nsz Git - cparser/blob - format_check.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2009 Matthias Braun <matze@braunis.de>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  18  * 02111-1307, USA.
  19  */
  20 #include <ctype.h>
  21 #include <wctype.h>
  22
  23 #include "adt/util.h"
  24 #include "format_check.h"
  25 #include "symbol_t.h"
  26 #include "ast_t.h"
  27 #include "entity_t.h"
  28 #include "diagnostic.h"
  29 #include "types.h"
  30 #include "type_t.h"
  31 #include "warning.h"
  32 #include "lang_features.h"
  33
  34 typedef enum format_flag_t {
  35         FMT_FLAG_NONE  = 0,
  36         FMT_FLAG_HASH  = 1U << 0,
  37         FMT_FLAG_ZERO  = 1U << 1,
  38         FMT_FLAG_MINUS = 1U << 2,
  39         FMT_FLAG_SPACE = 1U << 3,
  40         FMT_FLAG_PLUS  = 1U << 4,
  41         FMT_FLAG_TICK  = 1U << 5
  42 } format_flag_t;
  43
  44 typedef unsigned format_flags_t;
  45
  46 typedef enum format_length_modifier_t {
  47         FMT_MOD_NONE,
  48         FMT_MOD_L,
  49         FMT_MOD_hh,
  50         FMT_MOD_h,
  51         FMT_MOD_l,
  52         FMT_MOD_ll,
  53         FMT_MOD_j,
  54         FMT_MOD_t,
  55         FMT_MOD_z,
  56         FMT_MOD_q,
  57         /* only in microsoft mode */
  58         FMT_MOD_w,
  59         FMT_MOD_I,
  60         FMT_MOD_I32,
  61         FMT_MOD_I64
  62 } format_length_modifier_t;
  63
  64 typedef struct format_spec_t {
  65         const char    *name;     /**< name of the function */
  66         format_kind_t  fmt_kind; /**< kind */
  67         unsigned       fmt_idx;  /**< index of the format string */
  68         unsigned       arg_idx;  /**< index of the first argument */
  69 } format_spec_t;
  70
  71 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  72 {
  73         static const char* const names[] = {
  74                 [FMT_MOD_NONE] = "",
  75                 [FMT_MOD_L]    = "L",
  76                 [FMT_MOD_hh]   = "hh",
  77                 [FMT_MOD_h]    = "h",
  78                 [FMT_MOD_l]    = "l",
  79                 [FMT_MOD_ll]   = "ll",
  80                 [FMT_MOD_j]    = "j",
  81                 [FMT_MOD_t]    = "t",
  82                 [FMT_MOD_z]    = "z",
  83                 [FMT_MOD_q]    = "q",
  84                 /* only in microsoft mode */
  85                 [FMT_MOD_w]    = "w",
  86                 [FMT_MOD_I]    = "I",
  87                 [FMT_MOD_I32]  = "I32",
  88                 [FMT_MOD_I64]  = "I64"
  89         };
  90         assert(mod < lengthof(names));
  91         return names[mod];
  92 }
  93
  94 static void warn_invalid_length_modifier(const source_position_t *pos,
  95                                          const format_length_modifier_t mod,
  96                                          const wchar_rep_t conversion)
  97 {
  98         warningf(pos,
  99                 "invalid length modifier '%s' for conversion specifier '%%%c'",
 100                 get_length_modifier_name(mod), conversion
 101         );
 102 }
 103
 104 typedef struct vchar_t vchar_t;
 105 struct vchar_t {
 106         const void *string;   /**< the string */
 107         size_t     position;  /**< current position */
 108         size_t     size;      /**< size of the string */
 109
 110         /** return the first character of the string and setthe position to 0. */
 111         unsigned (*first)(vchar_t *self);
 112         /** return the next character of the string */
 113         unsigned (*next)(vchar_t *self);
 114         /** return non_zero if the given character is a digit */
 115         int (*is_digit)(unsigned vchar);
 116 };
 117
 118 static unsigned string_first(vchar_t *self)
 119 {
 120         self->position = 0;
 121         const string_t *string = self->string;
 122         return string->begin[0];
 123 }
 124
 125 static unsigned string_next(vchar_t *self)
 126 {
 127         ++self->position;
 128         const string_t *string = self->string;
 129         return string->begin[self->position];
 130 }
 131
 132 static int string_isdigit(unsigned vchar)
 133 {
 134         return isdigit(vchar);
 135 }
 136
 137 static unsigned wstring_first(vchar_t *self)
 138 {
 139         self->position = 0;
 140         const wide_string_t *wstring = self->string;
 141         return wstring->begin[0];
 142 }
 143
 144 static unsigned wstring_next(vchar_t *self)
 145 {
 146         ++self->position;
 147         const wide_string_t *wstring = self->string;
 148         return wstring->begin[self->position];
 149 }
 150
 151 static int wstring_isdigit(unsigned vchar)
 152 {
 153         return iswdigit(vchar);
 154 }
 155
 156 static bool atend(vchar_t *self)
 157 {
 158         return self->position + 1 == self->size;
 159 }
 160
 161 /**
 162  * Check printf-style format.
 163  */
 164 static ssize_t internal_check_printf_format(const expression_t *fmt_expr,
 165     const call_argument_t *arg, const format_spec_t *spec)
 166 {
 167         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 168                 fmt_expr = fmt_expr->unary.value;
 169         }
 170
 171         vchar_t vchar;
 172         switch (fmt_expr->kind) {
 173                 case EXPR_STRING_LITERAL:
 174                         vchar.string   = &fmt_expr->string.value;
 175                         vchar.size     = fmt_expr->string.value.size;
 176                         vchar.first    = string_first;
 177                         vchar.next     = string_next;
 178                         vchar.is_digit = string_isdigit;
 179                         break;
 180
 181                 case EXPR_WIDE_STRING_LITERAL:
 182                         vchar.string   = &fmt_expr->wide_string.value;
 183                         vchar.size     = fmt_expr->wide_string.value.size;
 184                         vchar.first    = wstring_first;
