86e7cf5ccb31d77974dfef38e9fb3fdb2f8b04a0
[deliverable/binutils-gdb.git] / gas / config / tc-i386.c
1 /* 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
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 INLINE unsigned int mode_from_disp_size PARAMS ((unsigned int));
67 static INLINE int fits_in_signed_byte PARAMS ((offsetT));
68 static INLINE int fits_in_unsigned_byte PARAMS ((offsetT));
69 static INLINE int fits_in_unsigned_word PARAMS ((offsetT));
70 static INLINE int fits_in_signed_word PARAMS ((offsetT));
71 static INLINE int fits_in_unsigned_long PARAMS ((offsetT));
72 static INLINE int fits_in_signed_long PARAMS ((offsetT));
73 static int smallest_imm_type PARAMS ((offsetT));
74 static offsetT offset_in_range PARAMS ((offsetT, int));
75 static int add_prefix PARAMS ((unsigned int));
76 static void set_code_flag PARAMS ((int));
77 static void set_16bit_gcc_code_flag PARAMS ((int));
78 static void set_intel_syntax PARAMS ((int));
79 static void set_cpu_arch PARAMS ((int));
80 #ifdef TE_PE
81 static void pe_directive_secrel PARAMS ((int));
82 #endif
83 static void signed_cons PARAMS ((int));
84 static char *output_invalid PARAMS ((int c));
85 static int i386_operand PARAMS ((char *operand_string));
86 static int i386_intel_operand PARAMS ((char *operand_string, int got_a_float));
87 static const reg_entry *parse_register PARAMS ((char *reg_string,
88 char **end_op));
89 static char *parse_insn PARAMS ((char *, char *));
90 static char *parse_operands PARAMS ((char *, const char *));
91 static void swap_operands PARAMS ((void));
92 static void swap_imm_operands PARAMS ((void));
93 static void optimize_imm PARAMS ((void));
94 static void optimize_disp PARAMS ((void));
95 static int match_template PARAMS ((void));
96 static int check_string PARAMS ((void));
97 static int process_suffix PARAMS ((void));
98 static int check_byte_reg PARAMS ((void));
99 static int check_long_reg PARAMS ((void));
100 static int check_qword_reg PARAMS ((void));
101 static int check_word_reg PARAMS ((void));
102 static int finalize_imm PARAMS ((void));
103 static int process_operands PARAMS ((void));
104 static const seg_entry *build_modrm_byte PARAMS ((void));
105 static void output_insn PARAMS ((void));
106 static void output_branch PARAMS ((void));
107 static void output_jump PARAMS ((void));
108 static void output_interseg_jump PARAMS ((void));
109 static void output_imm PARAMS ((fragS *insn_start_frag,
110 offsetT insn_start_off));
111 static void output_disp PARAMS ((fragS *insn_start_frag,
112 offsetT insn_start_off));
113 #ifndef I386COFF
114 static void s_bss PARAMS ((int));
115 #endif
116 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
117 static void handle_large_common (int small ATTRIBUTE_UNUSED);
118 #endif
119
120 static const char *default_arch = DEFAULT_ARCH;
121
122 /* 'md_assemble ()' gathers together information and puts it into a
123 i386_insn. */
124
125 union i386_op
126 {
127 expressionS *disps;
128 expressionS *imms;
129 const reg_entry *regs;
130 };
131
132 struct _i386_insn
133 {
134 /* TM holds the template for the insn were currently assembling. */
135 template tm;
136
137 /* SUFFIX holds the instruction mnemonic suffix if given.
138 (e.g. 'l' for 'movl') */
139 char suffix;
140
141 /* OPERANDS gives the number of given operands. */
142 unsigned int operands;
143
144 /* REG_OPERANDS, DISP_OPERANDS, MEM_OPERANDS, IMM_OPERANDS give the number
145 of given register, displacement, memory operands and immediate
146 operands. */
147 unsigned int reg_operands, disp_operands, mem_operands, imm_operands;
148
149 /* TYPES [i] is the type (see above #defines) which tells us how to
150 use OP[i] for the corresponding operand. */
151 unsigned int types[MAX_OPERANDS];
152
153 /* Displacement expression, immediate expression, or register for each
154 operand. */
155 union i386_op op[MAX_OPERANDS];
156
157 /* Flags for operands. */
158 unsigned int flags[MAX_OPERANDS];
159 #define Operand_PCrel 1
160
161 /* Relocation type for operand */
162 enum bfd_reloc_code_real reloc[MAX_OPERANDS];
163
164 /* BASE_REG, INDEX_REG, and LOG2_SCALE_FACTOR are used to encode
165 the base index byte below. */
166 const reg_entry *base_reg;
167 const reg_entry *index_reg;
168 unsigned int log2_scale_factor;
169
170 /* SEG gives the seg_entries of this insn. They are zero unless
171 explicit segment overrides are given. */
172 const seg_entry *seg[2];
173
174 /* PREFIX holds all the given prefix opcodes (usually null).
175 PREFIXES is the number of prefix opcodes. */
176 unsigned int prefixes;
177 unsigned char prefix[MAX_PREFIXES];
178
179 /* RM and SIB are the modrm byte and the sib byte where the
180 addressing modes of this insn are encoded. */
181
182 modrm_byte rm;
183 rex_byte rex;
184 sib_byte sib;
185 };
186
187 typedef struct _i386_insn i386_insn;
188
189 /* List of chars besides those in app.c:symbol_chars that can start an
190 operand. Used to prevent the scrubber eating vital white-space. */
191 const char extra_symbol_chars[] = "*%-(["
192 #ifdef LEX_AT
193 "@"
194 #endif
195 #ifdef LEX_QM
196 "?"
197 #endif
198 ;
199
200 #if (defined (TE_I386AIX) \
201 || ((defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)) \
202 && !defined (TE_GNU) \
203 && !defined (TE_LINUX) \
204 && !defined (TE_NETWARE) \
205 && !defined (TE_FreeBSD) \
206 && !defined (TE_NetBSD)))
207 /* This array holds the chars that always start a comment. If the
208 pre-processor is disabled, these aren't very useful. The option
209 --divide will remove '/' from this list. */
210 const char *i386_comment_chars = "#/";
211 #define SVR4_COMMENT_CHARS 1
212 #define PREFIX_SEPARATOR '\\'
213
214 #else
215 const char *i386_comment_chars = "#";
216 #define PREFIX_SEPARATOR '/'
217 #endif
218
219 /* This array holds the chars that only start a comment at the beginning of
220 a line. If the line seems to have the form '# 123 filename'
221 .line and .file directives will appear in the pre-processed output.
222 Note that input_file.c hand checks for '#' at the beginning of the
223 first line of the input file. This is because the compiler outputs
224 #NO_APP at the beginning of its output.
225 Also note that comments started like this one will always work if
226 '/' isn't otherwise defined. */
227 const char line_comment_chars[] = "#/";
228
229 const char line_separator_chars[] = ";";
230
231 /* Chars that can be used to separate mant from exp in floating point
232 nums. */
233 const char EXP_CHARS[] = "eE";
234
235 /* Chars that mean this number is a floating point constant
236 As in 0f12.456
237 or 0d1.2345e12. */
238 const char FLT_CHARS[] = "fFdDxX";
239
240 /* Tables for lexical analysis. */
241 static char mnemonic_chars[256];
242 static char register_chars[256];
243 static char operand_chars[256];
244 static char identifier_chars[256];
245 static char digit_chars[256];
246
247 /* Lexical macros. */
248 #define is_mnemonic_char(x) (mnemonic_chars[(unsigned char) x])
249 #define is_operand_char(x) (operand_chars[(unsigned char) x])
250 #define is_register_char(x) (register_chars[(unsigned char) x])
251 #define is_space_char(x) ((x) == ' ')
252 #define is_identifier_char(x) (identifier_chars[(unsigned char) x])
253 #define is_digit_char(x) (digit_chars[(unsigned char) x])
254
255 /* All non-digit non-letter characters that may occur in an operand. */
256 static char operand_special_chars[] = "%$-+(,)*._~/<>|&^!:[@]";
257
258 /* md_assemble() always leaves the strings it's passed unaltered. To
259 effect this we maintain a stack of saved characters that we've smashed
260 with '\0's (indicating end of strings for various sub-fields of the
261 assembler instruction). */
262 static char save_stack[32];
263 static char *save_stack_p;
264 #define END_STRING_AND_SAVE(s) \
265 do { *save_stack_p++ = *(s); *(s) = '\0'; } while (0)
266 #define RESTORE_END_STRING(s) \
267 do { *(s) = *--save_stack_p; } while (0)
268
269 /* The instruction we're assembling. */
270 static i386_insn i;
271
272 /* Possible templates for current insn. */
273 static const templates *current_templates;
274
275 /* Per instruction expressionS buffers: 2 displacements & 2 immediate max. */
276 static expressionS disp_expressions[2], im_expressions[2];
277
278 /* Current operand we are working on. */
279 static int this_operand;
280
281 /* We support four different modes. FLAG_CODE variable is used to distinguish
282 these. */
283
284 enum flag_code {
285 CODE_32BIT,
286 CODE_16BIT,
287 CODE_64BIT };
288 #define NUM_FLAG_CODE ((int) CODE_64BIT + 1)
289
290 static enum flag_code flag_code;
291 static unsigned int object_64bit;
292 static int use_rela_relocations = 0;
293
294 /* The names used to print error messages. */
295 static const char *flag_code_names[] =
296 {
297 "32",
298 "16",
299 "64"
300 };
301
302 /* 1 for intel syntax,
303 0 if att syntax. */
304 static int intel_syntax = 0;
305
306 /* 1 if register prefix % not required. */
307 static int allow_naked_reg = 0;
308
309 /* Used in 16 bit gcc mode to add an l suffix to call, ret, enter,
310 leave, push, and pop instructions so that gcc has the same stack
311 frame as in 32 bit mode. */
312 static char stackop_size = '\0';
313
314 /* Non-zero to optimize code alignment. */
315 int optimize_align_code = 1;
316
317 /* Non-zero to quieten some warnings. */
318 static int quiet_warnings = 0;
319
320 /* CPU name. */
321 static const char *cpu_arch_name = NULL;
322 static const char *cpu_sub_arch_name = NULL;
323
324 /* CPU feature flags. */
325 static unsigned int cpu_arch_flags = CpuUnknownFlags | CpuNo64;
326
327 /* If we have selected a cpu we are generating instructions for. */
328 static int cpu_arch_tune_set = 0;
329
330 /* Cpu we are generating instructions for. */
331 static enum processor_type cpu_arch_tune = PROCESSOR_UNKNOWN;
332
333 /* CPU feature flags of cpu we are generating instructions for. */
334 static unsigned int cpu_arch_tune_flags = 0;
335
336 /* CPU instruction set architecture used. */
337 static enum processor_type cpu_arch_isa = PROCESSOR_UNKNOWN;
338
339 /* CPU feature flags of instruction set architecture used. */
340 static unsigned int cpu_arch_isa_flags = 0;
341
342 /* If set, conditional jumps are not automatically promoted to handle
343 larger than a byte offset. */
344 static unsigned int no_cond_jump_promotion = 0;
345
346 /* Pre-defined "_GLOBAL_OFFSET_TABLE_". */
347 static symbolS *GOT_symbol;
348
349 /* The dwarf2 return column, adjusted for 32 or 64 bit. */
350 unsigned int x86_dwarf2_return_column;
351
352 /* The dwarf2 data alignment, adjusted for 32 or 64 bit. */
353 int x86_cie_data_alignment;
354
355 /* Interface to relax_segment.
356 There are 3 major relax states for 386 jump insns because the
357 different types of jumps add different sizes to frags when we're
358 figuring out what sort of jump to choose to reach a given label. */
359
360 /* Types. */
361 #define UNCOND_JUMP 0
362 #define COND_JUMP 1
363 #define COND_JUMP86 2
364
365 /* Sizes. */
366 #define CODE16 1
367 #define SMALL 0
368 #define SMALL16 (SMALL | CODE16)
369 #define BIG 2
370 #define BIG16 (BIG | CODE16)
371
372 #ifndef INLINE
373 #ifdef __GNUC__
374 #define INLINE __inline__
375 #else
376 #define INLINE
377 #endif
378 #endif
379
380 #define ENCODE_RELAX_STATE(type, size) \
381 ((relax_substateT) (((type) << 2) | (size)))
382 #define TYPE_FROM_RELAX_STATE(s) \
383 ((s) >> 2)
384 #define DISP_SIZE_FROM_RELAX_STATE(s) \
385 ((((s) & 3) == BIG ? 4 : (((s) & 3) == BIG16 ? 2 : 1)))
386
387 /* This table is used by relax_frag to promote short jumps to long
388 ones where necessary. SMALL (short) jumps may be promoted to BIG
389 (32 bit long) ones, and SMALL16 jumps to BIG16 (16 bit long). We
390 don't allow a short jump in a 32 bit code segment to be promoted to
391 a 16 bit offset jump because it's slower (requires data size
392 prefix), and doesn't work, unless the destination is in the bottom
393 64k of the code segment (The top 16 bits of eip are zeroed). */
394
395 const relax_typeS md_relax_table[] =
396 {
397 /* The fields are:
398 1) most positive reach of this state,
399 2) most negative reach of this state,
400 3) how many bytes this mode will have in the variable part of the frag
401 4) which index into the table to try if we can't fit into this one. */
402
403 /* UNCOND_JUMP states. */
404 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (UNCOND_JUMP, BIG)},
405 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (UNCOND_JUMP, BIG16)},
406 /* dword jmp adds 4 bytes to frag:
407 0 extra opcode bytes, 4 displacement bytes. */
408 {0, 0, 4, 0},
409 /* word jmp adds 2 byte2 to frag:
410 0 extra opcode bytes, 2 displacement bytes. */
411 {0, 0, 2, 0},
412
413 /* COND_JUMP states. */
414 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP, BIG)},
415 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP, BIG16)},
416 /* dword conditionals adds 5 bytes to frag:
417 1 extra opcode byte, 4 displacement bytes. */
418 {0, 0, 5, 0},
419 /* word conditionals add 3 bytes to frag:
420 1 extra opcode byte, 2 displacement bytes. */
421 {0, 0, 3, 0},
422
423 /* COND_JUMP86 states. */
424 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP86, BIG)},
425 {127 + 1, -128 + 1, 1, ENCODE_RELAX_STATE (COND_JUMP86, BIG16)},
426 /* dword conditionals adds 5 bytes to frag:
427 1 extra opcode byte, 4 displacement bytes. */
428 {0, 0, 5, 0},
429 /* word conditionals add 4 bytes to frag:
430 1 displacement byte and a 3 byte long branch insn. */
431 {0, 0, 4, 0}
432 };
433
434 static const arch_entry cpu_arch[] =
435 {
436 {"generic32", PROCESSOR_GENERIC32,
437 Cpu186|Cpu286|Cpu386},
438 {"generic64", PROCESSOR_GENERIC64,
439 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
440 |CpuMMX2|CpuSSE|CpuSSE2},
441 {"i8086", PROCESSOR_UNKNOWN,
442 0},
443 {"i186", PROCESSOR_UNKNOWN,
444 Cpu186},
445 {"i286", PROCESSOR_UNKNOWN,
446 Cpu186|Cpu286},
447 {"i386", PROCESSOR_GENERIC32,
448 Cpu186|Cpu286|Cpu386},
449 {"i486", PROCESSOR_I486,
450 Cpu186|Cpu286|Cpu386|Cpu486},
451 {"i586", PROCESSOR_PENTIUM,
452 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586},
453 {"i686", PROCESSOR_PENTIUMPRO,
454 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686},
455 {"pentium", PROCESSOR_PENTIUM,
456 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586},
457 {"pentiumpro",PROCESSOR_PENTIUMPRO,
458 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686},
459 {"pentiumii", PROCESSOR_PENTIUMPRO,
460 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuMMX},
461 {"pentiumiii",PROCESSOR_PENTIUMPRO,
462 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuMMX|CpuMMX2|CpuSSE},
463 {"pentium4", PROCESSOR_PENTIUM4,
464 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
465 |CpuMMX2|CpuSSE|CpuSSE2},
466 {"prescott", PROCESSOR_NOCONA,
467 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
468 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
469 {"nocona", PROCESSOR_NOCONA,
470 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
471 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
472 {"yonah", PROCESSOR_CORE,
473 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
474 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
475 {"core", PROCESSOR_CORE,
476 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
477 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
478 {"merom", PROCESSOR_CORE2,
479 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
480 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3},
481 {"core2", PROCESSOR_CORE2,
482 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuP4|CpuMMX
483 |CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3},
484 {"k6", PROCESSOR_K6,
485 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|CpuK6|CpuMMX},
486 {"k6_2", PROCESSOR_K6,
487 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|CpuK6|CpuMMX|Cpu3dnow},
488 {"athlon", PROCESSOR_ATHLON,
489 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
490 |CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA},
491 {"sledgehammer", PROCESSOR_K8,
492 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
493 |CpuSledgehammer|CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2},
494 {"opteron", PROCESSOR_K8,
495 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
496 |CpuSledgehammer|CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2},
497 {"k8", PROCESSOR_K8,
498 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6
499 |CpuSledgehammer|CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2},
500 {"amdfam10", PROCESSOR_AMDFAM10,
501 Cpu186|Cpu286|Cpu386|Cpu486|Cpu586|Cpu686|CpuK6|CpuSledgehammer
502 |CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA|CpuSSE|CpuSSE2|CpuSSE3|CpuSSE4a
503 |CpuABM},
504 {".mmx", PROCESSOR_UNKNOWN,
505 CpuMMX},
506 {".sse", PROCESSOR_UNKNOWN,
507 CpuMMX|CpuMMX2|CpuSSE},
508 {".sse2", PROCESSOR_UNKNOWN,
509 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2},
510 {".sse3", PROCESSOR_UNKNOWN,
511 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3},
512 {".ssse3", PROCESSOR_UNKNOWN,
513 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3},
514 {".3dnow", PROCESSOR_UNKNOWN,
515 CpuMMX|Cpu3dnow},
516 {".3dnowa", PROCESSOR_UNKNOWN,
517 CpuMMX|CpuMMX2|Cpu3dnow|Cpu3dnowA},
518 {".padlock", PROCESSOR_UNKNOWN,
519 CpuPadLock},
520 {".pacifica", PROCESSOR_UNKNOWN,
521 CpuSVME},
522 {".svme", PROCESSOR_UNKNOWN,
523 CpuSVME},
524 {".sse4a", PROCESSOR_UNKNOWN,
525 CpuMMX|CpuMMX2|CpuSSE|CpuSSE2|CpuSSE3|CpuSSE4a},
526 {".abm", PROCESSOR_UNKNOWN,
527 CpuABM}
528 };
529
530 const pseudo_typeS md_pseudo_table[] =
531 {
532 #if !defined(OBJ_AOUT) && !defined(USE_ALIGN_PTWO)
533 {"align", s_align_bytes, 0},
534 #else
535 {"align", s_align_ptwo, 0},
536 #endif
537 {"arch", set_cpu_arch, 0},
538 #ifndef I386COFF
539 {"bss", s_bss, 0},
540 #endif
541 {"ffloat", float_cons, 'f'},
542 {"dfloat", float_cons, 'd'},
543 {"tfloat", float_cons, 'x'},
544 {"value", cons, 2},
545 {"slong", signed_cons, 4},
546 {"noopt", s_ignore, 0},
547 {"optim", s_ignore, 0},
548 {"code16gcc", set_16bit_gcc_code_flag, CODE_16BIT},
549 {"code16", set_code_flag, CODE_16BIT},
550 {"code32", set_code_flag, CODE_32BIT},
551 {"code64", set_code_flag, CODE_64BIT},
552 {"intel_syntax", set_intel_syntax, 1},
553 {"att_syntax", set_intel_syntax, 0},
554 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
555 {"largecomm", handle_large_common, 0},
556 #else
557 {"file", (void (*) PARAMS ((int))) dwarf2_directive_file, 0},
558 {"loc", dwarf2_directive_loc, 0},
559 {"loc_mark_labels", dwarf2_directive_loc_mark_labels, 0},
560 #endif
561 #ifdef TE_PE
562 {"secrel32", pe_directive_secrel, 0},
563 #endif
564 {0, 0, 0}
565 };
566
567 /* For interface with expression (). */
568 extern char *input_line_pointer;
569
570 /* Hash table for instruction mnemonic lookup. */
571 static struct hash_control *op_hash;
572
573 /* Hash table for register lookup. */
574 static struct hash_control *reg_hash;
575 \f
576 void
577 i386_align_code (fragP, count)
578 fragS *fragP;
579 int count;
580 {
581 /* Various efficient no-op patterns for aligning code labels.
582 Note: Don't try to assemble the instructions in the comments.
583 0L and 0w are not legal. */
584 static const char f32_1[] =
585 {0x90}; /* nop */
586 static const char f32_2[] =
587 {0x66,0x90}; /* xchg %ax,%ax */
588 static const char f32_3[] =
589 {0x8d,0x76,0x00}; /* leal 0(%esi),%esi */
590 static const char f32_4[] =
591 {0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */
592 static const char f32_5[] =
593 {0x90, /* nop */
594 0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */
595 static const char f32_6[] =
596 {0x8d,0xb6,0x00,0x00,0x00,0x00}; /* leal 0L(%esi),%esi */
597 static const char f32_7[] =
598 {0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */
599 static const char f32_8[] =
600 {0x90, /* nop */
601 0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */
602 static const char f32_9[] =
603 {0x89,0xf6, /* movl %esi,%esi */
604 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
605 static const char f32_10[] =
606 {0x8d,0x76,0x00, /* leal 0(%esi),%esi */
607 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
608 static const char f32_11[] =
609 {0x8d,0x74,0x26,0x00, /* leal 0(%esi,1),%esi */
610 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
611 static const char f32_12[] =
612 {0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */
613 0x8d,0xbf,0x00,0x00,0x00,0x00}; /* leal 0L(%edi),%edi */
614 static const char f32_13[] =
615 {0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */
616 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
617 static const char f32_14[] =
618 {0x8d,0xb4,0x26,0x00,0x00,0x00,0x00, /* leal 0L(%esi,1),%esi */
619 0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */
620 static const char f32_15[] =
621 {0xeb,0x0d,0x90,0x90,0x90,0x90,0x90, /* jmp .+15; lotsa nops */
622 0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90};
623 static const char f16_3[] =
624 {0x8d,0x74,0x00}; /* lea 0(%esi),%esi */
625 static const char f16_4[] =
626 {0x8d,0xb4,0x00,0x00}; /* lea 0w(%si),%si */
627 static const char f16_5[] =
628 {0x90, /* nop */
629 0x8d,0xb4,0x00,0x00}; /* lea 0w(%si),%si */
630 static const char f16_6[] =
631 {0x89,0xf6, /* mov %si,%si */
632 0x8d,0xbd,0x00,0x00}; /* lea 0w(%di),%di */
633 static const char f16_7[] =
634 {0x8d,0x74,0x00, /* lea 0(%si),%si */
635 0x8d,0xbd,0x00,0x00}; /* lea 0w(%di),%di */
636 static const char f16_8[] =
637 {0x8d,0xb4,0x00,0x00, /* lea 0w(%si),%si */
638 0x8d,0xbd,0x00,0x00}; /* lea 0w(%di),%di */
639 static const char *const f32_patt[] = {
640 f32_1, f32_2, f32_3, f32_4, f32_5, f32_6, f32_7, f32_8,
641 f32_9, f32_10, f32_11, f32_12, f32_13, f32_14, f32_15
642 };
643 static const char *const f16_patt[] = {
644 f32_1, f32_2, f16_3, f16_4, f16_5, f16_6, f16_7, f16_8,
645 f32_15, f32_15, f32_15, f32_15, f32_15, f32_15, f32_15
646 };
647 /* nopl (%[re]ax) */
648 static const char alt_3[] =
649 {0x0f,0x1f,0x00};
650 /* nopl 0(%[re]ax) */
651 static const char alt_4[] =
652 {0x0f,0x1f,0x40,0x00};
653 /* nopl 0(%[re]ax,%[re]ax,1) */
654 static const char alt_5[] =
655 {0x0f,0x1f,0x44,0x00,0x00};
656 /* nopw 0(%[re]ax,%[re]ax,1) */
657 static const char alt_6[] =
658 {0x66,0x0f,0x1f,0x44,0x00,0x00};
659 /* nopl 0L(%[re]ax) */
660 static const char alt_7[] =
661 {0x0f,0x1f,0x80,0x00,0x00,0x00,0x00};
662 /* nopl 0L(%[re]ax,%[re]ax,1) */
663 static const char alt_8[] =
664 {0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
665 /* nopw 0L(%[re]ax,%[re]ax,1) */
666 static const char alt_9[] =
667 {0x66,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
668 /* nopw %cs:0L(%[re]ax,%[re]ax,1) */
669 static const char alt_10[] =
670 {0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
671 /* data16
672 nopw %cs:0L(%[re]ax,%[re]ax,1) */
673 static const char alt_long_11[] =
674 {0x66,
675 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
676 /* data16
677 data16
678 nopw %cs:0L(%[re]ax,%[re]ax,1) */
679 static const char alt_long_12[] =
680 {0x66,
681 0x66,
682 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
683 /* data16
684 data16
685 data16
686 nopw %cs:0L(%[re]ax,%[re]ax,1) */
687 static const char alt_long_13[] =
688 {0x66,
689 0x66,
690 0x66,
691 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
692 /* data16
693 data16
694 data16
695 data16
696 nopw %cs:0L(%[re]ax,%[re]ax,1) */
697 static const char alt_long_14[] =
698 {0x66,
699 0x66,
700 0x66,
701 0x66,
702 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
703 /* data16
704 data16
705 data16
706 data16
707 data16
708 nopw %cs:0L(%[re]ax,%[re]ax,1) */
709 static const char alt_long_15[] =
710 {0x66,
711 0x66,
712 0x66,
713 0x66,
714 0x66,
715 0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
716 /* nopl 0(%[re]ax,%[re]ax,1)
717 nopw 0(%[re]ax,%[re]ax,1) */
718 static const char alt_short_11[] =
719 {0x0f,0x1f,0x44,0x00,0x00,
720 0x66,0x0f,0x1f,0x44,0x00,0x00};
721 /* nopw 0(%[re]ax,%[re]ax,1)
722 nopw 0(%[re]ax,%[re]ax,1) */
723 static const char alt_short_12[] =
724 {0x66,0x0f,0x1f,0x44,0x00,0x00,
725 0x66,0x0f,0x1f,0x44,0x00,0x00};
726 /* nopw 0(%[re]ax,%[re]ax,1)
727 nopl 0L(%[re]ax) */
728 static const char alt_short_13[] =
729 {0x66,0x0f,0x1f,0x44,0x00,0x00,
730 0x0f,0x1f,0x80,0x00,0x00,0x00,0x00};
731 /* nopl 0L(%[re]ax)
732 nopl 0L(%[re]ax) */
733 static const char alt_short_14[] =
734 {0x0f,0x1f,0x80,0x00,0x00,0x00,0x00,
735 0x0f,0x1f,0x80,0x00,0x00,0x00,0x00};
736 /* nopl 0L(%[re]ax)
737 nopl 0L(%[re]ax,%[re]ax,1) */
738 static const char alt_short_15[] =
739 {0x0f,0x1f,0x80,0x00,0x00,0x00,0x00,
740 0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00};
741 static const char *const alt_short_patt[] = {
742 f32_1, f32_2, alt_3, alt_4, alt_5, alt_6, alt_7, alt_8,
743 alt_9, alt_10, alt_short_11, alt_short_12, alt_short_13,
744 alt_short_14, alt_short_15
745 };
746 static const char *const alt_long_patt[] = {
747 f32_1, f32_2, alt_3, alt_4, alt_5, alt_6, alt_7, alt_8,
748 alt_9, alt_10, alt_long_11, alt_long_12, alt_long_13,
749 alt_long_14, alt_long_15
750 };
751
752 if (count <= 0 || count > 15)
753 return;
754
755 /* We need to decide which NOP sequence to use for 32bit and
756 64bit. When -mtune= is used:
757
758 1. For PROCESSOR_I486, PROCESSOR_PENTIUM and PROCESSOR_GENERIC32,
759 f32_patt will be used.
760 2. For PROCESSOR_K8 and PROCESSOR_AMDFAM10 in 64bit, NOPs with 0x66 prefix will be used.
761 3. For PROCESSOR_CORE2, alt_long_patt will be used.
762 4. For PROCESSOR_PENTIUMPRO, PROCESSOR_PENTIUM4, PROCESSOR_NOCONA,
763 PROCESSOR_CORE, PROCESSOR_CORE2, PROCESSOR_K6, PROCESSOR_ATHLON
764 and PROCESSOR_GENERIC64, alt_short_patt will be used.
765
766 When -mtune= isn't used, alt_short_patt will be used if
767 cpu_arch_isa_flags has Cpu686. Otherwise, f32_patt will be used.
