nsz Git - cparser/blob - format_check.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2009 Matthias Braun <matze@braunis.de>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  18  * 02111-1307, USA.
  19  */
  20 #include <config.h>
  21
  22 #include <ctype.h>
  23
  24 #include "adt/util.h"
  25 #include "format_check.h"
  26 #include "symbol_t.h"
  27 #include "ast_t.h"
  28 #include "entity_t.h"
  29 #include "diagnostic.h"
  30 #include "types.h"
  31 #include "type_t.h"
  32 #include "warning.h"
  33 #include "lang_features.h"
  34
  35 typedef enum format_flag_t {
  36         FMT_FLAG_NONE  = 0,
  37         FMT_FLAG_HASH  = 1U << 0,
  38         FMT_FLAG_ZERO  = 1U << 1,
  39         FMT_FLAG_MINUS = 1U << 2,
  40         FMT_FLAG_SPACE = 1U << 3,
  41         FMT_FLAG_PLUS  = 1U << 4,
  42         FMT_FLAG_TICK  = 1U << 5
  43 } format_flag_t;
  44
  45 typedef unsigned format_flags_t;
  46
  47 typedef enum format_length_modifier_t {
  48         FMT_MOD_NONE,
  49         FMT_MOD_L,
  50         FMT_MOD_hh,
  51         FMT_MOD_h,
  52         FMT_MOD_l,
  53         FMT_MOD_ll,
  54         FMT_MOD_j,
  55         FMT_MOD_t,
  56         FMT_MOD_z,
  57         FMT_MOD_q,
  58         /* only in microsoft mode */
  59         FMT_MOD_w,
  60         FMT_MOD_I,
  61         FMT_MOD_I32,
  62         FMT_MOD_I64
  63 } format_length_modifier_t;
  64
  65 typedef struct format_spec_t {
  66         const char    *name;     /**< name of the function */
  67         format_kind_t  fmt_kind; /**< kind */
  68         unsigned       fmt_idx;  /**< index of the format string */
  69         unsigned       arg_idx;  /**< index of the first argument */
  70 } format_spec_t;
  71
  72 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  73 {
  74         static const char* const names[] = {
  75                 [FMT_MOD_NONE] = "",
  76                 [FMT_MOD_L]    = "L",
  77                 [FMT_MOD_hh]   = "hh",
  78                 [FMT_MOD_h]    = "h",
  79                 [FMT_MOD_l]    = "l",
  80                 [FMT_MOD_ll]   = "ll",
  81                 [FMT_MOD_j]    = "j",
  82                 [FMT_MOD_t]    = "t",
  83                 [FMT_MOD_z]    = "z",
  84                 [FMT_MOD_q]    = "q",
  85                 /* only in microsoft mode */
  86                 [FMT_MOD_w]    = "w",
  87                 [FMT_MOD_I]    = "I",
  88                 [FMT_MOD_I32]  = "I32",
  89                 [FMT_MOD_I64]  = "I64"
  90         };
  91         assert((size_t)mod < lengthof(names));
  92         return names[mod];
  93 }
  94
  95 static void warn_invalid_length_modifier(const source_position_t *pos,
  96                                          const format_length_modifier_t mod,
  97                                          const utf32 conversion)
  98 {
  99         char const *const lmod = get_length_modifier_name(mod);
 100         warningf(WARN_FORMAT, pos, "invalid length modifier '%s' for conversion specifier '%%%c'", lmod, conversion);
 101 }
 102
 103 /**
 104  * Check printf-style format. Returns number of expected arguments.
