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/strutil.h"
  25 #include "adt/util.h"
  26 #include "format_check.h"
  27 #include "symbol_t.h"
  28 #include "ast_t.h"
  29 #include "entity_t.h"
  30 #include "diagnostic.h"
  31 #include "parser.h"
  32 #include "types.h"
  33 #include "type_t.h"
  34 #include "warning.h"
  35 #include "lang_features.h"
  36
  37 typedef enum format_flag_t {
  38         FMT_FLAG_NONE  = 0,
  39         FMT_FLAG_HASH  = 1U << 0,
  40         FMT_FLAG_ZERO  = 1U << 1,
  41         FMT_FLAG_MINUS = 1U << 2,
  42         FMT_FLAG_SPACE = 1U << 3,
  43         FMT_FLAG_PLUS  = 1U << 4,
  44         FMT_FLAG_TICK  = 1U << 5
  45 } format_flag_t;
  46
  47 typedef unsigned format_flags_t;
  48
  49 typedef enum format_length_modifier_t {
  50         FMT_MOD_NONE,
  51         FMT_MOD_L,
  52         FMT_MOD_hh,
  53         FMT_MOD_h,
  54         FMT_MOD_l,
  55         FMT_MOD_ll,
  56         FMT_MOD_j,
  57         FMT_MOD_t,
  58         FMT_MOD_z,
  59         FMT_MOD_q,
  60         /* only in microsoft mode */
  61         FMT_MOD_w,
  62         FMT_MOD_I,
  63         FMT_MOD_I32,
  64         FMT_MOD_I64
  65 } format_length_modifier_t;
  66
  67 typedef struct format_spec_t {
  68         const char    *name;     /**< name of the function */
  69         format_kind_t  fmt_kind; /**< kind */
  70         unsigned       fmt_idx;  /**< index of the format string */
  71         unsigned       arg_idx;  /**< index of the first argument */
  72 } format_spec_t;
  73
  74 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  75 {
  76         static const char* const names[] = {
  77                 [FMT_MOD_NONE] = "",
  78                 [FMT_MOD_L]    = "L",
  79                 [FMT_MOD_hh]   = "hh",
  80                 [FMT_MOD_h]    = "h",
  81                 [FMT_MOD_l]    = "l",
  82                 [FMT_MOD_ll]   = "ll",
  83                 [FMT_MOD_j]    = "j",
  84                 [FMT_MOD_t]    = "t",
  85                 [FMT_MOD_z]    = "z",
  86                 [FMT_MOD_q]    = "q",
  87                 /* only in microsoft mode */
  88                 [FMT_MOD_w]    = "w",
  89                 [FMT_MOD_I]    = "I",
  90                 [FMT_MOD_I32]  = "I32",
  91                 [FMT_MOD_I64]  = "I64"
  92         };
  93         assert((size_t)mod < lengthof(names));
  94         return names[mod];
  95 }
  96
  97 static void warn_invalid_length_modifier(const source_position_t *pos,
  98                                          const format_length_modifier_t mod,
  99                                          const utf32 conversion)
 100 {
 101         char const *const lmod = get_length_modifier_name(mod);
 102         warningf(WARN_FORMAT, pos, "invalid length modifier '%s' for conversion specifier '%%%c'", lmod, conversion);
 103 }
 104
 105 /**
 106  * Check printf-style format. Returns number of expected arguments.
