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 <config.h>
  21
  22 #include <ctype.h>
  23
  24 #include "adt/util.h"
  25 #include "format_check.h"
  26 #include "symbol_t.h"
  27 #include "ast_t.h"
  28 #include "entity_t.h"
  29 #include "diagnostic.h"
  30 #include "types.h"
  31 #include "type_t.h"
  32 #include "warning.h"
  33 #include "lang_features.h"
  34
  35 typedef enum format_flag_t {
  36         FMT_FLAG_NONE  = 0,
  37         FMT_FLAG_HASH  = 1U << 0,
  38         FMT_FLAG_ZERO  = 1U << 1,
  39         FMT_FLAG_MINUS = 1U << 2,
  40         FMT_FLAG_SPACE = 1U << 3,
  41         FMT_FLAG_PLUS  = 1U << 4,
  42         FMT_FLAG_TICK  = 1U << 5
  43 } format_flag_t;
  44
  45 typedef unsigned format_flags_t;
  46
  47 typedef enum format_length_modifier_t {
  48         FMT_MOD_NONE,
  49         FMT_MOD_L,
  50         FMT_MOD_hh,
  51         FMT_MOD_h,
  52         FMT_MOD_l,
  53         FMT_MOD_ll,
  54         FMT_MOD_j,
  55         FMT_MOD_t,
  56         FMT_MOD_z,
  57         FMT_MOD_q,
  58         /* only in microsoft mode */
  59         FMT_MOD_w,
  60         FMT_MOD_I,
  61         FMT_MOD_I32,
  62         FMT_MOD_I64
  63 } format_length_modifier_t;
  64
  65 typedef struct format_spec_t {
  66         const char    *name;     /**< name of the function */
  67         format_kind_t  fmt_kind; /**< kind */
  68         unsigned       fmt_idx;  /**< index of the format string */
  69         unsigned       arg_idx;  /**< index of the first argument */
  70 } format_spec_t;
  71
  72 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  73 {
  74         static const char* const names[] = {
  75                 [FMT_MOD_NONE] = "",
  76                 [FMT_MOD_L]    = "L",
  77                 [FMT_MOD_hh]   = "hh",
  78                 [FMT_MOD_h]    = "h",
  79                 [FMT_MOD_l]    = "l",
  80                 [FMT_MOD_ll]   = "ll",
  81                 [FMT_MOD_j]    = "j",
  82                 [FMT_MOD_t]    = "t",
  83                 [FMT_MOD_z]    = "z",
  84                 [FMT_MOD_q]    = "q",
  85                 /* only in microsoft mode */
  86                 [FMT_MOD_w]    = "w",
  87                 [FMT_MOD_I]    = "I",
  88                 [FMT_MOD_I32]  = "I32",
  89                 [FMT_MOD_I64]  = "I64"
  90         };
  91         assert(mod < lengthof(names));
  92         return names[mod];
  93 }
  94
  95 static void warn_invalid_length_modifier(const source_position_t *pos,
  96                                          const format_length_modifier_t mod,
  97                                          const utf32 conversion)
  98 {
  99         warningf(pos,
 100                 "invalid length modifier '%s' for conversion specifier '%%%c'",
 101                 get_length_modifier_name(mod), conversion
 102         );
 103 }
 104
 105 /**
 106  * Check printf-style format.
