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