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