 105  */
 106 static int internal_check_printf_format(const expression_t *fmt_expr,
 107                                         const call_argument_t *arg,
 108                                         const format_spec_t *spec)
 109 {
 110         while (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 111                 fmt_expr = fmt_expr->unary.value;
 112         }
 113
 114         /*
 115          * gettext results in expressions like (X ? "format_string" : Y)
 116          * we assume the left part is the format string
 117          */
 118         if (fmt_expr->kind == EXPR_CONDITIONAL) {
 119                 conditional_expression_t const *const c = &fmt_expr->conditional;
 120                 expression_t             const *      t = c->true_expression;
 121                 if (t == NULL)
 122                         t = c->condition;
 123                 int const nt = internal_check_printf_format(t,                   arg, spec);
 124                 int const nf = internal_check_printf_format(c->false_expression, arg, spec);
 125                 return nt > nf ? nt : nf;
 126         }
 127
 128         if (fmt_expr->kind != EXPR_STRING_LITERAL
 129                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 130                 return -1;
 131
 132         const char *string = fmt_expr->literal.value.begin;
 133         size_t      size   = fmt_expr->literal.value.size;
 134         const char *c      = string;
 135
 136         const source_position_t *pos = &fmt_expr->base.source_position;
 137         unsigned num_fmt  = 0;
 138         unsigned num_args = 0;
 139         char     fmt;
 140         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 141                 if (fmt != '%')
 142                         continue;
 143                 fmt = *(++c);
 144
 145                 if (fmt == '\0') {
 146                         warningf(WARN_FORMAT, pos, "dangling %% in format string");
 147                         break;
 148                 }
 149                 if (fmt == '%')
 150                         continue;
 151
 152                 ++num_fmt;
 153                 ++num_args;
 154
 155                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 156                 if (fmt == '0') {
 157                         fmt = *(++c);
 158                         fmt_flags |= FMT_FLAG_ZERO;
 159                 }
 160
 161                 /* argument selector or minimum field width */
 162                 if (isdigit(fmt)) {
 163                         do {
 164                                 fmt = *(++c);
 165                         } while (isdigit(fmt));
 166
 167                         /* digit string was ... */
 168                         if (fmt == '$') {
 169                                 /* ... argument selector */
 170                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 171                                 /* TODO implement */
 172                                 return -1;
 173                         }
 174                         /* ... minimum field width */
 175                 } else {
 176                         /* flags */
 177                         for (;;) {
 178                                 format_flags_t flag;
 179                                 switch (fmt) {
 180                                         case '#':  flag = FMT_FLAG_HASH;  break;
 181                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 182                                         case '-':  flag = FMT_FLAG_MINUS; break;
 183                                         case '\'': flag = FMT_FLAG_TICK;  break;
 184
 185                                         case ' ':
 186                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 187                                                         warningf(WARN_FORMAT, pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 188                                                 }
 189                                                 flag = FMT_FLAG_SPACE;
 190                                                 break;
 191
 192                                         case '+':
 193                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 194                                                         warningf(WARN_FORMAT, pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 195                                                 }
 196                                                 flag = FMT_FLAG_PLUS;
 197                                                 break;
 198
 199                                         default: goto break_fmt_flags;
 200                                 }
 201                                 if (fmt_flags & flag) {
 202                                         warningf(WARN_FORMAT, pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 203                                 }
 204                                 fmt_flags |= flag;
 205                                 fmt = *(++c);
 206                         }
 207 break_fmt_flags:
 208
 209                         /* minimum field width */
 210                         if (fmt == '*') {
 211                                 ++num_args;
 212                                 fmt = *(++c);
 213                                 if (arg == NULL) {
 214                                         warningf(WARN_FORMAT, pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 215                                         return -1;
 216                                 }
 217                                 const type_t *const arg_type = arg->expression->base.type;
 218                                 if (arg_type != type_int) {
 219                                         warningf(WARN_FORMAT, pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 220                                 }
 221                                 arg = arg->next;
 222                         } else {
 223                                 while (isdigit(fmt)) {
 224                                         fmt = *(++c);
 225                                 }
 226                         }
 227                 }
 228
 229                 /* precision */
 230                 if (fmt == '.') {
 231                         if (fmt_flags & FMT_FLAG_ZERO) {
 232                                 warningf(WARN_FORMAT, pos, "'0' flag ignored with precision in conversion specification %u", num_fmt);
 233                         }
 234
 235                         ++num_args;
 236                         fmt = *(++c);
 237                         if (fmt == '*') {
 238                                 fmt = *(++c);
 239                                 if (arg == NULL) {
 240                                         warningf(WARN_FORMAT, pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 241                                         return -1;
 242                                 }
 243                                 const type_t *const arg_type = arg->expression->base.type;
 244                                 if (arg_type != type_int) {
 245                                         warningf(WARN_FORMAT, pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 246                                 }
 247                                 arg = arg->next;
 248                         } else {
 249                                 /* digit string may be omitted */
 250                                 while (isdigit(fmt)) {
 251                                         fmt = *(++c);
 252                                 }
 253                         }
 254                 }
 255
 256                 /* length modifier */
 257                 format_length_modifier_t fmt_mod;
 258                 switch (fmt) {
 259                         case 'h':
 260                                 fmt = *(++c);
 261                                 if (fmt == 'h') {
 262                                         fmt = *(++c);
