nasm.h

   1 /* nasm.h   main header file for the Netwide Assembler: inter-module interface
   2  *
   3  * The Netwide Assembler is copyright (C) 1996 Simon Tatham and
   4  * Julian Hall. All rights reserved. The software is
   5  * redistributable under the license given in the file "LICENSE"
   6  * distributed in the NASM archive.
   7  *
   8  * initial version: 27/iii/95 by Simon Tatham
   9  */
  10
  11 #ifndef NASM_NASM_H
  12 #define NASM_NASM_H
  13
  14 #include "compiler.h"
  15
  16 #include <stdio.h>
  17 #include <inttypes.h>
  18 #include "nasmlib.h"
  19 #include "preproc.h"
  20 #include "insnsi.h"             /* For enum opcode */
  21
  22 #define NO_SEG -1L              /* null segment value */
  23 #define SEG_ABS 0x40000000L     /* mask for far-absolute segments */
  24
  25 #ifndef FILENAME_MAX
  26 #define FILENAME_MAX 256
  27 #endif
  28
  29 #ifndef PREFIX_MAX
  30 #define PREFIX_MAX 10
  31 #endif
  32
  33 #ifndef POSTFIX_MAX
  34 #define POSTFIX_MAX 10
  35 #endif
  36
  37 #define IDLEN_MAX 4096
  38
  39 /*
  40  * Name pollution problems: <time.h> on Digital UNIX pulls in some
  41  * strange hardware header file which sees fit to define R_SP. We
  42  * undefine it here so as not to break the enum below.
  43  */
  44 #ifdef R_SP
  45 #undef R_SP
  46 #endif
  47
  48 /*
  49  * We must declare the existence of this structure type up here,
  50  * since we have to reference it before we define it...
  51  */
  52 struct ofmt;
  53
  54 /*
  55  * values for the `type' parameter to an output function.
  56  *
  57  * Exceptions are OUT_RELxADR, which denote an x-byte relocation
  58  * which will be a relative jump. For this we need to know the
  59  * distance in bytes from the start of the relocated record until
  60  * the end of the containing instruction. _This_ is what is stored
  61  * in the size part of the parameter, in this case.
  62  *
  63  * Also OUT_RESERVE denotes reservation of N bytes of BSS space,
  64  * and the contents of the "data" parameter is irrelevant.
  65  *
  66  * The "data" parameter for the output function points to a "int32_t",
  67  * containing the address in question, unless the type is
  68  * OUT_RAWDATA, in which case it points to an "uint8_t"
  69  * array.
  70  */
  71 enum out_type {
  72     OUT_RAWDATA,                /* Plain bytes */
  73     OUT_ADDRESS,                /* An address (symbol value) */
  74     OUT_RESERVE,                /* Reserved bytes (RESB et al) */
  75     OUT_REL2ADR,                /* 2-byte relative address */
  76     OUT_REL4ADR,                /* 4-byte relative address */
  77     OUT_REL8ADR,                /* 8-byte relative address */
  78 };
  79
  80 /*
  81  * -----------------------
  82  * Other function typedefs
  83  * -----------------------
  84  */
  85
  86 /*
  87  * A label-lookup function should look like this.
  88  */
  89 typedef bool (*lfunc) (char *label, int32_t *segment, int64_t *offset);
  90
  91 /*
  92  * And a label-definition function like this. The boolean parameter
  93  * `is_norm' states whether the label is a `normal' label (which
  94  * should affect the local-label system), or something odder like
  95  * an EQU or a segment-base symbol, which shouldn't.
  96  */
  97 typedef void (*ldfunc) (char *label, int32_t segment, int64_t offset,
  98                         char *special, bool is_norm, bool isextrn,
  99                         struct ofmt * ofmt, efunc error);
 100
 101 /*
 102  * List-file generators should look like this:
 103  */
 104 typedef struct {
 105     /*
 106      * Called to initialize the listing file generator. Before this
 107      * is called, the other routines will silently do nothing when
 108      * called. The `char *' parameter is the file name to write the
 109      * listing to.
 110      */
 111     void (*init) (char *, efunc);
 112
 113     /*
 114      * Called to clear stuff up and close the listing file.
 115      */
 116     void (*cleanup) (void);
 117
 118     /*
 119      * Called to output binary data. Parameters are: the offset;
 120      * the data; the data type. Data types are similar to the
 121      * output-format interface, only OUT_ADDRESS will _always_ be
 122      * displayed as if it's relocatable, so ensure that any non-
 123      * relocatable address has been converted to OUT_RAWDATA by
 124      * then. Note that OUT_RAWDATA,0 is a valid data type, and is a
 125      * dummy call used to give the listing generator an offset to
 126      * work with when doing things like uplevel(LIST_TIMES) or
 127      * uplevel(LIST_INCBIN).
 128      */
 129     void (*output) (int32_t, const void *, enum out_type, uint64_t);
 130
 131     /*
 132      * Called to send a text line to the listing generator. The
 133      * `int' parameter is LIST_READ or LIST_MACRO depending on
 134      * whether the line came directly from an input file or is the
 135      * result of a multi-line macro expansion.
 136      */
 137     void (*line) (int, char *);
 138
 139     /*
 140      * Called to change one of the various levelled mechanisms in
 141      * the listing generator. LIST_INCLUDE and LIST_MACRO can be
 142      * used to increase the nesting level of include files and
 143      * macro expansions; LIST_TIMES and LIST_INCBIN switch on the
 144      * two binary-output-suppression mechanisms for large-scale
 145      * pseudo-instructions.