 185                         vchar.next     = wstring_next;
 186                         vchar.is_digit = wstring_isdigit;
 187                         break;
 188
 189                 case EXPR_CONDITIONAL: {
 190                         conditional_expression_t const *const c = &fmt_expr->conditional;
 191                         expression_t             const *      t = c->true_expression;
 192                         if (t == NULL)
 193                                 t = c->condition;
 194                         ssize_t const nt = internal_check_printf_format(t,                   arg, spec);
 195                         ssize_t const nf = internal_check_printf_format(c->false_expression, arg, spec);
 196                         return nt > nf ? nt : nf;
 197                 }
 198
 199                 default:
 200                         return -1;
 201         }
 202
 203         const source_position_t *pos = &fmt_expr->base.source_position;
 204         unsigned fmt     = vchar.first(&vchar);
 205         unsigned num_fmt = 0;
 206         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 207                 if (fmt != '%')
 208                         continue;
 209                 fmt = vchar.next(&vchar);
 210
 211                 if (fmt == '%')
 212                         continue;
 213
 214                 ++num_fmt;
 215
 216                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 217                 if (fmt == '0') {
 218                         fmt = vchar.next(&vchar);
 219                         fmt_flags |= FMT_FLAG_ZERO;
 220                 }
 221
 222                 /* argument selector or minimum field width */
 223                 if (vchar.is_digit(fmt)) {
 224                         do {
 225                                 fmt = vchar.next(&vchar);
 226                         } while (vchar.is_digit(fmt));
 227
 228                         /* digit string was ... */
 229                         if (fmt == '$') {
 230                                 /* ... argument selector */
 231                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 232                                 /* TODO implement */
 233                                 return -1;
 234                         }
 235                         /* ... minimum field width */
 236                 } else {
 237                         /* flags */
 238                         for (;;) {
 239                                 format_flags_t flag;
 240                                 switch (fmt) {
 241                                         case '#':  flag = FMT_FLAG_HASH;  break;
 242                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 243                                         case '-':  flag = FMT_FLAG_MINUS; break;
 244                                         case '\'': flag = FMT_FLAG_TICK;  break;
 245
 246                                         case ' ':
 247                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 248                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 249                                                 }
 250                                                 flag = FMT_FLAG_SPACE;
 251                                                 break;
 252
 253                                         case '+':
 254                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 255                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 256                                                 }
 257                                                 flag = FMT_FLAG_PLUS;
 258                                                 break;
 259
 260                                         default: goto break_fmt_flags;
 261                                 }
 262                                 if (fmt_flags & flag) {
 263                                         warningf(pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 264                                 }
 265                                 fmt_flags |= flag;
 266                                 fmt = vchar.next(&vchar);
 267                         }
 268 break_fmt_flags:
 269
 270                         /* minimum field width */
 271                         if (fmt == '*') {
 272                                 fmt = vchar.next(&vchar);
 273                                 if (arg == NULL) {
 274                                         warningf(pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 275                                         return -1;
 276                                 }
 277                                 const type_t *const arg_type = arg->expression->base.type;
 278                                 if (arg_type != type_int) {
 279                                         warningf(pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 280                                 }
 281                                 arg = arg->next;
 282                         } else {
 283                                 while (vchar.is_digit(fmt)) {
 284                                         fmt = vchar.next(&vchar);
 285                                 }
 286                         }
 287                 }
 288
 289                 /* precision */
 290                 if (fmt == '.') {
 291                         fmt = vchar.next(&vchar);
 292                         if (fmt == '*') {
 293                                 fmt = vchar.next(&vchar);
 294                                 if (arg == NULL) {
 295                                         warningf(pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 296                                         return -1;
 297                                 }
 298                                 const type_t *const arg_type = arg->expression->base.type;
