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 "parser.h"
  31 #include "types.h"
  32 #include "type_t.h"
  33 #include "warning.h"
  34 #include "lang_features.h"
  35
  36 typedef enum format_flag_t {
  37         FMT_FLAG_NONE  = 0,
  38         FMT_FLAG_HASH  = 1U << 0,
  39         FMT_FLAG_ZERO  = 1U << 1,
  40         FMT_FLAG_MINUS = 1U << 2,
  41         FMT_FLAG_SPACE = 1U << 3,
  42         FMT_FLAG_PLUS  = 1U << 4,
  43         FMT_FLAG_TICK  = 1U << 5
  44 } format_flag_t;
  45
  46 typedef unsigned format_flags_t;
  47
  48 typedef enum format_length_modifier_t {
  49         FMT_MOD_NONE,
  50         FMT_MOD_L,
  51         FMT_MOD_hh,
  52         FMT_MOD_h,
  53         FMT_MOD_l,
  54         FMT_MOD_ll,
  55         FMT_MOD_j,
  56         FMT_MOD_t,
  57         FMT_MOD_z,
  58         FMT_MOD_q,
  59         /* only in microsoft mode */
  60         FMT_MOD_w,
  61         FMT_MOD_I,
  62         FMT_MOD_I32,
  63         FMT_MOD_I64
  64 } format_length_modifier_t;
  65
  66 typedef struct format_spec_t {
  67         const char    *name;     /**< name of the function */
  68         format_kind_t  fmt_kind; /**< kind */
  69         unsigned       fmt_idx;  /**< index of the format string */
  70         unsigned       arg_idx;  /**< index of the first argument */
  71 } format_spec_t;
  72
  73 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  74 {
  75         static const char* const names[] = {
  76                 [FMT_MOD_NONE] = "",
  77                 [FMT_MOD_L]    = "L",
  78                 [FMT_MOD_hh]   = "hh",
  79                 [FMT_MOD_h]    = "h",
  80                 [FMT_MOD_l]    = "l",
  81                 [FMT_MOD_ll]   = "ll",
  82                 [FMT_MOD_j]    = "j",
  83                 [FMT_MOD_t]    = "t",
  84                 [FMT_MOD_z]    = "z",
  85                 [FMT_MOD_q]    = "q",
  86                 /* only in microsoft mode */
  87                 [FMT_MOD_w]    = "w",
  88                 [FMT_MOD_I]    = "I",
  89                 [FMT_MOD_I32]  = "I32",
  90                 [FMT_MOD_I64]  = "I64"
  91         };
  92         assert((size_t)mod < lengthof(names));
  93         return names[mod];
  94 }
  95
  96 static void warn_invalid_length_modifier(const source_position_t *pos,
  97                                          const format_length_modifier_t mod,
  98                                          const utf32 conversion)
  99 {
 100         char const *const lmod = get_length_modifier_name(mod);
 101         warningf(WARN_FORMAT, pos, "invalid length modifier '%s' for conversion specifier '%%%c'", lmod, conversion);
 102 }
 103
 104 /**
 105  * Check printf-style format. Returns number of expected arguments.
