nsz Git - cparser/blob - format_check.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2008 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 "format_check.h"
  24 #include "symbol_t.h"
  25 #include "ast_t.h"
  26 #include "entity_t.h"
  27 #include "diagnostic.h"
  28 #include "types.h"
  29 #include "type_t.h"
  30 #include "warning.h"
  31 #include "lang_features.h"
  32
  33 typedef enum format_flag_t {
  34         FMT_FLAG_NONE  = 0,
  35         FMT_FLAG_HASH  = 1U << 0,
  36         FMT_FLAG_ZERO  = 1U << 1,
  37         FMT_FLAG_MINUS = 1U << 2,
  38         FMT_FLAG_SPACE = 1U << 3,
  39         FMT_FLAG_PLUS  = 1U << 4,
  40         FMT_FLAG_TICK  = 1U << 5
  41 } format_flag_t;
  42
  43 typedef unsigned format_flags_t;
  44
  45 typedef enum format_length_modifier_t {
  46         FMT_MOD_NONE,
  47         FMT_MOD_L,
  48         FMT_MOD_hh,
  49         FMT_MOD_h,
  50         FMT_MOD_l,
  51         FMT_MOD_ll,
  52         FMT_MOD_j,
  53         FMT_MOD_t,
  54         FMT_MOD_z,
  55         FMT_MOD_q,
  56         /* only in microsoft mode */
  57         FMT_MOD_w,
  58         FMT_MOD_I,
  59         FMT_MOD_I32,
  60         FMT_MOD_I64
  61 } format_length_modifier_t;
  62
  63 typedef struct format_spec_t {
  64         const char    *name;     /**< name of the function */
  65         format_kind_t  fmt_kind; /**< kind */
  66         unsigned       fmt_idx;  /**< index of the format string */
  67         unsigned       arg_idx;  /**< index of the first argument */
  68 } format_spec_t;
  69
  70 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  71 {
  72         static const char* const names[] = {
  73                 [FMT_MOD_NONE] = "",
  74                 [FMT_MOD_L]    = "L",
  75                 [FMT_MOD_hh]   = "hh",
  76                 [FMT_MOD_h]    = "h",
  77                 [FMT_MOD_l]    = "l",
  78                 [FMT_MOD_ll]   = "ll",
  79                 [FMT_MOD_j]    = "j",
  80                 [FMT_MOD_t]    = "t",
  81                 [FMT_MOD_z]    = "z",
  82                 [FMT_MOD_q]    = "q",
  83                 /* only in microsoft mode */
  84                 [FMT_MOD_w]    = "w",
  85                 [FMT_MOD_I]    = "I",
  86                 [FMT_MOD_I32]  = "I32",
  87                 [FMT_MOD_I64]  = "I64"
  88         };
  89         assert(mod < sizeof(names) / sizeof(*names));
  90         return names[mod];
  91 }
  92
  93 static void warn_invalid_length_modifier(const source_position_t *pos,
  94                                          const format_length_modifier_t mod,
  95                                          const wchar_rep_t conversion)
  96 {
  97         warningf(pos,
  98                 "invalid length modifier '%s' for conversion specifier '%%%c'",
  99                 get_length_modifier_name(mod), conversion
 100         );
 101 }
 102
 103 typedef struct vchar_t vchar_t;
 104 struct vchar_t {
 105         const void *string;   /**< the string */
 106         size_t     position;  /**< current position */
 107         size_t     size;      /**< size of the string */
 108
 109         /** return the first character of the string and setthe position to 0. */
 110         unsigned (*first)(vchar_t *self);
 111         /** return the next character of the string */
 112         unsigned (*next)(vchar_t *self);
 113         /** return non_zero if the given character is a digit */
 114         int (*is_digit)(unsigned vchar);
 115 };
 116
 117 static unsigned string_first(vchar_t *self) {
 118         self->position = 0;
 119         const string_t *string = self->string;
 120         return string->begin[0];
 121 }
 122
 123 static unsigned string_next(vchar_t *self) {
 124         ++self->position;
 125         const string_t *string = self->string;
 126         return string->begin[self->position];
 127 }
 128
 129 static int string_isdigit(unsigned vchar) {
 130         return isdigit(vchar);
 131 }
 132
 133 static unsigned wstring_first(vchar_t *self) {
 134         self->position = 0;
 135         const wide_string_t *wstring = self->string;
 136         return wstring->begin[0];
 137 }
 138
 139 static unsigned wstring_next(vchar_t *self) {
 140         ++self->position;
 141         const wide_string_t *wstring = self->string;
 142         return wstring->begin[self->position];
 143 }
 144
 145 static int wstring_isdigit(unsigned vchar) {
 146         return iswdigit(vchar);
 147 }
 148
 149 static bool atend(vchar_t *self) {
 150         return self->position + 1 == self->size;
 151 }
 152
 153 /**
 154  * Check printf-style format.