 263                                         fmt_mod = FMT_MOD_hh;
 264                                 } else {
 265                                         fmt_mod = FMT_MOD_h;
 266                                 }
 267                                 break;
 268
 269                         case 'l':
 270                                 fmt = *(++c);
 271                                 if (fmt == 'l') {
 272                                         fmt = *(++c);
 273                                         fmt_mod = FMT_MOD_ll;
 274                                 } else {
 275                                         fmt_mod = FMT_MOD_l;
 276                                 }
 277                                 break;
 278
 279                         case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
 280                         case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
 281                         case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
 282                         case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
 283                         case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 284                         /* microsoft mode */
 285                         case 'w':
 286                                 if (c_mode & _MS) {
 287                                         fmt = *(++c); fmt_mod = FMT_MOD_w;
 288                                 } else {
 289                                         fmt_mod = FMT_MOD_NONE;
 290                                 }
 291                                 break;
 292                         case 'I':
 293                                 if (c_mode & _MS) {
 294                                         fmt = *(++c); fmt_mod = FMT_MOD_I;
 295                                         if (fmt == '3') {
 296                                                 fmt = *(++c);
 297                                                 if (fmt == '2') {
 298                                                         fmt = *(++c);
 299                                                         fmt_mod = FMT_MOD_I32;
 300                                                 } else {
 301                                                         /* rewind */
 302                                                         fmt = *(--c);
 303                                                 }
 304                                         } else if (fmt == '6') {
 305                                                 fmt = *(++c);
 306                                                 if (fmt == '4') {
 307                                                         fmt = *(++c);
 308                                                         fmt_mod = FMT_MOD_I64;
 309                                                 } else {
 310                                                         /* rewind */
 311                                                         fmt = *(--c);
 312                                                 }
 313                                         }
 314                                 } else {
 315                                         fmt_mod = FMT_MOD_NONE;
 316                                 }
 317                                 break;
 318                         default:
 319                                 fmt_mod = FMT_MOD_NONE;
 320                                 break;
 321                 }
 322
 323
 324                 type_t            *expected_type;
 325                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 326                 format_flags_t     allowed_flags;
 327                 switch (fmt) {
 328                         case 'd':
 329                         case 'i':
 330                                 switch (fmt_mod) {
 331                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 332                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 333                                         case FMT_MOD_h:    expected_type = type_short;       break;
 334                                         case FMT_MOD_l:    expected_type = type_long;        break;
 335                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 336                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 337                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 338                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 339                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 340                                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 341                                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 342
 343                                         default:
 344                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 345                                                 goto next_arg;
 346                                 }
 347                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 348                                 break;
 349
 350                         case 'o':
 351                         case 'X':
 352                         case 'x':
 353                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 354                                 goto eval_fmt_mod_unsigned;
 355
 356                         case 'u':
 357                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 358 eval_fmt_mod_unsigned:
 359                                 switch (fmt_mod) {
 360                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 361                                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 362                                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 363                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 364                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 365                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 366                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 367                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 368                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 369                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 370                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 371
 372                                         default:
 373                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 374                                                 goto next_arg;
 375                                 }
 376                                 break;
 377
 378                         case 'A':
 379                         case 'a':
 380                         case 'E':
 381                         case 'e':
 382                         case 'F':
 383                         case 'f':
 384                         case 'G':
 385                         case 'g':
 386                                 switch (fmt_mod) {
 387                                         case FMT_MOD_l:    /* l modifier is ignored */
 388                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 389                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 390
 391                                         default:
 392                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 393                                                 goto next_arg;
 394                                 }
 395                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 396                                 break;
 397
 398                         case 'C':
 399                                 if (fmt_mod != FMT_MOD_NONE) {
 400                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 401                                         goto next_arg;
 402                                 }
 403                                 expected_type = type_wchar_t;