768
769 When -march= or .arch is used, we can't use anything beyond
770 cpu_arch_isa_flags. */
771
772 if (flag_code == CODE_16BIT)
773 {
774 memcpy (fragP->fr_literal + fragP->fr_fix,
775 f16_patt[count - 1], count);
776 if (count > 8)
777 /* Adjust jump offset. */
778 fragP->fr_literal[fragP->fr_fix + 1] = count - 2;
779 }
780 else if (flag_code == CODE_64BIT && cpu_arch_tune == PROCESSOR_K8)
781 {
782 int i;
783 int nnops = (count + 3) / 4;
784 int len = count / nnops;
785 int remains = count - nnops * len;
786 int pos = 0;
787
788 /* The recommended way to pad 64bit code is to use NOPs preceded
789 by maximally four 0x66 prefixes. Balance the size of nops. */
790 for (i = 0; i < remains; i++)
791 {
792 memset (fragP->fr_literal + fragP->fr_fix + pos, 0x66, len);
793 fragP->fr_literal[fragP->fr_fix + pos + len] = 0x90;
794 pos += len + 1;
795 }
796 for (; i < nnops; i++)
797 {
798 memset (fragP->fr_literal + fragP->fr_fix + pos, 0x66, len - 1);
799 fragP->fr_literal[fragP->fr_fix + pos + len - 1] = 0x90;
800 pos += len;
801 }
802 }
803 else
804 {
805 const char *const *patt = NULL;
806
807 if (cpu_arch_isa == PROCESSOR_UNKNOWN)
808 {
809 /* PROCESSOR_UNKNOWN means that all ISAs may be used. */
810 switch (cpu_arch_tune)
811 {
812 case PROCESSOR_UNKNOWN:
813 /* We use cpu_arch_isa_flags to check if we SHOULD
814 optimize for Cpu686. */
815 if ((cpu_arch_isa_flags & Cpu686) != 0)
816 patt = alt_short_patt;
817 else
818 patt = f32_patt;
819 break;
820 case PROCESSOR_CORE2:
821 patt = alt_long_patt;
822 break;
823 case PROCESSOR_PENTIUMPRO:
824 case PROCESSOR_PENTIUM4:
825 case PROCESSOR_NOCONA:
826 case PROCESSOR_CORE:
827 case PROCESSOR_K6:
828 case PROCESSOR_ATHLON:
829 case PROCESSOR_K8:
830 case PROCESSOR_GENERIC64:
831 case PROCESSOR_AMDFAM10:
832 patt = alt_short_patt;
833 break;
834 case PROCESSOR_I486:
835 case PROCESSOR_PENTIUM:
836 case PROCESSOR_GENERIC32:
837 patt = f32_patt;
838 break;
839 }
840 }
841 else
842 {
843 switch (cpu_arch_tune)
844 {
845 case PROCESSOR_UNKNOWN:
846 /* When cpu_arch_isa is net, cpu_arch_tune shouldn't be
847 PROCESSOR_UNKNOWN. */
848 abort ();
849 break;
850
851 case PROCESSOR_I486:
852 case PROCESSOR_PENTIUM:
853 case PROCESSOR_PENTIUMPRO:
854 case PROCESSOR_PENTIUM4:
855 case PROCESSOR_NOCONA:
856 case PROCESSOR_CORE:
857 case PROCESSOR_K6:
858 case PROCESSOR_ATHLON:
859 case PROCESSOR_K8:
860 case PROCESSOR_AMDFAM10:
861 case PROCESSOR_GENERIC32:
862 /* We use cpu_arch_isa_flags to check if we CAN optimize
863 for Cpu686. */
864 if ((cpu_arch_isa_flags & Cpu686) != 0)
865 patt = alt_short_patt;
866 else
867 patt = f32_patt;
868 break;
869 case PROCESSOR_CORE2:
870 if ((cpu_arch_isa_flags & Cpu686) != 0)
871 patt = alt_long_patt;
872 else
873 patt = f32_patt;
874 break;
875 case PROCESSOR_GENERIC64:
876 patt = alt_short_patt;
877 break;
878 }
879 }
880
881 memcpy (fragP->fr_literal + fragP->fr_fix,
882 patt[count - 1], count);
883 }
884 fragP->fr_var = count;
885 }
886
887 static INLINE unsigned int
888 mode_from_disp_size (t)
889 unsigned int t;
890 {
891 return (t & Disp8) ? 1 : (t & (Disp16 | Disp32 | Disp32S)) ? 2 : 0;
892 }
893
894 static INLINE int
895 fits_in_signed_byte (num)
896 offsetT num;
897 {
898 return (num >= -128) && (num <= 127);
899 }
900
901 static INLINE int
902 fits_in_unsigned_byte (num)
903 offsetT num;
904 {
905 return (num & 0xff) == num;
906 }
907
908 static INLINE int
909 fits_in_unsigned_word (num)
910 offsetT num;
911 {
912 return (num & 0xffff) == num;
913 }
914
915 static INLINE int
916 fits_in_signed_word (num)
917 offsetT num;
918 {
919 return (-32768 <= num) && (num <= 32767);
920 }
921
922 static INLINE int
923 fits_in_signed_long (num)
924 offsetT num ATTRIBUTE_UNUSED;
925 {
926 #ifndef BFD64
927 return 1;
928 #else
929 return (!(((offsetT) -1 << 31) & num)
930 || (((offsetT) -1 << 31) & num) == ((offsetT) -1 << 31));
931 #endif
932 } /* fits_in_signed_long() */
933
934 static INLINE int
935 fits_in_unsigned_long (num)
936 offsetT num ATTRIBUTE_UNUSED;
937 {
938 #ifndef BFD64
939 return 1;
940 #else
941 return (num & (((offsetT) 2 << 31) - 1)) == num;
942 #endif
943 } /* fits_in_unsigned_long() */
944
945 static int
946 smallest_imm_type (num)
947 offsetT num;
948 {
949 if (cpu_arch_flags != (Cpu186 | Cpu286 | Cpu386 | Cpu486 | CpuNo64))
950 {
951 /* This code is disabled on the 486 because all the Imm1 forms
952 in the opcode table are slower on the i486. They're the
953 versions with the implicitly specified single-position
954 displacement, which has another syntax if you really want to
955 use that form. */
956 if (num == 1)
957 return Imm1 | Imm8 | Imm8S | Imm16 | Imm32 | Imm32S | Imm64;
958 }
959 return (fits_in_signed_byte (num)
960 ? (Imm8S | Imm8 | Imm16 | Imm32 | Imm32S | Imm64)
961 : fits_in_unsigned_byte (num)
962 ? (Imm8 | Imm16 | Imm32 | Imm32S | Imm64)
963 : (fits_in_signed_word (num) || fits_in_unsigned_word (num))
964 ? (Imm16 | Imm32 | Imm32S | Imm64)
965 : fits_in_signed_long (num)
966 ? (Imm32 | Imm32S | Imm64)
967 : fits_in_unsigned_long (num)
968 ? (Imm32 | Imm64)
969 : Imm64);
970 }
971
972 static offsetT
973 offset_in_range (val, size)
974 offsetT val;
975 int size;
976 {
977 addressT mask;
978
979 switch (size)
980 {
981 case 1: mask = ((addressT) 1 << 8) - 1; break;
982 case 2: mask = ((addressT) 1 << 16) - 1; break;
983 case 4: mask = ((addressT) 2 << 31) - 1; break;
984 #ifdef BFD64
985 case 8: mask = ((addressT) 2 << 63) - 1; break;
986 #endif
987 default: abort ();
988 }
989
990 /* If BFD64, sign extend val. */
991 if (!use_rela_relocations)
992 if ((val & ~(((addressT) 2 << 31) - 1)) == 0)
993 val = (val ^ ((addressT) 1 << 31)) - ((addressT) 1 << 31);
994
995 if ((val & ~mask) != 0 && (val & ~mask) != ~mask)
996 {
997 char buf1[40], buf2[40];
998
999 sprint_value (buf1, val);
1000 sprint_value (buf2, val & mask);
1001 as_warn (_("%s shortened to %s"), buf1, buf2);
1002 }
1003 return val & mask;
1004 }
1005
1006 /* Returns 0 if attempting to add a prefix where one from the same
1007 class already exists, 1 if non rep/repne added, 2 if rep/repne
1008 added. */
1009 static int
1010 add_prefix (prefix)
1011 unsigned int prefix;
1012 {
1013 int ret = 1;
1014 unsigned int q;
1015
1016 if (prefix >= REX_OPCODE && prefix < REX_OPCODE + 16
1017 && flag_code == CODE_64BIT)
1018 {
1019 if ((i.prefix[REX_PREFIX] & prefix & REX_MODE64)
1020 || ((i.prefix[REX_PREFIX] & (REX_EXTX | REX_EXTY | REX_EXTZ))
1021 && (prefix & (REX_EXTX | REX_EXTY | REX_EXTZ))))
1022 ret = 0;
1023 q = REX_PREFIX;
1024 }
1025 else
1026 {
1027 switch (prefix)
1028 {
1029 default:
1030 abort ();
1031
1032 case CS_PREFIX_OPCODE:
1033 case DS_PREFIX_OPCODE:
1034 case ES_PREFIX_OPCODE:
1035 case FS_PREFIX_OPCODE:
1036 case GS_PREFIX_OPCODE:
1037 case SS_PREFIX_OPCODE:
1038 q = SEG_PREFIX;
1039 break;
1040
1041 case REPNE_PREFIX_OPCODE:
1042 case REPE_PREFIX_OPCODE:
1043 ret = 2;
1044 /* fall thru */
1045 case LOCK_PREFIX_OPCODE:
1046 q = LOCKREP_PREFIX;
1047 break;
1048
1049 case FWAIT_OPCODE:
1050 q = WAIT_PREFIX;
1051 break;
1052
1053 case ADDR_PREFIX_OPCODE:
1054 q = ADDR_PREFIX;
1055 break;
1056
1057 case DATA_PREFIX_OPCODE:
1058 q = DATA_PREFIX;
1059 break;
1060 }
1061 if (i.prefix[q] != 0)
1062 ret = 0;
1063 }
1064
1065 if (ret)
1066 {
1067 if (!i.prefix[q])
1068 ++i.prefixes;
1069 i.prefix[q] |= prefix;
1070 }
1071 else
1072 as_bad (_("same type of prefix used twice"));
1073
1074 return ret;
1075 }
1076
1077 static void
1078 set_code_flag (value)
1079 int value;
1080 {
1081 flag_code = value;
1082 cpu_arch_flags &= ~(Cpu64 | CpuNo64);
1083 cpu_arch_flags |= (flag_code == CODE_64BIT ? Cpu64 : CpuNo64);
1084 if (value == CODE_64BIT && !(cpu_arch_flags & CpuSledgehammer))
1085 {
1086 as_bad (_("64bit mode not supported on this CPU."));
1087 }
1088 if (value == CODE_32BIT && !(cpu_arch_flags & Cpu386))
1089 {
1090 as_bad (_("32bit mode not supported on this CPU."));
1091 }
1092 stackop_size = '\0';
1093 }
1094
1095 static void
1096 set_16bit_gcc_code_flag (new_code_flag)
1097 int new_code_flag;
1098 {
1099 flag_code = new_code_flag;
1100 cpu_arch_flags &= ~(Cpu64 | CpuNo64);
1101 cpu_arch_flags |= (flag_code == CODE_64BIT ? Cpu64 : CpuNo64);
1102 stackop_size = LONG_MNEM_SUFFIX;
1103 }
1104
1105 static void
1106 set_intel_syntax (syntax_flag)
1107 int syntax_flag;
1108 {
1109 /* Find out if register prefixing is specified. */
1110 int ask_naked_reg = 0;
1111
1112 SKIP_WHITESPACE ();
1113 if (!is_end_of_line[(unsigned char) *input_line_pointer])
1114 {
1115 char *string = input_line_pointer;
1116 int e = get_symbol_end ();
1117
1118 if (strcmp (string, "prefix") == 0)
1119 ask_naked_reg = 1;
1120 else if (strcmp (string, "noprefix") == 0)
1121 ask_naked_reg = -1;
1122 else
1123 as_bad (_("bad argument to syntax directive."));
1124 *input_line_pointer = e;
1125 }
1126 demand_empty_rest_of_line ();
1127
1128 intel_syntax = syntax_flag;
1129
1130 if (ask_naked_reg == 0)
1131 allow_naked_reg = (intel_syntax
1132 && (bfd_get_symbol_leading_char (stdoutput) != '\0'));
1133 else
1134 allow_naked_reg = (ask_naked_reg < 0);
1135
1136 identifier_chars['%'] = intel_syntax && allow_naked_reg ? '%' : 0;
1137 identifier_chars['$'] = intel_syntax ? '$' : 0;
1138 }
1139
1140 static void
1141 set_cpu_arch (dummy)
1142 int dummy ATTRIBUTE_UNUSED;
1143 {
1144 SKIP_WHITESPACE ();
1145
1146 if (!is_end_of_line[(unsigned char) *input_line_pointer])
1147 {
1148 char *string = input_line_pointer;
1149 int e = get_symbol_end ();
1150 unsigned int i;
1151
1152 for (i = 0; i < ARRAY_SIZE (cpu_arch); i++)
1153 {
1154 if (strcmp (string, cpu_arch[i].name) == 0)
1155 {
1156 if (*string != '.')
1157 {
1158 cpu_arch_name = cpu_arch[i].name;
1159 cpu_sub_arch_name = NULL;
1160 cpu_arch_flags = (cpu_arch[i].flags
1161 | (flag_code == CODE_64BIT ? Cpu64 : CpuNo64));
1162 cpu_arch_isa = cpu_arch[i].type;
1163 cpu_arch_isa_flags = cpu_arch[i].flags;
1164 if (!cpu_arch_tune_set)
1165 {
1166 cpu_arch_tune = cpu_arch_isa;
1167 cpu_arch_tune_flags = cpu_arch_isa_flags;
1168 }
1169 break;
1170 }
1171 if ((cpu_arch_flags | cpu_arch[i].flags) != cpu_arch_flags)
1172 {
1173 cpu_sub_arch_name = cpu_arch[i].name;
1174 cpu_arch_flags |= cpu_arch[i].flags;
1175 }
1176 *input_line_pointer = e;
1177 demand_empty_rest_of_line ();
1178 return;
1179 }
1180 }
1181 if (i >= ARRAY_SIZE (cpu_arch))
1182 as_bad (_("no such architecture: `%s'"), string);
1183
1184 *input_line_pointer = e;
1185 }
1186 else
1187 as_bad (_("missing cpu architecture"));
1188
1189 no_cond_jump_promotion = 0;
1190 if (*input_line_pointer == ','
1191 && !is_end_of_line[(unsigned char) input_line_pointer[1]])
1192 {
1193 char *string = ++input_line_pointer;
1194 int e = get_symbol_end ();
1195
1196 if (strcmp (string, "nojumps") == 0)
1197 no_cond_jump_promotion = 1;
1198 else if (strcmp (string, "jumps") == 0)
1199 ;
1200 else
1201 as_bad (_("no such architecture modifier: `%s'"), string);
1202
1203 *input_line_pointer = e;
1204 }
1205
1206 demand_empty_rest_of_line ();
1207 }
1208
1209 unsigned long
1210 i386_mach ()
1211 {
1212 if (!strcmp (default_arch, "x86_64"))
1213 return bfd_mach_x86_64;
1214 else if (!strcmp (default_arch, "i386"))
1215 return bfd_mach_i386_i386;
1216 else
1217 as_fatal (_("Unknown architecture"));
1218 }
1219 \f
1220 void
1221 md_begin ()
1222 {
1223 const char *hash_err;
1224
1225 /* Initialize op_hash hash table. */
1226 op_hash = hash_new ();
1227
1228 {
1229 const template *optab;
1230 templates *core_optab;
1231
1232 /* Setup for loop. */
1233 optab = i386_optab;
1234 core_optab = (templates *) xmalloc (sizeof (templates));
1235 core_optab->start = optab;
1236
1237 while (1)
1238 {
1239 ++optab;
1240 if (optab->name == NULL
1241 || strcmp (optab->name, (optab - 1)->name) != 0)
1242 {
1243 /* different name --> ship out current template list;
1244 add to hash table; & begin anew. */
1245 core_optab->end = optab;
1246 hash_err = hash_insert (op_hash,
1247 (optab - 1)->name,
1248 (PTR) core_optab);
1249 if (hash_err)
1250 {
1251 as_fatal (_("Internal Error: Can't hash %s: %s"),
1252 (optab - 1)->name,
1253 hash_err);
1254 }
1255 if (optab->name == NULL)
1256 break;
1257 core_optab = (templates *) xmalloc (sizeof (templates));
1258 core_optab->start = optab;
1259 }
1260 }
1261 }
1262
1263 /* Initialize reg_hash hash table. */
1264 reg_hash = hash_new ();
1265 {
1266 const reg_entry *regtab;
1267
1268 for (regtab = i386_regtab;
1269 regtab < i386_regtab + sizeof (i386_regtab) / sizeof (i386_regtab[0]);
1270 regtab++)
1271 {
1272 hash_err = hash_insert (reg_hash, regtab->reg_name, (PTR) regtab);
1273 if (hash_err)
1274 as_fatal (_("Internal Error: Can't hash %s: %s"),
1275 regtab->reg_name,
1276 hash_err);
1277 }
1278 }
1279
1280 /* Fill in lexical tables: mnemonic_chars, operand_chars. */
1281 {
1282 int c;
1283 char *p;
1284
1285 for (c = 0; c < 256; c++)
1286 {
1287 if (ISDIGIT (c))
1288 {
1289 digit_chars[c] = c;
1290 mnemonic_chars[c] = c;
1291 register_chars[c] = c;
1292 operand_chars[c] = c;
1293 }
1294 else if (ISLOWER (c))
1295 {
1296 mnemonic_chars[c] = c;
1297 register_chars[c] = c;
1298 operand_chars[c] = c;
1299 }
1300 else if (ISUPPER (c))
1301 {
1302 mnemonic_chars[c] = TOLOWER (c);
1303 register_chars[c] = mnemonic_chars[c];
1304 operand_chars[c] = c;
1305 }
1306
1307 if (ISALPHA (c) || ISDIGIT (c))
1308 identifier_chars[c] = c;
1309 else if (c >= 128)
1310 {
1311 identifier_chars[c] = c;
1312 operand_chars[c] = c;
1313 }
1314 }
1315
1316 #ifdef LEX_AT
1317 identifier_chars['@'] = '@';
1318 #endif
1319 #ifdef LEX_QM
1320 identifier_chars['?'] = '?';
1321 operand_chars['?'] = '?';
1322 #endif
1323 digit_chars['-'] = '-';
1324 mnemonic_chars['-'] = '-';
1325 identifier_chars['_'] = '_';
1326 identifier_chars['.'] = '.';
1327
1328 for (p = operand_special_chars; *p != '\0'; p++)
1329 operand_chars[(unsigned char) *p] = *p;
1330 }
1331
1332 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
1333 if (IS_ELF)
1334 {
1335 record_alignment (text_section, 2);
1336 record_alignment (data_section, 2);
1337 record_alignment (bss_section, 2);
1338 }
1339 #endif
1340
1341 if (flag_code == CODE_64BIT)
1342 {
1343 x86_dwarf2_return_column = 16;
1344 x86_cie_data_alignment = -8;
1345 }
1346 else
1347 {
1348 x86_dwarf2_return_column = 8;
1349 x86_cie_data_alignment = -4;
1350 }
1351 }
1352
1353 void
1354 i386_print_statistics (file)
1355 FILE *file;
1356 {
1357 hash_print_statistics (file, "i386 opcode", op_hash);
1358 hash_print_statistics (file, "i386 register", reg_hash);
1359 }
1360 \f
1361 #ifdef DEBUG386
1362
1363 /* Debugging routines for md_assemble. */
1364 static void pi PARAMS ((char *, i386_insn *));
1365 static void pte PARAMS ((template *));
1366 static void pt PARAMS ((unsigned int));
1367 static void pe PARAMS ((expressionS *));
1368 static void ps PARAMS ((symbolS *));
1369
1370 static void
1371 pi (line, x)
1372 char *line;
1373 i386_insn *x;
1374 {
1375 unsigned int i;
1376
1377 fprintf (stdout, "%s: template ", line);
1378 pte (&x->tm);
1379 fprintf (stdout, " address: base %s index %s scale %x\n",
1380 x->base_reg ? x->base_reg->reg_name : "none",
1381 x->index_reg ? x->index_reg->reg_name : "none",
1382 x->log2_scale_factor);
1383 fprintf (stdout, " modrm: mode %x reg %x reg/mem %x\n",
1384 x->rm.mode, x->rm.reg, x->rm.regmem);
1385 fprintf (stdout, " sib: base %x index %x scale %x\n",
1386 x->sib.base, x->sib.index, x->sib.scale);
1387 fprintf (stdout, " rex: 64bit %x extX %x extY %x extZ %x\n",
1388 (x->rex & REX_MODE64) != 0,
1389 (x->rex & REX_EXTX) != 0,
1390 (x->rex & REX_EXTY) != 0,
1391 (x->rex & REX_EXTZ) != 0);
1392 for (i = 0; i < x->operands; i++)
1393 {
1394 fprintf (stdout, " #%d: ", i + 1);
1395 pt (x->types[i]);
1396 fprintf (stdout, "\n");
1397 if (x->types[i]
1398 & (Reg | SReg2 | SReg3 | Control | Debug | Test | RegMMX | RegXMM))
1399 fprintf (stdout, "%s\n", x->op[i].regs->reg_name);
1400 if (x->types[i] & Imm)
1401 pe (x->op[i].imms);
1402 if (x->types[i] & Disp)
1403 pe (x->op[i].disps);
1404 }
1405 }
1406
1407 static void
1408 pte (t)
1409 template *t;
1410 {
1411 unsigned int i;
1412 fprintf (stdout, " %d operands ", t->operands);
1413 fprintf (stdout, "opcode %x ", t->base_opcode);
1414 if (t->extension_opcode != None)
1415 fprintf (stdout, "ext %x ", t->extension_opcode);
1416 if (t->opcode_modifier & D)
1417 fprintf (stdout, "D");
1418 if (t->opcode_modifier & W)
1419 fprintf (stdout, "W");
1420 fprintf (stdout, "\n");
1421 for (i = 0; i < t->operands; i++)
1422 {
1423 fprintf (stdout, " #%d type ", i + 1);
1424 pt (t->operand_types[i]);
1425 fprintf (stdout, "\n");
1426 }
1427 }
1428
1429 static void
1430 pe (e)
1431 expressionS *e;
1432 {
1433 fprintf (stdout, " operation %d\n", e->X_op);
1434 fprintf (stdout, " add_number %ld (%lx)\n",
1435 (long) e->X_add_number, (long) e->X_add_number);
1436 if (e->X_add_symbol)
1437 {
1438 fprintf (stdout, " add_symbol ");
1439 ps (e->X_add_symbol);
1440 fprintf (stdout, "\n");
1441 }
1442 if (e->X_op_symbol)
1443 {
1444 fprintf (stdout, " op_symbol ");
1445 ps (e->X_op_symbol);
1446 fprintf (stdout, "\n");
1447 }
1448 }
1449
1450 static void
1451 ps (s)
1452 symbolS *s;
1453 {
1454 fprintf (stdout, "%s type %s%s",
1455 S_GET_NAME (s),
1456 S_IS_EXTERNAL (s) ? "EXTERNAL " : "",
1457 segment_name (S_GET_SEGMENT (s)));
1458 }
1459
1460 static struct type_name
1461 {
1462 unsigned int mask;
1463 char *tname;
1464 }
1465 const type_names[] =
1466 {
1467 { Reg8, "r8" },
1468 { Reg16, "r16" },
1469 { Reg32, "r32" },
1470 { Reg64, "r64" },
1471 { Imm8, "i8" },
1472 { Imm8S, "i8s" },
1473 { Imm16, "i16" },
1474 { Imm32, "i32" },
1475 { Imm32S, "i32s" },
1476 { Imm64, "i64" },
1477 { Imm1, "i1" },
1478 { BaseIndex, "BaseIndex" },
1479 { Disp8, "d8" },
1480 { Disp16, "d16" },
1481 { Disp32, "d32" },
1482 { Disp32S, "d32s" },
1483 { Disp64, "d64" },
1484 { InOutPortReg, "InOutPortReg" },
1485 { ShiftCount, "ShiftCount" },
1486 { Control, "control reg" },
1487 { Test, "test reg" },
1488 { Debug, "debug reg" },
1489 { FloatReg, "FReg" },
1490 { FloatAcc, "FAcc" },
1491 { SReg2, "SReg2" },
1492 { SReg3, "SReg3" },
1493 { Acc, "Acc" },
1494 { JumpAbsolute, "Jump Absolute" },
1495 { RegMMX, "rMMX" },
1496 { RegXMM, "rXMM" },
1497 { EsSeg, "es" },
1498 { 0, "" }
1499 };
1500
1501 static void
1502 pt (t)
1503 unsigned int t;
1504 {
1505 const struct type_name *ty;
1506
1507 for (ty = type_names; ty->mask; ty++)
1508 if (t & ty->mask)
1509 fprintf (stdout, "%s, ", ty->tname);
1510 fflush (stdout);
1511 }
1512
1513 #endif /* DEBUG386 */
1514 \f
1515 static bfd_reloc_code_real_type
1516 reloc (unsigned int size,
1517 int pcrel,
1518 int sign,
1519 bfd_reloc_code_real_type other)
1520 {
1521 if (other != NO_RELOC)
1522 {
1523 reloc_howto_type *reloc;
1524
1525 if (size == 8)
1526 switch (other)
1527 {
1528 case BFD_RELOC_X86_64_GOT32:
1529 return BFD_RELOC_X86_64_GOT64;
1530 break;
1531 case BFD_RELOC_X86_64_PLTOFF64:
1532 return BFD_RELOC_X86_64_PLTOFF64;
1533 break;
1534 case BFD_RELOC_X86_64_GOTPC32:
1535 other = BFD_RELOC_X86_64_GOTPC64;
1536 break;
1537 case BFD_RELOC_X86_64_GOTPCREL:
1538 other = BFD_RELOC_X86_64_GOTPCREL64;
1539 break;
1540 case BFD_RELOC_X86_64_TPOFF32:
1541 other = BFD_RELOC_X86_64_TPOFF64;
1542 break;
1543 case BFD_RELOC_X86_64_DTPOFF32:
1544 other = BFD_RELOC_X86_64_DTPOFF64;
1545 break;
1546 default:
1547 break;
1548 }
1549
1550 /* Sign-checking 4-byte relocations in 16-/32-bit code is pointless. */
1551 if (size == 4 && flag_code != CODE_64BIT)
1552 sign = -1;
1553
1554 reloc = bfd_reloc_type_lookup (stdoutput, other);
1555 if (!reloc)
1556 as_bad (_("unknown relocation (%u)"), other);
1557 else if (size != bfd_get_reloc_size (reloc))
1558 as_bad (_("%u-byte relocation cannot be applied to %u-byte field"),
1559 bfd_get_reloc_size (reloc),
1560 size);
1561 else if (pcrel && !reloc->pc_relative)
1562 as_bad (_("non-pc-relative relocation for pc-relative field"));
1563 else if ((reloc->complain_on_overflow == complain_overflow_signed
1564 && !sign)
1565 || (reloc->complain_on_overflow == complain_overflow_unsigned
1566 && sign > 0))
1567 as_bad (_("relocated field and relocation type differ in signedness"));
1568 else
1569 return other;
1570 return NO_RELOC;
1571 }
1572
1573 if (pcrel)
1574 {
1575 if (!sign)
1576 as_bad (_("there are no unsigned pc-relative relocations"));
1577 switch (size)
1578 {
1579 case 1: return BFD_RELOC_8_PCREL;
1580 case 2: return BFD_RELOC_16_PCREL;
1581 case 4: return BFD_RELOC_32_PCREL;
1582 case 8: return BFD_RELOC_64_PCREL;
1583 }
1584 as_bad (_("cannot do %u byte pc-relative relocation"), size);
1585 }
1586 else
1587 {
1588 if (sign > 0)
1589 switch (size)
1590 {
1591 case 4: return BFD_RELOC_X86_64_32S;
1592 }
1593 else
1594 switch (size)
1595 {
1596 case 1: return BFD_RELOC_8;
1597 case 2: return BFD_RELOC_16;
1598 case 4: return BFD_RELOC_32;
1599 case 8: return BFD_RELOC_64;
1600 }
1601 as_bad (_("cannot do %s %u byte relocation"),
1602 sign > 0 ? "signed" : "unsigned", size);
1603 }
1604
1605 abort ();
1606 return BFD_RELOC_NONE;
1607 }
1608
1609 /* Here we decide which fixups can be adjusted to make them relative to
1610 the beginning of the section instead of the symbol. Basically we need
1611 to make sure that the dynamic relocations are done correctly, so in
1612 some cases we force the original symbol to be used. */
1613
1614 int
1615 tc_i386_fix_adjustable (fixP)
1616 fixS *fixP ATTRIBUTE_UNUSED;
1617 {
1618 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
1619 if (!IS_ELF)
1620 return 1;
1621
1622 /* Don't adjust pc-relative references to merge sections in 64-bit
1623 mode. */
1624 if (use_rela_relocations
1625 && (S_GET_SEGMENT (fixP->fx_addsy)->flags & SEC_MERGE) != 0
1626 && fixP->fx_pcrel)
1627 return 0;
1628
1629 /* The x86_64 GOTPCREL are represented as 32bit PCrel relocations
1630 and changed later by validate_fix. */
1631 if (GOT_symbol && fixP->fx_subsy == GOT_symbol
1632 && fixP->fx_r_type == BFD_RELOC_32_PCREL)
1633 return 0;
1634
1635 /* adjust_reloc_syms doesn't know about the GOT. */
1636 if (fixP->fx_r_type == BFD_RELOC_386_GOTOFF
1637 || fixP->fx_r_type == BFD_RELOC_386_PLT32
1638 || fixP->fx_r_type == BFD_RELOC_386_GOT32
1639 || fixP->fx_r_type == BFD_RELOC_386_TLS_GD
1640 || fixP->fx_r_type == BFD_RELOC_386_TLS_LDM
1641 || fixP->fx_r_type == BFD_RELOC_386_TLS_LDO_32
1642 || fixP->fx_r_type == BFD_RELOC_386_TLS_IE_32
1643 || fixP->fx_r_type == BFD_RELOC_386_TLS_IE
1644 || fixP->fx_r_type == BFD_RELOC_386_TLS_GOTIE
1645 || fixP->fx_r_type == BFD_RELOC_386_TLS_LE_32
1646 || fixP->fx_r_type == BFD_RELOC_386_TLS_LE
1647 || fixP->fx_r_type == BFD_RELOC_386_TLS_GOTDESC
1648 || fixP->fx_r_type == BFD_RELOC_386_TLS_DESC_CALL
1649 || fixP->fx_r_type == BFD_RELOC_X86_64_PLT32
1650 || fixP->fx_r_type == BFD_RELOC_X86_64_GOT32
1651 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTPCREL
1652 || fixP->fx_r_type == BFD_RELOC_X86_64_TLSGD
1653 || fixP->fx_r_type == BFD_RELOC_X86_64_TLSLD
1654 || fixP->fx_r_type == BFD_RELOC_X86_64_DTPOFF32
1655 || fixP->fx_r_type == BFD_RELOC_X86_64_DTPOFF64
1656 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTTPOFF
1657 || fixP->fx_r_type == BFD_RELOC_X86_64_TPOFF32
1658 || fixP->fx_r_type == BFD_RELOC_X86_64_TPOFF64
1659 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTOFF64
1660 || fixP->fx_r_type == BFD_RELOC_X86_64_GOTPC32_TLSDESC
1661 || fixP->fx_r_type == BFD_RELOC_X86_64_TLSDESC_CALL
1662 || fixP->fx_r_type == BFD_RELOC_VTABLE_INHERIT
1663 || fixP->fx_r_type == BFD_RELOC_VTABLE_ENTRY)
1664 return 0;
1665 #endif
1666 return 1;
1667 }
1668
1669 static int intel_float_operand PARAMS ((const char *mnemonic));
1670
1671 static int
1672 intel_float_operand (mnemonic)
1673 const char *mnemonic;
1674 {
1675 /* Note that the value returned is meaningful only for opcodes with (memory)
1676 operands, hence the code here is free to improperly handle opcodes that
1677 have no operands (for better performance and smaller code). */
1678
1679 if (mnemonic[0] != 'f')
1680 return 0; /* non-math */
1681
1682 switch (mnemonic[1])
1683 {
1684 /* fclex, fdecstp, fdisi, femms, feni, fincstp, finit, fsetpm, and
1685 the fs segment override prefix not currently handled because no
1686 call path can make opcodes without operands get here */
1687 case 'i':
1688 return 2 /* integer op */;
1689 case 'l':
1690 if (mnemonic[2] == 'd' && (mnemonic[3] == 'c' || mnemonic[3] == 'e'))
1691 return 3; /* fldcw/fldenv */
1692 break;
1693 case 'n':
1694 if (mnemonic[2] != 'o' /* fnop */)
1695 return 3; /* non-waiting control op */
1696 break;
1697 case 'r':
1698 if (mnemonic[2] == 's')
1699 return 3; /* frstor/frstpm */
1700 break;
1701 case 's':
1702 if (mnemonic[2] == 'a')
1703 return 3; /* fsave */
1704 if (mnemonic[2] == 't')
1705 {
1706 switch (mnemonic[3])
1707 {
1708 case 'c': /* fstcw */
1709 case 'd': /* fstdw */
1710 case 'e': /* fstenv */
1711 case 's': /* fsts[gw] */
1712 return 3;
1713 }
1714 }
1715 break;
1716 case 'x':
1717 if (mnemonic[2] == 'r' || mnemonic[2] == 's')
1718 return 0; /* fxsave/fxrstor are not really math ops */
1719 break;
1720 }
1721
1722 return 1;
1723 }
1724
1725 /* This is the guts of the machine-dependent assembler. LINE points to a
1726 machine dependent instruction. This function is supposed to emit
1727 the frags/bytes it assembles to. */
1728
1729 void
1730 md_assemble (line)
1731 char *line;
1732 {
1733 int j;
1734 char mnemonic[MAX_MNEM_SIZE];
1735
1736 /* Initialize globals. */
1737 memset (&i, '\0', sizeof (i));
1738 for (j = 0; j < MAX_OPERANDS; j++)
1739 i.reloc[j] = NO_RELOC;
1740 memset (disp_expressions, '\0', sizeof (disp_expressions));
1741 memset (im_expressions, '\0', sizeof (im_expressions));
1742 save_stack_p = save_stack;
1743
1744 /* First parse an instruction mnemonic & call i386_operand for the operands.
1745 We assume that the scrubber has arranged it so that line[0] is the valid
1746 start of a (possibly prefixed) mnemonic. */
1747
1748 line = parse_insn (line, mnemonic);
1749 if (line == NULL)
1750 return;
1751
1752 line = parse_operands (line, mnemonic);
1753 if (line == NULL)
1754 return;
1755
1756 /* The order of the immediates should be reversed
1757 for 2 immediates extrq and insertq instructions */
1758 if ((i.imm_operands == 2) &&
1759 ((strcmp (mnemonic, "extrq") == 0)
1760 || (strcmp (mnemonic, "insertq") == 0)))
1761 {
1762 swap_imm_operands ();
1763 /* "extrq" and insertq" are the only two instructions whose operands
1764 have to be reversed even though they have two immediate operands.
1765 */
1766 if (intel_syntax)
1767 swap_operands ();
1768 }
1769
1770 /* Now we've parsed the mnemonic into a set of templates, and have the
1771 operands at hand. */
1772
1773 /* All intel opcodes have reversed operands except for "bound" and
1774 "enter". We also don't reverse intersegment "jmp" and "call"
1775 instructions with 2 immediate operands so that the immediate segment
1776 precedes the offset, as it does when in AT&T mode. */
1777 if (intel_syntax && i.operands > 1
1778 && (strcmp (mnemonic, "bound") != 0)
1779 && (strcmp (mnemonic, "invlpga") != 0)
1780 && !((i.types[0] & Imm) && (i.types[1] & Imm)))
1781 swap_operands ();
1782
1783 if (i.imm_operands)
1784 optimize_imm ();
1785
1786 /* Don't optimize displacement for movabs since it only takes 64bit
1787 displacement. */
1788 if (i.disp_operands
1789 && (flag_code != CODE_64BIT
1790 || strcmp (mnemonic, "movabs") != 0))
1791 optimize_disp ();
1792
1793 /* Next, we find a template that matches the given insn,
1794 making sure the overlap of the given operands types is consistent
1795 with the template operand types. */
1796
1797 if (!match_template ())
1798 return;
1799
1800 if (intel_syntax)
1801 {
1802 /* Undo SYSV386_COMPAT brokenness when in Intel mode. See i386.h */
1803 if (SYSV386_COMPAT
1804 && (i.tm.base_opcode & 0xfffffde0) == 0xdce0)
1805 i.tm.base_opcode ^= FloatR;
1806
1807 /* Zap movzx and movsx suffix. The suffix may have been set from
1808 "word ptr" or "byte ptr" on the source operand, but we'll use
1809 the suffix later to choose the destination register. */
1810 if ((i.tm.base_opcode & ~9) == 0x0fb6)
1811 {
1812 if (i.reg_operands < 2
1813 && !i.suffix
1814 && (~i.tm.opcode_modifier
1815 & (No_bSuf
1816 | No_wSuf
1817 | No_lSuf
1818 | No_sSuf
1819 | No_xSuf
1820 | No_qSuf)))
1821 as_bad (_("ambiguous operand size for `%s'"), i.tm.name);
1822
1823 i.suffix = 0;
1824 }
1825 }
1826
1827 if (i.tm.opcode_modifier & FWait)
1828 if (!add_prefix (FWAIT_OPCODE))
1829 return;
1830
1831 /* Check string instruction segment overrides. */
1832 if ((i.tm.opcode_modifier & IsString) != 0 && i.mem_operands != 0)
1833 {
1834 if (!check_string ())
1835 return;
1836 }
1837
1838 if (!process_suffix ())
1839 return;
1840
1841 /* Make still unresolved immediate matches conform to size of immediate
1842 given in i.suffix. */
1843 if (!finalize_imm ())
1844 return;
1845
1846 if (i.types[0] & Imm1)
1847 i.imm_operands = 0; /* kludge for shift insns. */
1848 if (i.types[0] & ImplicitRegister)
1849 i.reg_operands--;
1850 if (i.types[1] & ImplicitRegister)
1851 i.reg_operands--;
1852 if (i.types[2] & ImplicitRegister)
1853 i.reg_operands--;
1854
1855 if (i.tm.opcode_modifier & ImmExt)
1856 {
1857 expressionS *exp;
1858
1859 if ((i.tm.cpu_flags & CpuSSE3) && i.operands > 0)
1860 {
1861 /* Streaming SIMD extensions 3 Instructions have the fixed
1862 operands with an opcode suffix which is coded in the same
1863 place as an 8-bit immediate field would be. Here we check
1864 those operands and remove them afterwards. */
1865 unsigned int x;
1866
1867 for (x = 0; x < i.operands; x++)
1868 if (i.op[x].regs->reg_num != x)
1869 as_bad (_("can't use register '%%%s' as operand %d in '%s'."),
1870 i.op[x].regs->reg_name, x + 1, i.tm.name);
1871 i.operands = 0;
1872 }
1873
1874 /* These AMD 3DNow! and Intel Katmai New Instructions have an
1875 opcode suffix which is coded in the same place as an 8-bit
1876 immediate field would be. Here we fake an 8-bit immediate
1877 operand from the opcode suffix stored in tm.extension_opcode. */
1878
1879 assert (i.imm_operands == 0 && i.operands <= 2 && 2 < MAX_OPERANDS);
1880
1881 exp = &im_expressions[i.imm_operands++];
1882 i.op[i.operands].imms = exp;
1883 i.types[i.operands++] = Imm8;
1884 exp->X_op = O_constant;
1885 exp->X_add_number = i.tm.extension_opcode;
1886 i.tm.extension_opcode = None;
1887 }
1888
1889 /* For insns with operands there are more diddles to do to the opcode. */
1890 if (i.operands)
1891 {
1892 if (!process_operands ())
1893 return;
1894 }
1895 else if (!quiet_warnings && (i.tm.opcode_modifier & Ugh) != 0)
1896 {
1897 /* UnixWare fsub no args is alias for fsubp, fadd -> faddp, etc. */
1898 as_warn (_("translating to `%sp'"), i.tm.name);
1899 }
1900
1901 /* Handle conversion of 'int $3' --> special int3 insn. */
1902 if (i.tm.base_opcode == INT_OPCODE && i.op[0].imms->X_add_number == 3)
1903 {
1904 i.tm.base_opcode = INT3_OPCODE;
1905 i.imm_operands = 0;
1906 }
1907
1908 if ((i.tm.opcode_modifier & (Jump | JumpByte | JumpDword))
1909 && i.op[0].disps->X_op == O_constant)
1910 {
1911 /* Convert "jmp constant" (and "call constant") to a jump (call) to
1912 the absolute address given by the constant. Since ix86 jumps and
1913 calls are pc relative, we need to generate a reloc. */
1914 i.op[0].disps->X_add_symbol = &abs_symbol;
1915 i.op[0].disps->X_op = O_symbol;
1916 }
1917
1918 if ((i.tm.opcode_modifier & Rex64) != 0)
1919 i.rex |= REX_MODE64;
1920
1921 /* For 8 bit registers we need an empty rex prefix. Also if the
1922 instruction already has a prefix, we need to convert old
1923 registers to new ones. */
1924
1925 if (((i.types[0] & Reg8) != 0
1926 && (i.op[0].regs->reg_flags & RegRex64) != 0)
1927 || ((i.types[1] & Reg8) != 0
1928 && (i.op[1].regs->reg_flags & RegRex64) != 0)
1929 || (((i.types[0] & Reg8) != 0 || (i.types[1] & Reg8) != 0)
1930 && i.rex != 0))
1931 {
1932 int x;
1933
1934 i.rex |= REX_OPCODE;
1935 for (x = 0; x < 2; x++)
1936 {
1937 /* Look for 8 bit operand that uses old registers. */
1938 if ((i.types[x] & Reg8) != 0
1939 && (i.op[x].regs->reg_flags & RegRex64) == 0)
1940 {
1941 /* In case it is "hi" register, give up. */
1942 if (i.op[x].regs->reg_num > 3)
1943 as_bad (_("can't encode register '%%%s' in an instruction requiring REX prefix."),
1944 i.op[x].regs->reg_name);
1945
1946 /* Otherwise it is equivalent to the extended register.