 155  */
 156 static void check_printf_format(const call_argument_t *arg, const format_spec_t *spec)
 157 {
 158         /* find format arg */
 159         unsigned idx = 0;
 160         for (; idx < spec->fmt_idx; ++idx)
 161                 arg = arg->next;
 162
 163         const expression_t *fmt_expr = arg->expression;
 164         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 165                 fmt_expr = fmt_expr->unary.value;
 166         }
 167
 168         vchar_t vchar;
 169         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 170                 vchar.string   = &fmt_expr->wide_string.value;
 171                 vchar.size     = fmt_expr->wide_string.value.size;
 172                 vchar.first    = wstring_first;
 173                 vchar.next     = wstring_next;
 174                 vchar.is_digit = wstring_isdigit;
 175         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 176                 vchar.string   = &fmt_expr->string.value;
 177                 vchar.size     = fmt_expr->string.value.size;
 178                 vchar.first    = string_first;
 179                 vchar.next     = string_next;
 180                 vchar.is_digit = string_isdigit;
 181         } else {
 182                 return;
 183         }
 184         /* find the real args */
 185         for(; idx < spec->arg_idx; ++idx)
 186                 arg = arg->next;
 187
 188         const source_position_t *pos = &fmt_expr->base.source_position;
 189         unsigned fmt     = vchar.first(&vchar);
 190         unsigned num_fmt = 0;
 191         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 192                 if (fmt != '%')
 193                         continue;
 194                 fmt = vchar.next(&vchar);
 195
 196                 if (fmt == '%')
 197                         continue;
 198
 199                 ++num_fmt;
 200
 201                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 202                 if (fmt == '0') {
 203                         fmt = vchar.next(&vchar);
 204                         fmt_flags |= FMT_FLAG_ZERO;
 205                 }
 206
 207                 /* argument selector or minimum field width */
 208                 if (vchar.is_digit(fmt)) {
 209                         do {
 210                                 fmt = vchar.next(&vchar);
 211                         } while (vchar.is_digit(fmt));
 212
 213                         /* digit string was ... */
 214                         if (fmt == '$') {
 215                                 /* ... argument selector */
 216                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 217                                 /* TODO implement */
 218                                 return;
 219                         }
 220                         /* ... minimum field width */
 221                 } else {
 222                         /* flags */
 223                         for (;;) {
 224                                 format_flags_t flag;
 225                                 switch (fmt) {
 226                                         case '#':  flag = FMT_FLAG_HASH;  break;
 227                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 228                                         case '-':  flag = FMT_FLAG_MINUS; break;
 229                                         case '\'': flag = FMT_FLAG_TICK;  break;
 230
 231                                         case ' ':
 232                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 233                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 234                                                 }
 235                                                 flag = FMT_FLAG_SPACE;
 236                                                 break;
 237
 238                                         case '+':
 239                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 240                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 241                                                 }
 242                                                 flag = FMT_FLAG_PLUS;
 243                                                 break;
 244
 245                                         default: goto break_fmt_flags;
 246                                 }
 247                                 if (fmt_flags & flag) {
 248                                         warningf(pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 249                                 }
 250                                 fmt_flags |= flag;
 251                                 fmt = vchar.next(&vchar);
 252                         }
 253 break_fmt_flags:
 254
 255                         /* minimum field width */
 256                         if (fmt == '*') {
 257                                 fmt = vchar.next(&vchar);
 258                                 if (arg == NULL) {
 259                                         warningf(pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 260                                         return;
 261                                 }
 262                                 const type_t *const arg_type = arg->expression->base.type;
 263                                 if (arg_type != type_int) {
 264                                         warningf(pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 265                                 }
 266                                 arg = arg->next;
 267                         } else {
 268                                 while (vchar.is_digit(fmt)) {
 269                                         fmt = vchar.next(&vchar);
 270                                 }
 271                         }
 272                 }
 273
 274                 /* precision */
 275                 if (fmt == '.') {
 276                         fmt = vchar.next(&vchar);
 277                         if (fmt == '*') {
 278                                 fmt = vchar.next(&vchar);
 279                                 if (arg == NULL) {
 280                                         warningf(pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 281                                         return;
 282                                 }
 283                                 const type_t *const arg_type = arg->expression->base.type;
 284                                 if (arg_type != type_int) {
 285                                         warningf(pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 286                                 }
 287                                 arg = arg->next;
 288                         } else {
 289                                 /* digit string may be omitted */
 290                                 while (vchar.is_digit(fmt)) {
 291                                         fmt = vchar.next(&vchar);
 292                                 }
 293                         }
 294                 }
 295
 296                 /* length modifier */
 297                 format_length_modifier_t fmt_mod;
 298                 switch (fmt) {
 299                         case 'h':
 300                                 fmt = vchar.next(&vchar);