 106  */
 107 static int internal_check_printf_format(const expression_t *fmt_expr,
 108                                         const call_argument_t *arg,
 109                                         const format_spec_t *spec)
 110 {
 111         while (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 112                 fmt_expr = fmt_expr->unary.value;
 113         }
 114
 115         /*
 116          * gettext results in expressions like (X ? "format_string" : Y)
 117          * we assume the left part is the format string
 118          */
 119         if (fmt_expr->kind == EXPR_CONDITIONAL) {
 120                 conditional_expression_t const *const c = &fmt_expr->conditional;
 121                 expression_t             const *      t = c->true_expression;
 122                 if (t == NULL)
 123                         t = c->condition;
 124                 int const nt = internal_check_printf_format(t,                   arg, spec);
 125                 int const nf = internal_check_printf_format(c->false_expression, arg, spec);
 126                 return nt > nf ? nt : nf;
 127         }
 128
 129         if (fmt_expr->kind != EXPR_STRING_LITERAL
 130                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 131                 return -1;
 132
 133         const char *string = fmt_expr->literal.value.begin;
 134         size_t      size   = fmt_expr->literal.value.size;
 135         const char *c      = string;
 136
 137         const source_position_t *pos = &fmt_expr->base.source_position;
 138         unsigned num_fmt  = 0;
 139         unsigned num_args = 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(WARN_FORMAT, pos, "dangling %% in format string");
 148                         break;
 149                 }
 150                 if (fmt == '%')
 151                         continue;
 152
 153                 ++num_fmt;
 154                 ++num_args;
 155
 156                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 157                 if (fmt == '0') {
 158                         fmt = *(++c);
 159                         fmt_flags |= FMT_FLAG_ZERO;
 160                 }
 161
 162                 /* argument selector or minimum field width */
 163                 if (isdigit(fmt)) {
 164                         do {
 165                                 fmt = *(++c);
 166                         } while (isdigit(fmt));
 167
 168                         /* digit string was ... */
 169                         if (fmt == '$') {
 170                                 /* ... argument selector */
 171                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 172                                 /* TODO implement */
 173                                 return -1;
 174                         }
 175                         /* ... minimum field width */
 176                 } else {
 177                         /* flags */
 178                         for (;;) {
 179                                 format_flags_t flag;
 180                                 switch (fmt) {
 181                                         case '#':  flag = FMT_FLAG_HASH;  break;
 182                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 183                                         case '-':  flag = FMT_FLAG_MINUS; break;
 184                                         case '\'': flag = FMT_FLAG_TICK;  break;
 185
 186                                         case ' ':
 187                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 188                                                         warningf(WARN_FORMAT, pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 189                                                 }
 190                                                 flag = FMT_FLAG_SPACE;
 191                                                 break;
 192
 193                                         case '+':
 194                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 195                                                         warningf(WARN_FORMAT, pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 196                                                 }
 197                                                 flag = FMT_FLAG_PLUS;
 198                                                 break;
 199
 200                                         default: goto break_fmt_flags;
 201                                 }
 202                                 if (fmt_flags & flag) {
 203                                         warningf(WARN_FORMAT, pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 204                                 }
 205                                 fmt_flags |= flag;
 206                                 fmt = *(++c);
 207                         }
 208 break_fmt_flags:
 209
 210                         /* minimum field width */
 211                         if (fmt == '*') {
 212                                 ++num_args;
 213                                 fmt = *(++c);
 214                                 if (arg == NULL) {
 215                                         warningf(WARN_FORMAT, pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 216                                         return -1;
 217                                 }
 218                                 const type_t *const arg_type = arg->expression->base.type;
 219                                 if (arg_type != type_int) {
 220                                         warningf(WARN_FORMAT, pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 221                                 }
 222                                 arg = arg->next;
 223                         } else {
 224                                 while (isdigit(fmt)) {
 225                                         fmt = *(++c);
 226                                 }
 227                         }
 228                 }
 229
 230                 /* precision */
 231                 if (fmt == '.') {
 232                         if (fmt_flags & FMT_FLAG_ZERO) {
 233                                 warningf(WARN_FORMAT, pos, "'0' flag ignored with precision in conversion specification %u", num_fmt);
 234                         }
 235
 236                         ++num_args;
 237                         fmt = *(++c);
 238                         if (fmt == '*') {
 239                                 fmt = *(++c);
 240                                 if (arg == NULL) {
 241                                         warningf(WARN_FORMAT, pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 242                                         return -1;
 243                                 }
 244                                 const type_t *const arg_type = arg->expression->base.type;
 245                                 if (arg_type != type_int) {
 246                                         warningf(WARN_FORMAT, pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 247                                 }
 248                                 arg = arg->next;
 249                         } else {
 250                                 /* digit string may be omitted */
 251                                 while (isdigit(fmt)) {
 252                                         fmt = *(++c);
 253                                 }
 254                         }
 255                 }
 256
 257                 /* length modifier */
 258                 format_length_modifier_t fmt_mod;
 259                 switch (fmt) {
 260                         case 'h':
 261                                 fmt = *(++c);
 262                                 if (fmt == 'h') {
 263                                         fmt = *(++c);
 264                                         fmt_mod = FMT_MOD_hh;
 265                                 } else {
 266                                         fmt_mod = FMT_MOD_h;
 267                                 }
 268                                 break;
 269
 270                         case 'l':
 271                                 fmt = *(++c);
 272                                 if (fmt == 'l') {
 273                                         fmt = *(++c);
 274                                         fmt_mod = FMT_MOD_ll;
