4 * Declarations for Target Values.
10 Discussion of new interface, proposals by Prof. Waite:
12 > 1. You say that you plan to replace the tv module. That replacement is
13 > absolutely essential for an ANSI C translator: Section 6.1.3.2 of the
14 > standard says that the representation of an integer_constant depends
15 > upon its value as well as any suffixes that are attached to it. The
16 > possible Firm modes for such a constant are i, I, l, and L. The
17 > current tv module provides only one integer conversion routine, and
18 > that requires conversion by the client. Since the type of the value
19 > argument is long, this may preclude the representation of an unsigned
22 > There is a similar problem with floating constants. Floating
23 > constants can be suffixed in C, and the mode depends upon the suffix.
24 > It can indicate that the constant is of type long double, which your
25 > current tv module is incapable of representing.
27 > Your tv module interface accepts two kinds of information: modes and
28 > values. Values obtained from the program text might be uninterpreted
29 > strings, strings interpreted as integers, and strings interpreted as
30 > reals. Values provided by the compiler are usually integers. Modes are
31 > always Firm modes. It seems to me that the tv module should provide
32 > tarval* constructors for three of the four kinds of values. Each of these
33 > constructors should have an ir_mode parameter and one or more parameters
34 > appropriate for the kind of value. As is currently the case, one
35 > constructor should be provided for both compiler-generated integers and
36 > source strings interpreted as integers. (This avoids problems of
37 > different conversion radices -- the client does the conversion.) For
38 > symmetry, the constructor for source strings interpreted as reals should
39 > accept a long double parameter and require the client to do the
49 # include "irnode.h" /* for pnc_number enum */
55 * tv -- TargetValue, short tarval.
56 * Internal representation for machine values.
59 * Christian von Roques
63 * Tarvals represent target machine values. They are typed by modes.
64 * Tarvals only represent values of mode_sort:
71 * In case of references the module accepts an entity to represent the
73 * Furthermore, computations and conversions of these values can
77 * This module is closely related to the irmode module, as the modes
78 * defined there are thoroughly used throughout the whole module.
79 * Also, the comparison functions rely on the definition of comparison
80 * values in the irnode module.
83 * The original tv module originated in the fiasco compiler written ...
84 * This is the new version, described in the tech report 1999-14 by ...
88 * irmode.h for the modes definitions
89 * irnode.h for the pnc_numbers table
91 * tarval_init1 and tarval_init2 for initialization of the
96 #ifndef _TARVAL_TYPEDEF_
97 #define _TARVAL_TYPEDEF_
98 typedef struct tarval tarval;
101 /* ************************ Constructors for tarvals ************************ */
104 * Constructor function for new tarvals.
106 * @param str The string representing the target value
107 * @param len The length of the string
108 * @param mode The mode requested for the result tarval
110 * This function creates a new tarval representing the value represented
111 * by a CString, aka char array. If a tarval representing this value already
112 * exists, this tarval is returned instead of a new one. So tarvals are
113 * directly comparable since their representation is unique.
115 * This function accepts the following strings:
117 * if mode is int_number:
118 * - 0(x|X)[0-9a-fA-F]+ (hexadecimal representation)
119 * - 0[0-7]* (octal representation)
120 * - (+|-)?[1-9][0-9]* (decimal representation)
122 * if mode if float_number:
123 * - (+|-)?(decimal int) (. (decimal int))? ((e|E)(+|-)?(decimal int))?
125 * if mode is boolean: true, True, TRUE ... False... 0, 1,
127 * if mode is reference: hexadecimal of decimal number as int
129 * if mode is character: hex or dec
131 * Leading and/or trailing spaces are ignored
134 * A tarval of proper type representing the requested value is returned.
135 * Tarvals are unique, so for any value/mode pair at most one tarval will
136 * exist, which will be returned upon further requests with an identical
140 * If the string is not representable in the given mode an assertion is
141 * thrown in assert build.
144 * irmode.h for predefined modes
145 * new_tarval_from_long()
146 * new_tarval_from_double()
148 tarval *new_tarval_from_str(const char *str, size_t len, ir_mode *mode);
151 * Constructor function for new tarvals
153 * @param l The long representing the value
154 * @param mode The mode requested for the result tarval
156 * This function creates a new tarval representing the value represented
157 * by a long integer. If a tarval representing this value already exists,
158 * this tarval is returned instead of a new one. So tarvals are directly
159 * comparable since their representation is unique.