 301                                 if (fmt == 'h') {
 302                                         fmt = vchar.next(&vchar);
 303                                         fmt_mod = FMT_MOD_hh;
 304                                 } else {
 305                                         fmt_mod = FMT_MOD_h;
 306                                 }
 307                                 break;
 308
 309                         case 'l':
 310                                 fmt = vchar.next(&vchar);
 311                                 if (fmt == 'l') {
 312                                         fmt = vchar.next(&vchar);
 313                                         fmt_mod = FMT_MOD_ll;
 314                                 } else {
 315                                         fmt_mod = FMT_MOD_l;
 316                                 }
 317                                 break;
 318
 319                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 320                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 321                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 322                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 323                         case 'q': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_q;    break;
 324                         /* microsoft mode */
 325                         case 'w':
 326                                 if (c_mode & _MS) {
 327                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 328                                 } else {
 329                                         fmt_mod = FMT_MOD_NONE;
 330                                 }
 331                                 break;
 332                         case 'I':
 333                                 if (c_mode & _MS) {
 334                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 335                                         if (fmt == '3') {
 336                                                 fmt = vchar.next(&vchar);
 337                                                 if (fmt == '2') {
 338                                                         fmt = vchar.next(&vchar);
 339                                                         fmt_mod = FMT_MOD_I32;
 340                                                 } else {
 341                                                         /* rewind */
 342                                                         --vchar.position;
 343                                                 }
 344                                         } else if (fmt == '6') {
 345                                                 fmt = vchar.next(&vchar);
 346                                                 if (fmt == '4') {
 347                                                         fmt = vchar.next(&vchar);
 348                                                         fmt_mod = FMT_MOD_I64;
 349                                                 } else {
 350                                                         /* rewind */
 351                                                         --vchar.position;
 352                                                 }
 353                                         }
 354                                 } else {
 355                                         fmt_mod = FMT_MOD_NONE;
 356                                 }
 357                                 break;
 358                         default:
 359                                 fmt_mod = FMT_MOD_NONE;
 360                                 break;
 361                 }
 362
 363                 if (fmt == '\0') {
 364                         warningf(pos, "dangling %% in format string");
 365                         break;
 366                 }
 367
 368                 type_t            *expected_type;
 369                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 370                 format_flags_t     allowed_flags;
 371                 switch (fmt) {
 372                         case 'd':
 373                         case 'i':
 374                                 switch (fmt_mod) {
 375                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 376                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 377                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 378                                         case FMT_MOD_l:    expected_type = type_long;      break;
 379                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 380                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 381                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 382                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 383                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 384                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 385                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 386
 387                                         default:
 388                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 389                                                 goto next_arg;
 390                                 }
 391                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 392                                 break;
 393
 394                         case 'o':
 395                         case 'X':
 396                         case 'x':
 397                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 398                                 goto eval_fmt_mod_unsigned;
 399
 400                         case 'u':
 401                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 402 eval_fmt_mod_unsigned:
 403                                 switch (fmt_mod) {
 404                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 405                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 406                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 407                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 408                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 409                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 410                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 411                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 412                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 413                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 414                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 415
 416                                         default:
 417                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 418                                                 goto next_arg;
 419                                 }
 420                                 break;
 421
 422                         case 'A':
 423                         case 'a':
 424                         case 'E':
 425                         case 'e':
 426                         case 'F':
 427                         case 'f':
 428                         case 'G':
 429                         case 'g':
 430                                 switch (fmt_mod) {
