645 intel_parser.op_string -= strlen (cur_token.str);
7646 free (cur_token.str);
7647 }
7648 cur_token = prev_token;
7649
7650 /* Forget prev_token. */
7651 prev_token.code = T_NIL;
7652 prev_token.reg = NULL;
7653 prev_token.str = NULL;
7654 }
7655
7656 int
7657 tc_x86_regname_to_dw2regnum (char *regname)
7658 {
7659 unsigned int regnum;
7660 unsigned int regnames_count;
7661 static const char *const regnames_32[] =
7662 {
7663 "eax", "ecx", "edx", "ebx",
7664 "esp", "ebp", "esi", "edi",
7665 "eip", "eflags", NULL,
7666 "st0", "st1", "st2", "st3",
7667 "st4", "st5", "st6", "st7",
7668 NULL, NULL,
7669 "xmm0", "xmm1", "xmm2", "xmm3",
7670 "xmm4", "xmm5", "xmm6", "xmm7",
7671 "mm0", "mm1", "mm2", "mm3",
7672 "mm4", "mm5", "mm6", "mm7",
7673 "fcw", "fsw", "mxcsr",
7674 "es", "cs", "ss", "ds", "fs", "gs", NULL, NULL,
7675 "tr", "ldtr"
7676 };
7677 static const char *const regnames_64[] =
7678 {
7679 "rax", "rdx", "rcx", "rbx",
7680 "rsi", "rdi", "rbp", "rsp",
7681 "r8", "r9", "r10", "r11",
7682 "r12", "r13", "r14", "r15",
7683 "rip",
7684 "xmm0", "xmm1", "xmm2", "xmm3",
7685 "xmm4", "xmm5", "xmm6", "xmm7",
7686 "xmm8", "xmm9", "xmm10", "xmm11",
7687 "xmm12", "xmm13", "xmm14", "xmm15",
7688 "st0", "st1", "st2", "st3",
7689 "st4", "st5", "st6", "st7",
7690 "mm0", "mm1", "mm2", "mm3",
7691 "mm4", "mm5", "mm6", "mm7",
7692 "rflags",
7693 "es", "cs", "ss", "ds", "fs", "gs", NULL, NULL,
7694 "fs.base", "gs.base", NULL, NULL,
7695 "tr", "ldtr",
7696 "mxcsr", "fcw", "fsw"
7697 };
7698 const char *const *regnames;
7699
7700 if (flag_code == CODE_64BIT)
7701 {
7702 regnames = regnames_64;
7703 regnames_count = ARRAY_SIZE (regnames_64);
7704 }
7705 else
7706 {
7707 regnames = regnames_32;
7708 regnames_count = ARRAY_SIZE (regnames_32);
7709 }
7710
7711 for (regnum = 0; regnum < regnames_count; regnum++)
7712 if (regnames[regnum] != NULL
7713 && strcmp (regname, regnames[regnum]) == 0)
7714 return regnum;
7715
7716 return -1;
7717 }
7718
7719 void
7720 tc_x86_frame_initial_instructions (void)
7721 {
7722 static unsigned int sp_regno;
7723
7724 if (!sp_regno)
7725 sp_regno = tc_x86_regname_to_dw2regnum (flag_code == CODE_64BIT
7726 ? "rsp" : "esp");
7727
7728 cfi_add_CFA_def_cfa (sp_regno, -x86_cie_data_alignment);
7729 cfi_add_CFA_offset (x86_dwarf2_return_column, x86_cie_data_alignment);
7730 }
7731
7732 int
7733 i386_elf_section_type (const char *str, size_t len)
7734 {
7735 if (flag_code == CODE_64BIT
7736 && len == sizeof ("unwind") - 1
7737 && strncmp (str, "unwind", 6) == 0)
7738 return SHT_X86_64_UNWIND;
7739
7740 return -1;
7741 }
7742
7743 #ifdef TE_PE
7744 void
7745 tc_pe_dwarf2_emit_offset (symbolS *symbol, unsigned int size)
7746 {
7747 expressionS expr;
7748
7749 expr.X_op = O_secrel;
7750 expr.X_add_symbol = symbol;
7751 expr.X_add_number = 0;
7752 emit_expr (&expr, size);
7753 }
7754 #endif
7755
7756 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
7757 /* For ELF on x86-64, add support for SHF_X86_64_LARGE. */
7758
7759 int
7760 x86_64_section_letter (int letter, char **ptr_msg)
7761 {
7762 if (flag_code == CODE_64BIT)
7763 {
7764 if (letter == 'l')
7765 return SHF_X86_64_LARGE;
7766
7767 *ptr_msg = _("Bad .section directive: want a,l,w,x,M,S,G,T in string");
7768 }
7769 else
7770 *ptr_msg = _("Bad .section directive: want a,w,x,M,S,G,T in string");
7771 return -1;
7772 }
7773
7774 int
7775 x86_64_section_word (char *str, size_t len)
7776 {
7777 if (len == 5 && flag_code == CODE_64BIT && CONST_STRNEQ (str, "large"))
7778 return SHF_X86_64_LARGE;
7779
7780 return -1;
7781 }
7782
7783 static void
7784 handle_large_common (int small ATTRIBUTE_UNUSED)
7785 {
7786 if (flag_code != CODE_64BIT)
7787 {
7788 s_comm_internal (0, elf_common_parse);
7789 as_warn (_(".largecomm supported only in 64bit mode, producing .comm"));
7790 }
7791 else
7792 {
7793 static segT lbss_section;
7794 asection *saved_com_section_ptr = elf_com_section_ptr;
7795 asection *saved_bss_section = bss_section;
7796
7797 if (lbss_section == NULL)
7798 {
7799 flagword applicable;
7800 segT seg = now_seg;
7801 subsegT subseg = now_subseg;
7802
7803 /* The .lbss section is for local .largecomm symbols. */
7804 lbss_section = subseg_new (".lbss", 0);
7805 applicable = bfd_applicable_section_flags (stdoutput);
7806 bfd_set_section_flags (stdoutput, lbss_section,
7807 applicable & SEC_ALLOC);
7808 seg_info (lbss_section)->bss = 1;
7809
7810 subseg_set (seg, subseg);
7811 }
7812
7813 elf_com_section_ptr = &_bfd_elf_large_com_section;
7814 bss_section = lbss_section;
7815
7816 s_comm_internal (0, elf_common_parse);
7817
7818 elf_com_section_ptr = saved_com_section_ptr;
7819 bss_section = saved_bss_section;
7820 }
7821 }
7822 #endif /* OBJ_ELF || OBJ_MAYBE_ELF */
This page took 0.194064 seconds and 4 git commands to generate.