 146      *
 147      * LIST_MACRO_NOLIST is synonymous with LIST_MACRO except that
 148      * it indicates the beginning of the expansion of a `nolist'
 149      * macro, so anything under that level won't be expanded unless
 150      * it includes another file.
 151      */
 152     void (*uplevel) (int);
 153
 154     /*
 155      * Reverse the effects of uplevel.
 156      */
 157     void (*downlevel) (int);
 158 } ListGen;
 159
 160 /*
 161  * Token types returned by the scanner, in addition to ordinary
 162  * ASCII character values, and zero for end-of-string.
 163  */
 164 enum token_type {               /* token types, other than chars */
 165     TOKEN_INVALID = -1,         /* a placeholder value */
 166     TOKEN_EOS = 0,              /* end of string */
 167     TOKEN_EQ = '=', TOKEN_GT = '>', TOKEN_LT = '<',     /* aliases */
 168     TOKEN_ID = 256,             /* identifier */
 169     TOKEN_NUM,                  /* numeric constant */
 170     TOKEN_ERRNUM,               /* malformed numeric constant */
 171     TOKEN_STR,                  /* string constant */
 172     TOKEN_ERRSTR,               /* unterminated string constant */
 173     TOKEN_FLOAT,                /* floating-point constant */
 174     TOKEN_REG,                  /* register name */
 175     TOKEN_INSN,                 /* instruction name */
 176     TOKEN_HERE, TOKEN_BASE,     /* $ and $$ */
 177     TOKEN_SPECIAL,              /* BYTE, WORD, DWORD, QWORD, FAR, NEAR, etc */
 178     TOKEN_PREFIX,               /* A32, O16, LOCK, REPNZ, TIMES, etc */
 179     TOKEN_SHL, TOKEN_SHR,       /* << and >> */
 180     TOKEN_SDIV, TOKEN_SMOD,     /* // and %% */
 181     TOKEN_GE, TOKEN_LE, TOKEN_NE,       /* >=, <= and <> (!= is same as <>) */
 182     TOKEN_DBL_AND, TOKEN_DBL_OR, TOKEN_DBL_XOR, /* &&, || and ^^ */
 183     TOKEN_SEG, TOKEN_WRT,       /* SEG and WRT */
 184     TOKEN_FLOATIZE,             /* __floatX__ */
 185     TOKEN_STRFUNC,              /* __utf16__, __utf32__ */
 186 };
 187
 188 enum floatize {
 189     FLOAT_8,
 190     FLOAT_16,
 191     FLOAT_32,
 192     FLOAT_64,
 193     FLOAT_80M,
 194     FLOAT_80E,
 195     FLOAT_128L,
 196     FLOAT_128H,
 197 };
 198
 199 /* Must match the list in string_transform(), in strfunc.c */
 200 enum strfunc {
 201     STRFUNC_UTF16,
 202     STRFUNC_UTF32,
 203 };
 204
 205 size_t string_transform(char *, size_t, char **, enum strfunc);
 206
 207 /*
 208  * The expression evaluator must be passed a scanner function; a
 209  * standard scanner is provided as part of nasmlib.c. The
 210  * preprocessor will use a different one. Scanners, and the
 211  * token-value structures they return, look like this.
 212  *
 213  * The return value from the scanner is always a copy of the
 214  * `t_type' field in the structure.
 215  */
 216 struct tokenval {
 217     enum token_type t_type;
 218     char *t_charptr;
 219     int64_t t_integer, t_inttwo;
 220 };
 221 typedef int (*scanner) (void *private_data, struct tokenval * tv);
 222
 223 struct location {
 224     int64_t offset;
 225     int32_t segment;
 226     int known;
 227 };
 228
 229 /*
 230  * Expression-evaluator datatype. Expressions, within the
 231  * evaluator, are stored as an array of these beasts, terminated by
 232  * a record with type==0. Mostly, it's a vector type: each type
 233  * denotes some kind of a component, and the value denotes the
 234  * multiple of that component present in the expression. The
 235  * exception is the WRT type, whose `value' field denotes the
 236  * segment to which the expression is relative. These segments will
 237  * be segment-base types, i.e. either odd segment values or SEG_ABS
 238  * types. So it is still valid to assume that anything with a
 239  * `value' field of zero is insignificant.
 240  */
 241 typedef struct {
 242     int32_t type;                  /* a register, or EXPR_xxx */
 243     int64_t value;                 /* must be >= 32 bits */
 244 } expr;
 245
 246 /*
 247  * Library routines to manipulate expression data types.
 248  */
 249 int is_reloc(expr *);
 250 int is_simple(expr *);
 251 int is_really_simple(expr *);
 252 int is_unknown(expr *);
 253 int is_just_unknown(expr *);
 254 int64_t reloc_value(expr *);
 255 int32_t reloc_seg(expr *);
 256 int32_t reloc_wrt(expr *);
 257
 258 /*
 259  * The evaluator can also return hints about which of two registers
 260  * used in an expression should be the base register. See also the
 261  * `operand' structure.
 262  */
 263 struct eval_hints {
 264     int64_t base;
 265     int type;
 266 };
 267
 268 /*
 269  * The actual expression evaluator function looks like this. When
 270  * called, it expects the first token of its expression to already
 271  * be in `*tv'; if it is not, set tv->t_type to TOKEN_INVALID and
 272  * it will start by calling the scanner.
 273  *
 274  * If a forward reference happens during evaluation, the evaluator
 275  * must set `*fwref' to true if `fwref' is non-NULL.