 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) {
 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                 return -1;
 132
 133         const char *string = fmt_expr->string_literal.value.begin;
 134         size_t      size   = fmt_expr->string_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 == '%')
 147                         continue;
 148
 149                 ++num_fmt;
 150                 ++num_args;
 151
 152                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 153                 if (fmt == '0') {
 154                         fmt = *(++c);
 155                         fmt_flags |= FMT_FLAG_ZERO;
 156                 }
 157
 158                 /* argument selector or minimum field width */
 159                 if (isdigit(fmt)) {
 160                         do {
 161                                 fmt = *(++c);
 162                         } while (isdigit(fmt));
 163
 164                         /* digit string was ... */
 165                         if (fmt == '$') {
 166                                 /* ... argument selector */
 167                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 168                                 /* TODO implement */
 169                                 return -1;
 170                         }
 171                         /* ... minimum field width */
 172                 } else {
 173                         /* flags */
 174                         for (;;) {
 175                                 format_flags_t flag;
 176                                 switch (fmt) {
 177                                         case '#':  flag = FMT_FLAG_HASH;  break;
 178                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 179                                         case '-':  flag = FMT_FLAG_MINUS; break;
 180                                         case '\'': flag = FMT_FLAG_TICK;  break;
 181
 182                                         case ' ':
 183                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 184                                                         warningf(WARN_FORMAT, pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 185                                                 }
 186                                                 flag = FMT_FLAG_SPACE;
 187                                                 break;
 188
 189                                         case '+':
 190                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 191                                                         warningf(WARN_FORMAT, pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 192                                                 }
 193                                                 flag = FMT_FLAG_PLUS;
 194                                                 break;
 195
 196                                         default: goto break_fmt_flags;
 197                                 }
 198                                 if (fmt_flags & flag) {
 199                                         warningf(WARN_FORMAT, pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 200                                 }
 201                                 fmt_flags |= flag;
 202                                 fmt = *(++c);
 203                         }
 204 break_fmt_flags:
 205
 206                         /* minimum field width */
 207                         if (fmt == '*') {
 208                                 ++num_args;
 209                                 fmt = *(++c);
 210                                 if (arg == NULL) {
 211                                         warningf(WARN_FORMAT, pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 212                                         return -1;
 213                                 }
 214                                 const type_t *const arg_type = arg->expression->base.type;
 215                                 if (arg_type != type_int) {
 216                                         warningf(WARN_FORMAT, pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 217                                 }
 218                                 arg = arg->next;
 219                         } else {
 220                                 while (isdigit(fmt)) {
 221                                         fmt = *(++c);
 222                                 }
 223                         }
 224                 }
 225
 226                 /* precision */
 227                 if (fmt == '.') {
 228                         if (fmt_flags & FMT_FLAG_ZERO) {
 229                                 warningf(WARN_FORMAT, pos, "'0' flag ignored with precision in conversion specification %u", num_fmt);
 230                         }
 231
 232                         ++num_args;
 233                         fmt = *(++c);
 234                         if (fmt == '*') {
 235                                 fmt = *(++c);
 236                                 if (arg == NULL) {
 237                                         warningf(WARN_FORMAT, pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 238                                         return -1;
 239                                 }
 240                                 const type_t *const arg_type = arg->expression->base.type;
 241                                 if (arg_type != type_int) {
 242                                         warningf(WARN_FORMAT, pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 243                                 }
 244                                 arg = arg->next;
 245                         } else {
 246                                 /* digit string may be omitted */
 247                                 while (isdigit(fmt)) {
 248                                         fmt = *(++c);
 249                                 }
 250                         }
 251                 }
 252
 253                 /* length modifier */
 254                 format_length_modifier_t fmt_mod;
 255                 switch (fmt) {
 256                         case 'h':
 257                                 fmt = *(++c);
 258                                 if (fmt == 'h') {
 259                                         fmt = *(++c);
 260                                         fmt_mod = FMT_MOD_hh;
 261                                 } else {
 262                                         fmt_mod = FMT_MOD_h;
 263                                 }
 264                                 break;
 265
 266                         case 'l':
 267                                 fmt = *(++c);
 268                                 if (fmt == 'l') {
 269                                         fmt = *(++c);
 270                                         fmt_mod = FMT_MOD_ll;
 271                                 } else {
 272                                         fmt_mod = FMT_MOD_l;
 273                                 }
 274                                 break;
 275
 276                         case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
 277                         case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
 278                         case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
 279                         case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
 280                         case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 281                         /* microsoft mode */
 282                         case 'w':
 283                                 if (c_mode & _MS) {
 284                                         fmt = *(++c); fmt_mod = FMT_MOD_w;
 285                                 } else {
 286                                         fmt_mod = FMT_MOD_NONE;
 287                                 }