 275                                 } else {
 276                                         fmt_mod = FMT_MOD_l;
 277                                 }
 278                                 break;
 279
 280                         case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
 281                         case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
 282                         case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
 283                         case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
 284                         case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 285                         /* microsoft mode */
 286                         case 'w':
 287                                 if (c_mode & _MS) {
 288                                         fmt = *(++c); fmt_mod = FMT_MOD_w;
 289                                 } else {
 290                                         fmt_mod = FMT_MOD_NONE;
 291                                 }
 292                                 break;
 293                         case 'I':
 294                                 if (c_mode & _MS) {
 295                                         fmt = *(++c); fmt_mod = FMT_MOD_I;
 296                                         if (fmt == '3') {
 297                                                 fmt = *(++c);
 298                                                 if (fmt == '2') {
 299                                                         fmt = *(++c);
 300                                                         fmt_mod = FMT_MOD_I32;
 301                                                 } else {
 302                                                         /* rewind */
 303                                                         fmt = *(--c);
 304                                                 }
 305                                         } else if (fmt == '6') {
 306                                                 fmt = *(++c);
 307                                                 if (fmt == '4') {
 308                                                         fmt = *(++c);
 309                                                         fmt_mod = FMT_MOD_I64;
 310                                                 } else {
 311                                                         /* rewind */
 312                                                         fmt = *(--c);
 313                                                 }
 314                                         }
 315                                 } else {
 316                                         fmt_mod = FMT_MOD_NONE;
 317                                 }
 318                                 break;
 319                         default:
 320                                 fmt_mod = FMT_MOD_NONE;
 321                                 break;
 322                 }
 323
 324
 325                 type_t            *expected_type;
 326                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 327                 format_flags_t     allowed_flags;
 328                 switch (fmt) {
 329                         case 'd':
 330                         case 'i':
 331                                 switch (fmt_mod) {
 332                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 333                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 334                                         case FMT_MOD_h:    expected_type = type_short;       break;
 335                                         case FMT_MOD_l:    expected_type = type_long;        break;
 336                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 337                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 338                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 339                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 340                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 341                                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 342                                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 343
 344                                         default:
 345                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 346                                                 goto next_arg;
 347                                 }
 348                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 349                                 break;
 350
 351                         case 'o':
 352                         case 'X':
 353                         case 'x':
 354                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 355                                 goto eval_fmt_mod_unsigned;
 356
 357                         case 'u':
 358                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 359 eval_fmt_mod_unsigned:
 360                                 switch (fmt_mod) {
 361                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 362                                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 363                                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 364                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 365                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 366                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 367                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 368                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 369                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 370                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 371                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 372
 373                                         default:
 374                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 375                                                 goto next_arg;
 376                                 }
 377                                 break;
 378
 379                         case 'A':
 380                         case 'a':
 381                         case 'E':
 382                         case 'e':
 383                         case 'F':
 384                         case 'f':
 385                         case 'G':
 386                         case 'g':
 387                                 switch (fmt_mod) {
 388                                         case FMT_MOD_l:    /* l modifier is ignored */
 389                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 390                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 391
 392                                         default:
 393                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 394                                                 goto next_arg;
 395                                 }
 396                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 397                                 break;
 398
 399                         case 'C':
 400                                 if (fmt_mod != FMT_MOD_NONE) {
 401                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 402                                         goto next_arg;
 403                                 }
 404                                 expected_type = type_wchar_t;
 405                                 allowed_flags = FMT_FLAG_NONE;
 406                                 break;
 407
 408                         case 'c':
 409                                 expected_type = type_int;
 410                                 switch (fmt_mod) {
 411                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 412                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 413                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 414
 415                                         default:
 416                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 417                                                 goto next_arg;
 418                                 }
 419                                 allowed_flags = FMT_FLAG_NONE;
 420                                 break;
 421
 422                         case 'S':