 276  *
 277  * `critical' is non-zero if the expression may not contain forward
 278  * references. The evaluator will report its own error if this
 279  * occurs; if `critical' is 1, the error will be "symbol not
 280  * defined before use", whereas if `critical' is 2, the error will
 281  * be "symbol undefined".
 282  *
 283  * If `critical' has bit 8 set (in addition to its main value: 0x101
 284  * and 0x102 correspond to 1 and 2) then an extended expression
 285  * syntax is recognised, in which relational operators such as =, <
 286  * and >= are accepted, as well as low-precedence logical operators
 287  * &&, ^^ and ||.
 288  *
 289  * If `hints' is non-NULL, it gets filled in with some hints as to
 290  * the base register in complex effective addresses.
 291  */
 292 #define CRITICAL 0x100
 293 typedef expr *(*evalfunc) (scanner sc, void *scprivate,
 294                            struct tokenval * tv, int *fwref, int critical,
 295                            efunc error, struct eval_hints * hints);
 296
 297 /*
 298  * Special values for expr->type.  These come after EXPR_REG_END
 299  * as defined in regs.h.
 300  */
 301
 302 #define EXPR_UNKNOWN    (EXPR_REG_END+1) /* forward references */
 303 #define EXPR_SIMPLE     (EXPR_REG_END+2)
 304 #define EXPR_WRT        (EXPR_REG_END+3)
 305 #define EXPR_SEGBASE    (EXPR_REG_END+4)
 306
 307 /*
 308  * Linked list of strings...
 309  */
 310 typedef struct string_list {
 311     struct string_list *next;
 312     char str[1];
 313 } StrList;
 314
 315 /*
 316  * preprocessors ought to look like this:
 317  */
 318 typedef struct preproc_ops {
 319     /*
 320      * Called at the start of a pass; given a file name, the number
 321      * of the pass, an error reporting function, an evaluator
 322      * function, and a listing generator to talk to.
 323      */
 324     void (*reset) (char *, int, efunc, evalfunc, ListGen *, StrList **);
 325
 326     /*
 327      * Called to fetch a line of preprocessed source. The line
 328      * returned has been malloc'ed, and so should be freed after
 329      * use.
 330      */
 331     char *(*getline) (void);
 332
 333     /*
 334      * Called at the end of a pass.
 335      */
 336     void (*cleanup) (int);
 337 } Preproc;
 338
 339 extern Preproc nasmpp;
 340
 341 /*
 342  * ----------------------------------------------------------------
 343  * Some lexical properties of the NASM source language, included
 344  * here because they are shared between the parser and preprocessor
 345  * ----------------------------------------------------------------
 346  */
 347
 348 /*
 349  * isidstart matches any character that may start an identifier, and isidchar
 350  * matches any character that may appear at places other than the start of an
 351  * identifier. E.g. a period may only appear at the start of an identifier
 352  * (for local labels), whereas a number may appear anywhere *but* at the
 353  * start.
 354  */
 355
 356 #define isidstart(c) ( nasm_isalpha(c) || (c)=='_' || (c)=='.' || (c)=='?' \
 357                                   || (c)=='@' )
 358 #define isidchar(c)  ( isidstart(c) || nasm_isdigit(c) || \
 359                        (c)=='$' || (c)=='#' || (c)=='~' )
 360
 361 /* Ditto for numeric constants. */
 362
 363 #define isnumstart(c)  ( nasm_isdigit(c) || (c)=='$' )
 364 #define isnumchar(c)   ( nasm_isalnum(c) || (c)=='_' )
 365
 366 /* This returns the numeric value of a given 'digit'. */
 367
 368 #define numvalue(c)  ((c)>='a' ? (c)-'a'+10 : (c)>='A' ? (c)-'A'+10 : (c)-'0')
 369
 370 /*
 371  * Data-type flags that get passed to listing-file routines.
 372  */
 373 enum {
 374     LIST_READ, LIST_MACRO, LIST_MACRO_NOLIST, LIST_INCLUDE,
 375     LIST_INCBIN, LIST_TIMES
 376 };
 377
 378 /*
 379  * -----------------------------------------------------------
 380  * Format of the `insn' structure returned from `parser.c' and
 381  * passed into `assemble.c'
 382  * -----------------------------------------------------------
 383  */
 384
 385 /*
 386  * Here we define the operand types. These are implemented as bit
 387  * masks, since some are subsets of others; e.g. AX in a MOV
 388  * instruction is a special operand type, whereas AX in other
 389  * contexts is just another 16-bit register. (Also, consider CL in
 390  * shift instructions, DX in OUT, etc.)
 391  *
 392  * The basic concept here is that
 393  *    (class & ~operand) == 0
 394  *
 395  * if and only if "operand" belongs to class type "class".