 404                                 allowed_flags = FMT_FLAG_NONE;
 405                                 break;
 406
 407                         case 'c':
 408                                 expected_type = type_int;
 409                                 switch (fmt_mod) {
 410                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 411                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 412                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 413
 414                                         default:
 415                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 416                                                 goto next_arg;
 417                                 }
 418                                 allowed_flags = FMT_FLAG_NONE;
 419                                 break;
 420
 421                         case 'S':
 422                                 if (fmt_mod != FMT_MOD_NONE) {
 423                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 424                                         goto next_arg;
 425                                 }
 426                                 expected_type = type_wchar_t_ptr;
 427                                 expected_qual = TYPE_QUALIFIER_CONST;
 428                                 allowed_flags = FMT_FLAG_MINUS;
 429                                 break;
 430
 431                         case 's':
 432                                 switch (fmt_mod) {
 433                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 434                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 435                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 436
 437                                         default:
 438                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 439                                                 goto next_arg;
 440                                 }
 441                                 expected_qual = TYPE_QUALIFIER_CONST;
 442                                 allowed_flags = FMT_FLAG_MINUS;
 443                                 break;
 444
 445                         case 'p':
 446                                 if (fmt_mod != FMT_MOD_NONE) {
 447                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 448                                         goto next_arg;
 449                                 }
 450                                 expected_type = type_void_ptr;
 451                                 allowed_flags = FMT_FLAG_NONE;
 452                                 break;
 453
 454                         case 'n':
 455                                 switch (fmt_mod) {
 456                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 457                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 458                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 459                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 460                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 461                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 462                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 463                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 464
 465                                         default:
 466                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 467                                                 goto next_arg;
 468                                 }
 469                                 allowed_flags = FMT_FLAG_NONE;
 470                                 break;
 471
 472                         default:
 473                                 warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at position %u", fmt, num_fmt);
 474                                 if (arg == NULL) {
 475                                         goto too_few_args;
 476                                 }
 477                                 goto next_arg;
 478                 }
 479
 480                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 481                 if (wrong_flags != 0) {
 482                         char  wrong[8];
 483                         char *p = wrong;
 484                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 485                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 486                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 487                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 488                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 489                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 490                         *p = '\0';
 491
 492                         warningf(WARN_FORMAT, pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 493                 }
 494
 495                 if (arg == NULL) {
 496 too_few_args:
 497                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 498                         return -1;
 499                 }
 500
 501                 { /* create a scope here to prevent warning about the jump to next_arg */
 502                         type_t *const arg_type           = arg->expression->base.type;
 503                         type_t *const arg_skip           = skip_typeref(arg_type);
 504                         type_t *const expected_type_skip = skip_typeref(expected_type);
 505
 506                         if (fmt == 'p') {
 507                                 /* allow any pointer type for %p, not just void */
 508                                 if (is_type_pointer(arg_skip))
 509                                         goto next_arg;
 510                         }
 511
 512                         if (is_type_pointer(expected_type_skip)) {
 513                                 if (is_type_pointer(arg_skip)) {
 514                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 515                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 516                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 517                                                 get_unqualified_type(arg_to) == exp_to) {
 518                                                 goto next_arg;
 519                                         }
 520                                 }
 521                         } else if (get_unqualified_type(arg_skip) == expected_type_skip) {
 522                                 goto next_arg;
 523                         } else if (arg->expression->kind == EXPR_UNARY_CAST_IMPLICIT) {
 524                                 expression_t const *const expr        = arg->expression->unary.value;
 525                                 type_t             *const unprom_type = skip_typeref(expr->base.type);
 526                                 if (get_unqualified_type(unprom_type) == expected_type_skip) {
 527                                         goto next_arg;
 528                                 }
 529                                 if (expected_type_skip == type_unsigned_int && !is_type_signed(unprom_type)) {
 530                                         goto next_arg;
 531                                 }
 532                         }
 533                         if (is_type_valid(arg_skip)) {
 534                                 source_position_t const *const apos = &arg->expression->base.source_position;
 535                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 536                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 537                         }
 538                 }
 539 next_arg:
 540                 arg = arg->next;
 541         }
 542         assert(fmt == '\0');
 543         if (c+1 < string + size) {
 544                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 545         }
 546         return num_args;
 547 }
 548
 549 /**
 550  * Check printf-style format.