 431                                         case FMT_MOD_l:    /* l modifier is ignored */
 432                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 433                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 434
 435                                         default:
 436                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 437                                                 goto next_arg;
 438                                 }
 439                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 440                                 break;
 441
 442                         case 'C':
 443                                 if (fmt_mod != FMT_MOD_NONE) {
 444                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 445                                         goto next_arg;
 446                                 }
 447                                 expected_type = type_wchar_t;
 448                                 allowed_flags = FMT_FLAG_NONE;
 449                                 break;
 450
 451                         case 'c':
 452                                 expected_type = type_int;
 453                                 switch (fmt_mod) {
 454                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 455                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 456                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 457
 458                                         default:
 459                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 460                                                 goto next_arg;
 461                                 }
 462                                 allowed_flags = FMT_FLAG_NONE;
 463                                 break;
 464
 465                         case 'S':
 466                                 if (fmt_mod != FMT_MOD_NONE) {
 467                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 468                                         goto next_arg;
 469                                 }
 470                                 expected_type = type_wchar_t_ptr;
 471                                 expected_qual = TYPE_QUALIFIER_CONST;
 472                                 allowed_flags = FMT_FLAG_MINUS;
 473                                 break;
 474
 475                         case 's':
 476                                 switch (fmt_mod) {
 477                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 478                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 479                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 480
 481                                         default:
 482                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 483                                                 goto next_arg;
 484                                 }
 485                                 expected_qual = TYPE_QUALIFIER_CONST;
 486                                 allowed_flags = FMT_FLAG_MINUS;
 487                                 break;
 488
 489                         case 'p':
 490                                 if (fmt_mod != FMT_MOD_NONE) {
 491                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 492                                         goto next_arg;
 493                                 }
 494                                 expected_type = type_void_ptr;
 495                                 allowed_flags = FMT_FLAG_NONE;
 496                                 break;
 497
 498                         case 'n':
 499                                 switch (fmt_mod) {
 500                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 501                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 502                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 503                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 504                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 505                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 506                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 507                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 508
 509                                         default:
 510                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 511                                                 goto next_arg;
 512                                 }
 513                                 allowed_flags = FMT_FLAG_NONE;
 514                                 break;
 515
 516                         default:
 517                                 warningf(pos, "encountered unknown conversion specifier '%%%C' at position %u", (wint_t)fmt, num_fmt);
 518                                 goto next_arg;
 519                 }
 520
 521                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 522                 if (wrong_flags != 0) {
 523                         char wrong[8];
 524                         int idx = 0;
 525                         if (wrong_flags & FMT_FLAG_HASH)  wrong[idx++] = '#';
 526                         if (wrong_flags & FMT_FLAG_ZERO)  wrong[idx++] = '0';
 527                         if (wrong_flags & FMT_FLAG_MINUS) wrong[idx++] = '-';
 528                         if (wrong_flags & FMT_FLAG_SPACE) wrong[idx++] = ' ';
 529                         if (wrong_flags & FMT_FLAG_PLUS)  wrong[idx++] = '+';
 530                         if (wrong_flags & FMT_FLAG_TICK)  wrong[idx++] = '\'';
 531                         wrong[idx] = '\0';
 532
 533                         warningf(pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 534                 }
 535
 536                 if (arg == NULL) {
 537                         warningf(pos, "too few arguments for format string");
 538                         return;
 539                 }
 540
 541                 {       /* create a scope here to prevent warning about the jump to next_arg */
 542                         type_t *const arg_type           = arg->expression->base.type;
 543                         type_t *const arg_skip           = skip_typeref(arg_type);
 544                         type_t *const expected_type_skip = skip_typeref(expected_type);
 545
 546                         if (fmt == 'p') {
 547                                 /* allow any pointer type for %p, not just void */
 548                                 if (is_type_pointer(arg_skip))
 549                                         goto next_arg;
 550                         }
 551
 552                         if (is_type_pointer(expected_type_skip)) {
 553                                 if (is_type_pointer(arg_skip)) {
 554                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 555                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 556                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 557                                                 get_unqualified_type(arg_to) == exp_to) {
 558                                                 goto next_arg;
 559                                         }
 560                                 }