 396  *
 397  * The bits are assigned as follows:
 398  *
 399  * Bits 0-7, 23, 29: sizes
 400  *  0:  8 bits (BYTE)
 401  *  1: 16 bits (WORD)
 402  *  2: 32 bits (DWORD)
 403  *  3: 64 bits (QWORD)
 404  *  4: 80 bits (TWORD)
 405  *  5: FAR
 406  *  6: NEAR
 407  *  7: SHORT
 408  * 23: 256 bits (YWORD)
 409  * 29: 128 bits (OWORD)
 410  *
 411  * Bits 8-11 modifiers
 412  *  8: TO
 413  *  9: COLON
 414  * 10: STRICT
 415  * 11: (reserved)
 416  *
 417  * Bits 12-15: type of operand
 418  * 12: REGISTER
 419  * 13: IMMEDIATE
 420  * 14: MEMORY (always has REGMEM attribute as well)
 421  * 15: REGMEM (valid EA operand)
 422  *
 423  * Bits 16-19, 28: subclasses
 424  * With REG_CDT:
 425  * 16: REG_CREG (CRx)
 426  * 17: REG_DREG (DRx)
 427  * 18: REG_TREG (TRx)
 428
 429  * With REG_GPR:
 430  * 16: REG_ACCUM  (AL, AX, EAX, RAX)
 431  * 17: REG_COUNT  (CL, CX, ECX, RCX)
 432  * 18: REG_DATA   (DL, DX, EDX, RDX)
 433  * 19: REG_HIGH   (AH, CH, DH, BH)
 434  * 28: REG_NOTACC (not REG_ACCUM)
 435  *
 436  * With REG_SREG:
 437  * 16: REG_CS
 438  * 17: REG_DESS (DS, ES, SS)
 439  * 18: REG_FSGS
 440  * 19: REG_SEG67
 441  *
 442  * With FPUREG:
 443  * 16: FPU0
 444  *
 445  * With XMMREG:
 446  * 16: XMM0
 447  *
 448  * With YMMREG:
 449  * 16: YMM0
 450  *
 451  * With MEMORY:
 452  * 16: MEM_OFFS (this is a simple offset)
 453  * 17: IP_REL (IP-relative offset)
 454  *
 455  * With IMMEDIATE:
 456  * 16: UNITY (1)
 457  * 17: BYTENESS16 (-128..127)
 458  * 18: BYTENESS32 (-128..127)
 459  * 19: BYTENESS64 (-128..127)
 460  *
 461  * Bits 20-22, 24-27: register classes
 462  * 20: REG_CDT (CRx, DRx, TRx)
 463  * 21: RM_GPR (REG_GPR) (integer register)
 464  * 22: REG_SREG
 465  * 24: FPUREG
 466  * 25: RM_MMX (MMXREG)
 467  * 26: RM_XMM (XMMREG)
 468  * 27: RM_YMM (YMMREG)
 469  *
 470  * Bit 31 is currently unallocated.
 471  *
 472  * 30: SAME_AS
 473  * Special flag only used in instruction patterns; means this operand
 474  * has to be identical to another operand.  Currently only supported
 475  * for registers.
 476  */
 477
 478 typedef uint32_t opflags_t;
 479
 480 /* Size, and other attributes, of the operand */
 481 #define BITS8           0x00000001U
 482 #define BITS16          0x00000002U
 483 #define BITS32          0x00000004U
 484 #define BITS64          0x00000008U   /* x64 and FPU only */
 485 #define BITS80          0x00000010U   /* FPU only */
 486 #define BITS128         0x20000000U
 487 #define BITS256         0x00800000U
 488 #define FAR             0x00000020U   /* grotty: this means 16:16 or */
 489                                        /* 16:32, like in CALL/JMP */
 490 #define NEAR            0x00000040U
 491 #define SHORT           0x00000080U   /* and this means what it says :) */
 492
 493 #define SIZE_MASK       0x208000FFU   /* all the size attributes */
 494
 495 /* Modifiers */
 496 #define MODIFIER_MASK   0x00000f00U
 497 #define TO              0x00000100U   /* reverse effect in FADD, FSUB &c */
 498 #define COLON           0x00000200U   /* operand is followed by a colon */
 499 #define STRICT          0x00000400U   /* do not optimize this operand */
 500
 501 /* Type of operand: memory reference, register, etc. */
 502 #define OPTYPE_MASK     0x0000f000U
 503 #define REGISTER        0x00001000U   /* register number in 'basereg' */
 504 #define IMMEDIATE       0x00002000U
 505 #define MEMORY          0x0000c000U
 506 #define REGMEM          0x00008000U   /* for r/m, ie EA, operands */
 507
 508 /* Register classes */
 509 #define REG_EA          0x00009000U   /* 'normal' reg, qualifies as EA */
 510 #define RM_GPR          0x00208000U   /* integer operand */
 511 #define REG_GPR         0x00209000U   /* integer register */
 512 #define REG8            0x00209001U   /*  8-bit GPR  */
 513 #define REG16           0x00209002U   /* 16-bit GPR */
 514 #define REG32           0x00209004U   /* 32-bit GPR */
 515 #define REG64           0x00209008U   /* 64-bit GPR */
 516 #define FPUREG          0x01001000U   /* floating point stack registers */
 517 #define FPU0            0x01011000U   /* FPU stack register zero */
 518 #define RM_MMX          0x02008000U   /* MMX operand */
 519 #define MMXREG          0x02009000U   /* MMX register */
 520 #define RM_XMM          0x04008000U   /* XMM (SSE) operand */
 521 #define XMMREG          0x04009000U   /* XMM (SSE) register */
 522 #define XMM0            0x04019000U   /* XMM register zero */
 523 #define RM_YMM          0x08008000U   /* YMM (AVX) operand */