 551  */
 552 static void check_printf_format(call_argument_t const *arg,
 553                                 format_spec_t const *const spec)
 554 {
 555         /* find format arg */
 556         size_t idx = 0;
 557         for (; idx < spec->fmt_idx; ++idx) {
 558                 if (arg == NULL)
 559                         return;
 560                 arg = arg->next;
 561         }
 562
 563         expression_t const *const fmt_expr = arg->expression;
 564
 565         /* find the real args */
 566         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 567                 arg = arg->next;
 568
 569         int const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
 570         if (num_fmt < 0)
 571                 return;
 572
 573         size_t num_args = 0;
 574         for (; arg != NULL; arg = arg->next)
 575                 ++num_args;
 576         if (num_args > (size_t)num_fmt) {
 577                 source_position_t const *const pos = &fmt_expr->base.source_position;
 578                 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" : "");
 579         }
 580 }
 581
 582 /**
 583  * Check scanf-style format.
 584  */
 585 static void check_scanf_format(const call_argument_t *arg,
 586                                const format_spec_t *spec)
 587 {
 588         /* find format arg */
 589         unsigned idx = 0;
 590         for (; idx < spec->fmt_idx; ++idx) {
 591                 if (arg == NULL)
 592                         return;
 593                 arg = arg->next;
 594         }
 595
 596         const expression_t *fmt_expr = arg->expression;
 597         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 598                 fmt_expr = fmt_expr->unary.value;
 599         }
 600
 601         if (fmt_expr->kind != EXPR_STRING_LITERAL
 602                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 603                 return;
 604
 605         const char *string = fmt_expr->literal.value.begin;
 606         size_t      size   = fmt_expr->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                 /* look for length modifiers */
 632                 format_length_modifier_t fmt_mod = FMT_MOD_NONE;
 633                 switch (fmt) {
 634                 case 'h':
 635                         fmt = *(++c);
 636                         if (fmt == 'h') {
 637                                 fmt = *(++c);
 638                                 fmt_mod = FMT_MOD_hh;
 639                         } else {
 640                                 fmt_mod = FMT_MOD_h;
 641                         }
 642                         break;
 643
 644                 case 'l':
 645                         fmt = *(++c);
 646                         if (fmt == 'l') {
 647                                 fmt = *(++c);
 648                                 fmt_mod = FMT_MOD_ll;
 649                         } else {
 650                                 fmt_mod = FMT_MOD_l;
 651                         }
 652                         break;
 653
 654                 case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
 655                 case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
 656                 case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
 657                 case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
 658                 /* microsoft mode */
 659                 case 'w':
 660                         if (c_mode & _MS) {
 661                                 fmt = *(++c);
 662                                 fmt_mod = FMT_MOD_w;
 663                         }
 664                         break;
 665                 case 'I':
 666                         if (c_mode & _MS) {
 667                                 fmt = *(++c);
 668                                 fmt_mod = FMT_MOD_I;
 669                                 if (fmt == '3') {
 670                                         fmt = *(++c);
 671                                         if (fmt == '2') {
 672                                                 fmt = *(++c);
 673                                                 fmt_mod = FMT_MOD_I32;
 674                                         } else {
 675                                                 /* rewind */
 676                                                 fmt = *(--c);
 677                                         }
 678                                 } else if (fmt == '6') {
 679                                         fmt = *(++c);
 680                                         if (fmt == '4') {
 681                                                 fmt = *(++c);
 682                                                 fmt_mod = FMT_MOD_I64;
 683                                         } else {
 684                                                 /* rewind */
 685                                                 fmt = *(--c);
 686                                         }
 687                                 }
 688                         }
 689                         break;
 690                 }
 691
 692                 if (fmt == '\0') {
 693                         warningf(WARN_FORMAT, pos, "dangling %% with conversion specififer in format string");
 694                         break;
 695                 }
 696
 697                 type_t *expected_type;
 698                 switch (fmt) {
 699                 case 'd':
 700                 case 'i':
 701                         switch (fmt_mod) {
 702                         case FMT_MOD_NONE: expected_type = type_int;         break;
 703                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 704                         case FMT_MOD_h:    expected_type = type_short;       break;
 705                         case FMT_MOD_l:    expected_type = type_long;        break;
 706                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 707                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 708                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 709                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 710                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 711                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 712                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 713
 714                         default:
 715                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 716                                 goto next_arg;
 717                         }
 718                         break;
 719