 299                                 if (arg_type != type_int) {
 300                                         warningf(pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 301                                 }
 302                                 arg = arg->next;
 303                         } else {
 304                                 /* digit string may be omitted */
 305                                 while (vchar.is_digit(fmt)) {
 306                                         fmt = vchar.next(&vchar);
 307                                 }
 308                         }
 309                 }
 310
 311                 /* length modifier */
 312                 format_length_modifier_t fmt_mod;
 313                 switch (fmt) {
 314                         case 'h':
 315                                 fmt = vchar.next(&vchar);
 316                                 if (fmt == 'h') {
 317                                         fmt = vchar.next(&vchar);
 318                                         fmt_mod = FMT_MOD_hh;
 319                                 } else {
 320                                         fmt_mod = FMT_MOD_h;
 321                                 }
 322                                 break;
 323
 324                         case 'l':
 325                                 fmt = vchar.next(&vchar);
 326                                 if (fmt == 'l') {
 327                                         fmt = vchar.next(&vchar);
 328                                         fmt_mod = FMT_MOD_ll;
 329                                 } else {
 330                                         fmt_mod = FMT_MOD_l;
 331                                 }
 332                                 break;
 333
 334                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 335                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 336                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 337                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 338                         case 'q': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_q;    break;
 339                         /* microsoft mode */
 340                         case 'w':
 341                                 if (c_mode & _MS) {
 342                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 343                                 } else {
 344                                         fmt_mod = FMT_MOD_NONE;
 345                                 }
 346                                 break;
 347                         case 'I':
 348                                 if (c_mode & _MS) {
 349                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 350                                         if (fmt == '3') {
 351                                                 fmt = vchar.next(&vchar);
 352                                                 if (fmt == '2') {
 353                                                         fmt = vchar.next(&vchar);
 354                                                         fmt_mod = FMT_MOD_I32;
 355                                                 } else {
 356                                                         /* rewind */
 357                                                         --vchar.position;
 358                                                 }
 359                                         } else if (fmt == '6') {
 360                                                 fmt = vchar.next(&vchar);
 361                                                 if (fmt == '4') {
 362                                                         fmt = vchar.next(&vchar);
 363                                                         fmt_mod = FMT_MOD_I64;
 364                                                 } else {
 365                                                         /* rewind */
 366                                                         --vchar.position;
 367                                                 }
 368                                         }
 369                                 } else {
 370                                         fmt_mod = FMT_MOD_NONE;
 371                                 }
 372                                 break;
 373                         default:
 374                                 fmt_mod = FMT_MOD_NONE;
 375                                 break;
 376                 }
 377
 378                 if (fmt == '\0') {
 379                         warningf(pos, "dangling %% in format string");
 380                         break;
 381                 }
 382
 383                 type_t            *expected_type;
 384                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 385                 format_flags_t     allowed_flags;
 386                 switch (fmt) {
 387                         case 'd':
 388                         case 'i':
 389                                 switch (fmt_mod) {
 390                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 391                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 392                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 393                                         case FMT_MOD_l:    expected_type = type_long;      break;
 394                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 395                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 396                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 397                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 398                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 399                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 400                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 401
 402                                         default:
 403                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 404                                                 goto next_arg;
 405                                 }
 406                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 407                                 break;
 408
 409                         case 'o':
 410                         case 'X':
 411                         case 'x':
 412                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 413                                 goto eval_fmt_mod_unsigned;
 414
 415                         case 'u':
 416                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 417 eval_fmt_mod_unsigned:
 418                                 switch (fmt_mod) {
 419                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 420                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 421                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 422                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 423                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 424                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 425                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 426                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 427                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 428                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 429                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 430
 431                                         default:
 432                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 433                                                 goto next_arg;
 434                                 }
 435                                 break;
 436
 437                         case 'A':
 438                         case 'a':
 439                         case 'E':
 440                         case 'e':
 441                         case 'F':
 442                         case 'f':
 443                         case 'G':
 444                         case 'g':
 445                                 switch (fmt_mod) {
 446                                         case FMT_MOD_l:    /* l modifier is ignored */
 447                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 448                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 449
 450                                         default:
 451                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 452                                                 goto next_arg;
 453                                 }
 454                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 455                                 break;
 456
 457                         case 'C':
 458                                 if (fmt_mod != FMT_MOD_NONE) {
 459                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 460                                         goto next_arg;
 461                                 }
 462                                 expected_type = type_wchar_t;
 463                                 allowed_flags = FMT_FLAG_NONE;
 464                                 break;
 465
 466                         case 'c':
 467                                 expected_type = type_int;
 468                                 switch (fmt_mod) {
 469                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 470                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 471                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 472
 473                                         default:
 474                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 475                                                 goto next_arg;
 476                                 }
 477                                 allowed_flags = FMT_FLAG_NONE;
 478                                 break;
 479
 480                         case 'S':
 481                                 if (fmt_mod != FMT_MOD_NONE) {
 482                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 483                                         goto next_arg;
 484                                 }
 485                                 expected_type = type_wchar_t_ptr;
 486                                 expected_qual = TYPE_QUALIFIER_CONST;
 487                                 allowed_flags = FMT_FLAG_MINUS;
 488                                 break;
 489
 490                         case 's':
 491                                 switch (fmt_mod) {
 492                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 493                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 494                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 495
 496                                         default:
 497                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 498                                                 goto next_arg;
 499                                 }
 500                                 expected_qual = TYPE_QUALIFIER_CONST;
 501                                 allowed_flags = FMT_FLAG_MINUS;
 502                                 break;
 503
 504                         case 'p':
 505                                 if (fmt_mod != FMT_MOD_NONE) {
 506                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 507                                         goto next_arg;
 508                                 }
 509                                 expected_type = type_void_ptr;
 510                                 allowed_flags = FMT_FLAG_NONE;
 511                                 break;
 512
 513                         case 'n':