 288                                 break;
 289                         case 'I':
 290                                 if (c_mode & _MS) {
 291                                         fmt = *(++c); fmt_mod = FMT_MOD_I;
 292                                         if (fmt == '3') {
 293                                                 fmt = *(++c);
 294                                                 if (fmt == '2') {
 295                                                         fmt = *(++c);
 296                                                         fmt_mod = FMT_MOD_I32;
 297                                                 } else {
 298                                                         /* rewind */
 299                                                         fmt = *(--c);
 300                                                 }
 301                                         } else if (fmt == '6') {
 302                                                 fmt = *(++c);
 303                                                 if (fmt == '4') {
 304                                                         fmt = *(++c);
 305                                                         fmt_mod = FMT_MOD_I64;
 306                                                 } else {
 307                                                         /* rewind */
 308                                                         fmt = *(--c);
 309                                                 }
 310                                         }
 311                                 } else {
 312                                         fmt_mod = FMT_MOD_NONE;
 313                                 }
 314                                 break;
 315                         default:
 316                                 fmt_mod = FMT_MOD_NONE;
 317                                 break;
 318                 }
 319
 320                 if (fmt == '\0') {
 321                         warningf(WARN_FORMAT, pos, "dangling %% in format string");
 322                         break;
 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) {
 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, "conversion '%%%s%c' at position %u specifies type '%T' but the argument has type '%T'", mod, (char)fmt, num_fmt, expected_type, arg_type);
 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) {
 599                 fmt_expr = fmt_expr->unary.value;
 600         }
 601
 602         if (fmt_expr->kind != EXPR_STRING_LITERAL)
 603                 return;
 604
 605         const char *string = fmt_expr->string_literal.value.begin;
 606         size_t      size   = fmt_expr->string_literal.value.size;
 607         const char *c      = string;
 608
 609         /* find the real args */
 610         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 611                 arg = arg->next;
 612
 613         const source_position_t *pos = &fmt_expr->base.source_position;
 614         unsigned num_fmt = 0;
 615         char     fmt;
 616         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 617                 if (fmt != '%')
 618                         continue;
 619                 fmt = *(++c);
 620                 if (fmt == '%')
 621                         continue;
 622
 623                 ++num_fmt;
 624
 625                 bool suppress_assignment = false;
 626                 if (fmt == '*') {
 627                         fmt = *++c;
 628                         suppress_assignment = true;
 629                 }
 630
 631                 size_t width = 0;
 632                 if ('0' <= fmt && fmt <= '9') {
 633                         do {
 634                                 width = width * 10 + (fmt - '0');
 635                                 fmt   = *++c;
 636                         } while ('0' <= fmt && fmt <= '9');
 637                         if (width == 0) {
 638                                 warningf(WARN_FORMAT, pos, "field width is zero at format %u", num_fmt);
 639                         }
 640                 }
 641
 642                 /* look for length modifiers */
 643                 format_length_modifier_t fmt_mod = FMT_MOD_NONE;
 644                 switch (fmt) {
 645                 case 'h':
 646                         fmt = *(++c);
 647                         if (fmt == 'h') {
 648                                 fmt = *(++c);
 649                                 fmt_mod = FMT_MOD_hh;
 650                         } else {
 651                                 fmt_mod = FMT_MOD_h;
 652                         }
 653                         break;
 654
 655                 case 'l':
 656                         fmt = *(++c);
 657                         if (fmt == 'l') {
 658                                 fmt = *(++c);
 659                                 fmt_mod = FMT_MOD_ll;
 660                         } else {
 661                                 fmt_mod = FMT_MOD_l;
 662                         }
 663                         break;
 664
 665                 case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
 666                 case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
 667                 case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
 668                 case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
 669                 /* microsoft mode */
 670                 case 'w':
 671                         if (c_mode & _MS) {
 672                                 fmt = *(++c);
 673                                 fmt_mod = FMT_MOD_w;
 674                         }
 675                         break;
 676                 case 'I':
 677                         if (c_mode & _MS) {
 678                                 fmt = *(++c);
 679                                 fmt_mod = FMT_MOD_I;
 680                                 if (fmt == '3') {
 681                                         fmt = *(++c);
 682                                         if (fmt == '2') {
 683                                                 fmt = *(++c);
 684                                                 fmt_mod = FMT_MOD_I32;
 685                                         } else {
 686                                                 /* rewind */
 687                                                 fmt = *(--c);
 688                                         }
 689                                 } else if (fmt == '6') {
 690                                         fmt = *(++c);
 691                                         if (fmt == '4') {
 692                                                 fmt = *(++c);
 693                                                 fmt_mod = FMT_MOD_I64;
 694                                         } else {
 695                                                 /* rewind */
 696                                                 fmt = *(--c);
 697                                         }
 698                                 }
 699                         }
 700                         break;
 701                 }
 702
 703                 if (fmt == '\0') {
 704                         warningf(WARN_FORMAT, pos, "dangling %% with conversion specififer in format string");
 705                         break;
 706                 }
 707
 708                 type_t *expected_type;
 709                 switch (fmt) {
 710                 case 'd':
 711                 case 'i':
 712                         switch (fmt_mod) {
 713                         case FMT_MOD_NONE: expected_type = type_int;         break;
 714                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 715                         case FMT_MOD_h:    expected_type = type_short;       break;
 716                         case FMT_MOD_l:    expected_type = type_long;        break;