 561                         } else {
 562                                 if (get_unqualified_type(arg_skip) == expected_type_skip) {
 563                                         goto next_arg;
 564                                 }
 565                         }
 566                         if (is_type_valid(arg_skip)) {
 567                                 warningf(pos,
 568                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 569                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 570                         }
 571                 }
 572 next_arg:
 573                 arg = arg->next;
 574         }
 575         if (!atend(&vchar)) {
 576                 warningf(pos, "format string contains NUL");
 577         }
 578         if (arg != NULL) {
 579                 unsigned num_args = num_fmt;
 580                 while (arg != NULL) {
 581                         ++num_args;
 582                         arg = arg->next;
 583                 }
 584                 warningf(pos, "%u argument%s but only %u format string%s",
 585                         num_args, num_args != 1 ? "s" : "",
 586                         num_fmt, num_fmt != 1 ? "s" : "");
 587         }
 588 }
 589
 590 /**
 591  * Check scanf-style format.
 592  */
 593 static void check_scanf_format(const call_argument_t *arg, const format_spec_t *spec)
 594 {
 595         /* find format arg */
 596         unsigned idx = 0;
 597         for (; idx < spec->fmt_idx; ++idx)
 598                 arg = arg->next;
 599
 600         const expression_t *fmt_expr = arg->expression;
 601         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 602                 fmt_expr = fmt_expr->unary.value;
 603         }
 604
 605         vchar_t vchar;
 606         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 607                 vchar.string   = &fmt_expr->wide_string.value;
 608                 vchar.size     = fmt_expr->wide_string.value.size;
 609                 vchar.first    = wstring_first;
 610                 vchar.next     = wstring_next;
 611                 vchar.is_digit = wstring_isdigit;
 612         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 613                 vchar.string   = &fmt_expr->string.value;
 614                 vchar.size     = fmt_expr->string.value.size;
 615                 vchar.first    = string_first;
 616                 vchar.next     = string_next;
 617                 vchar.is_digit = string_isdigit;
 618         } else {
 619                 return;
 620         }
 621         /* find the real args */
 622         for (; idx < spec->arg_idx; ++idx)
 623                 arg = arg->next;
 624
 625         const source_position_t *pos = &fmt_expr->base.source_position;
 626         unsigned fmt     = vchar.first(&vchar);
 627         unsigned num_fmt = 0;
 628         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 629                 if (fmt != '%')
 630                         continue;
 631                 fmt = vchar.next(&vchar);
 632
 633                 if (fmt == '%')
 634                         continue;
 635
 636                 ++num_fmt;
 637
 638                 /* length modifier */
 639                 format_length_modifier_t fmt_mod;
 640                 switch (fmt) {
 641                         case 'h':
 642                                 fmt = vchar.next(&vchar);
 643                                 if (fmt == 'h') {
 644                                         fmt = vchar.next(&vchar);
 645                                         fmt_mod = FMT_MOD_hh;
 646                                 } else {
 647                                         fmt_mod = FMT_MOD_h;
 648                                 }
 649                                 break;
 650
 651                         case 'l':
 652                                 fmt = vchar.next(&vchar);
 653                                 if (fmt == 'l') {
 654                                         fmt = vchar.next(&vchar);
 655                                         fmt_mod = FMT_MOD_ll;
 656                                 } else {
 657                                         fmt_mod = FMT_MOD_l;
 658                                 }
 659                                 break;
 660
 661                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 662                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 663                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 664                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 665                         /* microsoft mode */
 666                         case 'w':
 667                                 if (c_mode & _MS) {
 668                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 669                                 } else {
 670                                         fmt_mod = FMT_MOD_NONE;
 671                                 }
 672                                 break;
 673                         case 'I':
 674                                 if (c_mode & _MS) {
 675                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 676                                         if (fmt == '3') {
 677                                                 fmt = vchar.next(&vchar);
 678                                                 if (fmt == '2') {
 679                                                         fmt = vchar.next(&vchar);
 680                                                         fmt_mod = FMT_MOD_I32;
 681                                                 } else {
 682                                                         /* rewind */
 683                                                         --vchar.position;
 684                                                 }
 685                                         } else if (fmt == '6') {
 686                                                 fmt = vchar.next(&vchar);
 687                                                 if (fmt == '4') {
 688                                                         fmt = vchar.next(&vchar);
 689                                                         fmt_mod = FMT_MOD_I64;
 690                                                 } else {
 691                                                         /* rewind */
 692                                                         --vchar.position;
 693                                                 }
 694                                         }
 695                                 } else {
 696                                         fmt_mod = FMT_MOD_NONE;
 697                                 }
 698                                 break;
 699                         default:
 700                                 fmt_mod = FMT_MOD_NONE;
 701                                 break;
 702                 }
 703
 704                 if (fmt == '\0') {
 705                         warningf(pos, "dangling %% in format string");
 706                         break;
 707                 }
 708
 709                 type_t            *expected_type;
 710                 switch (fmt) {
 711                         case 'd':