 107  */
 108 static int internal_check_printf_format(const expression_t *fmt_expr,
 109                                         const call_argument_t *arg,
 110                                         const format_spec_t *spec)
 111 {
 112         while (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 113                 fmt_expr = fmt_expr->unary.value;
 114         }
 115
 116         /*
 117          * gettext results in expressions like (X ? "format_string" : Y)
 118          * we assume the left part is the format string
 119          */
 120         if (fmt_expr->kind == EXPR_CONDITIONAL) {
 121                 conditional_expression_t const *const c = &fmt_expr->conditional;
 122                 expression_t             const *      t = c->true_expression;
 123                 if (t == NULL)
 124                         t = c->condition;
 125                 int const nt = internal_check_printf_format(t,                   arg, spec);
 126                 int const nf = internal_check_printf_format(c->false_expression, arg, spec);
 127                 return nt > nf ? nt : nf;
 128         }
 129
 130         if (fmt_expr->kind != EXPR_STRING_LITERAL
 131                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 132                 return -1;
 133
 134         const char *string = fmt_expr->literal.value.begin;
 135         size_t      size   = fmt_expr->literal.value.size;
 136         const char *c      = string;
 137
 138         const source_position_t *pos = &fmt_expr->base.source_position;
 139         unsigned num_fmt = 0;
 140         char     fmt;
 141         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 142                 if (fmt != '%')
 143                         continue;
 144                 fmt = *(++c);
 145
 146                 if (fmt == '\0') {
 147                         warningf(pos, "dangling %% in format string");
 148                         break;
 149                 }
 150                 if (fmt == '%')
 151                         continue;
 152
 153                 ++num_fmt;
 154
 155                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 156                 if (fmt == '0') {
 157                         fmt = *(++c);
 158                         fmt_flags |= FMT_FLAG_ZERO;
 159                 }
 160
 161                 /* argument selector or minimum field width */
 162                 if (isdigit(fmt)) {
 163                         do {
 164                                 fmt = *(++c);
 165                         } while (isdigit(fmt));
 166
 167                         /* digit string was ... */
 168                         if (fmt == '$') {
 169                                 /* ... argument selector */
 170                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 171                                 /* TODO implement */
 172                                 return -1;
 173                         }
 174                         /* ... minimum field width */
 175                 } else {
 176                         /* flags */
 177                         for (;;) {
 178                                 format_flags_t flag;
 179                                 switch (fmt) {
 180                                         case '#':  flag = FMT_FLAG_HASH;  break;
 181                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 182                                         case '-':  flag = FMT_FLAG_MINUS; break;
 183                                         case '\'': flag = FMT_FLAG_TICK;  break;
 184
 185                                         case ' ':
 186                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 187                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 188                                                 }
 189                                                 flag = FMT_FLAG_SPACE;
 190                                                 break;
 191
 192                                         case '+':
 193                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 194                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 195                                                 }
 196                                                 flag = FMT_FLAG_PLUS;
 197                                                 break;
 198
 199                                         default: goto break_fmt_flags;
 200                                 }
 201                                 if (fmt_flags & flag) {
 202                                         warningf(pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 203                                 }
 204                                 fmt_flags |= flag;
 205                                 fmt = *(++c);
 206                         }
 207 break_fmt_flags:
 208
 209                         /* minimum field width */
 210                         if (fmt == '*') {
 211                                 fmt = *(++c);
 212                                 if (arg == NULL) {
 213                                         warningf(pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 214                                         return -1;
 215                                 }
 216                                 const type_t *const arg_type = arg->expression->base.type;
 217                                 if (arg_type != type_int) {
 218                                         warningf(pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 219                                 }
 220                                 arg = arg->next;
 221                         } else {
 222                                 while (isdigit(fmt)) {
 223                                         fmt = *(++c);
 224                                 }
 225                         }
 226                 }
 227
 228                 /* precision */
 229                 if (fmt == '.') {
 230                         fmt = *(++c);
 231                         if (fmt == '*') {
 232                                 fmt = *(++c);
 233                                 if (arg == NULL) {
 234                                         warningf(pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 235                                         return -1;
 236                                 }
 237                                 const type_t *const arg_type = arg->expression->base.type;
 238                                 if (arg_type != type_int) {
 239                                         warningf(pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 240                                 }
 241                                 arg = arg->next;
 242                         } else {
 243                                 /* digit string may be omitted */
 244                                 while (isdigit(fmt)) {
 245                                         fmt = *(++c);
 246                                 }
 247                         }
 248                 }
 249
 250                 /* length modifier */
 251                 format_length_modifier_t fmt_mod;
 252                 switch (fmt) {
 253                         case 'h':
 254                                 fmt = *(++c);
 255                                 if (fmt == 'h') {
 256                                         fmt = *(++c);
 257                                         fmt_mod = FMT_MOD_hh;
 258                                 } else {