 717                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 718                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 719                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 720                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 721                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 722                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 723                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 724
 725                         default:
 726                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 727                                 goto next_arg;
 728                         }
 729                         break;
 730
 731                 case 'o':
 732                 case 'X':
 733                 case 'x':
 734                 case 'u':
 735                         switch (fmt_mod) {
 736                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 737                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 738                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 739                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 740                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 741                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 742                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 743                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 744                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 745                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 746                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 747
 748                         default:
 749                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 750                                 goto next_arg;
 751                         }
 752                         break;
 753
 754                 case 'A':
 755                 case 'a':
 756                 case 'E':
 757                 case 'e':
 758                 case 'F':
 759                 case 'f':
 760                 case 'G':
 761                 case 'g':
 762                         switch (fmt_mod) {
 763                         case FMT_MOD_l:    expected_type = type_double;      break;
 764                         case FMT_MOD_NONE: expected_type = type_float;       break;
 765                         case FMT_MOD_L:    expected_type = type_long_double; break;
 766
 767                         default:
 768                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 769                                 goto next_arg;
 770                         }
 771                         break;
 772
 773                 case 'C':
 774                         if (fmt_mod != FMT_MOD_NONE) {
 775                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 776                                 goto next_arg;
 777                         }
 778                         expected_type = type_wchar_t;
 779                         goto check_c_width;
 780
 781                 case 'c': {
 782                         switch (fmt_mod) {
 783                         case FMT_MOD_NONE: expected_type = type_char;    break;
 784                         case FMT_MOD_l:    expected_type = type_wchar_t; break;
 785                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 786
 787                         default:
 788                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 789                                 goto next_arg;
 790                         }
 791
 792 check_c_width:
 793                         if (width == 0)
 794                                 width = 1;
 795                         if (!suppress_assignment && arg != NULL) {
 796                                 type_t *const type = skip_typeref(revert_automatic_type_conversion(arg->expression));
 797                                 if (is_type_array(type)       &&
 798                                     type->array.size_constant &&
 799                                     width > type->array.size) {
 800                                         warningf(WARN_FORMAT, pos, "target buffer '%T' is too small for %u characters at format %u", type, width, num_fmt);
 801                                 }
 802                         }
 803                         break;
 804                 }
 805
 806                 case 'S':
 807                         if (fmt_mod != FMT_MOD_NONE) {
 808                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 809                                 goto next_arg;
 810                         }
 811                         expected_type = type_wchar_t;
 812                         break;
 813
 814                 case 's':
 815                 case '[': {
 816                         switch (fmt_mod) {
 817                                 case FMT_MOD_NONE: expected_type = type_char;    break;
 818                                 case FMT_MOD_l:    expected_type = type_wchar_t; break;
 819                                 case FMT_MOD_w:    expected_type = type_wchar_t; break;
 820
 821                                 default:
 822                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 823                                         goto next_arg;
 824                         }
 825
 826                         if (!suppress_assignment &&
 827                             width != 0           &&
 828                             arg   != NULL) {
 829                                 type_t *const type = skip_typeref(revert_automatic_type_conversion(arg->expression));
 830                                 if (is_type_array(type)       &&
 831                                     type->array.size_constant &&
 832                                     width >= type->array.size) {
 833                                         warningf(WARN_FORMAT, pos, "target buffer '%T' is too small for %u characters and \\0 at format %u", type, width, num_fmt);
 834                                 }
 835                         }
 836                         break;
 837                 }
 838
 839                 case 'p':
 840                         if (fmt_mod != FMT_MOD_NONE) {
 841                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 842                                 goto next_arg;
 843                         }
 844                         expected_type = type_void;
 845                         break;
 846
 847                 case 'n': {
 848                         if (suppress_assignment) {
 849                                 warningf(WARN_FORMAT, pos, "conversion '%n' cannot be suppressed with '*' at format %u", num_fmt);
 850                         }
 851
 852                         switch (fmt_mod) {
 853                         case FMT_MOD_NONE: expected_type = type_int;         break;
 854                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 855                         case FMT_MOD_h:    expected_type = type_short;       break;
 856                         case FMT_MOD_l:    expected_type = type_long;        break;
 857                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 858                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 859                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 860                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 861
 862                         default:
 863                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 864                                 goto next_arg;
 865                         }
 866                         break;
 867                 }
 868
