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