 259                                         fmt_mod = FMT_MOD_h;
 260                                 }
 261                                 break;
 262
 263                         case 'l':
 264                                 fmt = *(++c);
 265                                 if (fmt == 'l') {
 266                                         fmt = *(++c);
 267                                         fmt_mod = FMT_MOD_ll;
 268                                 } else {
 269                                         fmt_mod = FMT_MOD_l;
 270                                 }
 271                                 break;
 272
 273                         case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
 274                         case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
 275                         case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
 276                         case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
 277                         case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 278                         /* microsoft mode */
 279                         case 'w':
 280                                 if (c_mode & _MS) {
 281                                         fmt = *(++c); fmt_mod = FMT_MOD_w;
 282                                 } else {
 283                                         fmt_mod = FMT_MOD_NONE;
 284                                 }
 285                                 break;
 286                         case 'I':
 287                                 if (c_mode & _MS) {
 288                                         fmt = *(++c); fmt_mod = FMT_MOD_I;
 289                                         if (fmt == '3') {
 290                                                 fmt = *(++c);
 291                                                 if (fmt == '2') {
 292                                                         fmt = *(++c);
 293                                                         fmt_mod = FMT_MOD_I32;
 294                                                 } else {
 295                                                         /* rewind */
 296                                                         fmt = *(--c);
 297                                                 }
 298                                         } else if (fmt == '6') {
 299                                                 fmt = *(++c);
 300                                                 if (fmt == '4') {
 301                                                         fmt = *(++c);
 302                                                         fmt_mod = FMT_MOD_I64;
 303                                                 } else {
 304                                                         /* rewind */
 305                                                         fmt = *(--c);
 306                                                 }
 307                                         }
 308                                 } else {
 309                                         fmt_mod = FMT_MOD_NONE;
 310                                 }
 311                                 break;
 312                         default:
 313                                 fmt_mod = FMT_MOD_NONE;
 314                                 break;
 315                 }
 316
 317
 318                 type_t            *expected_type;
 319                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 320                 format_flags_t     allowed_flags;
 321                 switch (fmt) {
 322                         case 'd':
 323                         case 'i':
 324                                 switch (fmt_mod) {
 325                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 326                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 327                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 328                                         case FMT_MOD_l:    expected_type = type_long;      break;
 329                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 330                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 331                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 332                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 333                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 334                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 335                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 336
 337                                         default:
 338                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 339                                                 goto next_arg;
 340                                 }
 341                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 342                                 break;
 343
 344                         case 'o':
 345                         case 'X':
 346                         case 'x':
 347                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 348                                 goto eval_fmt_mod_unsigned;
 349
 350                         case 'u':
 351                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 352 eval_fmt_mod_unsigned:
 353                                 switch (fmt_mod) {
 354                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 355                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 356                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 357                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 358                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 359                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 360                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 361                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 362                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 363                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 364                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 365
 366                                         default:
 367                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 368                                                 goto next_arg;
 369                                 }
 370                                 break;
 371
 372                         case 'A':
 373                         case 'a':
 374                         case 'E':
 375                         case 'e':
 376                         case 'F':
 377                         case 'f':
 378                         case 'G':
 379                         case 'g':
 380                                 switch (fmt_mod) {
 381                                         case FMT_MOD_l:    /* l modifier is ignored */
 382                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 383                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 384
 385                                         default:
 386                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 387                                                 goto next_arg;
 388                                 }
 389                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 390                                 break;
 391
 392                         case 'C':
 393                                 if (fmt_mod != FMT_MOD_NONE) {