 514                                 switch (fmt_mod) {
 515                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 516                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 517                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 518                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 519                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 520                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 521                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 522                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 523
 524                                         default:
 525                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 526                                                 goto next_arg;
 527                                 }
 528                                 allowed_flags = FMT_FLAG_NONE;
 529                                 break;
 530
 531                         default:
 532                                 warningf(pos, "encountered unknown conversion specifier '%%%C' at position %u", (wint_t)fmt, num_fmt);
 533                                 if (arg == NULL) {
 534                                         warningf(pos, "too few arguments for format string");
 535                                         return -1;
 536                                 }
 537                                 goto next_arg;
 538                 }
 539
 540                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 541                 if (wrong_flags != 0) {
 542                         char  wrong[8];
 543                         char *p = wrong;
 544                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 545                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 546                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 547                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 548                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 549                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 550                         *p = '\0';
 551
 552                         warningf(pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 553                 }
 554
 555                 if (arg == NULL) {
 556                         warningf(pos, "too few arguments for format string");
 557                         return -1;
 558                 }
 559
 560                 {       /* create a scope here to prevent warning about the jump to next_arg */
 561                         type_t *const arg_type           = arg->expression->base.type;
 562                         type_t *const arg_skip           = skip_typeref(arg_type);
 563                         type_t *const expected_type_skip = skip_typeref(expected_type);
 564
 565                         if (fmt == 'p') {
 566                                 /* allow any pointer type for %p, not just void */
 567                                 if (is_type_pointer(arg_skip))
 568                                         goto next_arg;
 569                         }
 570
 571                         if (is_type_pointer(expected_type_skip)) {
 572                                 if (is_type_pointer(arg_skip)) {
 573                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 574                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 575                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 576                                                 get_unqualified_type(arg_to) == exp_to) {
 577                                                 goto next_arg;
 578                                         }
 579                                 }
 580                         } else {
 581                                 if (get_unqualified_type(arg_skip) == expected_type_skip) {
 582                                         goto next_arg;
 583                                 }
 584                         }
 585                         if (is_type_valid(arg_skip)) {
 586                                 warningf(pos,
 587                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 588                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 589                         }
 590                 }
 591 next_arg:
 592                 arg = arg->next;
 593         }
 594         if (!atend(&vchar)) {
 595                 warningf(pos, "format string contains '\\0'");
 596         }
 597         return num_fmt;
 598 }
 599
 600 /**
 601  * Check printf-style format.
 602  */
 603 static void check_printf_format(call_argument_t const *arg, format_spec_t const *const spec)
 604 {
 605         /* find format arg */
 606         size_t idx = 0;
 607         for (; idx < spec->fmt_idx; ++idx) {
 608                 if (arg == NULL)
 609                         return;
 610                 arg = arg->next;
 611         }
 612
 613         expression_t const *const fmt_expr = arg->expression;
 614
 615         /* find the real args */
 616         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 617                 arg = arg->next;
 618
 619         ssize_t const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
 620         if (num_fmt < 0)
 621                 return;
 622
 623         size_t num_args = 0;
 624         for (; arg != NULL; arg = arg->next)
 625                 ++num_args;
 626         if (num_args > (size_t)num_fmt) {
 627                 warningf(&fmt_expr->base.source_position,
 628                         "%u argument%s but only %u format specifier%s",
 629                         num_args, num_args != 1 ? "s" : "",
 630                         num_fmt,  num_fmt  != 1 ? "s" : "");
 631         }
 632 }
 633
 634 /**
 635  * Check scanf-style format.