 423                                 if (fmt_mod != FMT_MOD_NONE) {
 424                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 425                                         goto next_arg;
 426                                 }
 427                                 expected_type = type_wchar_t_ptr;
 428                                 expected_qual = TYPE_QUALIFIER_CONST;
 429                                 allowed_flags = FMT_FLAG_MINUS;
 430                                 break;
 431
 432                         case 's':
 433                                 switch (fmt_mod) {
 434                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 435                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 436                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 437
 438                                         default:
 439                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 440                                                 goto next_arg;
 441                                 }
 442                                 expected_qual = TYPE_QUALIFIER_CONST;
 443                                 allowed_flags = FMT_FLAG_MINUS;
 444                                 break;
 445
 446                         case 'p':
 447                                 if (fmt_mod != FMT_MOD_NONE) {
 448                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 449                                         goto next_arg;
 450                                 }
 451                                 expected_type = type_void_ptr;
 452                                 allowed_flags = FMT_FLAG_NONE;
 453                                 break;
 454
 455                         case 'n':
 456                                 switch (fmt_mod) {
 457                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 458                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 459                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 460                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 461                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 462                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 463                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 464                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 465
 466                                         default:
 467                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 468                                                 goto next_arg;
 469                                 }
 470                                 allowed_flags = FMT_FLAG_NONE;
 471                                 break;
 472
 473                         default:
 474                                 warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at position %u", fmt, num_fmt);
 475                                 if (arg == NULL) {
 476                                         goto too_few_args;
 477                                 }
 478                                 goto next_arg;
 479                 }
 480
 481                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 482                 if (wrong_flags != 0) {
 483                         char  wrong[8];
 484                         char *p = wrong;
 485                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 486                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 487                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 488                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 489                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 490                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 491                         *p = '\0';
 492
 493                         warningf(WARN_FORMAT, pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 494                 }
 495
 496                 if (arg == NULL) {
 497 too_few_args:
 498                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 499                         return -1;
 500                 }
 501
 502                 { /* create a scope here to prevent warning about the jump to next_arg */
 503                         type_t *const arg_type           = arg->expression->base.type;
 504                         type_t *const arg_skip           = skip_typeref(arg_type);
 505                         type_t *const expected_type_skip = skip_typeref(expected_type);
 506
 507                         if (fmt == 'p') {
 508                                 /* allow any pointer type for %p, not just void */
 509                                 if (is_type_pointer(arg_skip))
 510                                         goto next_arg;
 511                         }
 512
 513                         if (is_type_pointer(expected_type_skip)) {
 514                                 if (is_type_pointer(arg_skip)) {
 515                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 516                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 517                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 518                                                 get_unqualified_type(arg_to) == exp_to) {
 519                                                 goto next_arg;
 520                                         }
 521                                 }
 522                         } else if (get_unqualified_type(arg_skip) == expected_type_skip) {
 523                                 goto next_arg;
 524                         } else if (arg->expression->kind == EXPR_UNARY_CAST_IMPLICIT) {
 525                                 expression_t const *const expr        = arg->expression->unary.value;
 526                                 type_t             *const unprom_type = skip_typeref(expr->base.type);
 527                                 if (get_unqualified_type(unprom_type) == expected_type_skip) {
 528                                         goto next_arg;
 529                                 }
 530                                 if (expected_type_skip == type_unsigned_int && !is_type_signed(unprom_type)) {
 531                                         goto next_arg;
 532                                 }
 533                         }
 534                         if (is_type_valid(arg_skip)) {
 535                                 source_position_t const *const apos = &arg->expression->base.source_position;
 536                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 537                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 538                         }
 539                 }
 540 next_arg:
 541                 arg = arg->next;
 542         }
 543         assert(fmt == '\0');
 544         if (c+1 < string + size) {
 545                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 546         }
 547         return num_args;
 548 }
 549
 550 /**
 551  * Check printf-style format.
 552  */
 553 static void check_printf_format(call_argument_t const *arg,
 554                                 format_spec_t const *const spec)
 555 {
 556         /* find format arg */
 557         size_t idx = 0;
 558         for (; idx < spec->fmt_idx; ++idx) {
 559                 if (arg == NULL)
 560                         return;
 561                 arg = arg->next;
 562         }
 563
 564         expression_t const *const fmt_expr = arg->expression;
 565
 566         /* find the real args */
 567         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 568                 arg = arg->next;
 569
 570         int const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
 571         if (num_fmt < 0)
 572                 return;
 573
 574         size_t num_args = 0;
 575         for (; arg != NULL; arg = arg->next)
 576                 ++num_args;
 577         if (num_args > (size_t)num_fmt) {
 578                 source_position_t const *const pos = &fmt_expr->base.source_position;
 579                 warningf(WARN_FORMAT, pos, "%u argument%s but only %u format specifier%s", num_args, num_args != 1 ? "s" : "", num_fmt,  num_fmt  != 1 ? "s" : "");
 580         }
 581 }
 582
 583 /**
 584  * Check scanf-style format.