 394                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 395                                         goto next_arg;
 396                                 }
 397                                 expected_type = type_wchar_t;
 398                                 allowed_flags = FMT_FLAG_NONE;
 399                                 break;
 400
 401                         case 'c':
 402                                 expected_type = type_int;
 403                                 switch (fmt_mod) {
 404                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 405                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 406                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 407
 408                                         default:
 409                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 410                                                 goto next_arg;
 411                                 }
 412                                 allowed_flags = FMT_FLAG_NONE;
 413                                 break;
 414
 415                         case 'S':
 416                                 if (fmt_mod != FMT_MOD_NONE) {
 417                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 418                                         goto next_arg;
 419                                 }
 420                                 expected_type = type_wchar_t_ptr;
 421                                 expected_qual = TYPE_QUALIFIER_CONST;
 422                                 allowed_flags = FMT_FLAG_MINUS;
 423                                 break;
 424
 425                         case 's':
 426                                 switch (fmt_mod) {
 427                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 428                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 429                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 430
 431                                         default:
 432                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 433                                                 goto next_arg;
 434                                 }
 435                                 expected_qual = TYPE_QUALIFIER_CONST;
 436                                 allowed_flags = FMT_FLAG_MINUS;
 437                                 break;
 438
 439                         case 'p':
 440                                 if (fmt_mod != FMT_MOD_NONE) {
 441                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 442                                         goto next_arg;
 443                                 }
 444                                 expected_type = type_void_ptr;
 445                                 allowed_flags = FMT_FLAG_NONE;
 446                                 break;
 447
 448                         case 'n':
 449                                 switch (fmt_mod) {
 450                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 451                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 452                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 453                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 454                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 455                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 456                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 457                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 458
 459                                         default:
 460                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 461                                                 goto next_arg;
 462                                 }
 463                                 allowed_flags = FMT_FLAG_NONE;
 464                                 break;
 465
 466                         default:
 467                                 warningf(pos, "encountered unknown conversion specifier '%%%c' at position %u", fmt, num_fmt);
 468                                 if (arg == NULL) {
 469                                         warningf(pos, "too few arguments for format string");
 470                                         return -1;
 471                                 }
 472                                 goto next_arg;
 473                 }
 474
 475                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 476                 if (wrong_flags != 0) {
 477                         char  wrong[8];
 478                         char *p = wrong;
 479                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 480                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 481                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 482                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 483                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 484                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 485                         *p = '\0';
 486
 487                         warningf(pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 488                 }
 489
 490                 if (arg == NULL) {
 491                         warningf(pos, "too few arguments for format string");
 492                         return -1;
 493                 }
 494
 495                 {       /* create a scope here to prevent warning about the jump to next_arg */
 496                         type_t *const arg_type           = arg->expression->base.type;
 497                         type_t *const arg_skip           = skip_typeref(arg_type);
 498                         type_t *const expected_type_skip = skip_typeref(expected_type);
 499
 500                         if (fmt == 'p') {
 501                                 /* allow any pointer type for %p, not just void */
 502                                 if (is_type_pointer(arg_skip))
 503                                         goto next_arg;
 504                         }
 505
 506                         if (is_type_pointer(expected_type_skip)) {
 507                                 if (is_type_pointer(arg_skip)) {
 508                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 509                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 510                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 511                                                 get_unqualified_type(arg_to) == exp_to) {
 512                                                 goto next_arg;
 513                                         }
 514                                 }
 515                         } else if (get_unqualified_type(arg_skip) == expected_type_skip) {
 516                                 goto next_arg;
 517                         }
 518                         if (is_type_valid(arg_skip)) {
 519                                 warningf(pos,
 520                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 521                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 522                         }
 523                 }
 524 next_arg:
 525                 arg = arg->next;
 526         }
 527         assert(fmt == '\0');
 528         if (c+1 < string + size) {
 529                 warningf(pos, "format string contains '\\0'");
 530         }
 531         return num_fmt;
 532 }
 533
 534 /**
 535  * Check printf-style format.