 524 #define YMMREG          0x08009000U   /* YMM (AVX) register */
 525 #define YMM0            0x08019000U   /* YMM register zero */
 526 #define REG_CDT         0x00101004U   /* CRn, DRn and TRn */
 527 #define REG_CREG        0x00111004U   /* CRn */
 528 #define REG_DREG        0x00121004U   /* DRn */
 529 #define REG_TREG        0x00141004U   /* TRn */
 530 #define REG_SREG        0x00401002U   /* any segment register */
 531 #define REG_CS          0x00411002U   /* CS */
 532 #define REG_DESS        0x00421002U   /* DS, ES, SS */
 533 #define REG_FSGS        0x00441002U   /* FS, GS */
 534 #define REG_SEG67       0x00481002U   /* Unimplemented segment registers */
 535
 536 #define REG_RIP         0x00801008U   /* RIP relative addressing */
 537 #define REG_EIP         0x00801004U   /* EIP relative addressing */
 538
 539 /* Special GPRs */
 540 #define REG_SMASK       0x100f0000U   /* a mask for the following */
 541 #define REG_ACCUM       0x00219000U   /* accumulator: AL, AX, EAX, RAX */
 542 #define REG_AL          0x00219001U
 543 #define REG_AX          0x00219002U
 544 #define REG_EAX         0x00219004U
 545 #define REG_RAX         0x00219008U
 546 #define REG_COUNT       0x10229000U   /* counter: CL, CX, ECX, RCX */
 547 #define REG_CL          0x10229001U
 548 #define REG_CX          0x10229002U
 549 #define REG_ECX         0x10229004U
 550 #define REG_RCX         0x10229008U
 551 #define REG_DL          0x10249001U   /* data: DL, DX, EDX, RDX */
 552 #define REG_DX          0x10249002U
 553 #define REG_EDX         0x10249004U
 554 #define REG_RDX         0x10249008U
 555 #define REG_HIGH        0x10289001U   /* high regs: AH, CH, DH, BH */
 556 #define REG_NOTACC      0x10000000U   /* non-accumulator register */
 557 #define REG8NA          0x10209001U   /*  8-bit non-acc GPR  */
 558 #define REG16NA         0x10209002U   /* 16-bit non-acc GPR */
 559 #define REG32NA         0x10209004U   /* 32-bit non-acc GPR */
 560 #define REG64NA         0x10209008U   /* 64-bit non-acc GPR */
 561
 562 /* special types of EAs */
 563 #define MEM_OFFS        0x0001c000U   /* simple [address] offset - absolute! */
 564 #define IP_REL          0x0002c000U   /* IP-relative offset */
 565
 566 /* memory which matches any type of r/m operand */
 567 #define MEMORY_ANY      (MEMORY|RM_GPR|RM_MMX|RM_XMM|RM_YMM)
 568
 569 /* special type of immediate operand */
 570 #define UNITY           0x00012000U   /* for shift/rotate instructions */
 571 #define SBYTE16         0x00022000U   /* for op r16,immediate instrs. */
 572 #define SBYTE32         0x00042000U   /* for op r32,immediate instrs. */
 573 #define SBYTE64         0x00082000U   /* for op r64,immediate instrs. */
 574 #define BYTENESS        0x000e0000U   /* for testing for byteness */
 575
 576 /* special flags */
 577 #define SAME_AS         0x40000000U
 578
 579 /* Register names automatically generated from regs.dat */
 580 #include "regs.h"
 581
 582 enum ccode {                    /* condition code names */
 583     C_A, C_AE, C_B, C_BE, C_C, C_E, C_G, C_GE, C_L, C_LE, C_NA, C_NAE,
 584     C_NB, C_NBE, C_NC, C_NE, C_NG, C_NGE, C_NL, C_NLE, C_NO, C_NP,
 585     C_NS, C_NZ, C_O, C_P, C_PE, C_PO, C_S, C_Z,
 586     C_none = -1
 587 };
 588
 589 /*
 590  * REX flags
 591  */
 592 #define REX_REAL        0x4f    /* Actual REX prefix bits */
 593 #define REX_B           0x01    /* ModRM r/m extension */
 594 #define REX_X           0x02    /* SIB index extension */
 595 #define REX_R           0x04    /* ModRM reg extension */
 596 #define REX_W           0x08    /* 64-bit operand size */
 597 #define REX_L           0x20    /* Use LOCK prefix instead of REX.R */
 598 #define REX_P           0x40    /* REX prefix present/required */
 599 #define REX_H           0x80    /* High register present, REX forbidden */
 600 #define REX_D           0x0100  /* Instruction uses DREX instead of REX */
 601 #define REX_OC          0x0200  /* DREX suffix has the OC0 bit set */
 602 #define REX_V           0x0400  /* Instruction uses VEX/XOP instead of REX */
 603 #define REX_NH          0x0800  /* Instruction which doesn't use high regs */
 604
 605 /*
 606  * REX_V "classes" (prefixes which behave like VEX)
 607  */
 608 enum vex_class {
 609     RV_VEX              = 0,    /* C4/C5 */
 610     RV_XOP              = 1     /* 8F */
 611 };
 612
 613 /*
 614  * Note that because segment registers may be used as instruction
 615  * prefixes, we must ensure the enumerations for prefixes and
 616  * register names do not overlap.