1947 Since the encoding doesn't change this is merely
1948 cosmetic cleanup for debug output. */
1949
1950 i.op[x].regs = i.op[x].regs + 8;
1951 }
1952 }
1953 }
1954
1955 if (i.rex != 0)
1956 add_prefix (REX_OPCODE | i.rex);
1957
1958 /* We are ready to output the insn. */
1959 output_insn ();
1960 }
1961
1962 static char *
1963 parse_insn (line, mnemonic)
1964 char *line;
1965 char *mnemonic;
1966 {
1967 char *l = line;
1968 char *token_start = l;
1969 char *mnem_p;
1970 int supported;
1971 const template *t;
1972
1973 /* Non-zero if we found a prefix only acceptable with string insns. */
1974 const char *expecting_string_instruction = NULL;
1975
1976 while (1)
1977 {
1978 mnem_p = mnemonic;
1979 while ((*mnem_p = mnemonic_chars[(unsigned char) *l]) != 0)
1980 {
1981 mnem_p++;
1982 if (mnem_p >= mnemonic + MAX_MNEM_SIZE)
1983 {
1984 as_bad (_("no such instruction: `%s'"), token_start);
1985 return NULL;
1986 }
1987 l++;
1988 }
1989 if (!is_space_char (*l)
1990 && *l != END_OF_INSN
1991 && (intel_syntax
1992 || (*l != PREFIX_SEPARATOR
1993 && *l != ',')))
1994 {
1995 as_bad (_("invalid character %s in mnemonic"),
1996 output_invalid (*l));
1997 return NULL;
1998 }
1999 if (token_start == l)
2000 {
2001 if (!intel_syntax && *l == PREFIX_SEPARATOR)
2002 as_bad (_("expecting prefix; got nothing"));
2003 else
2004 as_bad (_("expecting mnemonic; got nothing"));
2005 return NULL;
2006 }
2007
2008 /* Look up instruction (or prefix) via hash table. */
2009 current_templates = hash_find (op_hash, mnemonic);
2010
2011 if (*l != END_OF_INSN
2012 && (!is_space_char (*l) || l[1] != END_OF_INSN)
2013 && current_templates
2014 && (current_templates->start->opcode_modifier & IsPrefix))
2015 {
2016 if (current_templates->start->cpu_flags
2017 & (flag_code != CODE_64BIT ? Cpu64 : CpuNo64))
2018 {
2019 as_bad ((flag_code != CODE_64BIT
2020 ? _("`%s' is only supported in 64-bit mode")
2021 : _("`%s' is not supported in 64-bit mode")),
2022 current_templates->start->name);
2023 return NULL;
2024 }
2025 /* If we are in 16-bit mode, do not allow addr16 or data16.
2026 Similarly, in 32-bit mode, do not allow addr32 or data32. */
2027 if ((current_templates->start->opcode_modifier & (Size16 | Size32))
2028 && flag_code != CODE_64BIT
2029 && (((current_templates->start->opcode_modifier & Size32) != 0)
2030 ^ (flag_code == CODE_16BIT)))
2031 {
2032 as_bad (_("redundant %s prefix"),
2033 current_templates->start->name);
2034 return NULL;
2035 }
2036 /* Add prefix, checking for repeated prefixes. */
2037 switch (add_prefix (current_templates->start->base_opcode))
2038 {
2039 case 0:
2040 return NULL;
2041 case 2:
2042 expecting_string_instruction = current_templates->start->name;
2043 break;
2044 }
2045 /* Skip past PREFIX_SEPARATOR and reset token_start. */
2046 token_start = ++l;
2047 }
2048 else
2049 break;
2050 }
2051
2052 if (!current_templates)
2053 {
2054 /* See if we can get a match by trimming off a suffix. */
2055 switch (mnem_p[-1])
2056 {
2057 case WORD_MNEM_SUFFIX:
2058 if (intel_syntax && (intel_float_operand (mnemonic) & 2))
2059 i.suffix = SHORT_MNEM_SUFFIX;
2060 else
2061 case BYTE_MNEM_SUFFIX:
2062 case QWORD_MNEM_SUFFIX:
2063 i.suffix = mnem_p[-1];
2064 mnem_p[-1] = '\0';
2065 current_templates = hash_find (op_hash, mnemonic);
2066 break;
2067 case SHORT_MNEM_SUFFIX:
2068 case LONG_MNEM_SUFFIX:
2069 if (!intel_syntax)
2070 {
2071 i.suffix = mnem_p[-1];
2072 mnem_p[-1] = '\0';
2073 current_templates = hash_find (op_hash, mnemonic);
2074 }
2075 break;
2076
2077 /* Intel Syntax. */
2078 case 'd':
2079 if (intel_syntax)
2080 {
2081 if (intel_float_operand (mnemonic) == 1)
2082 i.suffix = SHORT_MNEM_SUFFIX;
2083 else
2084 i.suffix = LONG_MNEM_SUFFIX;
2085 mnem_p[-1] = '\0';
2086 current_templates = hash_find (op_hash, mnemonic);
2087 }
2088 break;
2089 }
2090 if (!current_templates)
2091 {
2092 as_bad (_("no such instruction: `%s'"), token_start);
2093 return NULL;
2094 }
2095 }
2096
2097 if (current_templates->start->opcode_modifier & (Jump | JumpByte))
2098 {
2099 /* Check for a branch hint. We allow ",pt" and ",pn" for
2100 predict taken and predict not taken respectively.
2101 I'm not sure that branch hints actually do anything on loop
2102 and jcxz insns (JumpByte) for current Pentium4 chips. They
2103 may work in the future and it doesn't hurt to accept them
2104 now. */
2105 if (l[0] == ',' && l[1] == 'p')
2106 {
2107 if (l[2] == 't')
2108 {
2109 if (!add_prefix (DS_PREFIX_OPCODE))
2110 return NULL;
2111 l += 3;
2112 }
2113 else if (l[2] == 'n')
2114 {
2115 if (!add_prefix (CS_PREFIX_OPCODE))
2116 return NULL;
2117 l += 3;
2118 }
2119 }
2120 }
2121 /* Any other comma loses. */
2122 if (*l == ',')
2123 {
2124 as_bad (_("invalid character %s in mnemonic"),
2125 output_invalid (*l));
2126 return NULL;
2127 }
2128
2129 /* Check if instruction is supported on specified architecture. */
2130 supported = 0;
2131 for (t = current_templates->start; t < current_templates->end; ++t)
2132 {
2133 if (!((t->cpu_flags & ~(Cpu64 | CpuNo64))
2134 & ~(cpu_arch_flags & ~(Cpu64 | CpuNo64))))
2135 supported |= 1;
2136 if (!(t->cpu_flags & (flag_code == CODE_64BIT ? CpuNo64 : Cpu64)))
2137 supported |= 2;
2138 }
2139 if (!(supported & 2))
2140 {
2141 as_bad (flag_code == CODE_64BIT
2142 ? _("`%s' is not supported in 64-bit mode")
2143 : _("`%s' is only supported in 64-bit mode"),
2144 current_templates->start->name);
2145 return NULL;
2146 }
2147 if (!(supported & 1))
2148 {
2149 as_warn (_("`%s' is not supported on `%s%s'"),
2150 current_templates->start->name,
2151 cpu_arch_name,
2152 cpu_sub_arch_name ? cpu_sub_arch_name : "");
2153 }
2154 else if ((Cpu386 & ~cpu_arch_flags) && (flag_code != CODE_16BIT))
2155 {
2156 as_warn (_("use .code16 to ensure correct addressing mode"));
2157 }
2158
2159 /* Check for rep/repne without a string instruction. */
2160 if (expecting_string_instruction)
2161 {
2162 static templates override;
2163
2164 for (t = current_templates->start; t < current_templates->end; ++t)
2165 if (t->opcode_modifier & IsString)
2166 break;
2167 if (t >= current_templates->end)
2168 {
2169 as_bad (_("expecting string instruction after `%s'"),
2170 expecting_string_instruction);
2171 return NULL;
2172 }
2173 for (override.start = t; t < current_templates->end; ++t)
2174 if (!(t->opcode_modifier & IsString))
2175 break;
2176 override.end = t;
2177 current_templates = &override;
2178 }
2179
2180 return l;
2181 }
2182
2183 static char *
2184 parse_operands (l, mnemonic)
2185 char *l;
2186 const char *mnemonic;
2187 {
2188 char *token_start;
2189
2190 /* 1 if operand is pending after ','. */
2191 unsigned int expecting_operand = 0;
2192
2193 /* Non-zero if operand parens not balanced. */
2194 unsigned int paren_not_balanced;
2195
2196 while (*l != END_OF_INSN)
2197 {
2198 /* Skip optional white space before operand. */
2199 if (is_space_char (*l))
2200 ++l;
2201 if (!is_operand_char (*l) && *l != END_OF_INSN)
2202 {
2203 as_bad (_("invalid character %s before operand %d"),
2204 output_invalid (*l),
2205 i.operands + 1);
2206 return NULL;
2207 }
2208 token_start = l; /* after white space */
2209 paren_not_balanced = 0;
2210 while (paren_not_balanced || *l != ',')
2211 {
2212 if (*l == END_OF_INSN)
2213 {
2214 if (paren_not_balanced)
2215 {
2216 if (!intel_syntax)
2217 as_bad (_("unbalanced parenthesis in operand %d."),
2218 i.operands + 1);
2219 else
2220 as_bad (_("unbalanced brackets in operand %d."),
2221 i.operands + 1);
2222 return NULL;
2223 }
2224 else
2225 break; /* we are done */
2226 }
2227 else if (!is_operand_char (*l) && !is_space_char (*l))
2228 {
2229 as_bad (_("invalid character %s in operand %d"),
2230 output_invalid (*l),
2231 i.operands + 1);
2232 return NULL;
2233 }
2234 if (!intel_syntax)
2235 {
2236 if (*l == '(')
2237 ++paren_not_balanced;
2238 if (*l == ')')
2239 --paren_not_balanced;
2240 }
2241 else
2242 {
2243 if (*l == '[')
2244 ++paren_not_balanced;
2245 if (*l == ']')
2246 --paren_not_balanced;
2247 }
2248 l++;
2249 }
2250 if (l != token_start)
2251 { /* Yes, we've read in another operand. */
2252 unsigned int operand_ok;
2253 this_operand = i.operands++;
2254 if (i.operands > MAX_OPERANDS)
2255 {
2256 as_bad (_("spurious operands; (%d operands/instruction max)"),
2257 MAX_OPERANDS);
2258 return NULL;
2259 }
2260 /* Now parse operand adding info to 'i' as we go along. */
2261 END_STRING_AND_SAVE (l);
2262
2263 if (intel_syntax)
2264 operand_ok =
2265 i386_intel_operand (token_start,
2266 intel_float_operand (mnemonic));
2267 else
2268 operand_ok = i386_operand (token_start);
2269
2270 RESTORE_END_STRING (l);
2271 if (!operand_ok)
2272 return NULL;
2273 }
2274 else
2275 {
2276 if (expecting_operand)
2277 {
2278 expecting_operand_after_comma:
2279 as_bad (_("expecting operand after ','; got nothing"));
2280 return NULL;
2281 }
2282 if (*l == ',')
2283 {
2284 as_bad (_("expecting operand before ','; got nothing"));
2285 return NULL;
2286 }
2287 }
2288
2289 /* Now *l must be either ',' or END_OF_INSN. */
2290 if (*l == ',')
2291 {
2292 if (*++l == END_OF_INSN)
2293 {
2294 /* Just skip it, if it's \n complain. */
2295 goto expecting_operand_after_comma;
2296 }
2297 expecting_operand = 1;
2298 }
2299 }
2300 return l;
2301 }
2302
2303 static void
2304 swap_imm_operands ()
2305 {
2306 union i386_op temp_op;
2307 unsigned int temp_type;
2308 enum bfd_reloc_code_real temp_reloc;
2309 int xchg1 = 0;
2310 int xchg2 = 1;
2311
2312 temp_type = i.types[xchg2];
2313 i.types[xchg2] = i.types[xchg1];
2314 i.types[xchg1] = temp_type;
2315 temp_op = i.op[xchg2];
2316 i.op[xchg2] = i.op[xchg1];
2317 i.op[xchg1] = temp_op;
2318 temp_reloc = i.reloc[xchg2];
2319 i.reloc[xchg2] = i.reloc[xchg1];
2320 i.reloc[xchg1] = temp_reloc;
2321 }
2322
2323
2324 static void
2325 swap_operands ()
2326 {
2327 union i386_op temp_op;
2328 unsigned int temp_type;
2329 enum bfd_reloc_code_real temp_reloc;
2330 int xchg1 = 0;
2331 int xchg2 = 0;
2332
2333 if (i.operands == 4)
2334 /* There will be two exchanges in a 4 operand instruction.
2335 First exchange is the done inside this block.(1st and 4rth operand)
2336 The next exchange is done outside this block.(2nd and 3rd operand) */
2337 {
2338 xchg1 = 0;
2339 xchg2 = 3;
2340 temp_type = i.types[xchg2];
2341 i.types[xchg2] = i.types[xchg1];
2342 i.types[xchg1] = temp_type;
2343 temp_op = i.op[xchg2];
2344 i.op[xchg2] = i.op[xchg1];
2345 i.op[xchg1] = temp_op;
2346 temp_reloc = i.reloc[xchg2];
2347 i.reloc[xchg2] = i.reloc[xchg1];
2348 i.reloc[xchg1] = temp_reloc;
2349 xchg1 = 1;
2350 xchg2 = 2;
2351 }
2352
2353 if (i.operands == 2)
2354 {
2355 xchg1 = 0;
2356 xchg2 = 1;
2357 }
2358 else if (i.operands == 3)
2359 {
2360 xchg1 = 0;
2361 xchg2 = 2;
2362 }
2363 temp_type = i.types[xchg2];
2364 i.types[xchg2] = i.types[xchg1];
2365 i.types[xchg1] = temp_type;
2366 temp_op = i.op[xchg2];
2367 i.op[xchg2] = i.op[xchg1];
2368 i.op[xchg1] = temp_op;
2369 temp_reloc = i.reloc[xchg2];
2370 i.reloc[xchg2] = i.reloc[xchg1];
2371 i.reloc[xchg1] = temp_reloc;
2372
2373 if (i.mem_operands == 2)
2374 {
2375 const seg_entry *temp_seg;
2376 temp_seg = i.seg[0];
2377 i.seg[0] = i.seg[1];
2378 i.seg[1] = temp_seg;
2379 }
2380 }
2381
2382 /* Try to ensure constant immediates are represented in the smallest
2383 opcode possible. */
2384 static void
2385 optimize_imm ()
2386 {
2387 char guess_suffix = 0;
2388 int op;
2389
2390 if (i.suffix)
2391 guess_suffix = i.suffix;
2392 else if (i.reg_operands)
2393 {
2394 /* Figure out a suffix from the last register operand specified.
2395 We can't do this properly yet, ie. excluding InOutPortReg,
2396 but the following works for instructions with immediates.
2397 In any case, we can't set i.suffix yet. */
2398 for (op = i.operands; --op >= 0;)
2399 if (i.types[op] & Reg)
2400 {
2401 if (i.types[op] & Reg8)
2402 guess_suffix = BYTE_MNEM_SUFFIX;
2403 else if (i.types[op] & Reg16)
2404 guess_suffix = WORD_MNEM_SUFFIX;
2405 else if (i.types[op] & Reg32)
2406 guess_suffix = LONG_MNEM_SUFFIX;
2407 else if (i.types[op] & Reg64)
2408 guess_suffix = QWORD_MNEM_SUFFIX;
2409 break;
2410 }
2411 }
2412 else if ((flag_code == CODE_16BIT) ^ (i.prefix[DATA_PREFIX] != 0))
2413 guess_suffix = WORD_MNEM_SUFFIX;
2414
2415 for (op = i.operands; --op >= 0;)
2416 if (i.types[op] & Imm)
2417 {
2418 switch (i.op[op].imms->X_op)
2419 {
2420 case O_constant:
2421 /* If a suffix is given, this operand may be shortened. */
2422 switch (guess_suffix)
2423 {
2424 case LONG_MNEM_SUFFIX:
2425 i.types[op] |= Imm32 | Imm64;
2426 break;
2427 case WORD_MNEM_SUFFIX:
2428 i.types[op] |= Imm16 | Imm32S | Imm32 | Imm64;
2429 break;
2430 case BYTE_MNEM_SUFFIX:
2431 i.types[op] |= Imm16 | Imm8 | Imm8S | Imm32S | Imm32 | Imm64;
2432 break;
2433 }
2434
2435 /* If this operand is at most 16 bits, convert it
2436 to a signed 16 bit number before trying to see
2437 whether it will fit in an even smaller size.
2438 This allows a 16-bit operand such as $0xffe0 to
2439 be recognised as within Imm8S range. */
2440 if ((i.types[op] & Imm16)
2441 && (i.op[op].imms->X_add_number & ~(offsetT) 0xffff) == 0)
2442 {
2443 i.op[op].imms->X_add_number =
2444 (((i.op[op].imms->X_add_number & 0xffff) ^ 0x8000) - 0x8000);
2445 }
2446 if ((i.types[op] & Imm32)
2447 && ((i.op[op].imms->X_add_number & ~(((offsetT) 2 << 31) - 1))
2448 == 0))
2449 {
2450 i.op[op].imms->X_add_number = ((i.op[op].imms->X_add_number
2451 ^ ((offsetT) 1 << 31))
2452 - ((offsetT) 1 << 31));
2453 }
2454 i.types[op] |= smallest_imm_type (i.op[op].imms->X_add_number);
2455
2456 /* We must avoid matching of Imm32 templates when 64bit
2457 only immediate is available. */
2458 if (guess_suffix == QWORD_MNEM_SUFFIX)
2459 i.types[op] &= ~Imm32;
2460 break;
2461
2462 case O_absent:
2463 case O_register:
2464 abort ();
2465
2466 /* Symbols and expressions. */
2467 default:
2468 /* Convert symbolic operand to proper sizes for matching, but don't
2469 prevent matching a set of insns that only supports sizes other
2470 than those matching the insn suffix. */
2471 {
2472 unsigned int mask, allowed = 0;
2473 const template *t;
2474
2475 for (t = current_templates->start; t < current_templates->end; ++t)
2476 allowed |= t->operand_types[op];
2477 switch (guess_suffix)
2478 {
2479 case QWORD_MNEM_SUFFIX:
2480 mask = Imm64 | Imm32S;
2481 break;
2482 case LONG_MNEM_SUFFIX:
2483 mask = Imm32;
2484 break;
2485 case WORD_MNEM_SUFFIX:
2486 mask = Imm16;
2487 break;
2488 case BYTE_MNEM_SUFFIX:
2489 mask = Imm8;
2490 break;
2491 default:
2492 mask = 0;
2493 break;
2494 }
2495 if (mask & allowed)
2496 i.types[op] &= mask;
2497 }
2498 break;
2499 }
2500 }
2501 }
2502
2503 /* Try to use the smallest displacement type too. */
2504 static void
2505 optimize_disp ()
2506 {
2507 int op;
2508
2509 for (op = i.operands; --op >= 0;)
2510 if (i.types[op] & Disp)
2511 {
2512 if (i.op[op].disps->X_op == O_constant)
2513 {
2514 offsetT disp = i.op[op].disps->X_add_number;
2515
2516 if ((i.types[op] & Disp16)
2517 && (disp & ~(offsetT) 0xffff) == 0)
2518 {
2519 /* If this operand is at most 16 bits, convert
2520 to a signed 16 bit number and don't use 64bit
2521 displacement. */
2522 disp = (((disp & 0xffff) ^ 0x8000) - 0x8000);
2523 i.types[op] &= ~Disp64;
2524 }
2525 if ((i.types[op] & Disp32)
2526 && (disp & ~(((offsetT) 2 << 31) - 1)) == 0)
2527 {
2528 /* If this operand is at most 32 bits, convert
2529 to a signed 32 bit number and don't use 64bit
2530 displacement. */
2531 disp &= (((offsetT) 2 << 31) - 1);
2532 disp = (disp ^ ((offsetT) 1 << 31)) - ((addressT) 1 << 31);
2533 i.types[op] &= ~Disp64;
2534 }
2535 if (!disp && (i.types[op] & BaseIndex))
2536 {
2537 i.types[op] &= ~Disp;
2538 i.op[op].disps = 0;
2539 i.disp_operands--;
2540 }
2541 else if (flag_code == CODE_64BIT)
2542 {
2543 if (fits_in_signed_long (disp))
2544 {
2545 i.types[op] &= ~Disp64;
2546 i.types[op] |= Disp32S;
2547 }
2548 if (fits_in_unsigned_long (disp))
2549 i.types[op] |= Disp32;
2550 }
2551 if ((i.types[op] & (Disp32 | Disp32S | Disp16))
2552 && fits_in_signed_byte (disp))
2553 i.types[op] |= Disp8;
2554 }
2555 else if (i.reloc[op] == BFD_RELOC_386_TLS_DESC_CALL
2556 || i.reloc[op] == BFD_RELOC_X86_64_TLSDESC_CALL)
2557 {
2558 fix_new_exp (frag_now, frag_more (0) - frag_now->fr_literal, 0,
2559 i.op[op].disps, 0, i.reloc[op]);
2560 i.types[op] &= ~Disp;
2561 }
2562 else
2563 /* We only support 64bit displacement on constants. */
2564 i.types[op] &= ~Disp64;
2565 }
2566 }
2567
2568 static int
2569 match_template ()
2570 {
2571 /* Points to template once we've found it. */
2572 const template *t;
2573 unsigned int overlap0, overlap1, overlap2, overlap3;
2574 unsigned int found_reverse_match;
2575 int suffix_check;
2576 unsigned int operand_types [MAX_OPERANDS];
2577 int addr_prefix_disp;
2578 unsigned int j;
2579
2580 #if MAX_OPERANDS != 4
2581 # error "MAX_OPERANDS must be 4."
2582 #endif
2583
2584 #define MATCH(overlap, given, template) \
2585 ((overlap & ~JumpAbsolute) \
2586 && (((given) & (BaseIndex | JumpAbsolute)) \
2587 == ((overlap) & (BaseIndex | JumpAbsolute))))
2588
2589 /* If given types r0 and r1 are registers they must be of the same type
2590 unless the expected operand type register overlap is null.
2591 Note that Acc in a template matches every size of reg. */
2592 #define CONSISTENT_REGISTER_MATCH(m0, g0, t0, m1, g1, t1) \
2593 (((g0) & Reg) == 0 || ((g1) & Reg) == 0 \
2594 || ((g0) & Reg) == ((g1) & Reg) \
2595 || ((((m0) & Acc) ? Reg : (t0)) & (((m1) & Acc) ? Reg : (t1)) & Reg) == 0 )
2596
2597 overlap0 = 0;
2598 overlap1 = 0;
2599 overlap2 = 0;
2600 overlap3 = 0;
2601 found_reverse_match = 0;
2602 for (j = 0; j < MAX_OPERANDS; j++)
2603 operand_types [j] = 0;
2604 addr_prefix_disp = -1;
2605 suffix_check = (i.suffix == BYTE_MNEM_SUFFIX
2606 ? No_bSuf
2607 : (i.suffix == WORD_MNEM_SUFFIX
2608 ? No_wSuf
2609 : (i.suffix == SHORT_MNEM_SUFFIX
2610 ? No_sSuf
2611 : (i.suffix == LONG_MNEM_SUFFIX
2612 ? No_lSuf
2613 : (i.suffix == QWORD_MNEM_SUFFIX
2614 ? No_qSuf
2615 : (i.suffix == LONG_DOUBLE_MNEM_SUFFIX
2616 ? No_xSuf : 0))))));
2617
2618 for (t = current_templates->start; t < current_templates->end; t++)
2619 {
2620 addr_prefix_disp = -1;
2621
2622 /* Must have right number of operands. */
2623 if (i.operands != t->operands)
2624 continue;
2625
2626 /* Check the suffix, except for some instructions in intel mode. */
2627 if ((t->opcode_modifier & suffix_check)
2628 && !(intel_syntax
2629 && (t->opcode_modifier & IgnoreSize)))
2630 continue;
2631
2632 for (j = 0; j < MAX_OPERANDS; j++)
2633 operand_types [j] = t->operand_types [j];
2634
2635 /* In general, don't allow 64-bit operands in 32-bit mode. */
2636 if (i.suffix == QWORD_MNEM_SUFFIX
2637 && flag_code != CODE_64BIT
2638 && (intel_syntax
2639 ? (!(t->opcode_modifier & IgnoreSize)
2640 && !intel_float_operand (t->name))
2641 : intel_float_operand (t->name) != 2)
2642 && (!(operand_types[0] & (RegMMX | RegXMM))
2643 || !(operand_types[t->operands > 1] & (RegMMX | RegXMM)))
2644 && (t->base_opcode != 0x0fc7
2645 || t->extension_opcode != 1 /* cmpxchg8b */))
2646 continue;
2647
2648 /* Do not verify operands when there are none. */
2649 else if (!t->operands)
2650 {
2651 if (t->cpu_flags & ~cpu_arch_flags)
2652 continue;
2653 /* We've found a match; break out of loop. */
2654 break;
2655 }
2656
2657 /* Address size prefix will turn Disp64/Disp32/Disp16 operand
2658 into Disp32/Disp16/Disp32 operand. */
2659 if (i.prefix[ADDR_PREFIX] != 0)
2660 {
2661 unsigned int DispOn = 0, DispOff = 0;
2662
2663 switch (flag_code)
2664 {
2665 case CODE_16BIT:
2666 DispOn = Disp32;
2667 DispOff = Disp16;
2668 break;
2669 case CODE_32BIT:
2670 DispOn = Disp16;
2671 DispOff = Disp32;
2672 break;
2673 case CODE_64BIT:
2674 DispOn = Disp32;
2675 DispOff = Disp64;
2676 break;
2677 }
2678
2679 for (j = 0; j < MAX_OPERANDS; j++)
2680 {
2681 /* There should be only one Disp operand. */
2682 if ((operand_types[j] & DispOff))
2683 {
2684 addr_prefix_disp = j;
2685 operand_types[j] |= DispOn;
2686 operand_types[j] &= ~DispOff;
2687 break;
2688 }
2689 }
2690 }
2691
2692 overlap0 = i.types[0] & operand_types[0];
2693 switch (t->operands)
2694 {
2695 case 1:
2696 if (!MATCH (overlap0, i.types[0], operand_types[0]))
2697 continue;
2698 break;
2699 case 2:
2700 case 3:
2701 case 4:
2702 overlap1 = i.types[1] & operand_types[1];
2703 if (!MATCH (overlap0, i.types[0], operand_types[0])
2704 || !MATCH (overlap1, i.types[1], operand_types[1])
2705 /* monitor in SSE3 is a very special case. The first
2706 register and the second register may have different
2707 sizes. */
2708 || !((t->base_opcode == 0x0f01
2709 && t->extension_opcode == 0xc8)
2710 || CONSISTENT_REGISTER_MATCH (overlap0, i.types[0],
2711 operand_types[0],
2712 overlap1, i.types[1],
2713 operand_types[1])))
2714 {
2715 /* Check if other direction is valid ... */
2716 if ((t->opcode_modifier & (D | FloatD)) == 0)
2717 continue;
2718
2719 /* Try reversing direction of operands. */
2720 overlap0 = i.types[0] & operand_types[1];
2721 overlap1 = i.types[1] & operand_types[0];
2722 if (!MATCH (overlap0, i.types[0], operand_types[1])
2723 || !MATCH (overlap1, i.types[1], operand_types[0])
2724 || !CONSISTENT_REGISTER_MATCH (overlap0, i.types[0],
2725 operand_types[1],
2726 overlap1, i.types[1],
2727 operand_types[0]))
2728 {
2729 /* Does not match either direction. */
2730 continue;
2731 }
2732 /* found_reverse_match holds which of D or FloatDR
2733 we've found. */
2734 found_reverse_match = t->opcode_modifier & (D | FloatDR);
2735 }
2736 else
2737 {
2738 /* Found a forward 2 operand match here. */
2739 switch (t->operands)
2740 {
2741 case 4:
2742 overlap3 = i.types[3] & operand_types[3];
2743 case 3:
2744 overlap2 = i.types[2] & operand_types[2];
2745 break;
2746 }
2747
2748 switch (t->operands)
2749 {
2750 case 4:
2751 if (!MATCH (overlap3, i.types[3], operand_types[3])
2752 || !CONSISTENT_REGISTER_MATCH (overlap2,
2753 i.types[2],
2754 operand_types[2],
2755 overlap3,
2756 i.types[3],
2757 operand_types[3]))
2758 continue;
2759 case 3:
2760 /* Here we make use of the fact that there are no
2761 reverse match 3 operand instructions, and all 3
2762 operand instructions only need to be checked for
2763 register consistency between operands 2 and 3. */
2764 if (!MATCH (overlap2, i.types[2], operand_types[2])
2765 || !CONSISTENT_REGISTER_MATCH (overlap1,
2766 i.types[1],
2767 operand_types[1],
2768 overlap2,
2769 i.types[2],
2770 operand_types[2]))
2771 continue;
2772 break;
2773 }
2774 }
2775 /* Found either forward/reverse 2, 3 or 4 operand match here:
2776 slip through to break. */
2777 }
2778 if (t->cpu_flags & ~cpu_arch_flags)
2779 {
2780 found_reverse_match = 0;
2781 continue;
2782 }
2783 /* We've found a match; break out of loop. */
2784 break;
2785 }
2786
2787 if (t == current_templates->end)
2788 {
2789 /* We found no match. */
2790 as_bad (_("suffix or operands invalid for `%s'"),
2791 current_templates->start->name);
2792 return 0;
2793 }
2794
2795 if (!quiet_warnings)
2796 {
2797 if (!intel_syntax
2798 && ((i.types[0] & JumpAbsolute)
2799 != (operand_types[0] & JumpAbsolute)))
2800 {
2801 as_warn (_("indirect %s without `*'"), t->name);
2802 }
2803
2804 if ((t->opcode_modifier & (IsPrefix | IgnoreSize))
2805 == (IsPrefix | IgnoreSize))
2806 {
2807 /* Warn them that a data or address size prefix doesn't
2808 affect assembly of the next line of code. */
2809 as_warn (_("stand-alone `%s' prefix"), t->name);
2810 }
2811 }
2812
2813 /* Copy the template we found. */
2814 i.tm = *t;
2815
2816 if (addr_prefix_disp != -1)
2817 i.tm.operand_types[addr_prefix_disp]
2818 = operand_types[addr_prefix_disp];
2819
2820 if (found_reverse_match)
2821 {
2822 /* If we found a reverse match we must alter the opcode
2823 direction bit. found_reverse_match holds bits to change
2824 (different for int & float insns). */
2825
2826 i.tm.base_opcode ^= found_reverse_match;
2827
2828 i.tm.operand_types[0] = operand_types[1];
2829 i.tm.operand_types[1] = operand_types[0];
2830 }
2831
2832 return 1;
2833 }
2834
2835 static int
2836 check_string ()
2837 {
2838 int mem_op = (i.types[0] & AnyMem) ? 0 : 1;
2839 if ((i.tm.operand_types[mem_op] & EsSeg) != 0)
2840 {
2841 if (i.seg[0] != NULL && i.seg[0] != &es)
2842 {
2843 as_bad (_("`%s' operand %d must use `%%es' segment"),
2844 i.tm.name,
2845 mem_op + 1);
2846 return 0;
2847 }
2848 /* There's only ever one segment override allowed per instruction.