 585  */
 586 static void check_scanf_format(const call_argument_t *arg,
 587                                const format_spec_t *spec)
 588 {
 589         /* find format arg */
 590         unsigned idx = 0;
 591         for (; idx < spec->fmt_idx; ++idx) {
 592                 if (arg == NULL)
 593                         return;
 594                 arg = arg->next;
 595         }
 596
 597         const expression_t *fmt_expr = arg->expression;
 598         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 599                 fmt_expr = fmt_expr->unary.value;
 600         }
 601
 602         if (fmt_expr->kind != EXPR_STRING_LITERAL
 603                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 604                 return;
 605
 606         const char *string = fmt_expr->literal.value.begin;
 607         size_t      size   = fmt_expr->literal.value.size;
 608         const char *c      = string;
 609
 610         /* find the real args */
 611         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 612                 arg = arg->next;
 613
 614         const source_position_t *pos = &fmt_expr->base.source_position;
 615         unsigned num_fmt = 0;
 616         char     fmt;
 617         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 618                 if (fmt != '%')
 619                         continue;
 620                 fmt = *(++c);
 621                 if (fmt == '%')
 622                         continue;
 623
 624                 ++num_fmt;
 625
 626                 bool suppress_assignment = false;
 627                 if (fmt == '*') {
 628                         fmt = *++c;
 629                         suppress_assignment = true;
 630                 }
 631
 632                 size_t width = 0;
 633                 if ('0' <= fmt && fmt <= '9') {
 634                         do {
 635                                 width = width * 10 + (fmt - '0');
 636                                 fmt   = *++c;
 637                         } while ('0' <= fmt && fmt <= '9');
 638                         if (width == 0) {
 639                                 warningf(WARN_FORMAT, pos, "field width is zero at format %u", num_fmt);
 640                         }
 641                 }
 642
 643                 /* look for length modifiers */
 644                 format_length_modifier_t fmt_mod = FMT_MOD_NONE;
 645                 switch (fmt) {
 646                 case 'h':
 647                         fmt = *(++c);
 648                         if (fmt == 'h') {
 649                                 fmt = *(++c);
 650                                 fmt_mod = FMT_MOD_hh;
 651                         } else {
 652                                 fmt_mod = FMT_MOD_h;
 653                         }
 654                         break;
 655
 656                 case 'l':
 657                         fmt = *(++c);
 658                         if (fmt == 'l') {
 659                                 fmt = *(++c);
 660                                 fmt_mod = FMT_MOD_ll;
 661                         } else {
 662                                 fmt_mod = FMT_MOD_l;
 663                         }
 664                         break;
 665
 666                 case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
 667                 case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
 668                 case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
 669                 case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
 670                 /* microsoft mode */
 671                 case 'w':
 672                         if (c_mode & _MS) {
 673                                 fmt = *(++c);
 674                                 fmt_mod = FMT_MOD_w;
 675                         }
 676                         break;
 677                 case 'I':
 678                         if (c_mode & _MS) {
 679                                 fmt = *(++c);
 680                                 fmt_mod = FMT_MOD_I;
 681                                 if (fmt == '3') {
 682                                         fmt = *(++c);
 683                                         if (fmt == '2') {
 684                                                 fmt = *(++c);
 685                                                 fmt_mod = FMT_MOD_I32;
 686                                         } else {
 687                                                 /* rewind */
 688                                                 fmt = *(--c);
 689                                         }
 690                                 } else if (fmt == '6') {
 691                                         fmt = *(++c);
 692                                         if (fmt == '4') {
 693                                                 fmt = *(++c);
 694                                                 fmt_mod = FMT_MOD_I64;
 695                                         } else {
 696                                                 /* rewind */
 697                                                 fmt = *(--c);
 698                                         }
 699                                 }
 700                         }
 701                         break;
 702                 }
 703
 704                 if (fmt == '\0') {
 705                         warningf(WARN_FORMAT, pos, "dangling %% with conversion specififer 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                 case 'u':
 736                         switch (fmt_mod) {
 737                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 738                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 739                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 740                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 741                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 742                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 743                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 744                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 745                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 746                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 747                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 748
 749                         default:
 750                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 751                                 goto next_arg;
 752                         }
 753                         break;
 754
 755                 case 'A':
 756                 case 'a':
 757                 case 'E':
 758                 case 'e':
 759                 case 'F':