 869                 default:
 870                         warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at format %u", fmt, num_fmt);
 871                         if (suppress_assignment)
 872                                 continue;
 873                         if (arg == NULL)
 874                                 goto too_few_args;
 875                         goto next_arg;
 876                 }
 877
 878                 if (suppress_assignment)
 879                         continue;
 880
 881                 if (arg == NULL) {
 882 too_few_args:
 883                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 884                         return;
 885                 }
 886
 887                 { /* create a scope here to prevent warning about the jump to next_arg */
 888                         type_t *const arg_type           = arg->expression->base.type;
 889                         type_t *const arg_skip           = skip_typeref(arg_type);
 890                         type_t *const expected_type_skip = skip_typeref(expected_type);
 891
 892                         if (! is_type_pointer(arg_skip))
 893                                 goto error_arg_type;
 894                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 895
 896                         if (fmt == 'p') {
 897                                 /* allow any pointer type for %p, not just void */
 898                                 if (is_type_pointer(ptr_skip))
 899                                         goto next_arg;
 900                         }
 901
 902                         /* do NOT allow const or restrict, all other should be ok */
 903                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 904                                 goto error_arg_type;
 905                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 906                         if (unqual_ptr == expected_type_skip) {
 907                                 goto next_arg;
 908                         } else if (expected_type_skip == type_char) {
 909                                 /* char matches with unsigned char AND signed char */
 910                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 911                                         goto next_arg;
 912                         }
 913 error_arg_type:
 914                         if (is_type_valid(arg_skip)) {
 915                                 source_position_t const *const apos = &arg->expression->base.source_position;
 916                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 917                                 warningf(WARN_FORMAT, apos, "conversion '%%%s%c' at position %u specifies type '%T*' but the argument has type '%T'", mod, (char)fmt, num_fmt, expected_type, arg_type);
 918                         }
 919                 }
 920 next_arg:
 921                 arg = arg->next;
 922         }
 923         assert(fmt == '\0');
 924         if (c+1 < string + size) {
 925                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 926         }
 927         if (arg != NULL) {
 928                 unsigned num_args = num_fmt;
 929                 while (arg != NULL) {
 930                         ++num_args;
 931                         arg = arg->next;
 932                 }
 933                 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" : "");
 934         }
 935 }
 936
 937 static const format_spec_t builtin_table[] = {
 938         { "printf",        FORMAT_PRINTF,   0, 1 },
 939         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 940         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 941         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 942         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 943         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 944         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 945         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 946         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 947         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 948
 949         { "scanf",         FORMAT_SCANF,    0, 1 },
 950         { "wscanf",        FORMAT_SCANF,    0, 1 },
 951         { "sscanf",        FORMAT_SCANF,    1, 2 },
 952         { "swscanf",       FORMAT_SCANF,    1, 2 },
 953         { "fscanf",        FORMAT_SCANF,    1, 2 },
 954         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 955
 956         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 957         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 958
 959         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 960
 961         /* MS extensions */
 962         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 963         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 964         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 965         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 966         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 967         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 968         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 969         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 970         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 971         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 972         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 973         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 974         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 975         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 976         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 977         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 978         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 979         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 980         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 981         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 982         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 983         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 984 };
 985
 986 void check_format(const call_expression_t *const call)
 987 {
 988         if (!is_warn_on(WARN_FORMAT))
 989                 return;
 990
 991         const expression_t *const func_expr = call->function;
 992         if (func_expr->kind != EXPR_REFERENCE)
 993                 return;
 994
 995         const entity_t        *const entity = func_expr->reference.entity;
 996         const call_argument_t *      arg    = call->arguments;
 997
 998         /*
 999          * For some functions we always check the format, even if it was not
1000          * specified. This allows to check format even in MS mode or without
1001          * header included.
1002          */
1003         const char *const name = entity->base.symbol->string;
1004         for (size_t i = 0; i < lengthof(builtin_table); ++i) {
1005                 if (streq(name, builtin_table[i].name)) {
1006                         switch (builtin_table[i].fmt_kind) {
1007                         case FORMAT_PRINTF:
1008                                 check_printf_format(arg, &builtin_table[i]);
1009                                 break;
1010                         case FORMAT_SCANF:
1011                                 check_scanf_format(arg, &builtin_table[i]);
1012                                 break;
1013                         case FORMAT_STRFTIME:
1014                         case FORMAT_STRFMON:
1015                                 /* TODO: implement other cases */
1016                                 break;
1017                         }
1018                         break;
1019                 }
1020         }
1021 }