 617  */
 618 enum prefixes {                 /* instruction prefixes */
 619     P_none = 0,
 620     PREFIX_ENUM_START = REG_ENUM_LIMIT,
 621     P_A16 = PREFIX_ENUM_START, P_A32, P_A64, P_ASP,
 622     P_LOCK, P_O16, P_O32, P_O64, P_OSP,
 623     P_REP, P_REPE, P_REPNE, P_REPNZ, P_REPZ, P_TIMES,
 624     P_WAIT,
 625     PREFIX_ENUM_LIMIT
 626 };
 627
 628 enum extop_type {               /* extended operand types */
 629     EOT_NOTHING,
 630     EOT_DB_STRING,              /* Byte string */
 631     EOT_DB_STRING_FREE,         /* Byte string which should be nasm_free'd*/
 632     EOT_DB_NUMBER,              /* Integer */
 633 };
 634
 635 enum ea_flags {                 /* special EA flags */
 636     EAF_BYTEOFFS =  1,          /* force offset part to byte size */
 637     EAF_WORDOFFS =  2,          /* force offset part to [d]word size */
 638     EAF_TIMESTWO =  4,          /* really do EAX*2 not EAX+EAX */
 639     EAF_REL      =  8,          /* IP-relative addressing */
 640     EAF_ABS      = 16,          /* non-IP-relative addressing */
 641     EAF_FSGS     = 32           /* fs/gs segment override present */
 642 };
 643
 644 enum eval_hint {                /* values for `hinttype' */
 645     EAH_NOHINT   = 0,           /* no hint at all - our discretion */
 646     EAH_MAKEBASE = 1,           /* try to make given reg the base */
 647     EAH_NOTBASE  = 2            /* try _not_ to make reg the base */
 648 };
 649
 650 typedef struct operand {        /* operand to an instruction */
 651     int32_t type;               /* type of operand */
 652     int disp_size;              /* 0 means default; 16; 32; 64 */
 653     enum reg_enum basereg, indexreg; /* address registers */
 654     int scale;                  /* index scale */
 655     int hintbase;
 656     enum eval_hint hinttype;    /* hint as to real base register */
 657     int32_t segment;            /* immediate segment, if needed */
 658     int64_t offset;             /* any immediate number */
 659     int32_t wrt;                /* segment base it's relative to */
 660     int eaflags;                /* special EA flags */
 661     int opflags;                /* see OPFLAG_* defines below */
 662 } operand;
 663
 664 #define OPFLAG_FORWARD          1       /* operand is a forward reference */
 665 #define OPFLAG_EXTERN           2       /* operand is an external reference */
 666 #define OPFLAG_UNKNOWN          4       /* operand is an unknown reference */
 667                                         /* (always a forward reference also) */
 668
 669 typedef struct extop {          /* extended operand */
 670     struct extop *next;         /* linked list */
 671     char *stringval;            /* if it's a string, then here it is */
 672     size_t stringlen;           /* ... and here's how long it is */
 673     int64_t offset;             /* ... it's given here ... */
 674     int32_t segment;            /* if it's a number/address, then... */
 675     int32_t wrt;                /* ... and here */
 676     enum extop_type type;       /* defined above */
 677 } extop;
 678
 679 /* Prefix positions: each type of prefix goes in a specific slot.
 680    This affects the final ordering of the assembled output, which
 681    shouldn't matter to the processor, but if you have stylistic
 682    preferences, you can change this.  REX prefixes are handled
 683    differently for the time being.
 684
 685    Note that LOCK and REP are in the same slot.  This is
 686    an x86 architectural constraint. */
 687 enum prefix_pos {
 688     PPS_WAIT,                   /* WAIT (technically not a prefix!) */
 689     PPS_LREP,                   /* Lock or REP prefix */
 690     PPS_SEG,                    /* Segment override prefix */
 691     PPS_OSIZE,                  /* Operand size prefix */
 692     PPS_ASIZE,                  /* Address size prefix */
 693     MAXPREFIX                   /* Total number of prefix slots */
 694 };
 695
 696 /* If you need to change this, also change it in insns.pl */
 697 #define MAX_OPERANDS 5
 698
 699 typedef struct insn {           /* an instruction itself */
 700     char *label;                /* the label defined, or NULL */
 701     enum prefixes prefixes[MAXPREFIX]; /* instruction prefixes, if any */
 702     enum opcode opcode;         /* the opcode - not just the string */
 703     enum ccode condition;       /* the condition code, if Jcc/SETcc */
 704     int operands;               /* how many operands? 0-3
 705                                  * (more if db et al) */
 706     int addr_size;              /* address size */
 707     operand oprs[MAX_OPERANDS]; /* the operands, defined as above */
 708     extop *eops;                /* extended operands */
 709     int eops_float;             /* true if DD and floating */
 710     int32_t times;              /* repeat count (TIMES prefix) */
 711     bool forw_ref;              /* is there a forward reference? */
 712     int rex;                    /* Special REX Prefix */
 713     int drexdst;                /* Destination register for DREX/VEX suffix */
 714     int vex_cm;                 /* Class and M field for VEX prefix */
 715     int vex_wlp;                /* W, P and L information for VEX prefix */
 716 } insn;
 717
 718 enum geninfo { GI_SWITCH };
 719 /*
 720  * ------------------------------------------------------------
 721  * The data structure defining an output format driver, and the
 722  * interfaces to the functions therein.
 723  * ------------------------------------------------------------
 724  */
 725
 726 struct ofmt {
 727     /*
 728      * This is a short (one-liner) description of the type of
 729      * output generated by the driver.
 730      */
 731     const char *fullname;
 732
 733     /*
 734      * This is a single keyword used to select the driver.
 735      */
 736     const char *shortname;
 737
 738
 739     /*
 740      * this is reserved for out module specific help.
 741      * It is set to NULL in all the out modules and is not implemented
 742      * in the main program
 743      */
 744     const char *helpstring;
 745
 746     /*
 747      * this is a pointer to the first element of the debug information
 748      */
 749     struct dfmt **debug_formats;
 750
 751     /*
 752      * and a pointer to the element that is being used
 753      * note: this is set to the default at compile time and changed if the
 754      * -F option is selected.  If developing a set of new debug formats for
 755      * an output format, be sure to set this to whatever default you want
 756      *
 757      */
 758     struct dfmt *current_dfmt;
 759
 760     /*
 761      * This, if non-NULL, is a NULL-terminated list of `char *'s
 762      * pointing to extra standard macros supplied by the object
 763      * format (e.g. a sensible initial default value of __SECT__,
 764      * and user-level equivalents for any format-specific
 765      * directives).