2849 This instruction possibly has a legal segment override on the
2850 second operand, so copy the segment to where non-string
2851 instructions store it, allowing common code. */
2852 i.seg[0] = i.seg[1];
2853 }
2854 else if ((i.tm.operand_types[mem_op + 1] & EsSeg) != 0)
2855 {
2856 if (i.seg[1] != NULL && i.seg[1] != &es)
2857 {
2858 as_bad (_("`%s' operand %d must use `%%es' segment"),
2859 i.tm.name,
2860 mem_op + 2);
2861 return 0;
2862 }
2863 }
2864 return 1;
2865 }
2866
2867 static int
2868 process_suffix (void)
2869 {
2870 /* If matched instruction specifies an explicit instruction mnemonic
2871 suffix, use it. */
2872 if (i.tm.opcode_modifier & (Size16 | Size32 | Size64))
2873 {
2874 if (i.tm.opcode_modifier & Size16)
2875 i.suffix = WORD_MNEM_SUFFIX;
2876 else if (i.tm.opcode_modifier & Size64)
2877 i.suffix = QWORD_MNEM_SUFFIX;
2878 else
2879 i.suffix = LONG_MNEM_SUFFIX;
2880 }
2881 else if (i.reg_operands)
2882 {
2883 /* If there's no instruction mnemonic suffix we try to invent one
2884 based on register operands. */
2885 if (!i.suffix)
2886 {
2887 /* We take i.suffix from the last register operand specified,
2888 Destination register type is more significant than source
2889 register type. */
2890 int op;
2891
2892 for (op = i.operands; --op >= 0;)
2893 if ((i.types[op] & Reg)
2894 && !(i.tm.operand_types[op] & InOutPortReg))
2895 {
2896 i.suffix = ((i.types[op] & Reg8) ? BYTE_MNEM_SUFFIX :
2897 (i.types[op] & Reg16) ? WORD_MNEM_SUFFIX :
2898 (i.types[op] & Reg64) ? QWORD_MNEM_SUFFIX :
2899 LONG_MNEM_SUFFIX);
2900 break;
2901 }
2902 }
2903 else if (i.suffix == BYTE_MNEM_SUFFIX)
2904 {
2905 if (!check_byte_reg ())
2906 return 0;
2907 }
2908 else if (i.suffix == LONG_MNEM_SUFFIX)
2909 {
2910 if (!check_long_reg ())
2911 return 0;
2912 }
2913 else if (i.suffix == QWORD_MNEM_SUFFIX)
2914 {
2915 if (!check_qword_reg ())
2916 return 0;
2917 }
2918 else if (i.suffix == WORD_MNEM_SUFFIX)
2919 {
2920 if (!check_word_reg ())
2921 return 0;
2922 }
2923 else if (intel_syntax && (i.tm.opcode_modifier & IgnoreSize))
2924 /* Do nothing if the instruction is going to ignore the prefix. */
2925 ;
2926 else
2927 abort ();
2928 }
2929 else if ((i.tm.opcode_modifier & DefaultSize)
2930 && !i.suffix
2931 /* exclude fldenv/frstor/fsave/fstenv */
2932 && (i.tm.opcode_modifier & No_sSuf))
2933 {
2934 i.suffix = stackop_size;
2935 }
2936 else if (intel_syntax
2937 && !i.suffix
2938 && ((i.tm.operand_types[0] & JumpAbsolute)
2939 || (i.tm.opcode_modifier & (JumpByte|JumpInterSegment))
2940 || (i.tm.base_opcode == 0x0f01 /* [ls][gi]dt */
2941 && i.tm.extension_opcode <= 3)))
2942 {
2943 switch (flag_code)
2944 {
2945 case CODE_64BIT:
2946 if (!(i.tm.opcode_modifier & No_qSuf))
2947 {
2948 i.suffix = QWORD_MNEM_SUFFIX;
2949 break;
2950 }
2951 case CODE_32BIT:
2952 if (!(i.tm.opcode_modifier & No_lSuf))
2953 i.suffix = LONG_MNEM_SUFFIX;
2954 break;
2955 case CODE_16BIT:
2956 if (!(i.tm.opcode_modifier & No_wSuf))
2957 i.suffix = WORD_MNEM_SUFFIX;
2958 break;
2959 }
2960 }
2961
2962 if (!i.suffix)
2963 {
2964 if (!intel_syntax)
2965 {
2966 if (i.tm.opcode_modifier & W)
2967 {
2968 as_bad (_("no instruction mnemonic suffix given and no register operands; can't size instruction"));
2969 return 0;
2970 }
2971 }
2972 else
2973 {
2974 unsigned int suffixes = (~i.tm.opcode_modifier
2975 & (No_bSuf
2976 | No_wSuf
2977 | No_lSuf
2978 | No_sSuf
2979 | No_xSuf
2980 | No_qSuf));
2981
2982 if ((i.tm.opcode_modifier & W)
2983 || ((suffixes & (suffixes - 1))
2984 && !(i.tm.opcode_modifier & (DefaultSize | IgnoreSize))))
2985 {
2986 as_bad (_("ambiguous operand size for `%s'"), i.tm.name);
2987 return 0;
2988 }
2989 }
2990 }
2991
2992 /* Change the opcode based on the operand size given by i.suffix;
2993 We don't need to change things for byte insns. */
2994
2995 if (i.suffix && i.suffix != BYTE_MNEM_SUFFIX)
2996 {
2997 /* It's not a byte, select word/dword operation. */
2998 if (i.tm.opcode_modifier & W)
2999 {
3000 if (i.tm.opcode_modifier & ShortForm)
3001 i.tm.base_opcode |= 8;
3002 else
3003 i.tm.base_opcode |= 1;
3004 }
3005
3006 /* Now select between word & dword operations via the operand
3007 size prefix, except for instructions that will ignore this
3008 prefix anyway. */
3009 if (i.tm.base_opcode == 0x0f01 && i.tm.extension_opcode == 0xc8)
3010 {
3011 /* monitor in SSE3 is a very special case. The default size
3012 of AX is the size of mode. The address size override
3013 prefix will change the size of AX. */
3014 if (i.op->regs[0].reg_type &
3015 (flag_code == CODE_32BIT ? Reg16 : Reg32))
3016 if (!add_prefix (ADDR_PREFIX_OPCODE))
3017 return 0;
3018 }
3019 else if (i.suffix != QWORD_MNEM_SUFFIX
3020 && i.suffix != LONG_DOUBLE_MNEM_SUFFIX
3021 && !(i.tm.opcode_modifier & (IgnoreSize | FloatMF))
3022 && ((i.suffix == LONG_MNEM_SUFFIX) == (flag_code == CODE_16BIT)
3023 || (flag_code == CODE_64BIT
3024 && (i.tm.opcode_modifier & JumpByte))))
3025 {
3026 unsigned int prefix = DATA_PREFIX_OPCODE;
3027
3028 if (i.tm.opcode_modifier & JumpByte) /* jcxz, loop */
3029 prefix = ADDR_PREFIX_OPCODE;
3030
3031 if (!add_prefix (prefix))
3032 return 0;
3033 }
3034
3035 /* Set mode64 for an operand. */
3036 if (i.suffix == QWORD_MNEM_SUFFIX
3037 && flag_code == CODE_64BIT
3038 && (i.tm.opcode_modifier & NoRex64) == 0)
3039 {
3040 /* Special case for xchg %rax,%rax. It is NOP and doesn't
3041 need rex64. */
3042 if (i.operands != 2
3043 || i.types [0] != (Acc | Reg64)
3044 || i.types [1] != (Acc | Reg64)
3045 || strcmp (i.tm.name, "xchg") != 0)
3046 i.rex |= REX_MODE64;
3047 }
3048
3049 /* Size floating point instruction. */
3050 if (i.suffix == LONG_MNEM_SUFFIX)
3051 if (i.tm.opcode_modifier & FloatMF)
3052 i.tm.base_opcode ^= 4;
3053 }
3054
3055 return 1;
3056 }
3057
3058 static int
3059 check_byte_reg (void)
3060 {
3061 int op;
3062
3063 for (op = i.operands; --op >= 0;)
3064 {
3065 /* If this is an eight bit register, it's OK. If it's the 16 or
3066 32 bit version of an eight bit register, we will just use the
3067 low portion, and that's OK too. */
3068 if (i.types[op] & Reg8)
3069 continue;
3070
3071 /* movzx and movsx should not generate this warning. */
3072 if (intel_syntax
3073 && (i.tm.base_opcode == 0xfb7
3074 || i.tm.base_opcode == 0xfb6
3075 || i.tm.base_opcode == 0x63
3076 || i.tm.base_opcode == 0xfbe
3077 || i.tm.base_opcode == 0xfbf))
3078 continue;
3079
3080 if ((i.types[op] & WordReg) && i.op[op].regs->reg_num < 4)
3081 {
3082 /* Prohibit these changes in the 64bit mode, since the
3083 lowering is more complicated. */
3084 if (flag_code == CODE_64BIT
3085 && (i.tm.operand_types[op] & InOutPortReg) == 0)
3086 {
3087 as_bad (_("Incorrect register `%%%s' used with `%c' suffix"),
3088 i.op[op].regs->reg_name,
3089 i.suffix);
3090 return 0;
3091 }
3092 #if REGISTER_WARNINGS
3093 if (!quiet_warnings
3094 && (i.tm.operand_types[op] & InOutPortReg) == 0)
3095 as_warn (_("using `%%%s' instead of `%%%s' due to `%c' suffix"),
3096 (i.op[op].regs + (i.types[op] & Reg16
3097 ? REGNAM_AL - REGNAM_AX
3098 : REGNAM_AL - REGNAM_EAX))->reg_name,
3099 i.op[op].regs->reg_name,
3100 i.suffix);
3101 #endif
3102 continue;
3103 }
3104 /* Any other register is bad. */
3105 if (i.types[op] & (Reg | RegMMX | RegXMM
3106 | SReg2 | SReg3
3107 | Control | Debug | Test
3108 | FloatReg | FloatAcc))
3109 {
3110 as_bad (_("`%%%s' not allowed with `%s%c'"),
3111 i.op[op].regs->reg_name,
3112 i.tm.name,
3113 i.suffix);
3114 return 0;
3115 }
3116 }
3117 return 1;
3118 }
3119
3120 static int
3121 check_long_reg ()
3122 {
3123 int op;
3124
3125 for (op = i.operands; --op >= 0;)
3126 /* Reject eight bit registers, except where the template requires
3127 them. (eg. movzb) */
3128 if ((i.types[op] & Reg8) != 0
3129 && (i.tm.operand_types[op] & (Reg16 | Reg32 | Acc)) != 0)
3130 {
3131 as_bad (_("`%%%s' not allowed with `%s%c'"),
3132 i.op[op].regs->reg_name,
3133 i.tm.name,
3134 i.suffix);
3135 return 0;
3136 }
3137 /* Warn if the e prefix on a general reg is missing. */
3138 else if ((!quiet_warnings || flag_code == CODE_64BIT)
3139 && (i.types[op] & Reg16) != 0
3140 && (i.tm.operand_types[op] & (Reg32 | Acc)) != 0)
3141 {
3142 /* Prohibit these changes in the 64bit mode, since the
3143 lowering is more complicated. */
3144 if (flag_code == CODE_64BIT)
3145 {
3146 as_bad (_("Incorrect register `%%%s' used with `%c' suffix"),
3147 i.op[op].regs->reg_name,
3148 i.suffix);
3149 return 0;
3150 }
3151 #if REGISTER_WARNINGS
3152 else
3153 as_warn (_("using `%%%s' instead of `%%%s' due to `%c' suffix"),
3154 (i.op[op].regs + REGNAM_EAX - REGNAM_AX)->reg_name,
3155 i.op[op].regs->reg_name,
3156 i.suffix);
3157 #endif
3158 }
3159 /* Warn if the r prefix on a general reg is missing. */
3160 else if ((i.types[op] & Reg64) != 0
3161 && (i.tm.operand_types[op] & (Reg32 | Acc)) != 0)
3162 {
3163 as_bad (_("Incorrect register `%%%s' used with `%c' suffix"),
3164 i.op[op].regs->reg_name,
3165 i.suffix);
3166 return 0;
3167 }
3168 return 1;
3169 }
3170
3171 static int
3172 check_qword_reg ()
3173 {
3174 int op;
3175
3176 for (op = i.operands; --op >= 0; )
3177 /* Reject eight bit registers, except where the template requires
3178 them. (eg. movzb) */
3179 if ((i.types[op] & Reg8) != 0
3180 && (i.tm.operand_types[op] & (Reg16 | Reg32 | Acc)) != 0)
3181 {
3182 as_bad (_("`%%%s' not allowed with `%s%c'"),
3183 i.op[op].regs->reg_name,
3184 i.tm.name,
3185 i.suffix);
3186 return 0;
3187 }
3188 /* Warn if the e prefix on a general reg is missing. */
3189 else if (((i.types[op] & Reg16) != 0
3190 || (i.types[op] & Reg32) != 0)
3191 && (i.tm.operand_types[op] & (Reg32 | Acc)) != 0)
3192 {
3193 /* Prohibit these changes in the 64bit mode, since the
3194 lowering is more complicated. */
3195 as_bad (_("Incorrect register `%%%s' used with `%c' suffix"),
3196 i.op[op].regs->reg_name,
3197 i.suffix);
3198 return 0;
3199 }
3200 return 1;
3201 }
3202
3203 static int
3204 check_word_reg ()
3205 {
3206 int op;
3207 for (op = i.operands; --op >= 0;)
3208 /* Reject eight bit registers, except where the template requires
3209 them. (eg. movzb) */
3210 if ((i.types[op] & Reg8) != 0
3211 && (i.tm.operand_types[op] & (Reg16 | Reg32 | Acc)) != 0)
3212 {
3213 as_bad (_("`%%%s' not allowed with `%s%c'"),
3214 i.op[op].regs->reg_name,
3215 i.tm.name,
3216 i.suffix);
3217 return 0;
3218 }
3219 /* Warn if the e prefix on a general reg is present. */
3220 else if ((!quiet_warnings || flag_code == CODE_64BIT)
3221 && (i.types[op] & Reg32) != 0
3222 && (i.tm.operand_types[op] & (Reg16 | Acc)) != 0)
3223 {
3224 /* Prohibit these changes in the 64bit mode, since the
3225 lowering is more complicated. */
3226 if (flag_code == CODE_64BIT)
3227 {
3228 as_bad (_("Incorrect register `%%%s' used with `%c' suffix"),
3229 i.op[op].regs->reg_name,
3230 i.suffix);
3231 return 0;
3232 }
3233 else
3234 #if REGISTER_WARNINGS
3235 as_warn (_("using `%%%s' instead of `%%%s' due to `%c' suffix"),
3236 (i.op[op].regs + REGNAM_AX - REGNAM_EAX)->reg_name,
3237 i.op[op].regs->reg_name,
3238 i.suffix);
3239 #endif
3240 }
3241 return 1;
3242 }
3243
3244 static int
3245 finalize_imm ()
3246 {
3247 unsigned int overlap0, overlap1, overlap2;
3248
3249 overlap0 = i.types[0] & i.tm.operand_types[0];
3250 if ((overlap0 & (Imm8 | Imm8S | Imm16 | Imm32 | Imm32S | Imm64))
3251 && overlap0 != Imm8 && overlap0 != Imm8S
3252 && overlap0 != Imm16 && overlap0 != Imm32S
3253 && overlap0 != Imm32 && overlap0 != Imm64)
3254 {
3255 if (i.suffix)
3256 {
3257 overlap0 &= (i.suffix == BYTE_MNEM_SUFFIX
3258 ? Imm8 | Imm8S
3259 : (i.suffix == WORD_MNEM_SUFFIX
3260 ? Imm16
3261 : (i.suffix == QWORD_MNEM_SUFFIX
3262 ? Imm64 | Imm32S
3263 : Imm32)));
3264 }
3265 else if (overlap0 == (Imm16 | Imm32S | Imm32)
3266 || overlap0 == (Imm16 | Imm32)
3267 || overlap0 == (Imm16 | Imm32S))
3268 {
3269 overlap0 = ((flag_code == CODE_16BIT) ^ (i.prefix[DATA_PREFIX] != 0)
3270 ? Imm16 : Imm32S);
3271 }
3272 if (overlap0 != Imm8 && overlap0 != Imm8S
3273 && overlap0 != Imm16 && overlap0 != Imm32S
3274 && overlap0 != Imm32 && overlap0 != Imm64)
3275 {
3276 as_bad (_("no instruction mnemonic suffix given; can't determine immediate size"));
3277 return 0;
3278 }
3279 }
3280 i.types[0] = overlap0;
3281
3282 overlap1 = i.types[1] & i.tm.operand_types[1];
3283 if ((overlap1 & (Imm8 | Imm8S | Imm16 | Imm32S | Imm32 | Imm64))
3284 && overlap1 != Imm8 && overlap1 != Imm8S
3285 && overlap1 != Imm16 && overlap1 != Imm32S
3286 && overlap1 != Imm32 && overlap1 != Imm64)
3287 {
3288 if (i.suffix)
3289 {
3290 overlap1 &= (i.suffix == BYTE_MNEM_SUFFIX
3291 ? Imm8 | Imm8S
3292 : (i.suffix == WORD_MNEM_SUFFIX
3293 ? Imm16
3294 : (i.suffix == QWORD_MNEM_SUFFIX
3295 ? Imm64 | Imm32S
3296 : Imm32)));
3297 }
3298 else if (overlap1 == (Imm16 | Imm32 | Imm32S)
3299 || overlap1 == (Imm16 | Imm32)
3300 || overlap1 == (Imm16 | Imm32S))
3301 {
3302 overlap1 = ((flag_code == CODE_16BIT) ^ (i.prefix[DATA_PREFIX] != 0)
3303 ? Imm16 : Imm32S);
3304 }
3305 if (overlap1 != Imm8 && overlap1 != Imm8S
3306 && overlap1 != Imm16 && overlap1 != Imm32S
3307 && overlap1 != Imm32 && overlap1 != Imm64)
3308 {
3309 as_bad (_("no instruction mnemonic suffix given; can't determine immediate size %x %c"),overlap1, i.suffix);
3310 return 0;
3311 }
3312 }
3313 i.types[1] = overlap1;
3314
3315 overlap2 = i.types[2] & i.tm.operand_types[2];
3316 assert ((overlap2 & Imm) == 0);
3317 i.types[2] = overlap2;
3318
3319 return 1;
3320 }
3321
3322 static int
3323 process_operands ()
3324 {
3325 /* Default segment register this instruction will use for memory
3326 accesses. 0 means unknown. This is only for optimizing out
3327 unnecessary segment overrides. */
3328 const seg_entry *default_seg = 0;
3329
3330 /* The imul $imm, %reg instruction is converted into
3331 imul $imm, %reg, %reg, and the clr %reg instruction
3332 is converted into xor %reg, %reg. */
3333 if (i.tm.opcode_modifier & regKludge)
3334 {
3335 unsigned int first_reg_op = (i.types[0] & Reg) ? 0 : 1;
3336 /* Pretend we saw the extra register operand. */
3337 assert (i.op[first_reg_op + 1].regs == 0);
3338 i.op[first_reg_op + 1].regs = i.op[first_reg_op].regs;
3339 i.types[first_reg_op + 1] = i.types[first_reg_op];
3340 i.reg_operands = 2;
3341 }
3342
3343 if (i.tm.opcode_modifier & ShortForm)
3344 {
3345 /* The register or float register operand is in operand 0 or 1. */
3346 unsigned int op = (i.types[0] & (Reg | FloatReg)) ? 0 : 1;
3347 /* Register goes in low 3 bits of opcode. */
3348 i.tm.base_opcode |= i.op[op].regs->reg_num;
3349 if ((i.op[op].regs->reg_flags & RegRex) != 0)
3350 i.rex |= REX_EXTZ;
3351 if (!quiet_warnings && (i.tm.opcode_modifier & Ugh) != 0)
3352 {
3353 /* Warn about some common errors, but press on regardless.
3354 The first case can be generated by gcc (<= 2.8.1). */
3355 if (i.operands == 2)
3356 {
3357 /* Reversed arguments on faddp, fsubp, etc. */
3358 as_warn (_("translating to `%s %%%s,%%%s'"), i.tm.name,
3359 i.op[1].regs->reg_name,
3360 i.op[0].regs->reg_name);
3361 }
3362 else
3363 {
3364 /* Extraneous `l' suffix on fp insn. */
3365 as_warn (_("translating to `%s %%%s'"), i.tm.name,
3366 i.op[0].regs->reg_name);
3367 }
3368 }
3369 }
3370 else if (i.tm.opcode_modifier & Modrm)
3371 {
3372 /* The opcode is completed (modulo i.tm.extension_opcode which
3373 must be put into the modrm byte). Now, we make the modrm and
3374 index base bytes based on all the info we've collected. */
3375
3376 default_seg = build_modrm_byte ();
3377 }
3378 else if (i.tm.opcode_modifier & (Seg2ShortForm | Seg3ShortForm))
3379 {
3380 if (i.tm.base_opcode == POP_SEG_SHORT
3381 && i.op[0].regs->reg_num == 1)
3382 {
3383 as_bad (_("you can't `pop %%cs'"));
3384 return 0;
3385 }
3386 i.tm.base_opcode |= (i.op[0].regs->reg_num << 3);
3387 if ((i.op[0].regs->reg_flags & RegRex) != 0)
3388 i.rex |= REX_EXTZ;
3389 }
3390 else if ((i.tm.base_opcode & ~(D | W)) == MOV_AX_DISP32)
3391 {
3392 default_seg = &ds;
3393 }
3394 else if ((i.tm.opcode_modifier & IsString) != 0)
3395 {
3396 /* For the string instructions that allow a segment override
3397 on one of their operands, the default segment is ds. */
3398 default_seg = &ds;
3399 }
3400
3401 if ((i.tm.base_opcode == 0x8d /* lea */
3402 || (i.tm.cpu_flags & CpuSVME))
3403 && i.seg[0] && !quiet_warnings)
3404 as_warn (_("segment override on `%s' is ineffectual"), i.tm.name);
3405
3406 /* If a segment was explicitly specified, and the specified segment
3407 is not the default, use an opcode prefix to select it. If we
3408 never figured out what the default segment is, then default_seg
3409 will be zero at this point, and the specified segment prefix will
3410 always be used. */
3411 if ((i.seg[0]) && (i.seg[0] != default_seg))
3412 {
3413 if (!add_prefix (i.seg[0]->seg_prefix))
3414 return 0;
3415 }
3416 return 1;
3417 }
3418
3419 static const seg_entry *
3420 build_modrm_byte ()
3421 {
3422 const seg_entry *default_seg = 0;
3423
3424 /* i.reg_operands MUST be the number of real register operands;
3425 implicit registers do not count. */
3426 if (i.reg_operands == 2)
3427 {
3428 unsigned int source, dest;
3429 source = ((i.types[0]
3430 & (Reg | RegMMX | RegXMM
3431 | SReg2 | SReg3
3432 | Control | Debug | Test))
3433 ? 0 : 1);
3434
3435 /* In 4 operands instructions with 2 immediate operands, the first
3436 two are immediate bytes and hence source operand will be in the
3437 next byte after the immediates */
3438 if ((i.operands == 4)&&(i.imm_operands=2)) source++;
3439 dest = source + 1;
3440
3441 i.rm.mode = 3;
3442 /* One of the register operands will be encoded in the i.tm.reg
3443 field, the other in the combined i.tm.mode and i.tm.regmem
3444 fields. If no form of this instruction supports a memory
3445 destination operand, then we assume the source operand may
3446 sometimes be a memory operand and so we need to store the
3447 destination in the i.rm.reg field. */
3448 if ((i.tm.operand_types[dest] & AnyMem) == 0)
3449 {
3450 i.rm.reg = i.op[dest].regs->reg_num;
3451 i.rm.regmem = i.op[source].regs->reg_num;
3452 if ((i.op[dest].regs->reg_flags & RegRex) != 0)
3453 i.rex |= REX_EXTX;
3454 if ((i.op[source].regs->reg_flags & RegRex) != 0)
3455 i.rex |= REX_EXTZ;
3456 }
3457 else
3458 {
3459 i.rm.reg = i.op[source].regs->reg_num;
3460 i.rm.regmem = i.op[dest].regs->reg_num;
3461 if ((i.op[dest].regs->reg_flags & RegRex) != 0)
3462 i.rex |= REX_EXTZ;
3463 if ((i.op[source].regs->reg_flags & RegRex) != 0)
3464 i.rex |= REX_EXTX;
3465 }
3466 if (flag_code != CODE_64BIT && (i.rex & (REX_EXTX | REX_EXTZ)))
3467 {
3468 if (!((i.types[0] | i.types[1]) & Control))
3469 abort ();
3470 i.rex &= ~(REX_EXTX | REX_EXTZ);
3471 add_prefix (LOCK_PREFIX_OPCODE);
3472 }
3473 }
3474 else
3475 { /* If it's not 2 reg operands... */
3476 if (i.mem_operands)
3477 {
3478 unsigned int fake_zero_displacement = 0;
3479 unsigned int op = ((i.types[0] & AnyMem)
3480 ? 0
3481 : (i.types[1] & AnyMem) ? 1 : 2);
3482
3483 default_seg = &ds;
3484
3485 if (i.base_reg == 0)
3486 {
3487 i.rm.mode = 0;
3488 if (!i.disp_operands)
3489 fake_zero_displacement = 1;
3490 if (i.index_reg == 0)
3491 {
3492 /* Operand is just <disp> */
3493 if (flag_code == CODE_64BIT)
3494 {
3495 /* 64bit mode overwrites the 32bit absolute
3496 addressing by RIP relative addressing and
3497 absolute addressing is encoded by one of the
3498 redundant SIB forms. */
3499 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3500 i.sib.base = NO_BASE_REGISTER;
3501 i.sib.index = NO_INDEX_REGISTER;
3502 i.types[op] = ((i.prefix[ADDR_PREFIX] == 0)
3503 ? Disp32S : Disp32);
3504 }
3505 else if ((flag_code == CODE_16BIT)
3506 ^ (i.prefix[ADDR_PREFIX] != 0))
3507 {
3508 i.rm.regmem = NO_BASE_REGISTER_16;
3509 i.types[op] = Disp16;
3510 }
3511 else
3512 {
3513 i.rm.regmem = NO_BASE_REGISTER;
3514 i.types[op] = Disp32;
3515 }
3516 }
3517 else /* !i.base_reg && i.index_reg */
3518 {
3519 i.sib.index = i.index_reg->reg_num;
3520 i.sib.base = NO_BASE_REGISTER;
3521 i.sib.scale = i.log2_scale_factor;
3522 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3523 i.types[op] &= ~Disp;
3524 if (flag_code != CODE_64BIT)
3525 i.types[op] |= Disp32; /* Must be 32 bit */
3526 else
3527 i.types[op] |= Disp32S;
3528 if ((i.index_reg->reg_flags & RegRex) != 0)
3529 i.rex |= REX_EXTY;
3530 }
3531 }
3532 /* RIP addressing for 64bit mode. */
3533 else if (i.base_reg->reg_type == BaseIndex)
3534 {
3535 i.rm.regmem = NO_BASE_REGISTER;
3536 i.types[op] &= ~ Disp;
3537 i.types[op] |= Disp32S;
3538 i.flags[op] |= Operand_PCrel;
3539 if (! i.disp_operands)
3540 fake_zero_displacement = 1;
3541 }
3542 else if (i.base_reg->reg_type & Reg16)
3543 {
3544 switch (i.base_reg->reg_num)
3545 {
3546 case 3: /* (%bx) */
3547 if (i.index_reg == 0)
3548 i.rm.regmem = 7;
3549 else /* (%bx,%si) -> 0, or (%bx,%di) -> 1 */
3550 i.rm.regmem = i.index_reg->reg_num - 6;
3551 break;
3552 case 5: /* (%bp) */
3553 default_seg = &ss;
3554 if (i.index_reg == 0)
3555 {
3556 i.rm.regmem = 6;
3557 if ((i.types[op] & Disp) == 0)
3558 {
3559 /* fake (%bp) into 0(%bp) */
3560 i.types[op] |= Disp8;
3561 fake_zero_displacement = 1;
3562 }
3563 }
3564 else /* (%bp,%si) -> 2, or (%bp,%di) -> 3 */
3565 i.rm.regmem = i.index_reg->reg_num - 6 + 2;
3566 break;
3567 default: /* (%si) -> 4 or (%di) -> 5 */
3568 i.rm.regmem = i.base_reg->reg_num - 6 + 4;
3569 }
3570 i.rm.mode = mode_from_disp_size (i.types[op]);
3571 }
3572 else /* i.base_reg and 32/64 bit mode */
3573 {
3574 if (flag_code == CODE_64BIT
3575 && (i.types[op] & Disp))
3576 i.types[op] = ((i.types[op] & Disp8)
3577 | (i.prefix[ADDR_PREFIX] == 0
3578 ? Disp32S : Disp32));
3579
3580 i.rm.regmem = i.base_reg->reg_num;
3581 if ((i.base_reg->reg_flags & RegRex) != 0)
3582 i.rex |= REX_EXTZ;
3583 i.sib.base = i.base_reg->reg_num;
3584 /* x86-64 ignores REX prefix bit here to avoid decoder
3585 complications. */
3586 if ((i.base_reg->reg_num & 7) == EBP_REG_NUM)
3587 {
3588 default_seg = &ss;
3589 if (i.disp_operands == 0)
3590 {
3591 fake_zero_displacement = 1;
3592 i.types[op] |= Disp8;
3593 }
3594 }
3595 else if (i.base_reg->reg_num == ESP_REG_NUM)
3596 {
3597 default_seg = &ss;
3598 }
3599 i.sib.scale = i.log2_scale_factor;
3600 if (i.index_reg == 0)
3601 {
3602 /* <disp>(%esp) becomes two byte modrm with no index
3603 register. We've already stored the code for esp
3604 in i.rm.regmem ie. ESCAPE_TO_TWO_BYTE_ADDRESSING.
3605 Any base register besides %esp will not use the
3606 extra modrm byte. */
3607 i.sib.index = NO_INDEX_REGISTER;
3608 #if !SCALE1_WHEN_NO_INDEX
3609 /* Another case where we force the second modrm byte. */
3610 if (i.log2_scale_factor)
3611 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3612 #endif
3613 }
3614 else
3615 {
3616 i.sib.index = i.index_reg->reg_num;
3617 i.rm.regmem = ESCAPE_TO_TWO_BYTE_ADDRESSING;
3618 if ((i.index_reg->reg_flags & RegRex) != 0)
3619 i.rex |= REX_EXTY;
3620 }
3621
3622 if (i.disp_operands
3623 && (i.reloc[op] == BFD_RELOC_386_TLS_DESC_CALL
3624 || i.reloc[op] == BFD_RELOC_X86_64_TLSDESC_CALL))
3625 i.rm.mode = 0;
3626 else
3627 i.rm.mode = mode_from_disp_size (i.types[op]);
3628 }
3629
3630 if (fake_zero_displacement)
3631 {
3632 /* Fakes a zero displacement assuming that i.types[op]
3633 holds the correct displacement size. */
3634 expressionS *exp;
3635
3636 assert (i.op[op].disps == 0);
3637 exp = &disp_expressions[i.disp_operands++];
3638 i.op[op].disps = exp;
3639 exp->X_op = O_constant;
3640 exp->X_add_number = 0;
3641 exp->X_add_symbol = (symbolS *) 0;
3642 exp->X_op_symbol = (symbolS *) 0;
3643 }
3644 }
3645
3646 /* Fill in i.rm.reg or i.rm.regmem field with register operand
3647 (if any) based on i.tm.extension_opcode. Again, we must be
3648 careful to make sure that segment/control/debug/test/MMX
3649 registers are coded into the i.rm.reg field. */
3650 if (i.reg_operands)
3651 {
3652 unsigned int op =
3653 ((i.types[0]
3654 & (Reg | RegMMX | RegXMM
3655 | SReg2 | SReg3
3656 | Control | Debug | Test))
3657 ? 0
3658 : ((i.types[1]
3659 & (Reg | RegMMX | RegXMM
3660 | SReg2 | SReg3
3661 | Control | Debug | Test))
3662 ? 1
3663 : 2));
3664 /* If there is an extension opcode to put here, the register
3665 number must be put into the regmem field. */
3666 if (i.tm.extension_opcode != None)
3667 {
3668 i.rm.regmem = i.op[op].regs->reg_num;
3669 if ((i.op[op].regs->reg_flags & RegRex) != 0)
3670 i.rex |= REX_EXTZ;
3671 }
3672 else
3673 {
3674 i.rm.reg = i.op[op].regs->reg_num;
3675 if ((i.op[op].regs->reg_flags & RegRex) != 0)
3676 i.rex |= REX_EXTX;
3677 }
3678
3679 /* Now, if no memory operand has set i.rm.mode = 0, 1, 2 we
3680 must set it to 3 to indicate this is a register operand
3681 in the regmem field. */
3682 if (!i.mem_operands)
3683 i.rm.mode = 3;
3684 }
3685
3686 /* Fill in i.rm.reg field with extension opcode (if any). */
3687 if (i.tm.extension_opcode != None)
3688 i.rm.reg = i.tm.extension_opcode;
3689 }
3690 return default_seg;
3691 }
3692
3693 static void
3694 output_branch ()
3695 {
3696 char *p;
3697 int code16;
3698 int prefix;
3699 relax_substateT subtype;
3700 symbolS *sym;
3701 offsetT off;
3702
3703 code16 = 0;
3704 if (flag_code == CODE_16BIT)
3705 code16 = CODE16;
3706
3707 prefix = 0;
3708 if (i.prefix[DATA_PREFIX] != 0)
3709 {
3710 prefix = 1;
3711 i.prefixes -= 1;
3712 code16 ^= CODE16;
3713 }
3714 /* Pentium4 branch hints. */
3715 if (i.prefix[SEG_PREFIX] == CS_PREFIX_OPCODE /* not taken */
3716 || i.prefix[SEG_PREFIX] == DS_PREFIX_OPCODE /* taken */)
3717 {
3718 prefix++;
3719 i.prefixes--;
3720 }
3721 if (i.prefix[REX_PREFIX] != 0)
3722 {
3723 prefix++;
3724 i.prefixes--;
3725 }
3726
3727 if (i.prefixes != 0 && !intel_syntax)
3728 as_warn (_("skipping prefixes on this instruction"));
3729
3730 /* It's always a symbol; End frag & setup for relax.
3731 Make sure there is enough room in this frag for the largest
3732 instruction we may generate in md_convert_frag. This is 2
3733 bytes for the opcode and room for the prefix and largest
3734 displacement. */
3735 frag_grow (prefix + 2 + 4);
3736 /* Prefix and 1 opcode byte go in fr_fix. */
3737 p = frag_more (prefix + 1);
3738 if (i.prefix[DATA_PREFIX] != 0)
3739 *p++ = DATA_PREFIX_OPCODE;
3740 if (i.prefix[SEG_PREFIX] == CS_PREFIX_OPCODE
3741 || i.prefix[SEG_PREFIX] == DS_PREFIX_OPCODE)
3742 *p++ = i.prefix[SEG_PREFIX];
3743 if (i.prefix[REX_PREFIX] != 0)
3744 *p++ = i.prefix[REX_PREFIX];
3745 *p = i.tm.base_opcode;
3746
3747 if ((unsigned char) *p == JUMP_PC_RELATIVE)
3748 subtype = ENCODE_RELAX_STATE (UNCOND_JUMP, SMALL);
3749 else if ((cpu_arch_flags & Cpu386) != 0)
3750 subtype = ENCODE_RELAX_STATE (COND_JUMP, SMALL);
3751 else
3752 subtype = ENCODE_RELAX_STATE (COND_JUMP86, SMALL);
3753 subtype |= code16;
3754
3755 sym = i.op[0].disps->X_add_symbol;
3756 off = i.op[0].disps->X_add_number;
3757
3758 if (i.op[0].disps->X_op != O_constant
3759 && i.op[0].disps->X_op != O_symbol)
3760 {
3761 /* Handle complex expressions. */
3762 sym = make_expr_symbol (i.op[0].disps);
3763 off = 0;
3764 }
3765
3766 /* 1 possible extra opcode + 4 byte displacement go in var part.
3767 Pass reloc in fr_var. */
3768 frag_var (rs_machine_dependent, 5, i.reloc[0], subtype, sym, off, p);
3769 }
3770
3771 static void
3772 output_jump ()
3773 {
3774 char *p;
3775 int size;
3776 fixS *fixP;
3777
3778 if (i.tm.opcode_modifier & JumpByte)
3779 {
3780 /* This is a loop or jecxz type instruction. */
3781 size = 1;
3782 if (i.prefix[ADDR_PREFIX] != 0)
3783 {
3784 FRAG_APPEND_1_CHAR (ADDR_PREFIX_OPCODE);
3785 i.prefixes -= 1;
3786 }
3787 /* Pentium4 branch hints. */
3788 if (i.prefix[SEG_PREFIX] == CS_PREFIX_OPCODE /* not taken */
3789 || i.prefix[SEG_PREFIX] == DS_PREFIX_OPCODE /* taken */)
3790 {
3791 FRAG_APPEND_1_CHAR (i.prefix[SEG_PREFIX]);
3792 i.prefixes--;
3793 }
3794 }
3795 else
3796 {
3797 int code16;
3798
3799 code16 = 0;
3800 if (flag_code == CODE_16BIT)
3801 code16 = CODE16;
3802
3803 if (i.prefix[DATA_PREFIX] != 0)
3804 {
3805 FRAG_APPEND_1_CHAR (DATA_PREFIX_OPCODE);
3806 i.prefixes -= 1;
3807 code16 ^= CODE16;
3808 }
3809
3810 size = 4;
3811 if (code16)
3812 size = 2;
3813 }
3814
3815 if (i.prefix[REX_PREFIX] != 0)
3816 {
3817 FRAG_APPEND_1_CHAR (i.prefix[REX_PREFIX]);
3818 i.prefixes -= 1;
3819 }
3820
3821 if (i.prefixes != 0 && !intel_syntax)
3822 as_warn (_("skipping prefixes on this instruction"));
3823
3824 p = frag_more (1 + size);
3825 *p++ = i.tm.base_opcode;
3826
3827 fixP = fix_new_exp (frag_now, p - frag_now->fr_literal, size,
3828 i.op[0].disps, 1, reloc (size, 1, 1, i.reloc[0]));
3829
3830 /* All jumps handled here are signed, but don't use a signed limit
3831 check for 32 and 16 bit jumps as we want to allow wrap around at
3832 4G and 64k respectively. */
3833 if (size == 1)
3834 fixP->fx_signed = 1;
3835 }
3836
3837 static void
3838 output_interseg_jump ()
3839 {
3840 char *p;
3841 int size;
3842 int prefix;
3843 int code16;
3844
3845 code16 = 0;
3846 if (flag_code == CODE_16BIT)
3847 code16 = CODE16;
3848
3849 prefix = 0;
3850 if (i.prefix[DATA_PREFIX] != 0)
3851 {
3852 prefix = 1;
3853 i.prefixes -= 1;
3854 code16 ^= CODE16;
3855 }
3856 if (i.prefix[REX_PREFIX] != 0)
3857 {
3858 prefix++;
3859 i.prefixes -= 1;
3860 }
3861
3862 size = 4;
3863 if (code16)
3864 size = 2;
3865
3866 if (i.prefixes != 0 && !intel_syntax)
3867 as_warn (_("skipping prefixes on this instruction"));
3868
3869 /* 1 opcode; 2 segment; offset */
3870 p = frag_more (prefix + 1 + 2 + size);
3871
3872 if (i.prefix[DATA_PREFIX] != 0)
3873 *p++ = DATA_PREFIX_OPCODE;
3874
3875 if (i.prefix[REX_PREFIX] != 0)
3876 *p++ = i.prefix[REX_PREFIX];
3877
3878 *p++ = i.tm.base_opcode;
3879 if (i.op[1].imms->X_op == O_constant)
3880 {
3881 offsetT n = i.op[1].imms->X_add_number;
3882
3883 if (size == 2
3884 && !fits_in_unsigned_word (n)
3885 && !fits_in_signed_word (n))
3886 {
3887 as_bad (_("16-bit jump out of range"));
3888 return;
3889 }
3890 md_number_to_chars (p, n, size);
3891 }
3892 else
3893 fix_new_exp (frag_now, p - frag_now->fr_literal, size,
3894 i.op[1].imms, 0, reloc (size, 0, 0, i.reloc[1]));
3895 if (i.op[0].imms->X_op != O_constant)
3896 as_bad (_("can't handle non absolute segment in `%s'"),
3897 i.tm.name);
3898 md_number_to_chars (p + size, (valueT) i.op[0].imms->X_add_number, 2);
3899 }
3900
3901 static void
3902 output_insn ()
3903 {
3904 fragS *insn_start_frag;
3905 offsetT insn_start_off;
3906
3907 /* Tie dwarf2 debug info to the address at the start of the insn.