 536  */
 537 static void check_printf_format(call_argument_t const *arg,
 538                                 format_spec_t const *const spec)
 539 {
 540         /* find format arg */
 541         size_t idx = 0;
 542         for (; idx < spec->fmt_idx; ++idx) {
 543                 if (arg == NULL)
 544                         return;
 545                 arg = arg->next;
 546         }
 547
 548         expression_t const *const fmt_expr = arg->expression;
 549
 550         /* find the real args */
 551         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 552                 arg = arg->next;
 553
 554         int const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
 555         if (num_fmt < 0)
 556                 return;
 557
 558         size_t num_args = 0;
 559         for (; arg != NULL; arg = arg->next)
 560                 ++num_args;
 561         if (num_args > (size_t)num_fmt) {
 562                 warningf(&fmt_expr->base.source_position,
 563                          "%u argument%s but only %u format specifier%s",
 564                          num_args, num_args != 1 ? "s" : "",
 565                          num_fmt,  num_fmt  != 1 ? "s" : "");
 566         }
 567 }
 568
 569 /**
 570  * Check scanf-style format.
 571  */
 572 static void check_scanf_format(const call_argument_t *arg,
 573                                const format_spec_t *spec)
 574 {
 575         /* find format arg */
 576         unsigned idx = 0;
 577         for (; idx < spec->fmt_idx; ++idx) {
 578                 if (arg == NULL)
 579                         return;
 580                 arg = arg->next;
 581         }
 582
 583         const expression_t *fmt_expr = arg->expression;
 584         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 585                 fmt_expr = fmt_expr->unary.value;
 586         }
 587
 588         if (fmt_expr->kind != EXPR_STRING_LITERAL
 589                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 590                 return;
 591
 592         const char *string = fmt_expr->literal.value.begin;
 593         size_t      size   = fmt_expr->literal.value.size;
 594         const char *c      = string;
 595
 596         /* find the real args */
 597         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 598                 arg = arg->next;
 599
 600         const source_position_t *pos = &fmt_expr->base.source_position;
 601         unsigned num_fmt = 0;
 602         char     fmt;
 603         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 604                 if (fmt != '%')
 605                         continue;
 606                 fmt = *(++c);
 607                 if (fmt == '\0') {
 608                         warningf(pos, "dangling '%%' in format string");
 609                         break;
 610                 }
 611                 if (fmt == '%')
 612                         continue;
 613
 614                 ++num_fmt;
 615
 616                 /* look for length modifiers */
 617                 format_length_modifier_t fmt_mod = FMT_MOD_NONE;
 618                 switch (fmt) {
 619                 case 'h':
 620                         fmt = *(++c);
 621                         if (fmt == 'h') {
 622                                 fmt = *(++c);
 623                                 fmt_mod = FMT_MOD_hh;
 624                         } else {
 625                                 fmt_mod = FMT_MOD_h;
 626                         }
 627                         break;
 628
 629                 case 'l':
 630                         fmt = *(++c);
 631                         if (fmt == 'l') {
 632                                 fmt = *(++c);
 633                                 fmt_mod = FMT_MOD_ll;
 634                         } else {
 635                                 fmt_mod = FMT_MOD_l;
 636                         }
 637                         break;
 638
 639                 case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
 640                 case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
 641                 case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
 642                 case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
 643                 /* microsoft mode */
 644                 case 'w':
 645                         if (c_mode & _MS) {
 646                                 fmt = *(++c);
 647                                 fmt_mod = FMT_MOD_w;
 648                         }
 649                         break;
 650                 case 'I':
 651                         if (c_mode & _MS) {
 652                                 fmt = *(++c);
 653                                 fmt_mod = FMT_MOD_I;
 654                                 if (fmt == '3') {
 655                                         fmt = *(++c);
 656                                         if (fmt == '2') {
 657                                                 fmt = *(++c);
 658                                                 fmt_mod = FMT_MOD_I32;
 659                                         } else {
 660                                                 /* rewind */
 661                                                 fmt = *(--c);
 662                                         }
 663                                 } else if (fmt == '6') {
 664                                         fmt = *(++c);
 665                                         if (fmt == '4') {
 666                                                 fmt = *(++c);
 667                                                 fmt_mod = FMT_MOD_I64;
 668                                         } else {
 669                                                 /* rewind */
 670                                                 fmt = *(--c);
 671                                         }
 672                                 }
 673                         }
 674                         break;
 675                 }
 676
 677                 if (fmt == '\0') {
 678                         warningf(pos, "dangling % with conversion specififer in format string");
 679                         break;
 680                 }
 681
 682                 type_t *expected_type;
 683                 switch (fmt) {
 684                 case 'd':
 685                 case 'i':
 686                         switch (fmt_mod) {
 687                         case FMT_MOD_NONE: expected_type = type_int;         break;
 688                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 689                         case FMT_MOD_h:    expected_type = type_short;       break;
 690                         case FMT_MOD_l:    expected_type = type_long;        break;