 712                         case 'i':
 713                                 switch (fmt_mod) {
 714                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 715                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 716                                         case FMT_MOD_h:    expected_type = type_short;       break;
 717                                         case FMT_MOD_l:    expected_type = type_long;        break;
 718                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 719                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 720                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 721                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 722                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 723                                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 724                                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 725
 726                                         default:
 727                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 728                                                 goto next_arg;
 729                                 }
 730                                 break;
 731
 732                         case 'o':
 733                         case 'X':
 734                         case 'x':
 735                                 goto eval_fmt_mod_unsigned;
 736
 737                         case 'u':
 738 eval_fmt_mod_unsigned:
 739                                 switch (fmt_mod) {
 740                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 741                                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 742                                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 743                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 744                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 745                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 746                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 747                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 748                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 749                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 750                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 751
 752                                         default:
 753                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 754                                                 goto next_arg;
 755                                 }
 756                                 break;
 757
 758                         case 'A':
 759                         case 'a':
 760                         case 'E':
 761                         case 'e':
 762                         case 'F':
 763                         case 'f':
 764                         case 'G':
 765                         case 'g':
 766                                 switch (fmt_mod) {
 767                                         case FMT_MOD_l:    /* l modifier is ignored */
 768                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 769                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 770
 771                                         default:
 772                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 773                                                 goto next_arg;
 774                                 }
 775                                 break;
 776
 777                         case 'C':
 778                                 if (fmt_mod != FMT_MOD_NONE) {
 779                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 780                                         goto next_arg;
 781                                 }
 782                                 expected_type = type_wchar_t;
 783                                 break;
 784
 785                         case 'c':
 786                                 expected_type = type_int;
 787                                 switch (fmt_mod) {
 788                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 789                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 790                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 791
 792                                         default:
 793                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 794                                                 goto next_arg;
 795                                 }
 796                                 break;
 797
 798                         case 'S':
 799                                 if (fmt_mod != FMT_MOD_NONE) {
 800                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 801                                         goto next_arg;
 802                                 }
 803                                 expected_type = type_wchar_t;
 804                                 break;
 805
 806                         case 's':
 807                         case '[':
 808                                 switch (fmt_mod) {
 809                                         case FMT_MOD_NONE: expected_type = type_char;    break;
 810                                         case FMT_MOD_l:    expected_type = type_wchar_t; break;
 811                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 812
 813                                         default:
 814                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 815                                                 goto next_arg;
 816                                 }
 817                                 break;
 818
 819                         case 'p':
 820                                 if (fmt_mod != FMT_MOD_NONE) {
 821                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 822                                         goto next_arg;
 823                                 }
 824                                 expected_type = type_void_ptr;
 825                                 break;
 826
 827                         case 'n':
 828                                 switch (fmt_mod) {
 829                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 830                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 831                                         case FMT_MOD_h:    expected_type = type_short;       break;
 832                                         case FMT_MOD_l:    expected_type = type_long;        break;
 833                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 834                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 835                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 836                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 837
 838                                         default:
 839                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 840                                                 goto next_arg;
 841                                 }
 842                                 break;
 843
 844                         default:
 845                                 warningf(pos, "encountered unknown conversion specifier '%%%C' at position %u", (wint_t)fmt, num_fmt);
 846                                 goto next_arg;
 847                 }
 848
 849                 if (arg == NULL) {
 850                         warningf(pos, "too few arguments for format string");
 851                         return;
 852                 }
 853
 854                 {       /* create a scope here to prevent warning about the jump to next_arg */
 855                         type_t *const arg_type           = arg->expression->base.type;
 856                         type_t *const arg_skip           = skip_typeref(arg_type);
 857                         type_t *const expected_type_skip = skip_typeref(expected_type);
 858
 859                         if (! is_type_pointer(arg_skip))
 860                                 goto error_arg_type;
 861                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 862
 863                         if (fmt == 'p') {
 864                                 /* allow any pointer type for %p, not just void */
 865                                 if (is_type_pointer(ptr_skip))
 866                                         goto next_arg;
 867                         }
 868
 869                         if (ptr_skip == expected_type_skip) {
 870                                 goto next_arg;
 871                         } else if (expected_type_skip == type_char) {
 872                                 /* char matches with unsigned char AND signed char */
 873                                 if (ptr_skip == type_signed_char || ptr_skip == type_unsigned_char)
 874                                         goto next_arg;
 875                         }
 876 error_arg_type:
 877                         if (is_type_valid(arg_skip)) {
 878                                 warningf(pos,
 879                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 880                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 881                         }
 882                 }
 883 next_arg:
 884                 arg = arg->next;
 885         }
 886         if (!atend(&vchar)) {
 887                 warningf(pos, "format string contains NUL");
 888         }
 889         if (arg != NULL) {
 890                 unsigned num_args = num_fmt;
 891                 while (arg != NULL) {
 892                         ++num_args;
 893                         arg = arg->next;
 894                 }
 895                 warningf(pos, "%u argument%s but only %u format string%s",
 896                         num_args, num_args != 1 ? "s" : "",
 897                         num_fmt, num_fmt != 1 ? "s" : "");
 898         }
 899 }
 900
 901 static const format_spec_t builtin_table[] = {
 902         { "printf",        FORMAT_PRINTF,   0, 1 },
 903         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 904         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 905         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 906         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 907         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 908         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 909         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 910         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 911         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 912
 913         { "scanf",         FORMAT_SCANF,    0, 1 },
 914         { "wscanf",        FORMAT_SCANF,    0, 1 },
 915         { "sscanf",        FORMAT_SCANF,    1, 2 },
 916         { "swscanf",       FORMAT_SCANF,    1, 2 },
 917         { "fscanf",        FORMAT_SCANF,    1, 2 },
 918         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 919
 920         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 921         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 922
 923         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 924
 925         /* MS extensions */
 926         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 927         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 928         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 929         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 930         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 931         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 932         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 933         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 934         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 935         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 936         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 937         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 938         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 939         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 940         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 941         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 942         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 943         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 944         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 945         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 946         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 947         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 948 };
 949
 950 void check_format(const call_expression_t *const call)
 951 {
 952         if (!warning.format)
 953                 return;
 954
 955         const expression_t *const func_expr = call->function;
 956         if (func_expr->kind != EXPR_REFERENCE)
 957                 return;
 958
 959         const entity_t        *const entity = func_expr->reference.entity;
 960         const call_argument_t *      arg    = call->arguments;
 961
 962         if (false) {
 963                 /* the declaration has a GNU format attribute, check it */
 964         } else {
 965                 /*
 966                  * For some functions we always check the format, even if it was not specified.
 967                  * This allows to check format even in MS mode or without header included.
 968                  */
 969                 const char *const name = entity->base.symbol->string;
 970                 for (size_t i = 0; i < sizeof(builtin_table) / sizeof(builtin_table[0]); ++i) {
 971                         if (strcmp(name, builtin_table[i].name) == 0) {
 972                                 switch (builtin_table[i].fmt_kind) {
 973                                 case FORMAT_PRINTF:
 974                                         check_printf_format(arg, &builtin_table[i]);
 975                                         break;
 976                                 case FORMAT_SCANF:
 977                                         check_scanf_format(arg, &builtin_table[i]);
 978                                         break;
 979                                 }
 980                                 break;
 981                         }
 982                 }
 983         }
 984 }