3908 We can't do this after the insn has been output as the current
3909 frag may have been closed off. eg. by frag_var. */
3910 dwarf2_emit_insn (0);
3911
3912 insn_start_frag = frag_now;
3913 insn_start_off = frag_now_fix ();
3914
3915 /* Output jumps. */
3916 if (i.tm.opcode_modifier & Jump)
3917 output_branch ();
3918 else if (i.tm.opcode_modifier & (JumpByte | JumpDword))
3919 output_jump ();
3920 else if (i.tm.opcode_modifier & JumpInterSegment)
3921 output_interseg_jump ();
3922 else
3923 {
3924 /* Output normal instructions here. */
3925 char *p;
3926 unsigned char *q;
3927 unsigned int prefix;
3928
3929 /* All opcodes on i386 have either 1 or 2 bytes. Supplemental
3930 Streaming SIMD extensions 3 Instructions have 3 bytes. We may
3931 use one more higher byte to specify a prefix the instruction
3932 requires. */
3933 if ((i.tm.cpu_flags & CpuSSSE3) != 0)
3934 {
3935 if (i.tm.base_opcode & 0xff000000)
3936 {
3937 prefix = (i.tm.base_opcode >> 24) & 0xff;
3938 goto check_prefix;
3939 }
3940 }
3941 else if ((i.tm.base_opcode & 0xff0000) != 0)
3942 {
3943 prefix = (i.tm.base_opcode >> 16) & 0xff;
3944 if ((i.tm.cpu_flags & CpuPadLock) != 0)
3945 {
3946 check_prefix:
3947 if (prefix != REPE_PREFIX_OPCODE
3948 || i.prefix[LOCKREP_PREFIX] != REPE_PREFIX_OPCODE)
3949 add_prefix (prefix);
3950 }
3951 else
3952 add_prefix (prefix);
3953 }
3954
3955 /* The prefix bytes. */
3956 for (q = i.prefix;
3957 q < i.prefix + sizeof (i.prefix) / sizeof (i.prefix[0]);
3958 q++)
3959 {
3960 if (*q)
3961 {
3962 p = frag_more (1);
3963 md_number_to_chars (p, (valueT) *q, 1);
3964 }
3965 }
3966
3967 /* Now the opcode; be careful about word order here! */
3968 if (fits_in_unsigned_byte (i.tm.base_opcode))
3969 {
3970 FRAG_APPEND_1_CHAR (i.tm.base_opcode);
3971 }
3972 else
3973 {
3974 if ((i.tm.cpu_flags & CpuSSSE3) != 0)
3975 {
3976 p = frag_more (3);
3977 *p++ = (i.tm.base_opcode >> 16) & 0xff;
3978 }
3979 else
3980 p = frag_more (2);
3981
3982 /* Put out high byte first: can't use md_number_to_chars! */
3983 *p++ = (i.tm.base_opcode >> 8) & 0xff;
3984 *p = i.tm.base_opcode & 0xff;
3985 }
3986
3987 /* Now the modrm byte and sib byte (if present). */
3988 if (i.tm.opcode_modifier & Modrm)
3989 {
3990 p = frag_more (1);
3991 md_number_to_chars (p,
3992 (valueT) (i.rm.regmem << 0
3993 | i.rm.reg << 3
3994 | i.rm.mode << 6),
3995 1);
3996 /* If i.rm.regmem == ESP (4)
3997 && i.rm.mode != (Register mode)
3998 && not 16 bit
3999 ==> need second modrm byte. */
4000 if (i.rm.regmem == ESCAPE_TO_TWO_BYTE_ADDRESSING
4001 && i.rm.mode != 3
4002 && !(i.base_reg && (i.base_reg->reg_type & Reg16) != 0))
4003 {
4004 p = frag_more (1);
4005 md_number_to_chars (p,
4006 (valueT) (i.sib.base << 0
4007 | i.sib.index << 3
4008 | i.sib.scale << 6),
4009 1);
4010 }
4011 }
4012
4013 if (i.disp_operands)
4014 output_disp (insn_start_frag, insn_start_off);
4015
4016 if (i.imm_operands)
4017 output_imm (insn_start_frag, insn_start_off);
4018 }
4019
4020 #ifdef DEBUG386
4021 if (flag_debug)
4022 {
4023 pi ("" /*line*/, &i);
4024 }
4025 #endif /* DEBUG386 */
4026 }
4027
4028 static void
4029 output_disp (fragS *insn_start_frag, offsetT insn_start_off)
4030 {
4031 char *p;
4032 unsigned int n;
4033
4034 for (n = 0; n < i.operands; n++)
4035 {
4036 if (i.types[n] & Disp)
4037 {
4038 if (i.op[n].disps->X_op == O_constant)
4039 {
4040 int size;
4041 offsetT val;
4042
4043 size = 4;
4044 if (i.types[n] & (Disp8 | Disp16 | Disp64))
4045 {
4046 size = 2;
4047 if (i.types[n] & Disp8)
4048 size = 1;
4049 if (i.types[n] & Disp64)
4050 size = 8;
4051 }
4052 val = offset_in_range (i.op[n].disps->X_add_number,
4053 size);
4054 p = frag_more (size);
4055 md_number_to_chars (p, val, size);
4056 }
4057 else
4058 {
4059 enum bfd_reloc_code_real reloc_type;
4060 int size = 4;
4061 int sign = 0;
4062 int pcrel = (i.flags[n] & Operand_PCrel) != 0;
4063
4064 /* The PC relative address is computed relative
4065 to the instruction boundary, so in case immediate
4066 fields follows, we need to adjust the value. */
4067 if (pcrel && i.imm_operands)
4068 {
4069 int imm_size = 4;
4070 unsigned int n1;
4071
4072 for (n1 = 0; n1 < i.operands; n1++)
4073 if (i.types[n1] & Imm)
4074 {
4075 if (i.types[n1] & (Imm8 | Imm8S | Imm16 | Imm64))
4076 {
4077 imm_size = 2;
4078 if (i.types[n1] & (Imm8 | Imm8S))
4079 imm_size = 1;
4080 if (i.types[n1] & Imm64)
4081 imm_size = 8;
4082 }
4083 break;
4084 }
4085 /* We should find the immediate. */
4086 if (n1 == i.operands)
4087 abort ();
4088 i.op[n].disps->X_add_number -= imm_size;
4089 }
4090
4091 if (i.types[n] & Disp32S)
4092 sign = 1;
4093
4094 if (i.types[n] & (Disp16 | Disp64))
4095 {
4096 size = 2;
4097 if (i.types[n] & Disp64)
4098 size = 8;
4099 }
4100
4101 p = frag_more (size);
4102 reloc_type = reloc (size, pcrel, sign, i.reloc[n]);
4103 if (GOT_symbol
4104 && GOT_symbol == i.op[n].disps->X_add_symbol
4105 && (((reloc_type == BFD_RELOC_32
4106 || reloc_type == BFD_RELOC_X86_64_32S
4107 || (reloc_type == BFD_RELOC_64
4108 && object_64bit))
4109 && (i.op[n].disps->X_op == O_symbol
4110 || (i.op[n].disps->X_op == O_add
4111 && ((symbol_get_value_expression
4112 (i.op[n].disps->X_op_symbol)->X_op)
4113 == O_subtract))))
4114 || reloc_type == BFD_RELOC_32_PCREL))
4115 {
4116 offsetT add;
4117
4118 if (insn_start_frag == frag_now)
4119 add = (p - frag_now->fr_literal) - insn_start_off;
4120 else
4121 {
4122 fragS *fr;
4123
4124 add = insn_start_frag->fr_fix - insn_start_off;
4125 for (fr = insn_start_frag->fr_next;
4126 fr && fr != frag_now; fr = fr->fr_next)
4127 add += fr->fr_fix;
4128 add += p - frag_now->fr_literal;
4129 }
4130
4131 if (!object_64bit)
4132 {
4133 reloc_type = BFD_RELOC_386_GOTPC;
4134 i.op[n].imms->X_add_number += add;
4135 }
4136 else if (reloc_type == BFD_RELOC_64)
4137 reloc_type = BFD_RELOC_X86_64_GOTPC64;
4138 else
4139 /* Don't do the adjustment for x86-64, as there
4140 the pcrel addressing is relative to the _next_
4141 insn, and that is taken care of in other code. */
4142 reloc_type = BFD_RELOC_X86_64_GOTPC32;
4143 }
4144 fix_new_exp (frag_now, p - frag_now->fr_literal, size,
4145 i.op[n].disps, pcrel, reloc_type);
4146 }
4147 }
4148 }
4149 }
4150
4151 static void
4152 output_imm (fragS *insn_start_frag, offsetT insn_start_off)
4153 {
4154 char *p;
4155 unsigned int n;
4156
4157 for (n = 0; n < i.operands; n++)
4158 {
4159 if (i.types[n] & Imm)
4160 {
4161 if (i.op[n].imms->X_op == O_constant)
4162 {
4163 int size;
4164 offsetT val;
4165
4166 size = 4;
4167 if (i.types[n] & (Imm8 | Imm8S | Imm16 | Imm64))
4168 {
4169 size = 2;
4170 if (i.types[n] & (Imm8 | Imm8S))
4171 size = 1;
4172 else if (i.types[n] & Imm64)
4173 size = 8;
4174 }
4175 val = offset_in_range (i.op[n].imms->X_add_number,
4176 size);
4177 p = frag_more (size);
4178 md_number_to_chars (p, val, size);
4179 }
4180 else
4181 {
4182 /* Not absolute_section.
4183 Need a 32-bit fixup (don't support 8bit
4184 non-absolute imms). Try to support other
4185 sizes ... */
4186 enum bfd_reloc_code_real reloc_type;
4187 int size = 4;
4188 int sign = 0;
4189
4190 if ((i.types[n] & (Imm32S))
4191 && (i.suffix == QWORD_MNEM_SUFFIX
4192 || (!i.suffix && (i.tm.opcode_modifier & No_lSuf))))
4193 sign = 1;
4194 if (i.types[n] & (Imm8 | Imm8S | Imm16 | Imm64))
4195 {
4196 size = 2;
4197 if (i.types[n] & (Imm8 | Imm8S))
4198 size = 1;
4199 if (i.types[n] & Imm64)
4200 size = 8;
4201 }
4202
4203 p = frag_more (size);
4204 reloc_type = reloc (size, 0, sign, i.reloc[n]);
4205
4206 /* This is tough to explain. We end up with this one if we
4207 * have operands that look like
4208 * "_GLOBAL_OFFSET_TABLE_+[.-.L284]". The goal here is to
4209 * obtain the absolute address of the GOT, and it is strongly
4210 * preferable from a performance point of view to avoid using
4211 * a runtime relocation for this. The actual sequence of
4212 * instructions often look something like:
4213 *
4214 * call .L66
4215 * .L66:
4216 * popl %ebx
4217 * addl $_GLOBAL_OFFSET_TABLE_+[.-.L66],%ebx
4218 *
4219 * The call and pop essentially return the absolute address
4220 * of the label .L66 and store it in %ebx. The linker itself
4221 * will ultimately change the first operand of the addl so
4222 * that %ebx points to the GOT, but to keep things simple, the
4223 * .o file must have this operand set so that it generates not
4224 * the absolute address of .L66, but the absolute address of
4225 * itself. This allows the linker itself simply treat a GOTPC
4226 * relocation as asking for a pcrel offset to the GOT to be
4227 * added in, and the addend of the relocation is stored in the
4228 * operand field for the instruction itself.
4229 *
4230 * Our job here is to fix the operand so that it would add
4231 * the correct offset so that %ebx would point to itself. The
4232 * thing that is tricky is that .-.L66 will point to the
4233 * beginning of the instruction, so we need to further modify
4234 * the operand so that it will point to itself. There are
4235 * other cases where you have something like:
4236 *
4237 * .long $_GLOBAL_OFFSET_TABLE_+[.-.L66]
4238 *
4239 * and here no correction would be required. Internally in
4240 * the assembler we treat operands of this form as not being
4241 * pcrel since the '.' is explicitly mentioned, and I wonder
4242 * whether it would simplify matters to do it this way. Who
4243 * knows. In earlier versions of the PIC patches, the
4244 * pcrel_adjust field was used to store the correction, but
4245 * since the expression is not pcrel, I felt it would be
4246 * confusing to do it this way. */
4247
4248 if ((reloc_type == BFD_RELOC_32
4249 || reloc_type == BFD_RELOC_X86_64_32S
4250 || reloc_type == BFD_RELOC_64)
4251 && GOT_symbol
4252 && GOT_symbol == i.op[n].imms->X_add_symbol
4253 && (i.op[n].imms->X_op == O_symbol
4254 || (i.op[n].imms->X_op == O_add
4255 && ((symbol_get_value_expression
4256 (i.op[n].imms->X_op_symbol)->X_op)
4257 == O_subtract))))
4258 {
4259 offsetT add;
4260
4261 if (insn_start_frag == frag_now)
4262 add = (p - frag_now->fr_literal) - insn_start_off;
4263 else
4264 {
4265 fragS *fr;
4266
4267 add = insn_start_frag->fr_fix - insn_start_off;
4268 for (fr = insn_start_frag->fr_next;
4269 fr && fr != frag_now; fr = fr->fr_next)
4270 add += fr->fr_fix;
4271 add += p - frag_now->fr_literal;
4272 }
4273
4274 if (!object_64bit)
4275 reloc_type = BFD_RELOC_386_GOTPC;
4276 else if (size == 4)
4277 reloc_type = BFD_RELOC_X86_64_GOTPC32;
4278 else if (size == 8)
4279 reloc_type = BFD_RELOC_X86_64_GOTPC64;
4280 i.op[n].imms->X_add_number += add;
4281 }
4282 fix_new_exp (frag_now, p - frag_now->fr_literal, size,
4283 i.op[n].imms, 0, reloc_type);
4284 }
4285 }
4286 }
4287 }
4288 \f
4289 /* x86_cons_fix_new is called via the expression parsing code when a
4290 reloc is needed. We use this hook to get the correct .got reloc. */
4291 static enum bfd_reloc_code_real got_reloc = NO_RELOC;
4292 static int cons_sign = -1;
4293
4294 void
4295 x86_cons_fix_new (fragS *frag,
4296 unsigned int off,
4297 unsigned int len,
4298 expressionS *exp)
4299 {
4300 enum bfd_reloc_code_real r = reloc (len, 0, cons_sign, got_reloc);
4301
4302 got_reloc = NO_RELOC;
4303
4304 #ifdef TE_PE
4305 if (exp->X_op == O_secrel)
4306 {
4307 exp->X_op = O_symbol;
4308 r = BFD_RELOC_32_SECREL;
4309 }
4310 #endif
4311
4312 fix_new_exp (frag, off, len, exp, 0, r);
4313 }
4314
4315 #if (!defined (OBJ_ELF) && !defined (OBJ_MAYBE_ELF)) || defined (LEX_AT)
4316 # define lex_got(reloc, adjust, types) NULL
4317 #else
4318 /* Parse operands of the form
4319 <symbol>@GOTOFF+<nnn>
4320 and similar .plt or .got references.
4321
4322 If we find one, set up the correct relocation in RELOC and copy the
4323 input string, minus the `@GOTOFF' into a malloc'd buffer for
4324 parsing by the calling routine. Return this buffer, and if ADJUST
4325 is non-null set it to the length of the string we removed from the
4326 input line. Otherwise return NULL. */
4327 static char *
4328 lex_got (enum bfd_reloc_code_real *reloc,
4329 int *adjust,
4330 unsigned int *types)
4331 {
4332 /* Some of the relocations depend on the size of what field is to
4333 be relocated. But in our callers i386_immediate and i386_displacement
4334 we don't yet know the operand size (this will be set by insn
4335 matching). Hence we record the word32 relocation here,
4336 and adjust the reloc according to the real size in reloc(). */
4337 static const struct {
4338 const char *str;
4339 const enum bfd_reloc_code_real rel[2];
4340 const unsigned int types64;
4341 } gotrel[] = {
4342 { "PLTOFF", { 0, BFD_RELOC_X86_64_PLTOFF64 }, Imm64 },
4343 { "PLT", { BFD_RELOC_386_PLT32, BFD_RELOC_X86_64_PLT32 }, Imm32|Imm32S|Disp32 },
4344 { "GOTPLT", { 0, BFD_RELOC_X86_64_GOTPLT64 }, Imm64|Disp64 },
4345 { "GOTOFF", { BFD_RELOC_386_GOTOFF, BFD_RELOC_X86_64_GOTOFF64 }, Imm64|Disp64 },
4346 { "GOTPCREL", { 0, BFD_RELOC_X86_64_GOTPCREL }, Imm32|Imm32S|Disp32 },
4347 { "TLSGD", { BFD_RELOC_386_TLS_GD, BFD_RELOC_X86_64_TLSGD }, Imm32|Imm32S|Disp32 },
4348 { "TLSLDM", { BFD_RELOC_386_TLS_LDM, 0 }, 0 },
4349 { "TLSLD", { 0, BFD_RELOC_X86_64_TLSLD }, Imm32|Imm32S|Disp32 },
4350 { "GOTTPOFF", { BFD_RELOC_386_TLS_IE_32, BFD_RELOC_X86_64_GOTTPOFF }, Imm32|Imm32S|Disp32 },
4351 { "TPOFF", { BFD_RELOC_386_TLS_LE_32, BFD_RELOC_X86_64_TPOFF32 }, Imm32|Imm32S|Imm64|Disp32|Disp64 },
4352 { "NTPOFF", { BFD_RELOC_386_TLS_LE, 0 }, 0 },
4353 { "DTPOFF", { BFD_RELOC_386_TLS_LDO_32, BFD_RELOC_X86_64_DTPOFF32 }, Imm32|Imm32S|Imm64|Disp32|Disp64 },
4354 { "GOTNTPOFF",{ BFD_RELOC_386_TLS_GOTIE, 0 }, 0 },
4355 { "INDNTPOFF",{ BFD_RELOC_386_TLS_IE, 0 }, 0 },
4356 { "GOT", { BFD_RELOC_386_GOT32, BFD_RELOC_X86_64_GOT32 }, Imm32|Imm32S|Disp32|Imm64 },
4357 { "TLSDESC", { BFD_RELOC_386_TLS_GOTDESC, BFD_RELOC_X86_64_GOTPC32_TLSDESC }, Imm32|Imm32S|Disp32 },
4358 { "TLSCALL", { BFD_RELOC_386_TLS_DESC_CALL, BFD_RELOC_X86_64_TLSDESC_CALL }, Imm32|Imm32S|Disp32 }
4359 };
4360 char *cp;
4361 unsigned int j;
4362
4363 if (!IS_ELF)
4364 return NULL;
4365
4366 for (cp = input_line_pointer; *cp != '@'; cp++)
4367 if (is_end_of_line[(unsigned char) *cp])
4368 return NULL;
4369
4370 for (j = 0; j < sizeof (gotrel) / sizeof (gotrel[0]); j++)
4371 {
4372 int len;
4373
4374 len = strlen (gotrel[j].str);
4375 if (strncasecmp (cp + 1, gotrel[j].str, len) == 0)
4376 {
4377 if (gotrel[j].rel[object_64bit] != 0)
4378 {
4379 int first, second;
4380 char *tmpbuf, *past_reloc;
4381
4382 *reloc = gotrel[j].rel[object_64bit];
4383 if (adjust)
4384 *adjust = len;
4385
4386 if (types)
4387 {
4388 if (flag_code != CODE_64BIT)
4389 *types = Imm32|Disp32;
4390 else
4391 *types = gotrel[j].types64;
4392 }
4393
4394 if (GOT_symbol == NULL)
4395 GOT_symbol = symbol_find_or_make (GLOBAL_OFFSET_TABLE_NAME);
4396
4397 /* Replace the relocation token with ' ', so that
4398 errors like foo@GOTOFF1 will be detected. */
4399
4400 /* The length of the first part of our input line. */
4401 first = cp - input_line_pointer;
4402
4403 /* The second part goes from after the reloc token until
4404 (and including) an end_of_line char. Don't use strlen
4405 here as the end_of_line char may not be a NUL. */
4406 past_reloc = cp + 1 + len;
4407 for (cp = past_reloc; !is_end_of_line[(unsigned char) *cp++]; )
4408 ;
4409 second = cp - past_reloc;
4410
4411 /* Allocate and copy string. The trailing NUL shouldn't
4412 be necessary, but be safe. */
4413 tmpbuf = xmalloc (first + second + 2);
4414 memcpy (tmpbuf, input_line_pointer, first);
4415 tmpbuf[first] = ' ';
4416 memcpy (tmpbuf + first + 1, past_reloc, second);
4417 tmpbuf[first + second + 1] = '\0';
4418 return tmpbuf;
4419 }
4420
4421 as_bad (_("@%s reloc is not supported with %d-bit output format"),
4422 gotrel[j].str, 1 << (5 + object_64bit));
4423 return NULL;
4424 }
4425 }
4426
4427 /* Might be a symbol version string. Don't as_bad here. */
4428 return NULL;
4429 }
4430
4431 void
4432 x86_cons (exp, size)
4433 expressionS *exp;
4434 int size;
4435 {
4436 if (size == 4 || (object_64bit && size == 8))
4437 {
4438 /* Handle @GOTOFF and the like in an expression. */
4439 char *save;
4440 char *gotfree_input_line;
4441 int adjust;
4442
4443 save = input_line_pointer;
4444 gotfree_input_line = lex_got (&got_reloc, &adjust, NULL);
4445 if (gotfree_input_line)
4446 input_line_pointer = gotfree_input_line;
4447
4448 expression (exp);
4449
4450 if (gotfree_input_line)
4451 {
4452 /* expression () has merrily parsed up to the end of line,
4453 or a comma - in the wrong buffer. Transfer how far
4454 input_line_pointer has moved to the right buffer. */
4455 input_line_pointer = (save
4456 + (input_line_pointer - gotfree_input_line)
4457 + adjust);
4458 free (gotfree_input_line);
4459 }
4460 }
4461 else
4462 expression (exp);
4463 }
4464 #endif
4465
4466 static void signed_cons (int size)
4467 {
4468 if (flag_code == CODE_64BIT)
4469 cons_sign = 1;
4470 cons (size);
4471 cons_sign = -1;
4472 }
4473
4474 #ifdef TE_PE
4475 static void
4476 pe_directive_secrel (dummy)
4477 int dummy ATTRIBUTE_UNUSED;
4478 {
4479 expressionS exp;
4480
4481 do
4482 {
4483 expression (&exp);
4484 if (exp.X_op == O_symbol)
4485 exp.X_op = O_secrel;
4486
4487 emit_expr (&exp, 4);
4488 }
4489 while (*input_line_pointer++ == ',');
4490
4491 input_line_pointer--;
4492 demand_empty_rest_of_line ();
4493 }
4494 #endif
4495
4496 static int
4497 i386_immediate (char *imm_start)
4498 {
4499 char *save_input_line_pointer;
4500 char *gotfree_input_line;
4501 segT exp_seg = 0;
4502 expressionS *exp;
4503 unsigned int types = ~0U;
4504
4505 if (i.imm_operands == MAX_IMMEDIATE_OPERANDS)
4506 {
4507 as_bad (_("only 1 or 2 immediate operands are allowed"));
4508 return 0;
4509 }
4510
4511 exp = &im_expressions[i.imm_operands++];
4512 i.op[this_operand].imms = exp;
4513
4514 if (is_space_char (*imm_start))
4515 ++imm_start;
4516
4517 save_input_line_pointer = input_line_pointer;
4518 input_line_pointer = imm_start;
4519
4520 gotfree_input_line = lex_got (&i.reloc[this_operand], NULL, &types);
4521 if (gotfree_input_line)
4522 input_line_pointer = gotfree_input_line;
4523
4524 exp_seg = expression (exp);
4525
4526 SKIP_WHITESPACE ();
4527 if (*input_line_pointer)
4528 as_bad (_("junk `%s' after expression"), input_line_pointer);
4529
4530 input_line_pointer = save_input_line_pointer;
4531 if (gotfree_input_line)
4532 free (gotfree_input_line);
4533
4534 if (exp->X_op == O_absent || exp->X_op == O_big)
4535 {
4536 /* Missing or bad expr becomes absolute 0. */
4537 as_bad (_("missing or invalid immediate expression `%s' taken as 0"),
4538 imm_start);
4539 exp->X_op = O_constant;
4540 exp->X_add_number = 0;
4541 exp->X_add_symbol = (symbolS *) 0;
4542 exp->X_op_symbol = (symbolS *) 0;
4543 }
4544 else if (exp->X_op == O_constant)
4545 {
4546 /* Size it properly later. */
4547 i.types[this_operand] |= Imm64;
4548 /* If BFD64, sign extend val. */
4549 if (!use_rela_relocations)
4550 if ((exp->X_add_number & ~(((addressT) 2 << 31) - 1)) == 0)
4551 exp->X_add_number = (exp->X_add_number ^ ((addressT) 1 << 31)) - ((addressT) 1 << 31);
4552 }
4553 #if (defined (OBJ_AOUT) || defined (OBJ_MAYBE_AOUT))
4554 else if (OUTPUT_FLAVOR == bfd_target_aout_flavour
4555 && exp_seg != absolute_section
4556 && exp_seg != text_section
4557 && exp_seg != data_section
4558 && exp_seg != bss_section
4559 && exp_seg != undefined_section
4560 && !bfd_is_com_section (exp_seg))
4561 {
4562 as_bad (_("unimplemented segment %s in operand"), exp_seg->name);
4563 return 0;
4564 }
4565 #endif
4566 else if (!intel_syntax && exp->X_op == O_register)
4567 {
4568 as_bad (_("illegal immediate register operand %s"), imm_start);
4569 return 0;
4570 }
4571 else
4572 {
4573 /* This is an address. The size of the address will be
4574 determined later, depending on destination register,
4575 suffix, or the default for the section. */
4576 i.types[this_operand] |= Imm8 | Imm16 | Imm32 | Imm32S | Imm64;
4577 i.types[this_operand] &= types;
4578 }
4579
4580 return 1;
4581 }
4582
4583 static char *i386_scale PARAMS ((char *));
4584
4585 static char *
4586 i386_scale (scale)
4587 char *scale;
4588 {
4589 offsetT val;
4590 char *save = input_line_pointer;
4591
4592 input_line_pointer = scale;
4593 val = get_absolute_expression ();
4594
4595 switch (val)
4596 {
4597 case 1:
4598 i.log2_scale_factor = 0;
4599 break;
4600 case 2:
4601 i.log2_scale_factor = 1;
4602 break;
4603 case 4:
4604 i.log2_scale_factor = 2;
4605 break;
4606 case 8:
4607 i.log2_scale_factor = 3;
4608 break;
4609 default:
4610 {
4611 char sep = *input_line_pointer;
4612
4613 *input_line_pointer = '\0';
4614 as_bad (_("expecting scale factor of 1, 2, 4, or 8: got `%s'"),
4615 scale);
4616 *input_line_pointer = sep;
4617 input_line_pointer = save;
4618 return NULL;
4619 }
4620 }
4621 if (i.log2_scale_factor != 0 && i.index_reg == 0)
4622 {
4623 as_warn (_("scale factor of %d without an index register"),
4624 1 << i.log2_scale_factor);
4625 #if SCALE1_WHEN_NO_INDEX
4626 i.log2_scale_factor = 0;
4627 #endif
4628 }
4629 scale = input_line_pointer;
4630 input_line_pointer = save;
4631 return scale;
4632 }
4633
4634 static int i386_displacement PARAMS ((char *, char *));
4635
4636 static int
4637 i386_displacement (disp_start, disp_end)
4638 char *disp_start;
4639 char *disp_end;
4640 {
4641 expressionS *exp;
4642 segT exp_seg = 0;
4643 char *save_input_line_pointer;
4644 char *gotfree_input_line;
4645 int bigdisp, override;
4646 unsigned int types = Disp;
4647
4648 if ((i.types[this_operand] & JumpAbsolute)
4649 || !(current_templates->start->opcode_modifier & (Jump | JumpDword)))
4650 {
4651 bigdisp = Disp32;
4652 override = (i.prefix[ADDR_PREFIX] != 0);
4653 }
4654 else
4655 {
4656 /* For PC-relative branches, the width of the displacement
4657 is dependent upon data size, not address size. */
4658 bigdisp = 0;
4659 override = (i.prefix[DATA_PREFIX] != 0);
4660 }
4661 if (flag_code == CODE_64BIT)
4662 {
4663 if (!bigdisp)
4664 bigdisp = ((override || i.suffix == WORD_MNEM_SUFFIX)
4665 ? Disp16
4666 : Disp32S | Disp32);
4667 else if (!override)
4668 bigdisp = Disp64 | Disp32S | Disp32;
4669 }
4670 else
4671 {
4672 if (!bigdisp)
4673 {
4674 if (!override)
4675 override = (i.suffix == (flag_code != CODE_16BIT
4676 ? WORD_MNEM_SUFFIX
4677 : LONG_MNEM_SUFFIX));
4678 bigdisp = Disp32;
4679 }
4680 if ((flag_code == CODE_16BIT) ^ override)
4681 bigdisp = Disp16;
4682 }
4683 i.types[this_operand] |= bigdisp;
4684
4685 exp = &disp_expressions[i.disp_operands];
4686 i.op[this_operand].disps = exp;
4687 i.disp_operands++;
4688 save_input_line_pointer = input_line_pointer;
4689 input_line_pointer = disp_start;
4690 END_STRING_AND_SAVE (disp_end);
4691
4692 #ifndef GCC_ASM_O_HACK
4693 #define GCC_ASM_O_HACK 0
4694 #endif
4695 #if GCC_ASM_O_HACK
4696 END_STRING_AND_SAVE (disp_end + 1);
4697 if ((i.types[this_operand] & BaseIndex) != 0
4698 && displacement_string_end[-1] == '+')
4699 {
4700 /* This hack is to avoid a warning when using the "o"
4701 constraint within gcc asm statements.
4702 For instance:
4703
4704 #define _set_tssldt_desc(n,addr,limit,type) \
4705 __asm__ __volatile__ ( \
4706 "movw %w2,%0\n\t" \
4707 "movw %w1,2+%0\n\t" \
4708 "rorl $16,%1\n\t" \
4709 "movb %b1,4+%0\n\t" \
4710 "movb %4,5+%0\n\t" \
4711 "movb $0,6+%0\n\t" \
4712 "movb %h1,7+%0\n\t" \
4713 "rorl $16,%1" \
4714 : "=o"(*(n)) : "q" (addr), "ri"(limit), "i"(type))
4715
4716 This works great except that the output assembler ends
4717 up looking a bit weird if it turns out that there is
4718 no offset. You end up producing code that looks like:
4719
4720 #APP
4721 movw $235,(%eax)
4722 movw %dx,2+(%eax)
4723 rorl $16,%edx
4724 movb %dl,4+(%eax)
4725 movb $137,5+(%eax)
4726 movb $0,6+(%eax)
4727 movb %dh,7+(%eax)
4728 rorl $16,%edx
4729 #NO_APP
4730
4731 So here we provide the missing zero. */
4732
4733 *displacement_string_end = '0';
4734 }
4735 #endif
4736 gotfree_input_line = lex_got (&i.reloc[this_operand], NULL, &types);
4737 if (gotfree_input_line)
4738 input_line_pointer = gotfree_input_line;
4739
4740 exp_seg = expression (exp);
4741
4742 SKIP_WHITESPACE ();
4743 if (*input_line_pointer)
4744 as_bad (_("junk `%s' after expression"), input_line_pointer);
4745 #if GCC_ASM_O_HACK
4746 RESTORE_END_STRING (disp_end + 1);
4747 #endif
4748 RESTORE_END_STRING (disp_end);
4749 input_line_pointer = save_input_line_pointer;
4750 if (gotfree_input_line)
4751 free (gotfree_input_line);
4752
4753 /* We do this to make sure that the section symbol is in
4754 the symbol table. We will ultimately change the relocation
4755 to be relative to the beginning of the section. */
4756 if (i.reloc[this_operand] == BFD_RELOC_386_GOTOFF
4757 || i.reloc[this_operand] == BFD_RELOC_X86_64_GOTPCREL
4758 || i.reloc[this_operand] == BFD_RELOC_X86_64_GOTOFF64)
4759 {
4760 if (exp->X_op != O_symbol)
4761 {
4762 as_bad (_("bad expression used with @%s"),
4763 (i.reloc[this_operand] == BFD_RELOC_X86_64_GOTPCREL
4764 ? "GOTPCREL"
4765 : "GOTOFF"));
4766 return 0;
4767 }
4768
4769 if (S_IS_LOCAL (exp->X_add_symbol)
4770 && S_GET_SEGMENT (exp->X_add_symbol) != undefined_section)
4771 section_symbol (S_GET_SEGMENT (exp->X_add_symbol));
4772 exp->X_op = O_subtract;
4773 exp->X_op_symbol = GOT_symbol;
4774 if (i.reloc[this_operand] == BFD_RELOC_X86_64_GOTPCREL)
4775 i.reloc[this_operand] = BFD_RELOC_32_PCREL;
4776 else if (i.reloc[this_operand] == BFD_RELOC_X86_64_GOTOFF64)
4777 i.reloc[this_operand] = BFD_RELOC_64;
4778 else
4779 i.reloc[this_operand] = BFD_RELOC_32;
4780 }
4781
4782 if (exp->X_op == O_absent || exp->X_op == O_big)
4783 {
4784 /* Missing or bad expr becomes absolute 0. */
4785 as_bad (_("missing or invalid displacement expression `%s' taken as 0"),
4786 disp_start);
4787 exp->X_op = O_constant;
4788 exp->X_add_number = 0;
4789 exp->X_add_symbol = (symbolS *) 0;
4790 exp->X_op_symbol = (symbolS *) 0;
4791 }
4792
4793 #if (defined (OBJ_AOUT) || defined (OBJ_MAYBE_AOUT))
4794 if (exp->X_op != O_constant
4795 && OUTPUT_FLAVOR == bfd_target_aout_flavour
4796 && exp_seg != absolute_section
4797 && exp_seg != text_section
4798 && exp_seg != data_section
4799 && exp_seg != bss_section
4800 && exp_seg != undefined_section
4801 && !bfd_is_com_section (exp_seg))
4802 {
4803 as_bad (_("unimplemented segment %s in operand"), exp_seg->name);
4804 return 0;
4805 }
4806 #endif
4807
4808 if (!(i.types[this_operand] & ~Disp))
4809 i.types[this_operand] &= types;
4810
4811 return 1;
4812 }
4813
4814 static int i386_index_check PARAMS ((const char *));
4815
4816 /* Make sure the memory operand we've been dealt is valid.