162 * A tarval of proper type representing the requested value is returned.
163 * Tarvals are unique, so for any value/mode pair at most one tarval will
164 * exist, which will be returned upon further requests with an identical
168 * If the long is not representable in the given mode an assertion is
169 * thrown in assert build.
172 * irmode.h for predefined modes
173 * new_tarval_from_str()
174 * new_tarval_from_double()
177 tarval *new_tarval_from_long(long l, ir_mode *mode);
180 * This returns a long int with the value represented value, or
181 * gibberish, depending on the size of long int and the size of the
182 * stored value. It works for e.g. 1 as mode_Ls, but might not work for
183 * get_mode_max(mode_Ls).
184 * This will overflow silently, so use only if you know what
185 * you are doing! (better check with tarval_is_long()...)
187 long tarval_to_long(tarval *tv);
190 * This validates if tarval_to_long() will return a satisfying
191 * result. I.e. if tv is an int_number and between min, max
192 * of long int (signed!)
194 int tarval_is_long(tarval *tv);
197 * Constructor function for new tarvals.
199 * @param d The (long) double representing the value
200 * @param mode The mode requested for the result tarval
202 * This function creates a new tarval representing the value represented
203 * by a (long) double. If a tarval representing this value already exists,
204 * this tarval is returned instead of a new one. So tarvals are directly
205 * comparable since their representation is unique.
206 * Only modes of sort float_number can be constructed this way.
209 * A tarval of proper type representing the requested value is returned.
210 * Tarvals are unique, so for any value/mode pair at most one tarval will
211 * exist, which will be returned upon further requests with an identical
215 * If the (long) double is not representable in the given mode an assertion
216 * is thrown. This will happen for any mode not of sort float_number.
219 * irmode.h for predefined values
220 * new_tarval_from_str()
221 * new_tarval_from_long()
223 tarval *new_tarval_from_double(long double d, ir_mode *mode);
226 * This returns a double with the value represented value, or
227 * gibberish, depending on the size of double and the size of the
229 * This will overflow silently, so use only if you know what
230 * you are doing! (better check with tarval_is_long...)
232 long double tarval_to_double(tarval *tv);
235 * This validates if tarval_to_double() will return a satisfying
236 * result. I.e. if tv is an float_number and between min, max
239 int tarval_is_double(tarval *tv);
242 * Construct a tarval that represents the address of the entity.
244 * The address must be constant, the entity must have as owner the global type.
246 tarval *new_tarval_from_entity (entity *ent, ir_mode *mode);
249 * Returns the associated entity of a tarval.
251 entity *tarval_to_entity(tarval *tv);
254 * Returns non-zero if a the given tarval represents an entity.
256 int tarval_is_entity(tarval *tv);
258 /** ********** Access routines for tarval fields ********** **/
266 * ir_mode *get_tarval_mode(tarval *tv)
270 * These are access function for tarval struct members. It is encouraged
271 * to use them instead of direct access to the struct fields.