 691                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 692                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 693                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 694                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 695                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 696                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 697                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 698
 699                         default:
 700                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 701                                 goto next_arg;
 702                         }
 703                         break;
 704
 705                 case 'o':
 706                 case 'X':
 707                 case 'x':
 708                 case 'u':
 709                         switch (fmt_mod) {
 710                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 711                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 712                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 713                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 714                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 715                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 716                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 717                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 718                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 719                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 720                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 721
 722                         default:
 723                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 724                                 goto next_arg;
 725                         }
 726                         break;
 727
 728                 case 'A':
 729                 case 'a':
 730                 case 'E':
 731                 case 'e':
 732                 case 'F':
 733                 case 'f':
 734                 case 'G':
 735                 case 'g':
 736                         switch (fmt_mod) {
 737                         case FMT_MOD_l:    expected_type = type_double;      break;
 738                         case FMT_MOD_NONE: expected_type = type_float;       break;
 739                         case FMT_MOD_L:    expected_type = type_long_double; break;
 740
 741                         default:
 742                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 743                                 goto next_arg;
 744                         }
 745                         break;
 746
 747                 case 'C':
 748                         if (fmt_mod != FMT_MOD_NONE) {
 749                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 750                                 goto next_arg;
 751                         }
 752                         expected_type = type_wchar_t;
 753                         break;
 754
 755                 case 'c':
 756                         expected_type = type_int;
 757                         switch (fmt_mod) {
 758                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 759                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 760                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 761
 762                         default:
 763                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 764                                 goto next_arg;
 765                         }
 766                         break;
 767
 768                 case 'S':
 769                         if (fmt_mod != FMT_MOD_NONE) {
 770                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 771                                 goto next_arg;
 772                         }
 773                         expected_type = type_wchar_t;
 774                         break;
 775
 776                 case 's':
 777                 case '[':
 778                         switch (fmt_mod) {
 779                                 case FMT_MOD_NONE: expected_type = type_char;    break;
 780                                 case FMT_MOD_l:    expected_type = type_wchar_t; break;
 781                                 case FMT_MOD_w:    expected_type = type_wchar_t; break;
 782
 783                                 default:
 784                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 785                                         goto next_arg;
 786                         }
 787                         break;
 788
 789                 case 'p':
 790                         if (fmt_mod != FMT_MOD_NONE) {
 791                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 792                                 goto next_arg;
 793                         }
 794                         expected_type = type_void_ptr;
 795                         break;
 796
 797                 case 'n':
 798                         switch (fmt_mod) {
 799                         case FMT_MOD_NONE: expected_type = type_int;         break;
 800                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 801                         case FMT_MOD_h:    expected_type = type_short;       break;
 802                         case FMT_MOD_l:    expected_type = type_long;        break;
 803                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 804                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 805                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 806                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 807
 808                         default:
 809                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 810                                 goto next_arg;
 811                         }
 812                         break;
 813
 814                 default:
 815                         warningf(pos, "encountered unknown conversion specifier '%%%c' at format %u",
 816                                  fmt, num_fmt);
 817                         if (arg == NULL) {
 818                                 warningf(pos, "too few arguments for format string");
 819                                 return;
 820                         }
 821                         goto next_arg;
 822                 }
 823
 824                 if (arg == NULL) {
 825                         warningf(pos, "too few arguments for format string");
 826                         return;
 827                 }
 828
 829                 {       /* create a scope here to prevent warning about the jump to next_arg */
 830                         type_t *const arg_type           = arg->expression->base.type;
 831                         type_t *const arg_skip           = skip_typeref(arg_type);
 832                         type_t *const expected_type_skip = skip_typeref(expected_type);
 833
 834                         if (! is_type_pointer(arg_skip))