4817 Return 1 on success, 0 on a failure. */
4818
4819 static int
4820 i386_index_check (operand_string)
4821 const char *operand_string;
4822 {
4823 int ok;
4824 #if INFER_ADDR_PREFIX
4825 int fudged = 0;
4826
4827 tryprefix:
4828 #endif
4829 ok = 1;
4830 if ((current_templates->start->cpu_flags & CpuSVME)
4831 && current_templates->end[-1].operand_types[0] == AnyMem)
4832 {
4833 /* Memory operands of SVME insns are special in that they only allow
4834 rAX as their memory address and ignore any segment override. */
4835 unsigned RegXX;
4836
4837 /* SKINIT is even more restrictive: it always requires EAX. */
4838 if (strcmp (current_templates->start->name, "skinit") == 0)
4839 RegXX = Reg32;
4840 else if (flag_code == CODE_64BIT)
4841 RegXX = i.prefix[ADDR_PREFIX] == 0 ? Reg64 : Reg32;
4842 else
4843 RegXX = ((flag_code == CODE_16BIT) ^ (i.prefix[ADDR_PREFIX] != 0)
4844 ? Reg16
4845 : Reg32);
4846 if (!i.base_reg
4847 || !(i.base_reg->reg_type & Acc)
4848 || !(i.base_reg->reg_type & RegXX)
4849 || i.index_reg
4850 || (i.types[0] & Disp))
4851 ok = 0;
4852 }
4853 else if (flag_code == CODE_64BIT)
4854 {
4855 unsigned RegXX = (i.prefix[ADDR_PREFIX] == 0 ? Reg64 : Reg32);
4856
4857 if ((i.base_reg
4858 && ((i.base_reg->reg_type & RegXX) == 0)
4859 && (i.base_reg->reg_type != BaseIndex
4860 || i.index_reg))
4861 || (i.index_reg
4862 && ((i.index_reg->reg_type & (RegXX | BaseIndex))
4863 != (RegXX | BaseIndex))))
4864 ok = 0;
4865 }
4866 else
4867 {
4868 if ((flag_code == CODE_16BIT) ^ (i.prefix[ADDR_PREFIX] != 0))
4869 {
4870 /* 16bit checks. */
4871 if ((i.base_reg
4872 && ((i.base_reg->reg_type & (Reg16 | BaseIndex | RegRex))
4873 != (Reg16 | BaseIndex)))
4874 || (i.index_reg
4875 && (((i.index_reg->reg_type & (Reg16 | BaseIndex))
4876 != (Reg16 | BaseIndex))
4877 || !(i.base_reg
4878 && i.base_reg->reg_num < 6
4879 && i.index_reg->reg_num >= 6
4880 && i.log2_scale_factor == 0))))
4881 ok = 0;
4882 }
4883 else
4884 {
4885 /* 32bit checks. */
4886 if ((i.base_reg
4887 && (i.base_reg->reg_type & (Reg32 | RegRex)) != Reg32)
4888 || (i.index_reg
4889 && ((i.index_reg->reg_type & (Reg32 | BaseIndex | RegRex))
4890 != (Reg32 | BaseIndex))))
4891 ok = 0;
4892 }
4893 }
4894 if (!ok)
4895 {
4896 #if INFER_ADDR_PREFIX
4897 if (i.prefix[ADDR_PREFIX] == 0)
4898 {
4899 i.prefix[ADDR_PREFIX] = ADDR_PREFIX_OPCODE;
4900 i.prefixes += 1;
4901 /* Change the size of any displacement too. At most one of
4902 Disp16 or Disp32 is set.
4903 FIXME. There doesn't seem to be any real need for separate
4904 Disp16 and Disp32 flags. The same goes for Imm16 and Imm32.
4905 Removing them would probably clean up the code quite a lot. */
4906 if (flag_code != CODE_64BIT && (i.types[this_operand] & (Disp16 | Disp32)))
4907 i.types[this_operand] ^= (Disp16 | Disp32);
4908 fudged = 1;
4909 goto tryprefix;
4910 }
4911 if (fudged)
4912 as_bad (_("`%s' is not a valid base/index expression"),
4913 operand_string);
4914 else
4915 #endif
4916 as_bad (_("`%s' is not a valid %s bit base/index expression"),
4917 operand_string,
4918 flag_code_names[flag_code]);
4919 }
4920 return ok;
4921 }
4922
4923 /* Parse OPERAND_STRING into the i386_insn structure I. Returns non-zero
4924 on error. */
4925
4926 static int
4927 i386_operand (operand_string)
4928 char *operand_string;
4929 {
4930 const reg_entry *r;
4931 char *end_op;
4932 char *op_string = operand_string;
4933
4934 if (is_space_char (*op_string))
4935 ++op_string;
4936
4937 /* We check for an absolute prefix (differentiating,
4938 for example, 'jmp pc_relative_label' from 'jmp *absolute_label'. */
4939 if (*op_string == ABSOLUTE_PREFIX)
4940 {
4941 ++op_string;
4942 if (is_space_char (*op_string))
4943 ++op_string;
4944 i.types[this_operand] |= JumpAbsolute;
4945 }
4946
4947 /* Check if operand is a register. */
4948 if ((r = parse_register (op_string, &end_op)) != NULL)
4949 {
4950 /* Check for a segment override by searching for ':' after a
4951 segment register. */
4952 op_string = end_op;
4953 if (is_space_char (*op_string))
4954 ++op_string;
4955 if (*op_string == ':' && (r->reg_type & (SReg2 | SReg3)))
4956 {
4957 switch (r->reg_num)
4958 {
4959 case 0:
4960 i.seg[i.mem_operands] = &es;
4961 break;
4962 case 1:
4963 i.seg[i.mem_operands] = &cs;
4964 break;
4965 case 2:
4966 i.seg[i.mem_operands] = &ss;
4967 break;
4968 case 3:
4969 i.seg[i.mem_operands] = &ds;
4970 break;
4971 case 4:
4972 i.seg[i.mem_operands] = &fs;
4973 break;
4974 case 5:
4975 i.seg[i.mem_operands] = &gs;
4976 break;
4977 }
4978
4979 /* Skip the ':' and whitespace. */
4980 ++op_string;
4981 if (is_space_char (*op_string))
4982 ++op_string;
4983
4984 if (!is_digit_char (*op_string)
4985 && !is_identifier_char (*op_string)
4986 && *op_string != '('
4987 && *op_string != ABSOLUTE_PREFIX)
4988 {
4989 as_bad (_("bad memory operand `%s'"), op_string);
4990 return 0;
4991 }
4992 /* Handle case of %es:*foo. */
4993 if (*op_string == ABSOLUTE_PREFIX)
4994 {
4995 ++op_string;
4996 if (is_space_char (*op_string))
4997 ++op_string;
4998 i.types[this_operand] |= JumpAbsolute;
4999 }
5000 goto do_memory_reference;
5001 }
5002 if (*op_string)
5003 {
5004 as_bad (_("junk `%s' after register"), op_string);
5005 return 0;
5006 }
5007 i.types[this_operand] |= r->reg_type & ~BaseIndex;
5008 i.op[this_operand].regs = r;
5009 i.reg_operands++;
5010 }
5011 else if (*op_string == REGISTER_PREFIX)
5012 {
5013 as_bad (_("bad register name `%s'"), op_string);
5014 return 0;
5015 }
5016 else if (*op_string == IMMEDIATE_PREFIX)
5017 {
5018 ++op_string;
5019 if (i.types[this_operand] & JumpAbsolute)
5020 {
5021 as_bad (_("immediate operand illegal with absolute jump"));
5022 return 0;
5023 }
5024 if (!i386_immediate (op_string))
5025 return 0;
5026 }
5027 else if (is_digit_char (*op_string)
5028 || is_identifier_char (*op_string)
5029 || *op_string == '(')
5030 {
5031 /* This is a memory reference of some sort. */
5032 char *base_string;
5033
5034 /* Start and end of displacement string expression (if found). */
5035 char *displacement_string_start;
5036 char *displacement_string_end;
5037
5038 do_memory_reference:
5039 if ((i.mem_operands == 1
5040 && (current_templates->start->opcode_modifier & IsString) == 0)
5041 || i.mem_operands == 2)
5042 {
5043 as_bad (_("too many memory references for `%s'"),
5044 current_templates->start->name);
5045 return 0;
5046 }
5047
5048 /* Check for base index form. We detect the base index form by
5049 looking for an ')' at the end of the operand, searching
5050 for the '(' matching it, and finding a REGISTER_PREFIX or ','
5051 after the '('. */
5052 base_string = op_string + strlen (op_string);
5053
5054 --base_string;
5055 if (is_space_char (*base_string))
5056 --base_string;
5057
5058 /* If we only have a displacement, set-up for it to be parsed later. */
5059 displacement_string_start = op_string;
5060 displacement_string_end = base_string + 1;
5061
5062 if (*base_string == ')')
5063 {
5064 char *temp_string;
5065 unsigned int parens_balanced = 1;
5066 /* We've already checked that the number of left & right ()'s are
5067 equal, so this loop will not be infinite. */
5068 do
5069 {
5070 base_string--;
5071 if (*base_string == ')')
5072 parens_balanced++;
5073 if (*base_string == '(')
5074 parens_balanced--;
5075 }
5076 while (parens_balanced);
5077
5078 temp_string = base_string;
5079
5080 /* Skip past '(' and whitespace. */
5081 ++base_string;
5082 if (is_space_char (*base_string))
5083 ++base_string;
5084
5085 if (*base_string == ','
5086 || ((i.base_reg = parse_register (base_string, &end_op)) != NULL))
5087 {
5088 displacement_string_end = temp_string;
5089
5090 i.types[this_operand] |= BaseIndex;
5091
5092 if (i.base_reg)
5093 {
5094 base_string = end_op;
5095 if (is_space_char (*base_string))
5096 ++base_string;
5097 }
5098
5099 /* There may be an index reg or scale factor here. */
5100 if (*base_string == ',')
5101 {
5102 ++base_string;
5103 if (is_space_char (*base_string))
5104 ++base_string;
5105
5106 if ((i.index_reg = parse_register (base_string, &end_op)) != NULL)
5107 {
5108 base_string = end_op;
5109 if (is_space_char (*base_string))
5110 ++base_string;
5111 if (*base_string == ',')
5112 {
5113 ++base_string;
5114 if (is_space_char (*base_string))
5115 ++base_string;
5116 }
5117 else if (*base_string != ')')
5118 {
5119 as_bad (_("expecting `,' or `)' after index register in `%s'"),
5120 operand_string);
5121 return 0;
5122 }
5123 }
5124 else if (*base_string == REGISTER_PREFIX)
5125 {
5126 as_bad (_("bad register name `%s'"), base_string);
5127 return 0;
5128 }
5129
5130 /* Check for scale factor. */
5131 if (*base_string != ')')
5132 {
5133 char *end_scale = i386_scale (base_string);
5134
5135 if (!end_scale)
5136 return 0;
5137
5138 base_string = end_scale;
5139 if (is_space_char (*base_string))
5140 ++base_string;
5141 if (*base_string != ')')
5142 {
5143 as_bad (_("expecting `)' after scale factor in `%s'"),
5144 operand_string);
5145 return 0;
5146 }
5147 }
5148 else if (!i.index_reg)
5149 {
5150 as_bad (_("expecting index register or scale factor after `,'; got '%c'"),
5151 *base_string);
5152 return 0;
5153 }
5154 }
5155 else if (*base_string != ')')
5156 {
5157 as_bad (_("expecting `,' or `)' after base register in `%s'"),
5158 operand_string);
5159 return 0;
5160 }
5161 }
5162 else if (*base_string == REGISTER_PREFIX)
5163 {
5164 as_bad (_("bad register name `%s'"), base_string);
5165 return 0;
5166 }
5167 }
5168
5169 /* If there's an expression beginning the operand, parse it,
5170 assuming displacement_string_start and
5171 displacement_string_end are meaningful. */
5172 if (displacement_string_start != displacement_string_end)
5173 {
5174 if (!i386_displacement (displacement_string_start,
5175 displacement_string_end))
5176 return 0;
5177 }
5178
5179 /* Special case for (%dx) while doing input/output op. */
5180 if (i.base_reg
5181 && i.base_reg->reg_type == (Reg16 | InOutPortReg)
5182 && i.index_reg == 0
5183 && i.log2_scale_factor == 0
5184 && i.seg[i.mem_operands] == 0
5185 && (i.types[this_operand] & Disp) == 0)
5186 {
5187 i.types[this_operand] = InOutPortReg;
5188 return 1;
5189 }
5190
5191 if (i386_index_check (operand_string) == 0)
5192 return 0;
5193 i.mem_operands++;
5194 }
5195 else
5196 {
5197 /* It's not a memory operand; argh! */
5198 as_bad (_("invalid char %s beginning operand %d `%s'"),
5199 output_invalid (*op_string),
5200 this_operand + 1,
5201 op_string);
5202 return 0;
5203 }
5204 return 1; /* Normal return. */
5205 }
5206 \f
5207 /* md_estimate_size_before_relax()
5208
5209 Called just before relax() for rs_machine_dependent frags. The x86
5210 assembler uses these frags to handle variable size jump
5211 instructions.
5212
5213 Any symbol that is now undefined will not become defined.
5214 Return the correct fr_subtype in the frag.
5215 Return the initial "guess for variable size of frag" to caller.
5216 The guess is actually the growth beyond the fixed part. Whatever
5217 we do to grow the fixed or variable part contributes to our
5218 returned value. */
5219
5220 int
5221 md_estimate_size_before_relax (fragP, segment)
5222 fragS *fragP;
5223 segT segment;
5224 {
5225 /* We've already got fragP->fr_subtype right; all we have to do is
5226 check for un-relaxable symbols. On an ELF system, we can't relax
5227 an externally visible symbol, because it may be overridden by a
5228 shared library. */
5229 if (S_GET_SEGMENT (fragP->fr_symbol) != segment
5230 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5231 || (IS_ELF
5232 && (S_IS_EXTERNAL (fragP->fr_symbol)
5233 || S_IS_WEAK (fragP->fr_symbol)))
5234 #endif
5235 )
5236 {
5237 /* Symbol is undefined in this segment, or we need to keep a
5238 reloc so that weak symbols can be overridden. */
5239 int size = (fragP->fr_subtype & CODE16) ? 2 : 4;
5240 enum bfd_reloc_code_real reloc_type;
5241 unsigned char *opcode;
5242 int old_fr_fix;
5243
5244 if (fragP->fr_var != NO_RELOC)
5245 reloc_type = fragP->fr_var;
5246 else if (size == 2)
5247 reloc_type = BFD_RELOC_16_PCREL;
5248 else
5249 reloc_type = BFD_RELOC_32_PCREL;
5250
5251 old_fr_fix = fragP->fr_fix;
5252 opcode = (unsigned char *) fragP->fr_opcode;
5253
5254 switch (TYPE_FROM_RELAX_STATE (fragP->fr_subtype))
5255 {
5256 case UNCOND_JUMP:
5257 /* Make jmp (0xeb) a (d)word displacement jump. */
5258 opcode[0] = 0xe9;
5259 fragP->fr_fix += size;
5260 fix_new (fragP, old_fr_fix, size,
5261 fragP->fr_symbol,
5262 fragP->fr_offset, 1,
5263 reloc_type);
5264 break;
5265
5266 case COND_JUMP86:
5267 if (size == 2
5268 && (!no_cond_jump_promotion || fragP->fr_var != NO_RELOC))
5269 {
5270 /* Negate the condition, and branch past an
5271 unconditional jump. */
5272 opcode[0] ^= 1;
5273 opcode[1] = 3;
5274 /* Insert an unconditional jump. */
5275 opcode[2] = 0xe9;
5276 /* We added two extra opcode bytes, and have a two byte
5277 offset. */
5278 fragP->fr_fix += 2 + 2;
5279 fix_new (fragP, old_fr_fix + 2, 2,
5280 fragP->fr_symbol,
5281 fragP->fr_offset, 1,
5282 reloc_type);
5283 break;
5284 }
5285 /* Fall through. */
5286
5287 case COND_JUMP:
5288 if (no_cond_jump_promotion && fragP->fr_var == NO_RELOC)
5289 {
5290 fixS *fixP;
5291
5292 fragP->fr_fix += 1;
5293 fixP = fix_new (fragP, old_fr_fix, 1,
5294 fragP->fr_symbol,
5295 fragP->fr_offset, 1,
5296 BFD_RELOC_8_PCREL);
5297 fixP->fx_signed = 1;
5298 break;
5299 }
5300
5301 /* This changes the byte-displacement jump 0x7N
5302 to the (d)word-displacement jump 0x0f,0x8N. */
5303 opcode[1] = opcode[0] + 0x10;
5304 opcode[0] = TWO_BYTE_OPCODE_ESCAPE;
5305 /* We've added an opcode byte. */
5306 fragP->fr_fix += 1 + size;
5307 fix_new (fragP, old_fr_fix + 1, size,
5308 fragP->fr_symbol,
5309 fragP->fr_offset, 1,
5310 reloc_type);
5311 break;
5312
5313 default:
5314 BAD_CASE (fragP->fr_subtype);
5315 break;
5316 }
5317 frag_wane (fragP);
5318 return fragP->fr_fix - old_fr_fix;
5319 }
5320
5321 /* Guess size depending on current relax state. Initially the relax
5322 state will correspond to a short jump and we return 1, because
5323 the variable part of the frag (the branch offset) is one byte
5324 long. However, we can relax a section more than once and in that
5325 case we must either set fr_subtype back to the unrelaxed state,
5326 or return the value for the appropriate branch. */
5327 return md_relax_table[fragP->fr_subtype].rlx_length;
5328 }
5329
5330 /* Called after relax() is finished.
5331
5332 In: Address of frag.
5333 fr_type == rs_machine_dependent.
5334 fr_subtype is what the address relaxed to.
5335
5336 Out: Any fixSs and constants are set up.
5337 Caller will turn frag into a ".space 0". */
5338
5339 void
5340 md_convert_frag (abfd, sec, fragP)
5341 bfd *abfd ATTRIBUTE_UNUSED;
5342 segT sec ATTRIBUTE_UNUSED;
5343 fragS *fragP;
5344 {
5345 unsigned char *opcode;
5346 unsigned char *where_to_put_displacement = NULL;
5347 offsetT target_address;
5348 offsetT opcode_address;
5349 unsigned int extension = 0;
5350 offsetT displacement_from_opcode_start;
5351
5352 opcode = (unsigned char *) fragP->fr_opcode;
5353
5354 /* Address we want to reach in file space. */
5355 target_address = S_GET_VALUE (fragP->fr_symbol) + fragP->fr_offset;
5356
5357 /* Address opcode resides at in file space. */
5358 opcode_address = fragP->fr_address + fragP->fr_fix;
5359
5360 /* Displacement from opcode start to fill into instruction. */
5361 displacement_from_opcode_start = target_address - opcode_address;
5362
5363 if ((fragP->fr_subtype & BIG) == 0)
5364 {
5365 /* Don't have to change opcode. */
5366 extension = 1; /* 1 opcode + 1 displacement */
5367 where_to_put_displacement = &opcode[1];
5368 }
5369 else
5370 {
5371 if (no_cond_jump_promotion
5372 && TYPE_FROM_RELAX_STATE (fragP->fr_subtype) != UNCOND_JUMP)
5373 as_warn_where (fragP->fr_file, fragP->fr_line, _("long jump required"));
5374
5375 switch (fragP->fr_subtype)
5376 {
5377 case ENCODE_RELAX_STATE (UNCOND_JUMP, BIG):
5378 extension = 4; /* 1 opcode + 4 displacement */
5379 opcode[0] = 0xe9;
5380 where_to_put_displacement = &opcode[1];
5381 break;
5382
5383 case ENCODE_RELAX_STATE (UNCOND_JUMP, BIG16):
5384 extension = 2; /* 1 opcode + 2 displacement */
5385 opcode[0] = 0xe9;
5386 where_to_put_displacement = &opcode[1];
5387 break;
5388
5389 case ENCODE_RELAX_STATE (COND_JUMP, BIG):
5390 case ENCODE_RELAX_STATE (COND_JUMP86, BIG):
5391 extension = 5; /* 2 opcode + 4 displacement */
5392 opcode[1] = opcode[0] + 0x10;
5393 opcode[0] = TWO_BYTE_OPCODE_ESCAPE;
5394 where_to_put_displacement = &opcode[2];
5395 break;
5396
5397 case ENCODE_RELAX_STATE (COND_JUMP, BIG16):
5398 extension = 3; /* 2 opcode + 2 displacement */
5399 opcode[1] = opcode[0] + 0x10;
5400 opcode[0] = TWO_BYTE_OPCODE_ESCAPE;
5401 where_to_put_displacement = &opcode[2];
5402 break;
5403
5404 case ENCODE_RELAX_STATE (COND_JUMP86, BIG16):
5405 extension = 4;
5406 opcode[0] ^= 1;
5407 opcode[1] = 3;
5408 opcode[2] = 0xe9;
5409 where_to_put_displacement = &opcode[3];
5410 break;
5411
5412 default:
5413 BAD_CASE (fragP->fr_subtype);
5414 break;
5415 }
5416 }
5417
5418 /* If size if less then four we are sure that the operand fits,
5419 but if it's 4, then it could be that the displacement is larger
5420 then -/+ 2GB. */
5421 if (DISP_SIZE_FROM_RELAX_STATE (fragP->fr_subtype) == 4
5422 && object_64bit
5423 && ((addressT) (displacement_from_opcode_start - extension
5424 + ((addressT) 1 << 31))
5425 > (((addressT) 2 << 31) - 1)))
5426 {
5427 as_bad_where (fragP->fr_file, fragP->fr_line,
5428 _("jump target out of range"));
5429 /* Make us emit 0. */
5430 displacement_from_opcode_start = extension;
5431 }
5432 /* Now put displacement after opcode. */
5433 md_number_to_chars ((char *) where_to_put_displacement,
5434 (valueT) (displacement_from_opcode_start - extension),
5435 DISP_SIZE_FROM_RELAX_STATE (fragP->fr_subtype));
5436 fragP->fr_fix += extension;
5437 }
5438 \f
5439 /* Size of byte displacement jmp. */
5440 int md_short_jump_size = 2;
5441
5442 /* Size of dword displacement jmp. */
5443 int md_long_jump_size = 5;
5444
5445 void
5446 md_create_short_jump (ptr, from_addr, to_addr, frag, to_symbol)
5447 char *ptr;
5448 addressT from_addr, to_addr;
5449 fragS *frag ATTRIBUTE_UNUSED;
5450 symbolS *to_symbol ATTRIBUTE_UNUSED;
5451 {
5452 offsetT offset;
5453
5454 offset = to_addr - (from_addr + 2);
5455 /* Opcode for byte-disp jump. */
5456 md_number_to_chars (ptr, (valueT) 0xeb, 1);
5457 md_number_to_chars (ptr + 1, (valueT) offset, 1);
5458 }
5459
5460 void
5461 md_create_long_jump (ptr, from_addr, to_addr, frag, to_symbol)
5462 char *ptr;
5463 addressT from_addr, to_addr;
5464 fragS *frag ATTRIBUTE_UNUSED;
5465 symbolS *to_symbol ATTRIBUTE_UNUSED;
5466 {
5467 offsetT offset;
5468
5469 offset = to_addr - (from_addr + 5);
5470 md_number_to_chars (ptr, (valueT) 0xe9, 1);
5471 md_number_to_chars (ptr + 1, (valueT) offset, 4);
5472 }
5473 \f
5474 /* Apply a fixup (fixS) to segment data, once it has been determined
5475 by our caller that we have all the info we need to fix it up.
5476
5477 On the 386, immediates, displacements, and data pointers are all in
5478 the same (little-endian) format, so we don't need to care about which
5479 we are handling. */
5480
5481 void
5482 md_apply_fix (fixP, valP, seg)
5483 /* The fix we're to put in. */
5484 fixS *fixP;
5485 /* Pointer to the value of the bits. */
5486 valueT *valP;
5487 /* Segment fix is from. */
5488 segT seg ATTRIBUTE_UNUSED;
5489 {
5490 char *p = fixP->fx_where + fixP->fx_frag->fr_literal;
5491 valueT value = *valP;
5492
5493 #if !defined (TE_Mach)
5494 if (fixP->fx_pcrel)
5495 {
5496 switch (fixP->fx_r_type)
5497 {
5498 default:
5499 break;
5500
5501 case BFD_RELOC_64:
5502 fixP->fx_r_type = BFD_RELOC_64_PCREL;
5503 break;
5504 case BFD_RELOC_32:
5505 case BFD_RELOC_X86_64_32S:
5506 fixP->fx_r_type = BFD_RELOC_32_PCREL;
5507 break;
5508 case BFD_RELOC_16:
5509 fixP->fx_r_type = BFD_RELOC_16_PCREL;
5510 break;
5511 case BFD_RELOC_8:
5512 fixP->fx_r_type = BFD_RELOC_8_PCREL;
5513 break;
5514 }
5515 }
5516
5517 if (fixP->fx_addsy != NULL
5518 && (fixP->fx_r_type == BFD_RELOC_32_PCREL
5519 || fixP->fx_r_type == BFD_RELOC_64_PCREL
5520 || fixP->fx_r_type == BFD_RELOC_16_PCREL
5521 || fixP->fx_r_type == BFD_RELOC_8_PCREL)
5522 && !use_rela_relocations)
5523 {
5524 /* This is a hack. There should be a better way to handle this.
5525 This covers for the fact that bfd_install_relocation will
5526 subtract the current location (for partial_inplace, PC relative
5527 relocations); see more below. */
5528 #ifndef OBJ_AOUT
5529 if (IS_ELF
5530 #ifdef TE_PE
5531 || OUTPUT_FLAVOR == bfd_target_coff_flavour
5532 #endif
5533 )
5534 value += fixP->fx_where + fixP->fx_frag->fr_address;
5535 #endif
5536 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5537 if (IS_ELF)
5538 {
5539 segT sym_seg = S_GET_SEGMENT (fixP->fx_addsy);
5540
5541 if ((sym_seg == seg
5542 || (symbol_section_p (fixP->fx_addsy)
5543 && sym_seg != absolute_section))
5544 && !generic_force_reloc (fixP))
5545 {
5546 /* Yes, we add the values in twice. This is because
5547 bfd_install_relocation subtracts them out again. I think
5548 bfd_install_relocation is broken, but I don't dare change
5549 it. FIXME. */
5550 value += fixP->fx_where + fixP->fx_frag->fr_address;
5551 }
5552 }
5553 #endif
5554 #if defined (OBJ_COFF) && defined (TE_PE)
5555 /* For some reason, the PE format does not store a
5556 section address offset for a PC relative symbol. */
5557 if (S_GET_SEGMENT (fixP->fx_addsy) != seg
5558 || S_IS_WEAK (fixP->fx_addsy))
5559 value += md_pcrel_from (fixP);
5560 #endif
5561 }
5562
5563 /* Fix a few things - the dynamic linker expects certain values here,
5564 and we must not disappoint it. */
5565 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5566 if (IS_ELF && fixP->fx_addsy)
5567 switch (fixP->fx_r_type)
5568 {
5569 case BFD_RELOC_386_PLT32:
5570 case BFD_RELOC_X86_64_PLT32:
5571 /* Make the jump instruction point to the address of the operand. At
5572 runtime we merely add the offset to the actual PLT entry. */
5573 value = -4;
5574 break;
5575
5576 case BFD_RELOC_386_TLS_GD:
5577 case BFD_RELOC_386_TLS_LDM:
5578 case BFD_RELOC_386_TLS_IE_32:
5579 case BFD_RELOC_386_TLS_IE:
5580 case BFD_RELOC_386_TLS_GOTIE:
5581 case BFD_RELOC_386_TLS_GOTDESC:
5582 case BFD_RELOC_X86_64_TLSGD:
5583 case BFD_RELOC_X86_64_TLSLD:
5584 case BFD_RELOC_X86_64_GOTTPOFF:
5585 case BFD_RELOC_X86_64_GOTPC32_TLSDESC:
5586 value = 0; /* Fully resolved at runtime. No addend. */
5587 /* Fallthrough */
5588 case BFD_RELOC_386_TLS_LE:
5589 case BFD_RELOC_386_TLS_LDO_32:
5590 case BFD_RELOC_386_TLS_LE_32:
5591 case BFD_RELOC_X86_64_DTPOFF32:
5592 case BFD_RELOC_X86_64_DTPOFF64:
5593 case BFD_RELOC_X86_64_TPOFF32:
5594 case BFD_RELOC_X86_64_TPOFF64:
5595 S_SET_THREAD_LOCAL (fixP->fx_addsy);
5596 break;
5597
5598 case BFD_RELOC_386_TLS_DESC_CALL:
5599 case BFD_RELOC_X86_64_TLSDESC_CALL:
5600 value = 0; /* Fully resolved at runtime. No addend. */
5601 S_SET_THREAD_LOCAL (fixP->fx_addsy);
5602 fixP->fx_done = 0;
5603 return;
5604
5605 case BFD_RELOC_386_GOT32:
5606 case BFD_RELOC_X86_64_GOT32:
5607 value = 0; /* Fully resolved at runtime. No addend. */
5608 break;
5609
5610 case BFD_RELOC_VTABLE_INHERIT:
5611 case BFD_RELOC_VTABLE_ENTRY:
5612 fixP->fx_done = 0;
5613 return;
5614
5615 default:
5616 break;
5617 }
5618 #endif /* defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) */
5619 *valP = value;
5620 #endif /* !defined (TE_Mach) */
5621
5622 /* Are we finished with this relocation now? */
5623 if (fixP->fx_addsy == NULL)
5624 fixP->fx_done = 1;
5625 else if (use_rela_relocations)
5626 {
5627 fixP->fx_no_overflow = 1;
5628 /* Remember value for tc_gen_reloc. */
5629 fixP->fx_addnumber = value;
5630 value = 0;
5631 }
5632
5633 md_number_to_chars (p, value, fixP->fx_size);
5634 }
5635 \f
5636 #define MAX_LITTLENUMS 6
5637
5638 /* Turn the string pointed to by litP into a floating point constant
5639 of type TYPE, and emit the appropriate bytes. The number of
5640 LITTLENUMS emitted is stored in *SIZEP. An error message is
5641 returned, or NULL on OK. */
5642
5643 char *
5644 md_atof (type, litP, sizeP)
5645 int type;
5646 char *litP;
5647 int *sizeP;
5648 {
5649 int prec;
5650 LITTLENUM_TYPE words[MAX_LITTLENUMS];
5651 LITTLENUM_TYPE *wordP;
5652 char *t;
5653
5654 switch (type)
5655 {
5656 case 'f':
5657 case 'F':
5658 prec = 2;
5659 break;
5660
5661 case 'd':
5662 case 'D':
5663 prec = 4;
5664 break;
5665
5666 case 'x':
5667 case 'X':
5668 prec = 5;
5669 break;
5670
5671 default:
5672 *sizeP = 0;
5673 return _("Bad call to md_atof ()");
5674 }
5675 t = atof_ieee (input_line_pointer, type, words);
5676 if (t)
5677 input_line_pointer = t;
5678
5679 *sizeP = prec * sizeof (LITTLENUM_TYPE);
5680 /* This loops outputs the LITTLENUMs in REVERSE order; in accord with
5681 the bigendian 386. */
5682 for (wordP = words + prec - 1; prec--;)
5683 {
5684 md_number_to_chars (litP, (valueT) (*wordP--), sizeof (LITTLENUM_TYPE));
5685 litP += sizeof (LITTLENUM_TYPE);
5686 }
5687 return 0;
5688 }
5689 \f
5690 static char output_invalid_buf[sizeof (unsigned char) * 2 + 6];
5691
5692 static char *
5693 output_invalid (c)
5694 int c;
5695 {
5696 if (ISPRINT (c))
5697 snprintf (output_invalid_buf, sizeof (output_invalid_buf),
5698 "'%c'", c);
5699 else
5700 snprintf (output_invalid_buf, sizeof (output_invalid_buf),
5701 "(0x%x)", (unsigned char) c);
5702 return output_invalid_buf;
5703 }
5704
5705 /* REG_STRING starts *before* REGISTER_PREFIX. */
5706
5707 static const reg_entry *
5708 parse_real_register (char *reg_string, char **end_op)
5709 {
5710 char *s = reg_string;
5711 char *p;
5712 char reg_name_given[MAX_REG_NAME_SIZE + 1];
5713 const reg_entry *r;
5714
5715 /* Skip possible REGISTER_PREFIX and possible whitespace. */
5716 if (*s == REGISTER_PREFIX)
5717 ++s;
5718
5719 if (is_space_char (*s))
5720 ++s;
5721
5722 p = reg_name_given;
5723 while ((*p++ = register_chars[(unsigned char) *s]) != '\0')
5724 {
5725 if (p >= reg_name_given + MAX_REG_NAME_SIZE)
5726 return (const reg_entry *) NULL;
5727 s++;
5728 }
5729
5730 /* For naked regs, make sure that we are not dealing with an identifier.