274 * tv - The tarval to access fields of
277 * get_tv_mode: The mode of the tarval
283 /** Returns the mode of the tarval. */
284 ir_mode *get_tarval_mode (tarval *tv);
286 /* Testing properties of the represented values */
289 * Returns 1 if tv is negative
291 * @param a the tarval
293 int tarval_is_negative(tarval *a);
296 * Returns 1 if tv is null
298 * @param a the tarval
300 int tarval_is_null(tarval *a);
302 /** The 'bad' tarval. */
303 extern tarval *tarval_bad;
304 /** Returns the 'bad tarval. */
305 tarval *get_tarval_bad(void);
307 /** The 'undefined' tarval. */
308 extern tarval *tarval_undefined;
309 /** Returns the 'undefined' tarval. */
310 tarval *get_tarval_undefined(void);
312 /** The mode_b tarval 'false'. */
313 extern tarval *tarval_b_false;
314 /** Returns the mode_b tarval 'false'. */
315 tarval *get_tarval_b_false(void);
317 /** The mode_b tarval 'true'. */
318 extern tarval *tarval_b_true;
319 /** Returns the mode_b tarval 'true'. */
320 tarval *get_tarval_b_true(void);
322 /** The 'void' pointer tarval. */
323 extern tarval *tarval_P_void;
324 /** Returns the 'void' pointer tarval. */
325 tarval *get_tarval_P_void(void);
327 /* These functions calculate and return a tarval representing the requested
329 * The functions get_mode_{Max,Min,...} return tarvals retrieved from these
330 * functions, but these are stored on initialization of the irmode module and
331 * therefore the irmode functions should be prefered to the functions below. */
333 /** Returns the maximum value of a given mode. */
334 tarval *get_tarval_max(ir_mode *mode);
336 /** Returns the minimum value of a given mode. */
337 tarval *get_tarval_min(ir_mode *mode);
339 /** Returns the 0 value (additive neutral) of a given mode. */
340 tarval *get_tarval_null(ir_mode *mode);
342 /** Returns the 1 value (multiplicative neutral) of a given mode. */
343 tarval *get_tarval_one(ir_mode *mode);
345 /** Return quite nan for float_number modes. */
346 tarval *get_tarval_nan(ir_mode *mode);
348 /** Return +inf for float_number modes. */
349 tarval *get_tarval_inf(ir_mode *mode);
351 /* ******************** Arithmethic operations on tarvals ******************** */
354 * Compares two tarvals
356 * Compare a with b and return a pnc_number describing the relation
357 * between a and b. This is either Uo, Lt, Eq, Gt, or False if a or b
358 * are symbolic pointers which can not be compared at all.
360 * @param a A tarval to be compared
361 * @param b A tarval to be compared
364 * The pnc_number best describing the relation between a and b is returned.
365 * This means the mode with the least bits set is returned, e.g. if the
366 * tarvals are equal the pnc_number 'Eq' is returned, not 'Ge' which
367 * indicates 'greater or equal'
370 * irnode.h for the definition of pnc_numbers
372 pnc_number tarval_cmp(tarval *a, tarval *b);
375 * Converts a tarval to another mode.
377 * Convert tarval 'src' to mode 'mode', this will suceed if and only if mode
378 * 'mode' is wider than the mode of src, as defined in the firm documentation
379 * and as returned by the function mode_is_smaller defined in irmode.h.
381 * @param src The tarval to convert
382 * @param mode Tho mode to convert to
385 * If a tarval of mode 'mode' with the result of the conversion of the 'src'
386 * tarvals value already exists, it will be returned, else a new tarval is
387 * constructed and returned
390 * Illegal conversations will trigger an assertion
393 * FIRM documentation for conversion rules
394 * mode_is_smaller defined in irmode.h
396 tarval *tarval_convert_to(tarval *src, ir_mode *m);
399 * These function implement basic computations representable as opcodes
405 * a - the tarval to operate on
408 * a - the first operand tarval
409 * b - the second operand tarval
412 * If neccessary a new tarval is constructed for the resulting value,
413 * or the one already carrying the computation result is retrieved and
414 * returned as result.
417 * The order the arguments are given in is important, imagine postfix
419 * Illegal operations will trigger an assertion.
420 * The sort member of the struct mode defines which operations are valid
423 /** Negation of a tarval. */
424 tarval *tarval_neg(tarval *a);
426 /** Addition of two tarvals. */
427 tarval *tarval_add(tarval *a, tarval *b);
429 /** Subtraction from a tarval. */
430 tarval *tarval_sub(tarval *a, tarval *b);
432 /** Multiplication of tarvals. */
433 tarval *tarval_mul(tarval *a, tarval *b);
435 /** 'Exact' division. */
436 tarval *tarval_quo(tarval *a, tarval *b);
438 /** Integer division. */
439 tarval *tarval_div(tarval *a, tarval *b);
441 /** Remainder of integer division. */
442 tarval *tarval_mod(tarval *a, tarval *b);
444 /** Absolute value. */
445 tarval *tarval_abs(tarval *a);
448 tarval *tarval_and(tarval *a, tarval *b);
451 tarval *tarval_or(tarval *a, tarval *b);
453 /** Bitwise exclusive or. */
454 tarval *tarval_eor(tarval *a, tarval *b);
457 tarval *tarval_shl(tarval *a, tarval *b);
459 /** Unsigned (logical) right shift. */
460 tarval *tarval_shr(tarval *a, tarval *b);
462 /** Signed (arithmetic) right shift. */
463 tarval *tarval_shrs(tarval *a, tarval *b);
466 tarval *tarval_rot(tarval *a, tarval *b);
468 /* *********** Output of tarvals *********** */
471 * The output mode for tarval values.