 766      */
 767     macros_t *stdmac;
 768
 769     /*
 770      * This procedure is called at the start of an output session.
 771      * It tells the output format what file it will be writing to,
 772      * what routine to report errors through, and how to interface
 773      * to the label manager and expression evaluator if necessary.
 774      * It also gives it a chance to do other initialisation.
 775      */
 776     void (*init) (FILE * fp, efunc error, ldfunc ldef, evalfunc eval);
 777
 778     /*
 779      * This procedure is called to pass generic information to the
 780      * object file.  The first parameter gives the information type
 781      * (currently only command line switches)
 782      * and the second parameter gives the value.  This function returns
 783      * 1 if recognized, 0 if unrecognized
 784      */
 785     int (*setinfo) (enum geninfo type, char **string);
 786
 787     /*
 788      * This procedure is called by assemble() to write actual
 789      * generated code or data to the object file. Typically it
 790      * doesn't have to actually _write_ it, just store it for
 791      * later.
 792      *
 793      * The `type' argument specifies the type of output data, and
 794      * usually the size as well: its contents are described below.
 795      */
 796     void (*output) (int32_t segto, const void *data,
 797                     enum out_type type, uint64_t size,
 798                     int32_t segment, int32_t wrt);
 799
 800     /*
 801      * This procedure is called once for every symbol defined in
 802      * the module being assembled. It gives the name and value of
 803      * the symbol, in NASM's terms, and indicates whether it has
 804      * been declared to be global. Note that the parameter "name",
 805      * when passed, will point to a piece of static storage
 806      * allocated inside the label manager - it's safe to keep using
 807      * that pointer, because the label manager doesn't clean up
 808      * until after the output driver has.
 809      *
 810      * Values of `is_global' are: 0 means the symbol is local; 1
 811      * means the symbol is global; 2 means the symbol is common (in
 812      * which case `offset' holds the _size_ of the variable).
 813      * Anything else is available for the output driver to use
 814      * internally.
 815      *
 816      * This routine explicitly _is_ allowed to call the label
 817      * manager to define further symbols, if it wants to, even
 818      * though it's been called _from_ the label manager. That much
 819      * re-entrancy is guaranteed in the label manager. However, the
 820      * label manager will in turn call this routine, so it should
 821      * be prepared to be re-entrant itself.
 822      *
 823      * The `special' parameter contains special information passed
 824      * through from the command that defined the label: it may have
 825      * been an EXTERN, a COMMON or a GLOBAL. The distinction should
 826      * be obvious to the output format from the other parameters.
 827      */
 828     void (*symdef) (char *name, int32_t segment, int64_t offset,
 829                     int is_global, char *special);
 830
 831     /*
 832      * This procedure is called when the source code requests a
 833      * segment change. It should return the corresponding segment
 834      * _number_ for the name, or NO_SEG if the name is not a valid
 835      * segment name.
 836      *
 837      * It may also be called with NULL, in which case it is to
 838      * return the _default_ section number for starting assembly in.
 839      *
 840      * It is allowed to modify the string it is given a pointer to.
 841      *
 842      * It is also allowed to specify a default instruction size for
 843      * the segment, by setting `*bits' to 16 or 32. Or, if it
 844      * doesn't wish to define a default, it can leave `bits' alone.
 845      */
 846     int32_t (*section) (char *name, int pass, int *bits);
 847
 848     /*
 849      * This procedure is called to modify the segment base values
 850      * returned from the SEG operator. It is given a segment base
 851      * value (i.e. a segment value with the low bit set), and is
 852      * required to produce in return a segment value which may be
 853      * different. It can map segment bases to absolute numbers by
 854      * means of returning SEG_ABS types.
 855      *
 856      * It should return NO_SEG if the segment base cannot be
 857      * determined; the evaluator (which calls this routine) is
 858      * responsible for throwing an error condition if that occurs
 859      * in pass two or in a critical expression.
 860      */
 861     int32_t (*segbase) (int32_t segment);
 862
 863     /*
 864      * This procedure is called to allow the output driver to
 865      * process its own specific directives. When called, it has the
 866      * directive word in `directive' and the parameter string in
 867      * `value'. It is called in both assembly passes, and `pass'
 868      * will be either 1 or 2.
 869      *
 870      * This procedure should return zero if it does not _recognise_
 871      * the directive, so that the main program can report an error.
 872      * If it recognises the directive but then has its own errors,
 873      * it should report them itself and then return non-zero. It
 874      * should also return non-zero if it correctly processes the
 875      * directive.
 876      */
 877     int (*directive) (char *directive, char *value, int pass);
 878
 879     /*
 880      * This procedure is called before anything else - even before
 881      * the "init" routine - and is passed the name of the input
 882      * file from which this output file is being generated. It
 883      * should return its preferred name for the output file in
 884      * `outname', if outname[0] is not '\0', and do nothing to
 885      * `outname' otherwise. Since it is called before the driver is
 886      * properly initialized, it has to be passed its error handler
 887      * separately.
 888      *
 889      * This procedure may also take its own copy of the input file
 890      * name for use in writing the output file: it is _guaranteed_
 891      * that it will be called before the "init" routine.
 892      *
 893      * The parameter `outname' points to an area of storage
 894      * guaranteed to be at least FILENAME_MAX in size.