5731 This prevents confusing an identifier like `eax_var' with register
5732 `eax'. */
5733 if (allow_naked_reg && identifier_chars[(unsigned char) *s])
5734 return (const reg_entry *) NULL;
5735
5736 *end_op = s;
5737
5738 r = (const reg_entry *) hash_find (reg_hash, reg_name_given);
5739
5740 /* Handle floating point regs, allowing spaces in the (i) part. */
5741 if (r == i386_regtab /* %st is first entry of table */)
5742 {
5743 if (is_space_char (*s))
5744 ++s;
5745 if (*s == '(')
5746 {
5747 ++s;
5748 if (is_space_char (*s))
5749 ++s;
5750 if (*s >= '0' && *s <= '7')
5751 {
5752 r = &i386_float_regtab[*s - '0'];
5753 ++s;
5754 if (is_space_char (*s))
5755 ++s;
5756 if (*s == ')')
5757 {
5758 *end_op = s + 1;
5759 return r;
5760 }
5761 }
5762 /* We have "%st(" then garbage. */
5763 return (const reg_entry *) NULL;
5764 }
5765 }
5766
5767 if (r != NULL
5768 && ((r->reg_flags & (RegRex64 | RegRex)) | (r->reg_type & Reg64)) != 0
5769 && (r->reg_type != Control || !(cpu_arch_flags & CpuSledgehammer))
5770 && flag_code != CODE_64BIT)
5771 return (const reg_entry *) NULL;
5772
5773 return r;
5774 }
5775
5776 /* REG_STRING starts *before* REGISTER_PREFIX. */
5777
5778 static const reg_entry *
5779 parse_register (char *reg_string, char **end_op)
5780 {
5781 const reg_entry *r;
5782
5783 if (*reg_string == REGISTER_PREFIX || allow_naked_reg)
5784 r = parse_real_register (reg_string, end_op);
5785 else
5786 r = NULL;
5787 if (!r)
5788 {
5789 char *save = input_line_pointer;
5790 char c;
5791 symbolS *symbolP;
5792
5793 input_line_pointer = reg_string;
5794 c = get_symbol_end ();
5795 symbolP = symbol_find (reg_string);
5796 if (symbolP && S_GET_SEGMENT (symbolP) == reg_section)
5797 {
5798 const expressionS *e = symbol_get_value_expression (symbolP);
5799
5800 know (e->X_op == O_register);
5801 know (e->X_add_number >= 0 && (valueT) e->X_add_number < ARRAY_SIZE (i386_regtab));
5802 r = i386_regtab + e->X_add_number;
5803 *end_op = input_line_pointer;
5804 }
5805 *input_line_pointer = c;
5806 input_line_pointer = save;
5807 }
5808 return r;
5809 }
5810
5811 int
5812 i386_parse_name (char *name, expressionS *e, char *nextcharP)
5813 {
5814 const reg_entry *r;
5815 char *end = input_line_pointer;
5816
5817 *end = *nextcharP;
5818 r = parse_register (name, &input_line_pointer);
5819 if (r && end <= input_line_pointer)
5820 {
5821 *nextcharP = *input_line_pointer;
5822 *input_line_pointer = 0;
5823 e->X_op = O_register;
5824 e->X_add_number = r - i386_regtab;
5825 return 1;
5826 }
5827 input_line_pointer = end;
5828 *end = 0;
5829 return 0;
5830 }
5831
5832 void
5833 md_operand (expressionS *e)
5834 {
5835 if (*input_line_pointer == REGISTER_PREFIX)
5836 {
5837 char *end;
5838 const reg_entry *r = parse_real_register (input_line_pointer, &end);
5839
5840 if (r)
5841 {
5842 e->X_op = O_register;
5843 e->X_add_number = r - i386_regtab;
5844 input_line_pointer = end;
5845 }
5846 }
5847 }
5848
5849 \f
5850 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5851 const char *md_shortopts = "kVQ:sqn";
5852 #else
5853 const char *md_shortopts = "qn";
5854 #endif
5855
5856 #define OPTION_32 (OPTION_MD_BASE + 0)
5857 #define OPTION_64 (OPTION_MD_BASE + 1)
5858 #define OPTION_DIVIDE (OPTION_MD_BASE + 2)
5859 #define OPTION_MARCH (OPTION_MD_BASE + 3)
5860 #define OPTION_MTUNE (OPTION_MD_BASE + 4)
5861
5862 struct option md_longopts[] =
5863 {
5864 {"32", no_argument, NULL, OPTION_32},
5865 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) || defined(TE_PEP)
5866 {"64", no_argument, NULL, OPTION_64},
5867 #endif
5868 {"divide", no_argument, NULL, OPTION_DIVIDE},
5869 {"march", required_argument, NULL, OPTION_MARCH},
5870 {"mtune", required_argument, NULL, OPTION_MTUNE},
5871 {NULL, no_argument, NULL, 0}
5872 };
5873 size_t md_longopts_size = sizeof (md_longopts);
5874
5875 int
5876 md_parse_option (int c, char *arg)
5877 {
5878 unsigned int i;
5879
5880 switch (c)
5881 {
5882 case 'n':
5883 optimize_align_code = 0;
5884 break;
5885
5886 case 'q':
5887 quiet_warnings = 1;
5888 break;
5889
5890 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
5891 /* -Qy, -Qn: SVR4 arguments controlling whether a .comment section
5892 should be emitted or not. FIXME: Not implemented. */
5893 case 'Q':
5894 break;
5895
5896 /* -V: SVR4 argument to print version ID. */
5897 case 'V':
5898 print_version_id ();
5899 break;
5900
5901 /* -k: Ignore for FreeBSD compatibility. */
5902 case 'k':
5903 break;
5904
5905 case 's':
5906 /* -s: On i386 Solaris, this tells the native assembler to use
5907 .stab instead of .stab.excl. We always use .stab anyhow. */
5908 break;
5909 #endif
5910 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF) || defined(TE_PEP)
5911 case OPTION_64:
5912 {
5913 const char **list, **l;
5914
5915 list = bfd_target_list ();
5916 for (l = list; *l != NULL; l++)
5917 if (CONST_STRNEQ (*l, "elf64-x86-64")
5918 || strcmp (*l, "coff-x86-64") == 0
5919 || strcmp (*l, "pe-x86-64") == 0
5920 || strcmp (*l, "pei-x86-64") == 0)
5921 {
5922 default_arch = "x86_64";
5923 break;
5924 }
5925 if (*l == NULL)
5926 as_fatal (_("No compiled in support for x86_64"));
5927 free (list);
5928 }
5929 break;
5930 #endif
5931
5932 case OPTION_32:
5933 default_arch = "i386";
5934 break;
5935
5936 case OPTION_DIVIDE:
5937 #ifdef SVR4_COMMENT_CHARS
5938 {
5939 char *n, *t;
5940 const char *s;
5941
5942 n = (char *) xmalloc (strlen (i386_comment_chars) + 1);
5943 t = n;
5944 for (s = i386_comment_chars; *s != '\0'; s++)
5945 if (*s != '/')
5946 *t++ = *s;
5947 *t = '\0';
5948 i386_comment_chars = n;
5949 }
5950 #endif
5951 break;
5952
5953 case OPTION_MARCH:
5954 if (*arg == '.')
5955 as_fatal (_("Invalid -march= option: `%s'"), arg);
5956 for (i = 0; i < ARRAY_SIZE (cpu_arch); i++)
5957 {
5958 if (strcmp (arg, cpu_arch [i].name) == 0)
5959 {
5960 cpu_arch_isa = cpu_arch[i].type;
5961 cpu_arch_isa_flags = cpu_arch[i].flags;
5962 if (!cpu_arch_tune_set)
5963 {
5964 cpu_arch_tune = cpu_arch_isa;
5965 cpu_arch_tune_flags = cpu_arch_isa_flags;
5966 }
5967 break;
5968 }
5969 }
5970 if (i >= ARRAY_SIZE (cpu_arch))
5971 as_fatal (_("Invalid -march= option: `%s'"), arg);
5972 break;
5973
5974 case OPTION_MTUNE:
5975 if (*arg == '.')
5976 as_fatal (_("Invalid -mtune= option: `%s'"), arg);
5977 for (i = 0; i < ARRAY_SIZE (cpu_arch); i++)
5978 {
5979 if (strcmp (arg, cpu_arch [i].name) == 0)
5980 {
5981 cpu_arch_tune_set = 1;
5982 cpu_arch_tune = cpu_arch [i].type;
5983 cpu_arch_tune_flags = cpu_arch[i].flags;
5984 break;
5985 }
5986 }
5987 if (i >= ARRAY_SIZE (cpu_arch))
5988 as_fatal (_("Invalid -mtune= option: `%s'"), arg);
5989 break;
5990
5991 default:
5992 return 0;
5993 }
5994 return 1;
5995 }
5996
5997 void
5998 md_show_usage (stream)
5999 FILE *stream;
6000 {
6001 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
6002 fprintf (stream, _("\
6003 -Q ignored\n\
6004 -V print assembler version number\n\
6005 -k ignored\n"));
6006 #endif
6007 fprintf (stream, _("\
6008 -n Do not optimize code alignment\n\
6009 -q quieten some warnings\n"));
6010 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
6011 fprintf (stream, _("\
6012 -s ignored\n"));
6013 #endif
6014 #ifdef SVR4_COMMENT_CHARS
6015 fprintf (stream, _("\
6016 --divide do not treat `/' as a comment character\n"));
6017 #else
6018 fprintf (stream, _("\
6019 --divide ignored\n"));
6020 #endif
6021 fprintf (stream, _("\
6022 -march=CPU/-mtune=CPU generate code/optimize for CPU, where CPU is one of:\n\
6023 i386, i486, pentium, pentiumpro, pentium4, nocona,\n\
6024 core, core2, k6, athlon, k8, generic32, generic64\n"));
6025
6026 }
6027
6028 #if defined(TE_PEP)
6029 const char *
6030 x86_64_target_format (void)
6031 {
6032 if (strcmp (default_arch, "x86_64") == 0)
6033 {
6034 set_code_flag (CODE_64BIT);
6035 return COFF_TARGET_FORMAT;
6036 }
6037 else if (strcmp (default_arch, "i386") == 0)
6038 {
6039 set_code_flag (CODE_32BIT);
6040 return "coff-i386";
6041 }
6042
6043 as_fatal (_("Unknown architecture"));
6044 return NULL;
6045 }
6046 #endif
6047
6048 #if ((defined (OBJ_MAYBE_COFF) && defined (OBJ_MAYBE_AOUT)) \
6049 || defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF))
6050
6051 /* Pick the target format to use. */
6052
6053 const char *
6054 i386_target_format ()
6055 {
6056 if (!strcmp (default_arch, "x86_64"))
6057 {
6058 set_code_flag (CODE_64BIT);
6059 if (cpu_arch_isa_flags == 0)
6060 cpu_arch_isa_flags = Cpu186|Cpu286|Cpu386|Cpu486
6061 |Cpu586|Cpu686|CpuP4|CpuMMX|CpuMMX2
6062 |CpuSSE|CpuSSE2;
6063 if (cpu_arch_tune_flags == 0)
6064 cpu_arch_tune_flags = Cpu186|Cpu286|Cpu386|Cpu486
6065 |Cpu586|Cpu686|CpuP4|CpuMMX|CpuMMX2
6066 |CpuSSE|CpuSSE2;
6067 }
6068 else if (!strcmp (default_arch, "i386"))
6069 {
6070 set_code_flag (CODE_32BIT);
6071 if (cpu_arch_isa_flags == 0)
6072 cpu_arch_isa_flags = Cpu186|Cpu286|Cpu386;
6073 if (cpu_arch_tune_flags == 0)
6074 cpu_arch_tune_flags = Cpu186|Cpu286|Cpu386;
6075 }
6076 else
6077 as_fatal (_("Unknown architecture"));
6078 switch (OUTPUT_FLAVOR)
6079 {
6080 #ifdef OBJ_MAYBE_AOUT
6081 case bfd_target_aout_flavour:
6082 return AOUT_TARGET_FORMAT;
6083 #endif
6084 #ifdef OBJ_MAYBE_COFF
6085 case bfd_target_coff_flavour:
6086 return "coff-i386";
6087 #endif
6088 #if defined (OBJ_MAYBE_ELF) || defined (OBJ_ELF)
6089 case bfd_target_elf_flavour:
6090 {
6091 if (flag_code == CODE_64BIT)
6092 {
6093 object_64bit = 1;
6094 use_rela_relocations = 1;
6095 }
6096 return flag_code == CODE_64BIT ? ELF_TARGET_FORMAT64 : ELF_TARGET_FORMAT;
6097 }
6098 #endif
6099 default:
6100 abort ();
6101 return NULL;
6102 }
6103 }
6104
6105 #endif /* OBJ_MAYBE_ more than one */
6106
6107 #if (defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF))
6108 void i386_elf_emit_arch_note ()
6109 {
6110 if (IS_ELF && cpu_arch_name != NULL)
6111 {
6112 char *p;
6113 asection *seg = now_seg;
6114 subsegT subseg = now_subseg;
6115 Elf_Internal_Note i_note;
6116 Elf_External_Note e_note;
6117 asection *note_secp;
6118 int len;
6119
6120 /* Create the .note section. */
6121 note_secp = subseg_new (".note", 0);
6122 bfd_set_section_flags (stdoutput,
6123 note_secp,
6124 SEC_HAS_CONTENTS | SEC_READONLY);
6125
6126 /* Process the arch string. */
6127 len = strlen (cpu_arch_name);
6128
6129 i_note.namesz = len + 1;
6130 i_note.descsz = 0;
6131 i_note.type = NT_ARCH;
6132 p = frag_more (sizeof (e_note.namesz));
6133 md_number_to_chars (p, (valueT) i_note.namesz, sizeof (e_note.namesz));
6134 p = frag_more (sizeof (e_note.descsz));
6135 md_number_to_chars (p, (valueT) i_note.descsz, sizeof (e_note.descsz));
6136 p = frag_more (sizeof (e_note.type));
6137 md_number_to_chars (p, (valueT) i_note.type, sizeof (e_note.type));
6138 p = frag_more (len + 1);
6139 strcpy (p, cpu_arch_name);
6140
6141 frag_align (2, 0, 0);
6142
6143 subseg_set (seg, subseg);
6144 }
6145 }
6146 #endif
6147 \f
6148 symbolS *
6149 md_undefined_symbol (name)
6150 char *name;
6151 {
6152 if (name[0] == GLOBAL_OFFSET_TABLE_NAME[0]
6153 && name[1] == GLOBAL_OFFSET_TABLE_NAME[1]
6154 && name[2] == GLOBAL_OFFSET_TABLE_NAME[2]
6155 && strcmp (name, GLOBAL_OFFSET_TABLE_NAME) == 0)
6156 {
6157 if (!GOT_symbol)
6158 {
6159 if (symbol_find (name))
6160 as_bad (_("GOT already in symbol table"));
6161 GOT_symbol = symbol_new (name, undefined_section,
6162 (valueT) 0, &zero_address_frag);
6163 };
6164 return GOT_symbol;
6165 }
6166 return 0;
6167 }
6168
6169 /* Round up a section size to the appropriate boundary. */
6170
6171 valueT
6172 md_section_align (segment, size)
6173 segT segment ATTRIBUTE_UNUSED;
6174 valueT size;
6175 {
6176 #if (defined (OBJ_AOUT) || defined (OBJ_MAYBE_AOUT))
6177 if (OUTPUT_FLAVOR == bfd_target_aout_flavour)
6178 {
6179 /* For a.out, force the section size to be aligned. If we don't do
6180 this, BFD will align it for us, but it will not write out the
6181 final bytes of the section. This may be a bug in BFD, but it is
6182 easier to fix it here since that is how the other a.out targets
6183 work. */
6184 int align;
6185
6186 align = bfd_get_section_alignment (stdoutput, segment);
6187 size = ((size + (1 << align) - 1) & ((valueT) -1 << align));
6188 }
6189 #endif
6190
6191 return size;
6192 }
6193
6194 /* On the i386, PC-relative offsets are relative to the start of the
6195 next instruction. That is, the address of the offset, plus its
6196 size, since the offset is always the last part of the insn. */
6197
6198 long
6199 md_pcrel_from (fixP)
6200 fixS *fixP;
6201 {
6202 return fixP->fx_size + fixP->fx_where + fixP->fx_frag->fr_address;
6203 }
6204
6205 #ifndef I386COFF
6206
6207 static void
6208 s_bss (ignore)
6209 int ignore ATTRIBUTE_UNUSED;
6210 {
6211 int temp;
6212
6213 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
6214 if (IS_ELF)
6215 obj_elf_section_change_hook ();
6216 #endif
6217 temp = get_absolute_expression ();
6218 subseg_set (bss_section, (subsegT) temp);
6219 demand_empty_rest_of_line ();
6220 }
6221
6222 #endif
6223
6224 void
6225 i386_validate_fix (fixp)
6226 fixS *fixp;
6227 {
6228 if (fixp->fx_subsy && fixp->fx_subsy == GOT_symbol)
6229 {
6230 if (fixp->fx_r_type == BFD_RELOC_32_PCREL)
6231 {
6232 if (!object_64bit)
6233 abort ();
6234 fixp->fx_r_type = BFD_RELOC_X86_64_GOTPCREL;
6235 }
6236 else
6237 {
6238 if (!object_64bit)
6239 fixp->fx_r_type = BFD_RELOC_386_GOTOFF;
6240 else
6241 fixp->fx_r_type = BFD_RELOC_X86_64_GOTOFF64;
6242 }
6243 fixp->fx_subsy = 0;
6244 }
6245 }
6246
6247 arelent *
6248 tc_gen_reloc (section, fixp)
6249 asection *section ATTRIBUTE_UNUSED;
6250 fixS *fixp;
6251 {
6252 arelent *rel;
6253 bfd_reloc_code_real_type code;
6254
6255 switch (fixp->fx_r_type)
6256 {
6257 case BFD_RELOC_X86_64_PLT32:
6258 case BFD_RELOC_X86_64_GOT32:
6259 case BFD_RELOC_X86_64_GOTPCREL:
6260 case BFD_RELOC_386_PLT32:
6261 case BFD_RELOC_386_GOT32:
6262 case BFD_RELOC_386_GOTOFF:
6263 case BFD_RELOC_386_GOTPC:
6264 case BFD_RELOC_386_TLS_GD:
6265 case BFD_RELOC_386_TLS_LDM:
6266 case BFD_RELOC_386_TLS_LDO_32:
6267 case BFD_RELOC_386_TLS_IE_32:
6268 case BFD_RELOC_386_TLS_IE:
6269 case BFD_RELOC_386_TLS_GOTIE:
6270 case BFD_RELOC_386_TLS_LE_32:
6271 case BFD_RELOC_386_TLS_LE:
6272 case BFD_RELOC_386_TLS_GOTDESC:
6273 case BFD_RELOC_386_TLS_DESC_CALL:
6274 case BFD_RELOC_X86_64_TLSGD:
6275 case BFD_RELOC_X86_64_TLSLD:
6276 case BFD_RELOC_X86_64_DTPOFF32:
6277 case BFD_RELOC_X86_64_DTPOFF64:
6278 case BFD_RELOC_X86_64_GOTTPOFF:
6279 case BFD_RELOC_X86_64_TPOFF32:
6280 case BFD_RELOC_X86_64_TPOFF64:
6281 case BFD_RELOC_X86_64_GOTOFF64:
6282 case BFD_RELOC_X86_64_GOTPC32:
6283 case BFD_RELOC_X86_64_GOT64:
6284 case BFD_RELOC_X86_64_GOTPCREL64:
6285 case BFD_RELOC_X86_64_GOTPC64:
6286 case BFD_RELOC_X86_64_GOTPLT64:
6287 case BFD_RELOC_X86_64_PLTOFF64:
6288 case BFD_RELOC_X86_64_GOTPC32_TLSDESC:
6289 case BFD_RELOC_X86_64_TLSDESC_CALL:
6290 case BFD_RELOC_RVA:
6291 case BFD_RELOC_VTABLE_ENTRY:
6292 case BFD_RELOC_VTABLE_INHERIT:
6293 #ifdef TE_PE
6294 case BFD_RELOC_32_SECREL:
6295 #endif
6296 code = fixp->fx_r_type;
6297 break;
6298 case BFD_RELOC_X86_64_32S:
6299 if (!fixp->fx_pcrel)
6300 {
6301 /* Don't turn BFD_RELOC_X86_64_32S into BFD_RELOC_32. */
6302 code = fixp->fx_r_type;
6303 break;
6304 }
6305 default:
6306 if (fixp->fx_pcrel)
6307 {
6308 switch (fixp->fx_size)
6309 {
6310 default:
6311 as_bad_where (fixp->fx_file, fixp->fx_line,
6312 _("can not do %d byte pc-relative relocation"),
6313 fixp->fx_size);
6314 code = BFD_RELOC_32_PCREL;
6315 break;
6316 case 1: code = BFD_RELOC_8_PCREL; break;
6317 case 2: code = BFD_RELOC_16_PCREL; break;
6318 case 4: code = BFD_RELOC_32_PCREL; break;
6319 #ifdef BFD64
6320 case 8: code = BFD_RELOC_64_PCREL; break;
6321 #endif
6322 }
6323 }
6324 else
6325 {
6326 switch (fixp->fx_size)
6327 {
6328 default:
6329 as_bad_where (fixp->fx_file, fixp->fx_line,
6330 _("can not do %d byte relocation"),
6331 fixp->fx_size);
6332 code = BFD_RELOC_32;
6333 break;
6334 case 1: code = BFD_RELOC_8; break;
6335 case 2: code = BFD_RELOC_16; break;
6336 case 4: code = BFD_RELOC_32; break;
6337 #ifdef BFD64
6338 case 8: code = BFD_RELOC_64; break;
6339 #endif
6340 }
6341 }
6342 break;
6343 }
6344
6345 if ((code == BFD_RELOC_32
6346 || code == BFD_RELOC_32_PCREL
6347 || code == BFD_RELOC_X86_64_32S)
6348 && GOT_symbol
6349 && fixp->fx_addsy == GOT_symbol)
6350 {
6351 if (!object_64bit)
6352 code = BFD_RELOC_386_GOTPC;
6353 else
6354 code = BFD_RELOC_X86_64_GOTPC32;
6355 }
6356 if ((code == BFD_RELOC_64 || code == BFD_RELOC_64_PCREL)
6357 && GOT_symbol
6358 && fixp->fx_addsy == GOT_symbol)
6359 {
6360 code = BFD_RELOC_X86_64_GOTPC64;
6361 }
6362
6363 rel = (arelent *) xmalloc (sizeof (arelent));
6364 rel->sym_ptr_ptr = (asymbol **) xmalloc (sizeof (asymbol *));
6365 *rel->sym_ptr_ptr = symbol_get_bfdsym (fixp->fx_addsy);
6366
6367 rel->address = fixp->fx_frag->fr_address + fixp->fx_where;
6368
6369 if (!use_rela_relocations)
6370 {
6371 /* HACK: Since i386 ELF uses Rel instead of Rela, encode the
6372 vtable entry to be used in the relocation's section offset. */
6373 if (fixp->fx_r_type == BFD_RELOC_VTABLE_ENTRY)
6374 rel->address = fixp->fx_offset;
6375
6376 rel->addend = 0;
6377 }
6378 /* Use the rela in 64bit mode. */
6379 else
6380 {
6381 if (!fixp->fx_pcrel)
6382 rel->addend = fixp->fx_offset;
6383 else
6384 switch (code)
6385 {
6386 case BFD_RELOC_X86_64_PLT32:
6387 case BFD_RELOC_X86_64_GOT32:
6388 case BFD_RELOC_X86_64_GOTPCREL:
6389 case BFD_RELOC_X86_64_TLSGD:
6390 case BFD_RELOC_X86_64_TLSLD:
6391 case BFD_RELOC_X86_64_GOTTPOFF:
6392 case BFD_RELOC_X86_64_GOTPC32_TLSDESC:
6393 case BFD_RELOC_X86_64_TLSDESC_CALL:
6394 rel->addend = fixp->fx_offset - fixp->fx_size;
6395 break;
6396 default:
6397 rel->addend = (section->vma
6398 - fixp->fx_size
6399 + fixp->fx_addnumber
6400 + md_pcrel_from (fixp));
6401 break;
6402 }
6403 }
6404
6405 rel->howto = bfd_reloc_type_lookup (stdoutput, code);
6406 if (rel->howto == NULL)
6407 {
6408 as_bad_where (fixp->fx_file, fixp->fx_line,
6409 _("cannot represent relocation type %s"),
6410 bfd_get_reloc_code_name (code));
6411 /* Set howto to a garbage value so that we can keep going. */
6412 rel->howto = bfd_reloc_type_lookup (stdoutput, BFD_RELOC_32);
6413 assert (rel->howto != NULL);
6414 }
6415
6416 return rel;
6417 }
6418
6419 \f
6420 /* Parse operands using Intel syntax. This implements a recursive descent
6421 parser based on the BNF grammar published in Appendix B of the MASM 6.1
6422 Programmer's Guide.
6423
6424 FIXME: We do not recognize the full operand grammar defined in the MASM
6425 documentation. In particular, all the structure/union and
6426 high-level macro operands are missing.
6427
6428 Uppercase words are terminals, lower case words are non-terminals.
6429 Objects surrounded by double brackets '[[' ']]' are optional. Vertical
6430 bars '|' denote choices. Most grammar productions are implemented in
6431 functions called 'intel_<production>'.
6432
6433 Initial production is 'expr'.
6434
6435 addOp + | -
6436
6437 alpha [a-zA-Z]
6438
6439 binOp & | AND | \| | OR | ^ | XOR
6440
6441 byteRegister AL | AH | BL | BH | CL | CH | DL | DH
6442
6443 constant digits [[ radixOverride ]]
6444
6445 dataType BYTE | WORD | DWORD | FWORD | QWORD | TBYTE | OWORD | XMMWORD
6446
6447 digits decdigit
6448 | digits decdigit
6449 | digits hexdigit
6450
6451 decdigit [0-9]
6452
6453 e04 e04 addOp e05
6454 | e05
6455
6456 e05 e05 binOp e06
6457 | e06
6458
6459 e06 e06 mulOp e09
6460 | e09
6461
6462 e09 OFFSET e10
6463 | SHORT e10
6464 | + e10
6465 | - e10
6466 | ~ e10
6467 | NOT e10
6468 | e09 PTR e10
6469 | e09 : e10
6470 | e10
6471
6472 e10 e10 [ expr ]
6473 | e11
6474
6475 e11 ( expr )
6476 | [ expr ]
6477 | constant
6478 | dataType
6479 | id
6480 | $
6481 | register
6482
6483 => expr expr cmpOp e04
6484 | e04
6485
6486 gpRegister AX | EAX | BX | EBX | CX | ECX | DX | EDX
6487 | BP | EBP | SP | ESP | DI | EDI | SI | ESI
6488
6489 hexdigit a | b | c | d | e | f
6490 | A | B | C | D | E | F
6491
6492 id alpha
6493 | id alpha
6494 | id decdigit
6495
6496 mulOp * | / | % | MOD | << | SHL | >> | SHR
6497
6498 quote " | '
6499
6500 register specialRegister
6501 | gpRegister
6502 | byteRegister
6503
6504 segmentRegister CS | DS | ES | FS | GS | SS
6505
6506 specialRegister CR0 | CR2 | CR3 | CR4
6507 | DR0 | DR1 | DR2 | DR3 | DR6 | DR7
6508 | TR3 | TR4 | TR5 | TR6 | TR7
6509
6510 We simplify the grammar in obvious places (e.g., register parsing is
6511 done by calling parse_register) and eliminate immediate left recursion
6512 to implement a recursive-descent parser.
6513
6514 expr e04 expr'
6515
6516 expr' cmpOp e04 expr'
6517 | Empty
6518
6519 e04 e05 e04'
6520
6521 e04' addOp e05 e04'
6522 | Empty
6523
6524 e05 e06 e05'
6525
6526 e05' binOp e06 e05'
6527 | Empty
6528
6529 e06 e09 e06'
6530
6531 e06' mulOp e09 e06'
6532 | Empty
6533
6534 e09 OFFSET e10 e09'
6535 | SHORT e10'
6536 | + e10'
6537 | - e10'
6538 | ~ e10'
6539 | NOT e10'
6540 | e10 e09'
6541
6542 e09' PTR e10 e09'
6543 | : e10 e09'
6544 | Empty
6545
6546 e10 e11 e10'
6547
6548 e10' [ expr ] e10'
6549 | Empty
6550
6551 e11 ( expr )
6552 | [ expr ]
6553 | BYTE
6554 | WORD
6555 | DWORD
6556 | FWORD
6557 | QWORD
6558 | TBYTE
6559 | OWORD
6560 | XMMWORD
6561 | .