 636  */
 637 static void check_scanf_format(const call_argument_t *arg, const format_spec_t *spec)
 638 {
 639         /* find format arg */
 640         unsigned idx = 0;
 641         for (; idx < spec->fmt_idx; ++idx) {
 642                 if (arg == NULL)
 643                         return;
 644                 arg = arg->next;
 645         }
 646
 647         const expression_t *fmt_expr = arg->expression;
 648         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 649                 fmt_expr = fmt_expr->unary.value;
 650         }
 651
 652         vchar_t vchar;
 653         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 654                 vchar.string   = &fmt_expr->wide_string.value;
 655                 vchar.size     = fmt_expr->wide_string.value.size;
 656                 vchar.first    = wstring_first;
 657                 vchar.next     = wstring_next;
 658                 vchar.is_digit = wstring_isdigit;
 659         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 660                 vchar.string   = &fmt_expr->string.value;
 661                 vchar.size     = fmt_expr->string.value.size;
 662                 vchar.first    = string_first;
 663                 vchar.next     = string_next;
 664                 vchar.is_digit = string_isdigit;
 665         } else {
 666                 return;
 667         }
 668         /* find the real args */
 669         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 670                 arg = arg->next;
 671
 672         const source_position_t *pos = &fmt_expr->base.source_position;
 673         unsigned fmt     = vchar.first(&vchar);
 674         unsigned num_fmt = 0;
 675         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 676                 if (fmt != '%')
 677                         continue;
 678                 fmt = vchar.next(&vchar);
 679
 680                 if (fmt == '%')
 681                         continue;
 682
 683                 ++num_fmt;
 684
 685                 /* length modifier */
 686                 format_length_modifier_t fmt_mod;
 687                 switch (fmt) {
 688                         case 'h':
 689                                 fmt = vchar.next(&vchar);
 690                                 if (fmt == 'h') {
 691                                         fmt = vchar.next(&vchar);
 692                                         fmt_mod = FMT_MOD_hh;
 693                                 } else {
 694                                         fmt_mod = FMT_MOD_h;
 695                                 }
 696                                 break;
 697
 698                         case 'l':
 699                                 fmt = vchar.next(&vchar);
 700                                 if (fmt == 'l') {
 701                                         fmt = vchar.next(&vchar);
 702                                         fmt_mod = FMT_MOD_ll;
 703                                 } else {
 704                                         fmt_mod = FMT_MOD_l;
 705                                 }
 706                                 break;
 707
 708                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 709                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 710                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 711                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 712                         /* microsoft mode */
 713                         case 'w':
 714                                 if (c_mode & _MS) {
 715                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 716                                 } else {
 717                                         fmt_mod = FMT_MOD_NONE;
 718                                 }
 719                                 break;
 720                         case 'I':
 721                                 if (c_mode & _MS) {
 722                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 723                                         if (fmt == '3') {
 724                                                 fmt = vchar.next(&vchar);
 725                                                 if (fmt == '2') {
 726                                                         fmt = vchar.next(&vchar);
 727                                                         fmt_mod = FMT_MOD_I32;
 728                                                 } else {
 729                                                         /* rewind */
 730                                                         --vchar.position;
 731                                                 }
 732                                         } else if (fmt == '6') {
 733                                                 fmt = vchar.next(&vchar);
 734                                                 if (fmt == '4') {
 735                                                         fmt = vchar.next(&vchar);
 736                                                         fmt_mod = FMT_MOD_I64;
 737                                                 } else {
 738                                                         /* rewind */
 739                                                         --vchar.position;
 740                                                 }
 741                                         }
 742                                 } else {
 743                                         fmt_mod = FMT_MOD_NONE;
 744                                 }
 745                                 break;
 746                         default:
 747                                 fmt_mod = FMT_MOD_NONE;
 748                                 break;
 749                 }
 750
 751                 if (fmt == '\0') {
 752                         warningf(pos, "dangling %% in format string");
 753                         break;
 754                 }
 755
 756                 type_t            *expected_type;
 757                 switch (fmt) {
 758                         case 'd':
 759                         case 'i':
 760                                 switch (fmt_mod) {
 761                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 762                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 763                                         case FMT_MOD_h:    expected_type = type_short;       break;
 764                                         case FMT_MOD_l:    expected_type = type_long;        break;
 765                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 766                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 767                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 768                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 769                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 770                                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 771                                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 772
 773                                         default:
 774                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 775                                                 goto next_arg;
 776                                 }
 777                                 break;
 778
 779                         case 'o':
 780                         case 'X':
 781                         case 'x':
 782                                 goto eval_fmt_mod_unsigned;
 783
 784                         case 'u':
 785 eval_fmt_mod_unsigned:
 786                                 switch (fmt_mod) {
 787                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 788                                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 789                                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 790                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 791                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 792                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 793                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 794                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 795                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 796                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 797                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 798
 799                                         default:
 800                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 801                                                 goto next_arg;
 802                                 }
 803                                 break;
 804
 805                         case 'A':
 806                         case 'a':
 807                         case 'E':
 808                         case 'e':
 809                         case 'F':
 810                         case 'f':
 811                         case 'G':
 812                         case 'g':
 813                                 switch (fmt_mod) {
 814                                         case FMT_MOD_l:    /* l modifier is ignored */
 815                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 816                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 817
 818                                         default:
 819                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 820                                                 goto next_arg;
 821                                 }
 822                                 break;
 823
 824                         case 'C':
 825                                 if (fmt_mod != FMT_MOD_NONE) {
 826                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 827                                         goto next_arg;
 828                                 }
 829                                 expected_type = type_wchar_t;
 830                                 break;
 831
 832                         case 'c':
 833                                 expected_type = type_int;
 834                                 switch (fmt_mod) {
 835                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 836                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 837                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 838
 839                                         default:
 840                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 841                                                 goto next_arg;
 842                                 }
 843                                 break;
 844
 845                         case 'S':
 846                                 if (fmt_mod != FMT_MOD_NONE) {
 847                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 848                                         goto next_arg;
 849                                 }
 850                                 expected_type = type_wchar_t;
 851                                 break;
 852
 853                         case 's':
 854                         case '[':
 855                                 switch (fmt_mod) {
 856                                         case FMT_MOD_NONE: expected_type = type_char;    break;
 857                                         case FMT_MOD_l:    expected_type = type_wchar_t; break;
 858                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 859
 860                                         default:
 861                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 862                                                 goto next_arg;
 863                                 }
 864                                 break;
 865
 866                         case 'p':
 867                                 if (fmt_mod != FMT_MOD_NONE) {
 868                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 869                                         goto next_arg;
 870                                 }
 871                                 expected_type = type_void_ptr;
 872                                 break;
 873
 874                         case 'n':
 875                                 switch (fmt_mod) {
 876                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 877                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 878                                         case FMT_MOD_h:    expected_type = type_short;       break;
 879                                         case FMT_MOD_l:    expected_type = type_long;        break;
 880                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 881                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 882                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 883                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 884
 885                                         default:
 886                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 887                                                 goto next_arg;
 888                                 }
 889                                 break;
 890
 891                         default:
 892                                 warningf(pos, "encountered unknown conversion specifier '%%%C' at position %u", (wint_t)fmt, num_fmt);
 893                                 if (arg == NULL) {
 894                                         warningf(pos, "too few arguments for format string");
 895                                         return;
 896                                 }
 897                                 goto next_arg;
 898                 }
 899
 900                 if (arg == NULL) {
 901                         warningf(pos, "too few arguments for format string");
 902                         return;
 903                 }
 904
 905                 {       /* create a scope here to prevent warning about the jump to next_arg */
 906                         type_t *const arg_type           = arg->expression->base.type;