 895      */
 896     void (*filename) (char *inname, char *outname, efunc error);
 897
 898     /*
 899      * This procedure is called after assembly finishes, to allow
 900      * the output driver to clean itself up and free its memory.
 901      * Typically, it will also be the point at which the object
 902      * file actually gets _written_.
 903      *
 904      * One thing the cleanup routine should always do is to close
 905      * the output file pointer.
 906      */
 907     void (*cleanup) (int debuginfo);
 908 };
 909
 910
 911 /*
 912  * ------------------------------------------------------------
 913  * The data structure defining a debug format driver, and the
 914  * interfaces to the functions therein.
 915  * ------------------------------------------------------------
 916  */
 917
 918 struct dfmt {
 919
 920     /*
 921      * This is a short (one-liner) description of the type of
 922      * output generated by the driver.
 923      */
 924     const char *fullname;
 925
 926     /*
 927      * This is a single keyword used to select the driver.
 928      */
 929     const char *shortname;
 930
 931     /*
 932      * init - called initially to set up local pointer to object format,
 933      * void pointer to implementation defined data, file pointer (which
 934      * probably won't be used, but who knows?), and error function.
 935      */
 936     void (*init) (struct ofmt * of, void *id, FILE * fp, efunc error);
 937
 938     /*
 939      * linenum - called any time there is output with a change of
 940      * line number or file.
 941      */
 942     void (*linenum) (const char *filename, int32_t linenumber, int32_t segto);
 943
 944     /*
 945      * debug_deflabel - called whenever a label is defined. Parameters
 946      * are the same as to 'symdef()' in the output format. This function
 947      * would be called before the output format version.
 948      */
 949
 950     void (*debug_deflabel) (char *name, int32_t segment, int64_t offset,
 951                             int is_global, char *special);
 952     /*
 953      * debug_directive - called whenever a DEBUG directive other than 'LINE'
 954      * is encountered. 'directive' contains the first parameter to the
 955      * DEBUG directive, and params contains the rest. For example,
 956      * 'DEBUG VAR _somevar:int' would translate to a call to this
 957      * function with 'directive' equal to "VAR" and 'params' equal to
 958      * "_somevar:int".
 959      */
 960     void (*debug_directive) (const char *directive, const char *params);
 961
 962     /*
 963      * typevalue - called whenever the assembler wishes to register a type
 964      * for the last defined label.  This routine MUST detect if a type was
 965      * already registered and not re-register it.
 966      */
 967     void (*debug_typevalue) (int32_t type);
 968
 969     /*
 970      * debug_output - called whenever output is required
 971      * 'type' is the type of info required, and this is format-specific
 972      */
 973     void (*debug_output) (int type, void *param);
 974
 975     /*
 976      * cleanup - called after processing of file is complete
 977      */
 978     void (*cleanup) (void);
 979
 980 };
 981 /*
 982  * The type definition macros
 983  * for debugging
 984  *
 985  * low 3 bits: reserved
 986  * next 5 bits: type
 987  * next 24 bits: number of elements for arrays (0 for labels)
 988  */
 989
 990 #define TY_UNKNOWN 0x00
 991 #define TY_LABEL   0x08
 992 #define TY_BYTE    0x10
 993 #define TY_WORD    0x18
 994 #define TY_DWORD   0x20
 995 #define TY_FLOAT   0x28
 996 #define TY_QWORD   0x30
 997 #define TY_TBYTE   0x38
 998 #define TY_OWORD   0x40
 999 #define TY_YWORD   0x48
1000 #define TY_COMMON  0xE0
1001 #define TY_SEG     0xE8
1002 #define TY_EXTERN  0xF0
1003 #define TY_EQU     0xF8
1004
1005 #define TYM_TYPE(x) ((x) & 0xF8)
1006 #define TYM_ELEMENTS(x) (((x) & 0xFFFFFF00) >> 8)
1007
1008 #define TYS_ELEMENTS(x)  ((x) << 8)
1009
1010 /*
1011  * -----
1012  * Special tokens
1013  * -----
1014  */
1015
1016 enum special_tokens {
1017     SPECIAL_ENUM_START = PREFIX_ENUM_LIMIT,
1018     S_ABS = SPECIAL_ENUM_START,
1019     S_BYTE, S_DWORD, S_FAR, S_LONG, S_NEAR, S_NOSPLIT,
1020     S_OWORD, S_QWORD, S_REL, S_SHORT, S_STRICT, S_TO, S_TWORD, S_WORD, S_YWORD,
1021     SPECIAL_ENUM_LIMIT
1022 };
1023
1024 /*
1025  * -----
1026  * Global modes
1027  * -----
1028  */
1029
1030 /*
1031  * This declaration passes the "pass" number to all other modules
1032  * "pass0" assumes the values: 0, 0, ..., 0, 1, 2
1033  * where 0 = optimizing pass
1034  *       1 = pass 1
1035  *       2 = pass 2
1036  */
1037
1038 extern int pass0;
1039 extern int passn;               /* Actual pass number */
1040
1041 extern bool tasm_compatible_mode;
1042 extern int optimizing;
1043 extern int globalbits;          /* 16, 32 or 64-bit mode */
1044 extern int globalrel;           /* default to relative addressing? */
1045 extern int maxbits;             /* max bits supported by output */
1046
1047 /*
1048  * NASM version strings, defined in ver.c
1049  */
1050 extern const char nasm_version[];
1051 extern const char nasm_date[];
1052 extern const char nasm_compile_options[];
1053 extern const char nasm_comment[];
1054 extern const char nasm_signature[];
1055
1056 #endif