6562 | $
6563 | register
6564 | id
6565 | constant */
6566
6567 /* Parsing structure for the intel syntax parser. Used to implement the
6568 semantic actions for the operand grammar. */
6569 struct intel_parser_s
6570 {
6571 char *op_string; /* The string being parsed. */
6572 int got_a_float; /* Whether the operand is a float. */
6573 int op_modifier; /* Operand modifier. */
6574 int is_mem; /* 1 if operand is memory reference. */
6575 int in_offset; /* >=1 if parsing operand of offset. */
6576 int in_bracket; /* >=1 if parsing operand in brackets. */
6577 const reg_entry *reg; /* Last register reference found. */
6578 char *disp; /* Displacement string being built. */
6579 char *next_operand; /* Resume point when splitting operands. */
6580 };
6581
6582 static struct intel_parser_s intel_parser;
6583
6584 /* Token structure for parsing intel syntax. */
6585 struct intel_token
6586 {
6587 int code; /* Token code. */
6588 const reg_entry *reg; /* Register entry for register tokens. */
6589 char *str; /* String representation. */
6590 };
6591
6592 static struct intel_token cur_token, prev_token;
6593
6594 /* Token codes for the intel parser. Since T_SHORT is already used
6595 by COFF, undefine it first to prevent a warning. */
6596 #define T_NIL -1
6597 #define T_CONST 1
6598 #define T_REG 2
6599 #define T_BYTE 3
6600 #define T_WORD 4
6601 #define T_DWORD 5
6602 #define T_FWORD 6
6603 #define T_QWORD 7
6604 #define T_TBYTE 8
6605 #define T_XMMWORD 9
6606 #undef T_SHORT
6607 #define T_SHORT 10
6608 #define T_OFFSET 11
6609 #define T_PTR 12
6610 #define T_ID 13
6611 #define T_SHL 14
6612 #define T_SHR 15
6613
6614 /* Prototypes for intel parser functions. */
6615 static int intel_match_token PARAMS ((int code));
6616 static void intel_get_token PARAMS ((void));
6617 static void intel_putback_token PARAMS ((void));
6618 static int intel_expr PARAMS ((void));
6619 static int intel_e04 PARAMS ((void));
6620 static int intel_e05 PARAMS ((void));
6621 static int intel_e06 PARAMS ((void));
6622 static int intel_e09 PARAMS ((void));
6623 static int intel_bracket_expr PARAMS ((void));
6624 static int intel_e10 PARAMS ((void));
6625 static int intel_e11 PARAMS ((void));
6626
6627 static int
6628 i386_intel_operand (operand_string, got_a_float)
6629 char *operand_string;
6630 int got_a_float;
6631 {
6632 int ret;
6633 char *p;
6634
6635 p = intel_parser.op_string = xstrdup (operand_string);
6636 intel_parser.disp = (char *) xmalloc (strlen (operand_string) + 1);
6637
6638 for (;;)
6639 {
6640 /* Initialize token holders. */
6641 cur_token.code = prev_token.code = T_NIL;
6642 cur_token.reg = prev_token.reg = NULL;
6643 cur_token.str = prev_token.str = NULL;
6644
6645 /* Initialize parser structure. */
6646 intel_parser.got_a_float = got_a_float;
6647 intel_parser.op_modifier = 0;
6648 intel_parser.is_mem = 0;
6649 intel_parser.in_offset = 0;
6650 intel_parser.in_bracket = 0;
6651 intel_parser.reg = NULL;
6652 intel_parser.disp[0] = '\0';
6653 intel_parser.next_operand = NULL;
6654
6655 /* Read the first token and start the parser. */
6656 intel_get_token ();
6657 ret = intel_expr ();
6658
6659 if (!ret)
6660 break;
6661
6662 if (cur_token.code != T_NIL)
6663 {
6664 as_bad (_("invalid operand for '%s' ('%s' unexpected)"),
6665 current_templates->start->name, cur_token.str);
6666 ret = 0;
6667 }
6668 /* If we found a memory reference, hand it over to i386_displacement
6669 to fill in the rest of the operand fields. */
6670 else if (intel_parser.is_mem)
6671 {
6672 if ((i.mem_operands == 1
6673 && (current_templates->start->opcode_modifier & IsString) == 0)
6674 || i.mem_operands == 2)
6675 {
6676 as_bad (_("too many memory references for '%s'"),
6677 current_templates->start->name);
6678 ret = 0;
6679 }
6680 else
6681 {
6682 char *s = intel_parser.disp;
6683 i.mem_operands++;
6684
6685 if (!quiet_warnings && intel_parser.is_mem < 0)
6686 /* See the comments in intel_bracket_expr. */
6687 as_warn (_("Treating `%s' as memory reference"), operand_string);
6688
6689 /* Add the displacement expression. */
6690 if (*s != '\0')
6691 ret = i386_displacement (s, s + strlen (s));
6692 if (ret)
6693 {
6694 /* Swap base and index in 16-bit memory operands like
6695 [si+bx]. Since i386_index_check is also used in AT&T
6696 mode we have to do that here. */
6697 if (i.base_reg
6698 && i.index_reg
6699 && (i.base_reg->reg_type & Reg16)
6700 && (i.index_reg->reg_type & Reg16)
6701 && i.base_reg->reg_num >= 6
6702 && i.index_reg->reg_num < 6)
6703 {
6704 const reg_entry *base = i.index_reg;
6705
6706 i.index_reg = i.base_reg;
6707 i.base_reg = base;
6708 }
6709 ret = i386_index_check (operand_string);
6710 }
6711 }
6712 }
6713
6714 /* Constant and OFFSET expressions are handled by i386_immediate. */
6715 else if ((intel_parser.op_modifier & (1 << T_OFFSET))
6716 || intel_parser.reg == NULL)
6717 ret = i386_immediate (intel_parser.disp);
6718
6719 if (intel_parser.next_operand && this_operand >= MAX_OPERANDS - 1)
6720 ret = 0;
6721 if (!ret || !intel_parser.next_operand)
6722 break;
6723 intel_parser.op_string = intel_parser.next_operand;
6724 this_operand = i.operands++;
6725 }
6726
6727 free (p);
6728 free (intel_parser.disp);
6729
6730 return ret;
6731 }
6732
6733 #define NUM_ADDRESS_REGS (!!i.base_reg + !!i.index_reg)
6734
6735 /* expr e04 expr'
6736
6737 expr' cmpOp e04 expr'
6738 | Empty */
6739 static int
6740 intel_expr ()
6741 {
6742 /* XXX Implement the comparison operators. */
6743 return intel_e04 ();
6744 }
6745
6746 /* e04 e05 e04'
6747
6748 e04' addOp e05 e04'
6749 | Empty */
6750 static int
6751 intel_e04 ()
6752 {
6753 int nregs = -1;
6754
6755 for (;;)
6756 {
6757 if (!intel_e05())
6758 return 0;
6759
6760 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
6761 i.base_reg = i386_regtab + REGNAM_AL; /* al is invalid as base */
6762
6763 if (cur_token.code == '+')
6764 nregs = -1;
6765 else if (cur_token.code == '-')
6766 nregs = NUM_ADDRESS_REGS;
6767 else
6768 return 1;
6769
6770 strcat (intel_parser.disp, cur_token.str);
6771 intel_match_token (cur_token.code);
6772 }
6773 }
6774
6775 /* e05 e06 e05'
6776
6777 e05' binOp e06 e05'
6778 | Empty */
6779 static int
6780 intel_e05 ()
6781 {
6782 int nregs = ~NUM_ADDRESS_REGS;
6783
6784 for (;;)
6785 {
6786 if (!intel_e06())
6787 return 0;
6788
6789 if (cur_token.code == '&' || cur_token.code == '|' || cur_token.code == '^')
6790 {
6791 char str[2];
6792
6793 str[0] = cur_token.code;
6794 str[1] = 0;
6795 strcat (intel_parser.disp, str);
6796 }
6797 else
6798 break;
6799
6800 intel_match_token (cur_token.code);
6801
6802 if (nregs < 0)
6803 nregs = ~nregs;
6804 }
6805 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
6806 i.base_reg = i386_regtab + REGNAM_AL + 1; /* cl is invalid as base */
6807 return 1;
6808 }
6809
6810 /* e06 e09 e06'
6811
6812 e06' mulOp e09 e06'
6813 | Empty */
6814 static int
6815 intel_e06 ()
6816 {
6817 int nregs = ~NUM_ADDRESS_REGS;
6818
6819 for (;;)
6820 {
6821 if (!intel_e09())
6822 return 0;
6823
6824 if (cur_token.code == '*' || cur_token.code == '/' || cur_token.code == '%')
6825 {
6826 char str[2];
6827
6828 str[0] = cur_token.code;
6829 str[1] = 0;
6830 strcat (intel_parser.disp, str);
6831 }
6832 else if (cur_token.code == T_SHL)
6833 strcat (intel_parser.disp, "<<");
6834 else if (cur_token.code == T_SHR)
6835 strcat (intel_parser.disp, ">>");
6836 else
6837 break;
6838
6839 intel_match_token (cur_token.code);
6840
6841 if (nregs < 0)
6842 nregs = ~nregs;
6843 }
6844 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
6845 i.base_reg = i386_regtab + REGNAM_AL + 2; /* dl is invalid as base */
6846 return 1;
6847 }
6848
6849 /* e09 OFFSET e09
6850 | SHORT e09
6851 | + e09
6852 | - e09
6853 | ~ e09
6854 | NOT e09
6855 | e10 e09'
6856
6857 e09' PTR e10 e09'
6858 | : e10 e09'
6859 | Empty */
6860 static int
6861 intel_e09 ()
6862 {
6863 int nregs = ~NUM_ADDRESS_REGS;
6864 int in_offset = 0;
6865
6866 for (;;)
6867 {
6868 /* Don't consume constants here. */
6869 if (cur_token.code == '+' || cur_token.code == '-')
6870 {
6871 /* Need to look one token ahead - if the next token
6872 is a constant, the current token is its sign. */
6873 int next_code;
6874
6875 intel_match_token (cur_token.code);
6876 next_code = cur_token.code;
6877 intel_putback_token ();
6878 if (next_code == T_CONST)
6879 break;
6880 }
6881
6882 /* e09 OFFSET e09 */
6883 if (cur_token.code == T_OFFSET)
6884 {
6885 if (!in_offset++)
6886 ++intel_parser.in_offset;
6887 }
6888
6889 /* e09 SHORT e09 */
6890 else if (cur_token.code == T_SHORT)
6891 intel_parser.op_modifier |= 1 << T_SHORT;
6892
6893 /* e09 + e09 */
6894 else if (cur_token.code == '+')
6895 strcat (intel_parser.disp, "+");
6896
6897 /* e09 - e09
6898 | ~ e09
6899 | NOT e09 */
6900 else if (cur_token.code == '-' || cur_token.code == '~')
6901 {
6902 char str[2];
6903
6904 if (nregs < 0)
6905 nregs = ~nregs;
6906 str[0] = cur_token.code;
6907 str[1] = 0;
6908 strcat (intel_parser.disp, str);
6909 }
6910
6911 /* e09 e10 e09' */
6912 else
6913 break;
6914
6915 intel_match_token (cur_token.code);
6916 }
6917
6918 for (;;)
6919 {
6920 if (!intel_e10 ())
6921 return 0;
6922
6923 /* e09' PTR e10 e09' */
6924 if (cur_token.code == T_PTR)
6925 {
6926 char suffix;
6927
6928 if (prev_token.code == T_BYTE)
6929 suffix = BYTE_MNEM_SUFFIX;
6930
6931 else if (prev_token.code == T_WORD)
6932 {
6933 if (current_templates->start->name[0] == 'l'
6934 && current_templates->start->name[2] == 's'
6935 && current_templates->start->name[3] == 0)
6936 suffix = BYTE_MNEM_SUFFIX; /* so it will cause an error */
6937 else if (intel_parser.got_a_float == 2) /* "fi..." */
6938 suffix = SHORT_MNEM_SUFFIX;
6939 else
6940 suffix = WORD_MNEM_SUFFIX;
6941 }
6942
6943 else if (prev_token.code == T_DWORD)
6944 {
6945 if (current_templates->start->name[0] == 'l'
6946 && current_templates->start->name[2] == 's'
6947 && current_templates->start->name[3] == 0)
6948 suffix = WORD_MNEM_SUFFIX;
6949 else if (flag_code == CODE_16BIT
6950 && (current_templates->start->opcode_modifier
6951 & (Jump | JumpDword)))
6952 suffix = LONG_DOUBLE_MNEM_SUFFIX;
6953 else if (intel_parser.got_a_float == 1) /* "f..." */
6954 suffix = SHORT_MNEM_SUFFIX;
6955 else
6956 suffix = LONG_MNEM_SUFFIX;
6957 }
6958
6959 else if (prev_token.code == T_FWORD)
6960 {
6961 if (current_templates->start->name[0] == 'l'
6962 && current_templates->start->name[2] == 's'
6963 && current_templates->start->name[3] == 0)
6964 suffix = LONG_MNEM_SUFFIX;
6965 else if (!intel_parser.got_a_float)
6966 {
6967 if (flag_code == CODE_16BIT)
6968 add_prefix (DATA_PREFIX_OPCODE);
6969 suffix = LONG_DOUBLE_MNEM_SUFFIX;
6970 }
6971 else
6972 suffix = BYTE_MNEM_SUFFIX; /* so it will cause an error */
6973 }
6974
6975 else if (prev_token.code == T_QWORD)
6976 {
6977 if (intel_parser.got_a_float == 1) /* "f..." */
6978 suffix = LONG_MNEM_SUFFIX;
6979 else
6980 suffix = QWORD_MNEM_SUFFIX;
6981 }
6982
6983 else if (prev_token.code == T_TBYTE)
6984 {
6985 if (intel_parser.got_a_float == 1)
6986 suffix = LONG_DOUBLE_MNEM_SUFFIX;
6987 else
6988 suffix = BYTE_MNEM_SUFFIX; /* so it will cause an error */
6989 }
6990
6991 else if (prev_token.code == T_XMMWORD)
6992 {
6993 /* XXX ignored for now, but accepted since gcc uses it */
6994 suffix = 0;
6995 }
6996
6997 else
6998 {
6999 as_bad (_("Unknown operand modifier `%s'"), prev_token.str);
7000 return 0;
7001 }
7002
7003 /* Operands for jump/call using 'ptr' notation denote absolute
7004 addresses. */
7005 if (current_templates->start->opcode_modifier & (Jump | JumpDword))
7006 i.types[this_operand] |= JumpAbsolute;
7007
7008 if (current_templates->start->base_opcode == 0x8d /* lea */)
7009 ;
7010 else if (!i.suffix)
7011 i.suffix = suffix;
7012 else if (i.suffix != suffix)
7013 {
7014 as_bad (_("Conflicting operand modifiers"));
7015 return 0;
7016 }
7017
7018 }
7019
7020 /* e09' : e10 e09' */
7021 else if (cur_token.code == ':')
7022 {
7023 if (prev_token.code != T_REG)
7024 {
7025 /* While {call,jmp} SSSS:OOOO is MASM syntax only when SSSS is a
7026 segment/group identifier (which we don't have), using comma
7027 as the operand separator there is even less consistent, since
7028 there all branches only have a single operand. */
7029 if (this_operand != 0
7030 || intel_parser.in_offset
7031 || intel_parser.in_bracket
7032 || (!(current_templates->start->opcode_modifier
7033 & (Jump|JumpDword|JumpInterSegment))
7034 && !(current_templates->start->operand_types[0]
7035 & JumpAbsolute)))
7036 return intel_match_token (T_NIL);
7037 /* Remember the start of the 2nd operand and terminate 1st
7038 operand here.
7039 XXX This isn't right, yet (when SSSS:OOOO is right operand of
7040 another expression), but it gets at least the simplest case
7041 (a plain number or symbol on the left side) right. */
7042 intel_parser.next_operand = intel_parser.op_string;
7043 *--intel_parser.op_string = '\0';
7044 return intel_match_token (':');
7045 }
7046 }
7047
7048 /* e09' Empty */
7049 else
7050 break;
7051
7052 intel_match_token (cur_token.code);
7053
7054 }
7055
7056 if (in_offset)
7057 {
7058 --intel_parser.in_offset;
7059 if (nregs < 0)
7060 nregs = ~nregs;
7061 if (NUM_ADDRESS_REGS > nregs)
7062 {
7063 as_bad (_("Invalid operand to `OFFSET'"));
7064 return 0;
7065 }
7066 intel_parser.op_modifier |= 1 << T_OFFSET;
7067 }
7068
7069 if (nregs >= 0 && NUM_ADDRESS_REGS > nregs)
7070 i.base_reg = i386_regtab + REGNAM_AL + 3; /* bl is invalid as base */
7071 return 1;
7072 }
7073
7074 static int
7075 intel_bracket_expr ()
7076 {
7077 int was_offset = intel_parser.op_modifier & (1 << T_OFFSET);
7078 const char *start = intel_parser.op_string;
7079 int len;
7080
7081 if (i.op[this_operand].regs)
7082 return intel_match_token (T_NIL);
7083
7084 intel_match_token ('[');
7085
7086 /* Mark as a memory operand only if it's not already known to be an
7087 offset expression. If it's an offset expression, we need to keep
7088 the brace in. */
7089 if (!intel_parser.in_offset)
7090 {
7091 ++intel_parser.in_bracket;
7092
7093 /* Operands for jump/call inside brackets denote absolute addresses. */
7094 if (current_templates->start->opcode_modifier & (Jump | JumpDword))
7095 i.types[this_operand] |= JumpAbsolute;
7096
7097 /* Unfortunately gas always diverged from MASM in a respect that can't
7098 be easily fixed without risking to break code sequences likely to be
7099 encountered (the testsuite even check for this): MASM doesn't consider
7100 an expression inside brackets unconditionally as a memory reference.
7101 When that is e.g. a constant, an offset expression, or the sum of the
7102 two, this is still taken as a constant load. gas, however, always
7103 treated these as memory references. As a compromise, we'll try to make
7104 offset expressions inside brackets work the MASM way (since that's
7105 less likely to be found in real world code), but make constants alone
7106 continue to work the traditional gas way. In either case, issue a
7107 warning. */
7108 intel_parser.op_modifier &= ~was_offset;
7109 }
7110 else
7111 strcat (intel_parser.disp, "[");
7112
7113 /* Add a '+' to the displacement string if necessary. */
7114 if (*intel_parser.disp != '\0'
7115 && *(intel_parser.disp + strlen (intel_parser.disp) - 1) != '+')
7116 strcat (intel_parser.disp, "+");
7117
7118 if (intel_expr ()
7119 && (len = intel_parser.op_string - start - 1,
7120 intel_match_token (']')))
7121 {
7122 /* Preserve brackets when the operand is an offset expression. */
7123 if (intel_parser.in_offset)
7124 strcat (intel_parser.disp, "]");
7125 else
7126 {
7127 --intel_parser.in_bracket;
7128 if (i.base_reg || i.index_reg)
7129 intel_parser.is_mem = 1;
7130 if (!intel_parser.is_mem)
7131 {
7132 if (!(intel_parser.op_modifier & (1 << T_OFFSET)))
7133 /* Defer the warning until all of the operand was parsed. */
7134 intel_parser.is_mem = -1;
7135 else if (!quiet_warnings)
7136 as_warn (_("`[%.*s]' taken to mean just `%.*s'"), len, start, len, start);
7137 }
7138 }
7139 intel_parser.op_modifier |= was_offset;
7140
7141 return 1;
7142 }
7143 return 0;
7144 }
7145
7146 /* e10 e11 e10'
7147
7148 e10' [ expr ] e10'
7149 | Empty */
7150 static int
7151 intel_e10 ()
7152 {
7153 if (!intel_e11 ())
7154 return 0;
7155
7156 while (cur_token.code == '[')
7157 {
7158 if (!intel_bracket_expr ())
7159 return 0;
7160 }
7161
7162 return 1;
7163 }
7164
7165 /* e11 ( expr )
7166 | [ expr ]
7167 | BYTE
7168 | WORD
7169 | DWORD
7170 | FWORD
7171 | QWORD
7172 | TBYTE
7173 | OWORD
7174 | XMMWORD
7175 | $
7176 | .
7177 | register
7178 | id
7179 | constant */
7180 static int
7181 intel_e11 ()
7182 {
7183 switch (cur_token.code)
7184 {
7185 /* e11 ( expr ) */
7186 case '(':
7187 intel_match_token ('(');
7188 strcat (intel_parser.disp, "(");
7189
7190 if (intel_expr () && intel_match_token (')'))
7191 {
7192 strcat (intel_parser.disp, ")");
7193 return 1;
7194 }
7195 return 0;
7196
7197 /* e11 [ expr ] */
7198 case '[':
7199 return intel_bracket_expr ();
7200
7201 /* e11 $
7202 | . */
7203 case '.':
7204 strcat (intel_parser.disp, cur_token.str);
7205 intel_match_token (cur_token.code);
7206
7207 /* Mark as a memory operand only if it's not already known to be an
7208 offset expression. */
7209 if (!intel_parser.in_offset)
7210 intel_parser.is_mem = 1;
7211
7212 return 1;
7213
7214 /* e11 register */
7215 case T_REG:
7216 {
7217 const reg_entry *reg = intel_parser.reg = cur_token.reg;
7218
7219 intel_match_token (T_REG);
7220
7221 /* Check for segment change. */
7222 if (cur_token.code == ':')
7223 {
7224 if (!(reg->reg_type & (SReg2 | SReg3)))
7225 {
7226 as_bad (_("`%s' is not a valid segment register"), reg->reg_name);
7227 return 0;
7228 }
7229 else if (i.seg[i.mem_operands])
7230 as_warn (_("Extra segment override ignored"));
7231 else
7232 {
7233 if (!intel_parser.in_offset)
7234 intel_parser.is_mem = 1;
7235 switch (reg->reg_num)
7236 {
7237 case 0:
7238 i.seg[i.mem_operands] = &es;
7239 break;
7240 case 1:
7241 i.seg[i.mem_operands] = &cs;
7242 break;
7243 case 2:
7244 i.seg[i.mem_operands] = &ss;
7245 break;
7246 case 3:
7247 i.seg[i.mem_operands] = &ds;
7248 break;
7249 case 4:
7250 i.seg[i.mem_operands] = &fs;
7251 break;
7252 case 5:
7253 i.seg[i.mem_operands] = &gs;
7254 break;
7255 }
7256 }
7257 }
7258
7259 /* Not a segment register. Check for register scaling. */
7260 else if (cur_token.code == '*')
7261 {
7262 if (!intel_parser.in_bracket)
7263 {
7264 as_bad (_("Register scaling only allowed in memory operands"));
7265 return 0;
7266 }
7267
7268 if (reg->reg_type & Reg16) /* Disallow things like [si*1]. */
7269 reg = i386_regtab + REGNAM_AX + 4; /* sp is invalid as index */
7270 else if (i.index_reg)
7271 reg = i386_regtab + REGNAM_EAX + 4; /* esp is invalid as index */
7272
7273 /* What follows must be a valid scale. */
7274 intel_match_token ('*');
7275 i.index_reg = reg;
7276 i.types[this_operand] |= BaseIndex;
7277
7278 /* Set the scale after setting the register (otherwise,
7279 i386_scale will complain) */
7280 if (cur_token.code == '+' || cur_token.code == '-')
7281 {
7282 char *str, sign = cur_token.code;
7283 intel_match_token (cur_token.code);
7284 if (cur_token.code != T_CONST)
7285 {
7286 as_bad (_("Syntax error: Expecting a constant, got `%s'"),
7287 cur_token.str);
7288 return 0;
7289 }
7290 str = (char *) xmalloc (strlen (cur_token.str) + 2);
7291 strcpy (str + 1, cur_token.str);
7292 *str = sign;
7293 if (!i386_scale (str))
7294 return 0;
7295 free (str);
7296 }
7297 else if (!i386_scale (cur_token.str))
7298 return 0;
7299 intel_match_token (cur_token.code);
7300 }
7301
7302 /* No scaling. If this is a memory operand, the register is either a
7303 base register (first occurrence) or an index register (second
7304 occurrence). */
7305 else if (intel_parser.in_bracket)
7306 {
7307
7308 if (!i.base_reg)
7309 i.base_reg = reg;
7310 else if (!i.index_reg)
7311 i.index_reg = reg;
7312 else
7313 {
7314 as_bad (_("Too many register references in memory operand"));
7315 return 0;
7316 }
7317
7318 i.types[this_operand] |= BaseIndex;
7319 }
7320
7321 /* It's neither base nor index. */
7322 else if (!intel_parser.in_offset && !intel_parser.is_mem)
7323 {
7324 i.types[this_operand] |= reg->reg_type & ~BaseIndex;
7325 i.op[this_operand].regs = reg;
7326 i.reg_operands++;
7327 }
7328 else
7329 {
7330 as_bad (_("Invalid use of register"));
7331 return 0;
7332 }
7333
7334 /* Since registers are not part of the displacement string (except
7335 when we're parsing offset operands), we may need to remove any
7336 preceding '+' from the displacement string. */
7337 if (*intel_parser.disp != '\0'
7338 && !intel_parser.in_offset)
7339 {
7340 char *s = intel_parser.disp;
7341 s += strlen (s) - 1;
7342 if (*s == '+')
7343 *s = '\0';
7344 }
7345
7346 return 1;
7347 }
7348
7349 /* e11 BYTE
7350 | WORD
7351 | DWORD
7352 | FWORD
7353 | QWORD
7354 | TBYTE
7355 | OWORD
7356 | XMMWORD */
7357 case T_BYTE:
7358 case T_WORD:
7359 case T_DWORD:
7360 case T_FWORD:
7361 case T_QWORD:
7362 case T_TBYTE:
7363 case T_XMMWORD:
7364 intel_match_token (cur_token.code);
7365
7366 if (cur_token.code == T_PTR)
7367 return 1;
7368
7369 /* It must have been an identifier. */
7370 intel_putback_token ();
7371 cur_token.code = T_ID;
7372 /* FALLTHRU */
7373
7374 /* e11 id
7375 | constant */
7376 case T_ID:
7377 if (!intel_parser.in_offset && intel_parser.is_mem <= 0)
7378 {
7379 symbolS *symbolP;
7380
7381 /* The identifier represents a memory reference only if it's not
7382 preceded by an offset modifier and if it's not an equate. */
7383 symbolP = symbol_find(cur_token.str);
7384 if (!symbolP || S_GET_SEGMENT(symbolP) != absolute_section)
7385 intel_parser.is_mem = 1;
7386 }
7387 /* FALLTHRU */
7388
7389 case T_CONST:
7390 case '-':
7391 case '+':
7392 {
7393 char *save_str, sign = 0;
7394
7395 /* Allow constants that start with `+' or `-'. */
7396 if (cur_token.code == '-' || cur_token.code == '+')
7397 {
7398 sign = cur_token.code;
7399 intel_match_token (cur_token.code);
7400 if (cur_token.code != T_CONST)
7401 {
7402 as_bad (_("Syntax error: Expecting a constant, got `%s'"),
7403 cur_token.str);
7404 return 0;
7405 }
7406 }
7407
7408 save_str = (char *) xmalloc (strlen (cur_token.str) + 2);
7409 strcpy (save_str + !!sign, cur_token.str);
7410 if (sign)
7411 *save_str = sign;
7412
7413 /* Get the next token to check for register scaling. */
7414 intel_match_token (cur_token.code);
7415
7416 /* Check if this constant is a scaling factor for an index register. */
7417 if (cur_token.code == '*')
7418 {
7419 if (intel_match_token ('*') && cur_token.code == T_REG)
7420 {
7421 const reg_entry *reg = cur_token.reg;
7422
7423 if (!intel_parser.in_bracket)
7424 {
7425 as_bad (_("Register scaling only allowed in memory operands"));
7426 return 0;
7427 }
7428
7429 if (reg->reg_type & Reg16) /* Disallow things like [1*si]. */
7430 reg = i386_regtab + REGNAM_AX + 4; /* sp is invalid as index */
7431 else if (i.index_reg)
7432 reg = i386_regtab + REGNAM_EAX + 4; /* esp is invalid as index */
7433
7434 /* The constant is followed by `* reg', so it must be
7435 a valid scale. */
7436 i.index_reg = reg;
7437 i.types[this_operand] |= BaseIndex;
7438
7439 /* Set the scale after setting the register (otherwise,
7440 i386_scale will complain) */
7441 if (!i386_scale (save_str))
7442 return 0;
7443 intel_match_token (T_REG);
7444
7445 /* Since registers are not part of the displacement
7446 string, we may need to remove any preceding '+' from
7447 the displacement string. */
7448 if (*intel_parser.disp != '\0')
7449 {
7450 char *s = intel_parser.disp;
7451 s += strlen (s) - 1;
7452 if (*s == '+')
7453 *s = '\0';
7454 }
7455
7456 free (save_str);
7457
7458 return 1;
7459 }
7460
7461 /* The constant was not used for register scaling. Since we have
7462 already consumed the token following `*' we now need to put it
7463 back in the stream. */
7464 intel_putback_token ();
7465 }
7466
7467 /* Add the constant to the displacement string. */
7468 strcat (intel_parser.disp, save_str);
7469 free (save_str);
7470
7471 return 1;
7472 }
7473 }
7474
7475 as_bad (_("Unrecognized token '%s'"), cur_token.str);
7476 return 0;
7477 }
7478
7479 /* Match the given token against cur_token. If they match, read the next
7480 token from the operand string. */
7481 static int
7482 intel_match_token (code)
7483 int code;
7484 {
7485 if (cur_token.code == code)
7486 {
7487 intel_get_token ();
7488 return 1;
7489 }
7490 else
7491 {
7492 as_bad (_("Unexpected token `%s'"), cur_token.str);
7493 return 0;
7494 }
7495 }
7496
7497 /* Read a new token from intel_parser.op_string and store it in cur_token. */
7498 static void
7499 intel_get_token ()
7500 {
7501 char *end_op;
7502 const reg_entry *reg;
7503 struct intel_token new_token;
7504
7505 new_token.code = T_NIL;
7506 new_token.reg = NULL;
7507 new_token.str = NULL;
7508
7509 /* Free the memory allocated to the previous token and move
7510 cur_token to prev_token. */
7511 if (prev_token.str)
7512 free (prev_token.str);
7513
7514 prev_token = cur_token;
7515
7516 /* Skip whitespace. */
7517 while (is_space_char (*intel_parser.op_string))
7518 intel_parser.op_string++;
7519
7520 /* Return an empty token if we find nothing else on the line. */
7521 if (*intel_parser.op_string == '\0')
7522 {
7523 cur_token = new_token;
7524 return;
7525 }
7526
7527 /* The new token cannot be larger than the remainder of the operand
7528 string. */
7529 new_token.str = (char *) xmalloc (strlen (intel_parser.op_string) + 1);
7530 new_token.str[0] = '\0';
7531
7532 if (strchr ("0123456789", *intel_parser.op_string))
7533 {
7534 char *p = new_token.str;
7535 char *q = intel_parser.op_string;
7536 new_token.code = T_CONST;
7537
7538 /* Allow any kind of identifier char to encompass floating point and
7539 hexadecimal numbers. */
7540 while (is_identifier_char (*q))
7541 *p++ = *q++;
7542 *p = '\0';
7543
7544 /* Recognize special symbol names [0-9][bf]. */
7545 if (strlen (intel_parser.op_string) == 2
7546 && (intel_parser.op_string[1] == 'b'
7547 || intel_parser.op_string[1] == 'f'))
7548 new_token.code = T_ID;
7549 }
7550
7551 else if ((reg = parse_register (intel_parser.op_string, &end_op)) != NULL)
7552 {
7553 size_t len = end_op - intel_parser.op_string;
7554
7555 new_token.code = T_REG;
7556 new_token.reg = reg;
7557
7558 memcpy (new_token.str, intel_parser.op_string, len);
7559 new_token.str[len] = '\0';
7560 }
7561
7562 else if (is_identifier_char (*intel_parser.op_string))
7563 {
7564 char *p = new_token.str;
7565 char *q = intel_parser.op_string;
7566
7567 /* A '.' or '$' followed by an identifier char is an identifier.
7568 Otherwise, it's operator '.' followed by an expression. */
7569 if ((*q == '.' || *q == '$') && !is_identifier_char (*(q + 1)))
7570 {
7571 new_token.code = '.';
7572 new_token.str[0] = '.';
7573 new_token.str[1] = '\0';
7574 }
7575 else
7576 {
7577 while (is_identifier_char (*q) || *q == '@')
7578 *p++ = *q++;
7579 *p = '\0';
7580
7581 if (strcasecmp (new_token.str, "NOT") == 0)
7582 new_token.code = '~';
7583
7584 else if (strcasecmp (new_token.str, "MOD") == 0)
7585 new_token.code = '%';
7586
7587 else if (strcasecmp (new_token.str, "AND") == 0)
7588 new_token.code = '&';
7589
7590 else if (strcasecmp (new_token.str, "OR") == 0)
7591 new_token.code = '|';
7592
7593 else if (strcasecmp (new_token.str, "XOR") == 0)
7594 new_token.code = '^';
7595
7596 else if (strcasecmp (new_token.str, "SHL") == 0)
7597 new_token.code = T_SHL;
7598
7599 else if (strcasecmp (new_token.str, "SHR") == 0)
7600 new_token.code = T_SHR;
7601
7602 else if (strcasecmp (new_token.str, "BYTE") == 0)
7603 new_token.code = T_BYTE;
7604
7605 else if (strcasecmp (new_token.str, "WORD") == 0)
7606 new_token.code = T_WORD;
7607
7608 else if (strcasecmp (new_token.str, "DWORD") == 0)
7609 new_token.code = T_DWORD;
7610
7611 else if (strcasecmp (new_token.str, "FWORD") == 0)
7612 new_token.code = T_FWORD;
7613
7614 else if (strcasecmp (new_token.str, "QWORD") == 0)
7615 new_token.code = T_QWORD;
7616
7617 else if (strcasecmp (new_token.str, "TBYTE") == 0
7618 /* XXX remove (gcc still uses it) */
7619 || strcasecmp (new_token.str, "XWORD") == 0)
7620 new_token.code = T_TBYTE;
7621
7622 else if (strcasecmp (new_token.str, "XMMWORD") == 0
7623 || strcasecmp (new_token.str, "OWORD") == 0)
7624 new_token.code = T_XMMWORD;
7625
7626 else if (strcasecmp (new_token.str, "PTR") == 0)
7627 new_token.code = T_PTR;
7628
7629 else if (strcasecmp (new_token.str, "SHORT") == 0)
7630 new_token.code = T_SHORT;
7631
7632 else if (strcasecmp (new_token.str, "OFFSET") == 0)
7633 {
7634 new_token.code = T_OFFSET;
7635
7636 /* ??? This is not mentioned in the MASM grammar but gcc
7637 makes use of it with -mintel-syntax. OFFSET may be
7638 followed by FLAT: */
7639 if (strncasecmp (q, " FLAT:", 6) == 0)
7640 strcat (new_token.str, " FLAT:");
7641 }
7642
7643 /* ??? This is not mentioned in the MASM grammar. */
7644 else if (strcasecmp (new_token.str, "FLAT") == 0)
7645 {
7646 new_token.code = T_OFFSET;
7647 if (*q == ':')
7648 strcat (new_token.str, ":");
7649 else
7650 as_bad (_("`:' expected"));
7651 }
7652
7653 else
7654 new_token.code = T_ID;
7655 }
7656 }
7657
7658 else if (strchr ("+-/*%|&^:[]()~", *intel_parser.op_string))
7659 {
7660 new_token.code = *intel_parser.op_string;
7661 new_token.str[0] = *intel_parser.op_string;
7662 new_token.str[1] = '\0';
7663 }
7664
7665 else if (strchr ("<>", *intel_parser.op_string)
7666 && *intel_parser.op_string == *(intel_parser.op_string + 1))
7667 {
7668 new_token.code = *intel_parser.op_string == '<' ? T_SHL : T_SHR;
7669 new_token.str[0] = *intel_parser.op_string;
7670 new_token.str[1] = *intel_parser.op_string;
7671 new_token.str[2] = '\0';
7672 }
7673
7674 else
7675 as_bad (_("Unrecognized token `%s'"), intel_parser.op_string);
7676
7677 intel_parser.op_string += strlen (new_token.str);
7678 cur_token = new_token;
7679 }
7680
7681 /* Put cur_token back into the token stream and make cur_token point to
7682 prev_token. */
7683 static void
7684 intel_putback_token ()
7685 {
7686 if (cur_token.code != T_NIL)
7687 {
7688 intel_parser.op_string -= strlen (cur_token.str);
7689 free (cur_token.str);
7690 }
7691 cur_token = prev_token;
7692
7693 /* Forget prev_token. */
7694 prev_token.code = T_NIL;
7695 prev_token.reg = NULL;
7696 prev_token.str = NULL;
7697 }
7698
7699 int
7700 tc_x86_regname_to_dw2regnum (char *regname)
7701 {
7702 unsigned int regnum;
7703 unsigned int regnames_count;
7704 static const char *const regnames_32[] =
7705 {
7706 "eax", "ecx", "edx", "ebx",
7707 "esp", "ebp", "esi", "edi",
7708 "eip", "eflags", NULL,
7709 "st0", "st1", "st2", "st3",
7710 "st4", "st5", "st6", "st7",
7711 NULL, NULL,
7712 "xmm0", "xmm1", "xmm2", "xmm3",
7713 "xmm4", "xmm5", "xmm6", "xmm7",
7714 "mm0", "mm1", "mm2", "mm3",
7715 "mm4", "mm5", "mm6", "mm7",
7716 "fcw", "fsw", "mxcsr",
7717 "es", "cs", "ss", "ds", "fs", "gs", NULL, NULL,
7718 "tr", "ldtr"
7719 };
7720 static const char *const regnames_64[] =
7721 {
7722 "rax", "rdx", "rcx", "rbx",
7723 "rsi", "rdi", "rbp", "rsp",
7724 "r8", "r9", "r10", "r11",
7725 "r12", "r13", "r14", "r15",
7726 "rip",
7727 "xmm0", "xmm1", "xmm2", "xmm3",
7728 "xmm4", "xmm5", "xmm6", "xmm7",
7729 "xmm8", "xmm9", "xmm10", "xmm11",
7730 "xmm12", "xmm13", "xmm14", "xmm15",
7731 "st0", "st1", "st2", "st3",
7732 "st4", "st5", "st6", "st7",
7733 "mm0", "mm1", "mm2", "mm3",
7734 "mm4", "mm5", "mm6", "mm7",
7735 "rflags",
7736 "es", "cs", "ss", "ds", "fs", "gs", NULL, NULL,
7737 "fs.base", "gs.base", NULL, NULL,
7738 "tr", "ldtr",
7739 "mxcsr", "fcw", "fsw"
7740 };
7741 const char *const *regnames;
7742
7743 if (flag_code == CODE_64BIT)
7744 {
7745 regnames = regnames_64;
7746 regnames_count = ARRAY_SIZE (regnames_64);
7747 }
7748 else
7749 {
7750 regnames = regnames_32;
7751 regnames_count = ARRAY_SIZE (regnames_32);
7752 }
7753
7754 for (regnum = 0; regnum < regnames_count; regnum++)
7755 if (regnames[regnum] != NULL
7756 && strcmp (regname, regnames[regnum]) == 0)
7757 return regnum;
7758
7759 return -1;
7760 }
7761
7762 void
7763 tc_x86_frame_initial_instructions (void)
7764 {
7765 static unsigned int sp_regno;
7766
7767 if (!sp_regno)
7768 sp_regno = tc_x86_regname_to_dw2regnum (flag_code == CODE_64BIT
7769 ? "rsp" : "esp");
7770
7771 cfi_add_CFA_def_cfa (sp_regno, -x86_cie_data_alignment);
7772 cfi_add_CFA_offset (x86_dwarf2_return_column, x86_cie_data_alignment);
7773 }
7774
7775 int
7776 i386_elf_section_type (const char *str, size_t len)
7777 {
7778 if (flag_code == CODE_64BIT
7779 && len == sizeof ("unwind") - 1
7780 && strncmp (str, "unwind", 6) == 0)
7781 return SHT_X86_64_UNWIND;
7782
7783 return -1;
7784 }
7785
7786 #ifdef TE_PE
7787 void
7788 tc_pe_dwarf2_emit_offset (symbolS *symbol, unsigned int size)
7789 {
7790 expressionS expr;
7791
7792 expr.X_op = O_secrel;
7793 expr.X_add_symbol = symbol;
7794 expr.X_add_number = 0;
7795 emit_expr (&expr, size);
7796 }
7797 #endif
7798
7799 #if defined (OBJ_ELF) || defined (OBJ_MAYBE_ELF)
7800 /* For ELF on x86-64, add support for SHF_X86_64_LARGE. */
7801
7802 int
7803 x86_64_section_letter (int letter, char **ptr_msg)
7804 {
7805 if (flag_code == CODE_64BIT)
7806 {
7807 if (letter == 'l')
7808 return SHF_X86_64_LARGE;
7809
7810 *ptr_msg = _("Bad .section directive: want a,l,w,x,M,S,G,T in string");
7811 }
7812 else
7813 *ptr_msg = _("Bad .section directive: want a,w,x,M,S,G,T in string");
7814 return -1;
7815 }
7816
7817 int
7818 x86_64_section_word (char *str, size_t len)
7819 {
7820 if (len == 5 && flag_code == CODE_64BIT && CONST_STRNEQ (str, "large"))
7821 return SHF_X86_64_LARGE;
7822
7823 return -1;
7824 }
7825
7826 static void
7827 handle_large_common (int small ATTRIBUTE_UNUSED)
7828 {
7829 if (flag_code != CODE_64BIT)
7830 {
7831 s_comm_internal (0, elf_common_parse);
7832 as_warn (_(".largecomm supported only in 64bit mode, producing .comm"));
7833 }
7834 else
7835 {
7836 static segT lbss_section;
7837 asection *saved_com_section_ptr = elf_com_section_ptr;
7838 asection *saved_bss_section = bss_section;
7839
7840 if (lbss_section == NULL)
7841 {
7842 flagword applicable;
7843 segT seg = now_seg;
7844 subsegT subseg = now_subseg;
7845
7846 /* The .lbss section is for local .largecomm symbols. */
7847 lbss_section = subseg_new (".lbss", 0);
7848 applicable = bfd_applicable_section_flags (stdoutput);
7849 bfd_set_section_flags (stdoutput, lbss_section,
7850 applicable & SEC_ALLOC);
7851 seg_info (lbss_section)->bss = 1;
7852
7853 subseg_set (seg, subseg);
7854 }
7855
7856 elf_com_section_ptr = &_bfd_elf_large_com_section;
7857 bss_section = lbss_section;
7858
7859 s_comm_internal (0, elf_common_parse);
7860
7861 elf_com_section_ptr = saved_com_section_ptr;
7862 bss_section = saved_bss_section;
7863 }
7864 }
7865 #endif /* OBJ_ELF || OBJ_MAYBE_ELF */
This page took 0.199702 seconds and 4 git commands to generate.