 907                         type_t *const arg_skip           = skip_typeref(arg_type);
 908                         type_t *const expected_type_skip = skip_typeref(expected_type);
 909
 910                         if (! is_type_pointer(arg_skip))
 911                                 goto error_arg_type;
 912                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 913
 914                         if (fmt == 'p') {
 915                                 /* allow any pointer type for %p, not just void */
 916                                 if (is_type_pointer(ptr_skip))
 917                                         goto next_arg;
 918                         }
 919
 920                         /* do NOT allow const or restrict, all other should be ok */
 921                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 922                                 goto error_arg_type;
 923                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 924                         if (unqual_ptr == expected_type_skip) {
 925                                 goto next_arg;
 926                         } else if (expected_type_skip == type_char) {
 927                                 /* char matches with unsigned char AND signed char */
 928                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 929                                         goto next_arg;
 930                         }
 931 error_arg_type:
 932                         if (is_type_valid(arg_skip)) {
 933                                 warningf(pos,
 934                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 935                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 936                         }
 937                 }
 938 next_arg:
 939                 arg = arg->next;
 940         }
 941         if (!atend(&vchar)) {
 942                 warningf(pos, "format string contains '\\0'");
 943         }
 944         if (arg != NULL) {
 945                 unsigned num_args = num_fmt;
 946                 while (arg != NULL) {
 947                         ++num_args;
 948                         arg = arg->next;
 949                 }
 950                 warningf(pos, "%u argument%s but only %u format specifier%s",
 951                         num_args, num_args != 1 ? "s" : "",
 952                         num_fmt, num_fmt != 1 ? "s" : "");
 953         }
 954 }
 955
 956 static const format_spec_t builtin_table[] = {
 957         { "printf",        FORMAT_PRINTF,   0, 1 },
 958         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 959         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 960         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 961         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 962         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 963         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 964         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 965         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 966         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 967
 968         { "scanf",         FORMAT_SCANF,    0, 1 },
 969         { "wscanf",        FORMAT_SCANF,    0, 1 },
 970         { "sscanf",        FORMAT_SCANF,    1, 2 },
 971         { "swscanf",       FORMAT_SCANF,    1, 2 },
 972         { "fscanf",        FORMAT_SCANF,    1, 2 },
 973         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 974
 975         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 976         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 977
 978         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 979
 980         /* MS extensions */
 981         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 982         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 983         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 984         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 985         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 986         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 987         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 988         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 989         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 990         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 991         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 992         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 993         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 994         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 995         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 996         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 997         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 998         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 999         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
1000         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
1001         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
1002         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
1003 };
1004
1005 void check_format(const call_expression_t *const call)
1006 {
1007         if (!warning.format)
1008                 return;
1009
1010         const expression_t *const func_expr = call->function;
1011         if (func_expr->kind != EXPR_REFERENCE)
1012                 return;
1013
1014         const entity_t        *const entity = func_expr->reference.entity;
1015         const call_argument_t *      arg    = call->arguments;
1016
1017         if (false) {
1018                 /* the declaration has a GNU format attribute, check it */
1019         } else {
1020                 /*
1021                  * For some functions we always check the format, even if it was not specified.
1022                  * This allows to check format even in MS mode or without header included.
1023                  */
1024                 const char *const name = entity->base.symbol->string;
1025                 for (size_t i = 0; i < lengthof(builtin_table); ++i) {
1026                         if (strcmp(name, builtin_table[i].name) == 0) {
1027                                 switch (builtin_table[i].fmt_kind) {
1028                                 case FORMAT_PRINTF:
1029                                         check_printf_format(arg, &builtin_table[i]);
1030                                         break;
1031                                 case FORMAT_SCANF:
1032                                         check_scanf_format(arg, &builtin_table[i]);
1033                                         break;
1034                                 case FORMAT_STRFTIME:
1035                                 case FORMAT_STRFMON:
1036                                         /* TODO: implement other cases */
1037                                         break;
1038                                 }
1039                                 break;
1040                         }
1041                 }
1042         }
1043 }