 720                 case 'o':
 721                 case 'X':
 722                 case 'x':
 723                 case 'u':
 724                         switch (fmt_mod) {
 725                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 726                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 727                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 728                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 729                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 730                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 731                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 732                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 733                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 734                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 735                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 736
 737                         default:
 738                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 739                                 goto next_arg;
 740                         }
 741                         break;
 742
 743                 case 'A':
 744                 case 'a':
 745                 case 'E':
 746                 case 'e':
 747                 case 'F':
 748                 case 'f':
 749                 case 'G':
 750                 case 'g':
 751                         switch (fmt_mod) {
 752                         case FMT_MOD_l:    expected_type = type_double;      break;
 753                         case FMT_MOD_NONE: expected_type = type_float;       break;
 754                         case FMT_MOD_L:    expected_type = type_long_double; break;
 755
 756                         default:
 757                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 758                                 goto next_arg;
 759                         }
 760                         break;
 761
 762                 case 'C':
 763                         if (fmt_mod != FMT_MOD_NONE) {
 764                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 765                                 goto next_arg;
 766                         }
 767                         expected_type = type_wchar_t;
 768                         break;
 769
 770                 case 'c':
 771                         expected_type = type_int;
 772                         switch (fmt_mod) {
 773                         case FMT_MOD_NONE: expected_type = type_char;    break;
 774                         case FMT_MOD_l:    expected_type = type_wchar_t; break;
 775                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 776
 777                         default:
 778                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 779                                 goto next_arg;
 780                         }
 781                         break;
 782
 783                 case 'S':
 784                         if (fmt_mod != FMT_MOD_NONE) {
 785                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 786                                 goto next_arg;
 787                         }
 788                         expected_type = type_wchar_t;
 789                         break;
 790
 791                 case 's':
 792                 case '[':
 793                         switch (fmt_mod) {
 794                                 case FMT_MOD_NONE: expected_type = type_char;    break;
 795                                 case FMT_MOD_l:    expected_type = type_wchar_t; break;
 796                                 case FMT_MOD_w:    expected_type = type_wchar_t; break;
 797
 798                                 default:
 799                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 800                                         goto next_arg;
 801                         }
 802                         break;
 803
 804                 case 'p':
 805                         if (fmt_mod != FMT_MOD_NONE) {
 806                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 807                                 goto next_arg;
 808                         }
 809                         expected_type = type_void_ptr;
 810                         break;
 811
 812                 case 'n': {
 813                         if (suppress_assignment) {
 814                                 warningf(WARN_FORMAT, pos, "conversion '%n' cannot be suppressed with '*' at format %u", num_fmt);
 815                         }
 816
 817                         switch (fmt_mod) {
 818                         case FMT_MOD_NONE: expected_type = type_int;         break;
 819                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 820                         case FMT_MOD_h:    expected_type = type_short;       break;
 821                         case FMT_MOD_l:    expected_type = type_long;        break;
 822                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 823                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 824                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 825                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 826
 827                         default:
 828                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 829                                 goto next_arg;
 830                         }
 831                         break;
 832                 }
 833
 834                 default:
 835                         warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at format %u", fmt, num_fmt);
 836                         if (suppress_assignment)
 837                                 continue;
 838                         if (arg == NULL)
 839                                 goto too_few_args;
 840                         goto next_arg;
 841                 }
 842
 843                 if (suppress_assignment)
 844                         continue;
 845
 846                 if (arg == NULL) {
 847 too_few_args:
 848                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 849                         return;
 850                 }
 851
 852                 { /* create a scope here to prevent warning about the jump to next_arg */
 853                         type_t *const arg_type           = arg->expression->base.type;
 854                         type_t *const arg_skip           = skip_typeref(arg_type);
 855                         type_t *const expected_type_skip = skip_typeref(expected_type);
 856