 835                                 goto error_arg_type;
 836                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 837
 838                         if (fmt == 'p') {
 839                                 /* allow any pointer type for %p, not just void */
 840                                 if (is_type_pointer(ptr_skip))
 841                                         goto next_arg;
 842                         }
 843
 844                         /* do NOT allow const or restrict, all other should be ok */
 845                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 846                                 goto error_arg_type;
 847                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 848                         if (unqual_ptr == expected_type_skip) {
 849                                 goto next_arg;
 850                         } else if (expected_type_skip == type_char) {
 851                                 /* char matches with unsigned char AND signed char */
 852                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 853                                         goto next_arg;
 854                         }
 855 error_arg_type:
 856                         if (is_type_valid(arg_skip)) {
 857                                 warningf(pos,
 858                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 859                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 860                         }
 861                 }
 862 next_arg:
 863                 arg = arg->next;
 864         }
 865         assert(fmt == '\0');
 866         if (c+1 < string + size) {
 867                 warningf(pos, "format string contains '\\0'");
 868         }
 869         if (arg != NULL) {
 870                 unsigned num_args = num_fmt;
 871                 while (arg != NULL) {
 872                         ++num_args;
 873                         arg = arg->next;
 874                 }
 875                 warningf(pos, "%u argument%s but only %u format specifier%s",
 876                          num_args, num_args != 1 ? "s" : "",
 877                          num_fmt, num_fmt != 1 ? "s" : "");
 878         }
 879 }
 880
 881 static const format_spec_t builtin_table[] = {
 882         { "printf",        FORMAT_PRINTF,   0, 1 },
 883         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 884         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 885         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 886         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 887         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 888         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 889         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 890         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 891         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 892
 893         { "scanf",         FORMAT_SCANF,    0, 1 },
 894         { "wscanf",        FORMAT_SCANF,    0, 1 },
 895         { "sscanf",        FORMAT_SCANF,    1, 2 },
 896         { "swscanf",       FORMAT_SCANF,    1, 2 },
 897         { "fscanf",        FORMAT_SCANF,    1, 2 },
 898         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 899
 900         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 901         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 902
 903         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 904
 905         /* MS extensions */
 906         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 907         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 908         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 909         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 910         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 911         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 912         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 913         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 914         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 915         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 916         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 917         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 918         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 919         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 920         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 921         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 922         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 923         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 924         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 925         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 926         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 927         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 928 };
 929
 930 void check_format(const call_expression_t *const call)
 931 {
 932         if (!warning.format)
 933                 return;
 934
 935         const expression_t *const func_expr = call->function;
 936         if (func_expr->kind != EXPR_REFERENCE)
 937                 return;
 938
 939         const entity_t        *const entity = func_expr->reference.entity;
 940         const call_argument_t *      arg    = call->arguments;
 941
 942         if (false) {
 943                 /* the declaration has a GNU format attribute, check it */
 944         } else {
 945                 /*
 946                  * For some functions we always check the format, even if it was not
 947                  * specified. This allows to check format even in MS mode or without
 948                  * header included.
 949                  */
 950                 const char *const name = entity->base.symbol->string;
 951                 for (size_t i = 0; i < lengthof(builtin_table); ++i) {
 952                         if (strcmp(name, builtin_table[i].name) == 0) {
 953                                 switch (builtin_table[i].fmt_kind) {
 954                                 case FORMAT_PRINTF:
 955                                         check_printf_format(arg, &builtin_table[i]);
 956                                         break;
 957                                 case FORMAT_SCANF:
 958                                         check_scanf_format(arg, &builtin_table[i]);
 959                                         break;
 960                                 case FORMAT_STRFTIME:
 961                                 case FORMAT_STRFMON:
 962                                         /* TODO: implement other cases */
 963                                         break;
 964                                 }
 965                                 break;
 966                         }
 967                 }
 968         }
 969 }