 760                 case 'f':
 761                 case 'G':
 762                 case 'g':
 763                         switch (fmt_mod) {
 764                         case FMT_MOD_l:    expected_type = type_double;      break;
 765                         case FMT_MOD_NONE: expected_type = type_float;       break;
 766                         case FMT_MOD_L:    expected_type = type_long_double; break;
 767
 768                         default:
 769                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 770                                 goto next_arg;
 771                         }
 772                         break;
 773
 774                 case 'C':
 775                         if (fmt_mod != FMT_MOD_NONE) {
 776                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 777                                 goto next_arg;
 778                         }
 779                         expected_type = type_wchar_t;
 780                         goto check_c_width;
 781
 782                 case 'c': {
 783                         switch (fmt_mod) {
 784                         case FMT_MOD_NONE: expected_type = type_char;    break;
 785                         case FMT_MOD_l:    expected_type = type_wchar_t; break;
 786                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 787
 788                         default:
 789                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 790                                 goto next_arg;
 791                         }
 792
 793 check_c_width:
 794                         if (width == 0)
 795                                 width = 1;
 796                         if (!suppress_assignment && arg != NULL) {
 797                                 type_t *const type = skip_typeref(revert_automatic_type_conversion(arg->expression));
 798                                 if (is_type_array(type)       &&
 799                                     type->array.size_constant &&
 800                                     width > type->array.size) {
 801                                         warningf(WARN_FORMAT, pos, "target buffer '%T' is too small for %u characters at format %u", type, width, num_fmt);
 802                                 }
 803                         }
 804                         break;
 805                 }
 806
 807                 case 'S':
 808                         if (fmt_mod != FMT_MOD_NONE) {
 809                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 810                                 goto next_arg;
 811                         }
 812                         expected_type = type_wchar_t;
 813                         break;
 814
 815                 case 's':
 816                 case '[': {
 817                         switch (fmt_mod) {
 818                                 case FMT_MOD_NONE: expected_type = type_char;    break;
 819                                 case FMT_MOD_l:    expected_type = type_wchar_t; break;
 820                                 case FMT_MOD_w:    expected_type = type_wchar_t; break;
 821
 822                                 default:
 823                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 824                                         goto next_arg;
 825                         }
 826
 827                         if (!suppress_assignment &&
 828                             width != 0           &&
 829                             arg   != NULL) {
 830                                 type_t *const type = skip_typeref(revert_automatic_type_conversion(arg->expression));
 831                                 if (is_type_array(type)       &&
 832                                     type->array.size_constant &&
 833                                     width >= type->array.size) {
 834                                         warningf(WARN_FORMAT, pos, "target buffer '%T' is too small for %u characters and \\0 at format %u", type, width, num_fmt);
 835                                 }
 836                         }
 837                         break;
 838                 }
 839
 840                 case 'p':
 841                         if (fmt_mod != FMT_MOD_NONE) {
 842                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 843                                 goto next_arg;
 844                         }
 845                         expected_type = type_void_ptr;
 846                         break;
 847
 848                 case 'n': {
 849                         if (suppress_assignment) {
 850                                 warningf(WARN_FORMAT, pos, "conversion '%n' cannot be suppressed with '*' at format %u", num_fmt);
 851                         }
 852
 853                         switch (fmt_mod) {
 854                         case FMT_MOD_NONE: expected_type = type_int;         break;
 855                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 856                         case FMT_MOD_h:    expected_type = type_short;       break;
 857                         case FMT_MOD_l:    expected_type = type_long;        break;
 858                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 859                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 860                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 861                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 862
 863                         default:
 864                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 865                                 goto next_arg;
 866                         }
 867                         break;
 868                 }
 869
 870                 default:
 871                         warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at format %u", fmt, num_fmt);
 872                         if (suppress_assignment)
 873                                 continue;
 874                         if (arg == NULL)
 875                                 goto too_few_args;
 876                         goto next_arg;
 877                 }
 878
 879                 if (suppress_assignment)
 880                         continue;
 881
 882                 if (arg == NULL) {
 883 too_few_args:
 884                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 885                         return;
 886                 }
 887
 888                 { /* create a scope here to prevent warning about the jump to next_arg */
 889                         type_t *const arg_type           = arg->expression->base.type;
 890                         type_t *const arg_skip           = skip_typeref(arg_type);
 891                         type_t *const expected_type_skip = skip_typeref(expected_type);