 857                         if (! is_type_pointer(arg_skip))
 858                                 goto error_arg_type;
 859                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 860
 861                         if (fmt == 'p') {
 862                                 /* allow any pointer type for %p, not just void */
 863                                 if (is_type_pointer(ptr_skip))
 864                                         goto next_arg;
 865                         }
 866
 867                         /* do NOT allow const or restrict, all other should be ok */
 868                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 869                                 goto error_arg_type;
 870                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 871                         if (unqual_ptr == expected_type_skip) {
 872                                 goto next_arg;
 873                         } else if (expected_type_skip == type_char) {
 874                                 /* char matches with unsigned char AND signed char */
 875                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 876                                         goto next_arg;
 877                         }
 878 error_arg_type:
 879                         if (is_type_valid(arg_skip)) {
 880                                 source_position_t const *const apos = &arg->expression->base.source_position;
 881                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 882                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 883                         }
 884                 }
 885 next_arg:
 886                 arg = arg->next;
 887         }
 888         assert(fmt == '\0');
 889         if (c+1 < string + size) {
 890                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 891         }
 892         if (arg != NULL) {
 893                 unsigned num_args = num_fmt;
 894                 while (arg != NULL) {
 895                         ++num_args;
 896                         arg = arg->next;
 897                 }
 898                 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" : "");
 899         }
 900 }
 901
 902 static const format_spec_t builtin_table[] = {
 903         { "printf",        FORMAT_PRINTF,   0, 1 },
 904         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 905         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 906         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 907         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 908         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 909         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 910         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 911         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 912         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 913
 914         { "scanf",         FORMAT_SCANF,    0, 1 },
 915         { "wscanf",        FORMAT_SCANF,    0, 1 },
 916         { "sscanf",        FORMAT_SCANF,    1, 2 },
 917         { "swscanf",       FORMAT_SCANF,    1, 2 },
 918         { "fscanf",        FORMAT_SCANF,    1, 2 },
 919         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 920
 921         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 922         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 923
 924         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 925
 926         /* MS extensions */
 927         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 928         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 929         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 930         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 931         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 932         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 933         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 934         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 935         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 936         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 937         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 938         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 939         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 940         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 941         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 942         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 943         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 944         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 945         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 946         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 947         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 948         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 949 };
 950
 951 void check_format(const call_expression_t *const call)
 952 {
 953         if (!is_warn_on(WARN_FORMAT))
 954                 return;
 955
 956         const expression_t *const func_expr = call->function;
 957         if (func_expr->kind != EXPR_REFERENCE)
 958                 return;
 959
 960         const entity_t        *const entity = func_expr->reference.entity;
 961         const call_argument_t *      arg    = call->arguments;
 962
 963         /*
 964          * For some functions we always check the format, even if it was not
 965          * specified. This allows to check format even in MS mode or without
 966          * header included.
 967          */
 968         const char *const name = entity->base.symbol->string;
 969         for (size_t i = 0; i < lengthof(builtin_table); ++i) {
 970                 if (strcmp(name, builtin_table[i].name) == 0) {
 971                         switch (builtin_table[i].fmt_kind) {
 972                         case FORMAT_PRINTF:
 973                                 check_printf_format(arg, &builtin_table[i]);
 974                                 break;
 975                         case FORMAT_SCANF:
 976                                 check_scanf_format(arg, &builtin_table[i]);
 977                                 break;
 978                         case FORMAT_STRFTIME:
 979                         case FORMAT_STRFMON:
 980                                 /* TODO: implement other cases */
 981                                 break;
 982                         }
 983                         break;
 984                 }
 985         }
 986 }