473 * Some modes allow more that one representation, for instance integers
474 * can be represented hex or decimal. Of course it would be enough to have
475 * one and let every backend convert it into the 'right' one.
476 * However, we can do this in the tarval much simplier...
479 TVO_NATIVE, /**< the default output mode, depends on the mode */
480 TVO_HEX, /**< use hex representation, always possible */
481 TVO_DECIMAL, /**< use decimal representation */
482 TVO_OCTAL, /**< use octal representation */
483 TVO_BINARY, /**< use binary representation */
484 TVO_FLOAT, /**< use floating point representation (i.e 1.342e-2)*/
485 TVO_HEXFLOAT, /**< use hexadecimal floating point representation (i.e 0x1.ea32p-12)*/
489 * This structure contains helper information to format the output
490 * of a tarval of an mode.
492 typedef struct tarval_mode_info {
493 tv_output_mode mode_output; /**< if != TVO_NATIVE select a special mode */
494 const char *mode_prefix; /**< if set, this prefix will be printed
495 before a value of this mode */
496 const char *mode_suffix; /**< if set, this suffix will be printed
497 after a value of this mode */
501 * Specify the output options of one mode.
503 * This functions stores the modinfo, so DO NOT DESTROY it.
505 * @param mode a ir_mode that should be associated
506 * @param modeinfo the output format info
508 * @return zero on success.
510 int tarval_set_mode_output_option(ir_mode *mode, const tarval_mode_info *modeinfo);
513 * Returns the output options of one mode.
515 * This functions returns the modinfo of a given mode.
517 * @param mode a ir_mode that should be associated
519 * @return the output option
521 const tarval_mode_info *tarval_get_mode_output_option(ir_mode *mode);
524 * Returns Bit representation of a tarval value, as string of '0' and '1'
526 * @param tv The tarval
528 * This function returns a printable bit representation of any value
529 * stored as tarval. This representation is a null terminated C string.
532 * As usual in C a pointer to a char is returned. The length of the
533 * returned string if fixed, just read as many chars as the mode defines
537 * The string is allocated using malloc() and is free()ed on the next call
539 * The string consists of the ascii characters '0' and '1' and is
543 * irmode.h for the definition of the ir_mode struct
544 * the size member of aforementioned struct
546 char *tarval_bitpattern(tarval *tv);
549 * Returns the bitpattern of the bytes_ofs byte.
551 * This function succeeds even if the mode of the tarval uses lesser bits
552 * than requested, in that case the bitpattern is filled with zero bits.
554 * To query a 32bit value the following code can be used:
556 * val0 = tarval_sub_bits(tv, 0);
557 * val1 = tarval_sub_bits(tv, 1);
558 * val2 = tarval_sub_bits(tv, 2);
559 * val3 = tarval_sub_bits(tv, 3);
561 * Because this is the bit representation of the target machine, only the following
562 * operations are legal on the result:
564 * - concatenation (endian dependance MUST be handled by the CALLER)
565 * - bitwise logical operations to select/mask bits
567 * @param tv the tarval
568 * @param byte_ofs the byte offset
571 * The result of this funcion is undefined if the mode is neither integer nor float.
573 unsigned char tarval_sub_bits(tarval *tv, unsigned byte_ofs);
576 * Identifying some tarvals ???
579 * - 0 for additive neutral,
580 * - +1 for multiplicative neutral,
581 * - -1 for bitwise-and neutral
585 * This function is deprecated and its use strongly discouraged.
586 * Implemented for completeness.
588 long tarval_classify(tarval *tv);
591 * Initialization of the tarval module.
593 * Call before init_mode().
595 void init_tarval_1(void);
598 * Initialization of the tarval module.
600 * Call after init_mode().
602 void init_tarval_2(void);
605 * Output of tarvals to a buffer.
607 int tarval_snprintf(char *buf, size_t buflen, tarval *tv);
610 * Output of tarvals to stdio.
612 int tarval_printf(tarval *tv);