 892
 893                         if (! is_type_pointer(arg_skip))
 894                                 goto error_arg_type;
 895                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 896
 897                         if (fmt == 'p') {
 898                                 /* allow any pointer type for %p, not just void */
 899                                 if (is_type_pointer(ptr_skip))
 900                                         goto next_arg;
 901                         }
 902
 903                         /* do NOT allow const or restrict, all other should be ok */
 904                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 905                                 goto error_arg_type;
 906                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 907                         if (unqual_ptr == expected_type_skip) {
 908                                 goto next_arg;
 909                         } else if (expected_type_skip == type_char) {
 910                                 /* char matches with unsigned char AND signed char */
 911                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 912                                         goto next_arg;
 913                         }
 914 error_arg_type:
 915                         if (is_type_valid(arg_skip)) {
 916                                 source_position_t const *const apos = &arg->expression->base.source_position;
 917                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 918                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 919                         }
 920                 }
 921 next_arg:
 922                 arg = arg->next;
 923         }
 924         assert(fmt == '\0');
 925         if (c+1 < string + size) {
 926                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 927         }
 928         if (arg != NULL) {
 929                 unsigned num_args = num_fmt;
 930                 while (arg != NULL) {
 931                         ++num_args;
 932                         arg = arg->next;
 933                 }
 934                 warningf(WARN_FORMAT, pos, "%u argument%s but only %u format specifier%s", num_args, num_args != 1 ? "s" : "", num_fmt, num_fmt != 1 ? "s" : "");
 935         }
 936 }
 937
 938 static const format_spec_t builtin_table[] = {
 939         { "printf",        FORMAT_PRINTF,   0, 1 },
 940         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 941         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 942         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 943         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 944         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 945         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 946         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 947         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 948         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 949
 950         { "scanf",         FORMAT_SCANF,    0, 1 },
 951         { "wscanf",        FORMAT_SCANF,    0, 1 },
 952         { "sscanf",        FORMAT_SCANF,    1, 2 },
 953         { "swscanf",       FORMAT_SCANF,    1, 2 },
 954         { "fscanf",        FORMAT_SCANF,    1, 2 },
 955         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 956
 957         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 958         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 959
 960         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 961
 962         /* MS extensions */
 963         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 964         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 965         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 966         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 967         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 968         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 969         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 970         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 971         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 972         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 973         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 974         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 975         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 976         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 977         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 978         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 979         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 980         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 981         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 982         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 983         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 984         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 985 };
 986
 987 void check_format(const call_expression_t *const call)
 988 {
 989         if (!is_warn_on(WARN_FORMAT))
 990                 return;
 991
 992         const expression_t *const func_expr = call->function;
 993         if (func_expr->kind != EXPR_REFERENCE)
 994                 return;
 995
 996         const entity_t        *const entity = func_expr->reference.entity;
 997         const call_argument_t *      arg    = call->arguments;
 998
 999         /*
1000          * For some functions we always check the format, even if it was not
1001          * specified. This allows to check format even in MS mode or without
1002          * header included.
1003          */
1004         const char *const name = entity->base.symbol->string;
1005         for (size_t i = 0; i < lengthof(builtin_table); ++i) {
1006                 if (strcmp(name, builtin_table[i].name) == 0) {
1007                         switch (builtin_table[i].fmt_kind) {
1008                         case FORMAT_PRINTF:
1009                                 check_printf_format(arg, &builtin_table[i]);
1010                                 break;
1011                         case FORMAT_SCANF:
1012                                 check_scanf_format(arg, &builtin_table[i]);
1013                                 break;
1014                         case FORMAT_STRFTIME:
1015                         case FORMAT_STRFMON:
1016                                 /* TODO: implement other cases */
1017                                 break;
1018                         }
1019                         break;
1020                 }
1021         }
1022 }