nsz Git - cparser/blob - format_check.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2008 Matthias Braun <matze@braunis.de>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  18  * 02111-1307, USA.
  19  */
  20 #include <ctype.h>
  21 #include <wctype.h>
  22
  23 #include "format_check.h"
  24 #include "symbol_t.h"
  25 #include "ast_t.h"
  26 #include "entity_t.h"
  27 #include "diagnostic.h"
  28 #include "types.h"
  29 #include "type_t.h"
  30 #include "warning.h"
  31 #include "lang_features.h"
  32
  33 typedef enum format_flag_t {
  34         FMT_FLAG_NONE  = 0,
  35         FMT_FLAG_HASH  = 1U << 0,
  36         FMT_FLAG_ZERO  = 1U << 1,
  37         FMT_FLAG_MINUS = 1U << 2,
  38         FMT_FLAG_SPACE = 1U << 3,
  39         FMT_FLAG_PLUS  = 1U << 4,
  40         FMT_FLAG_TICK  = 1U << 5
  41 } format_flag_t;
  42
  43 typedef unsigned format_flags_t;
  44
  45 typedef enum format_length_modifier_t {
  46         FMT_MOD_NONE,
  47         FMT_MOD_L,
  48         FMT_MOD_hh,
  49         FMT_MOD_h,
  50         FMT_MOD_l,
  51         FMT_MOD_ll,
  52         FMT_MOD_j,
  53         FMT_MOD_t,
  54         FMT_MOD_z,
  55         FMT_MOD_q,
  56         /* only in microsoft mode */
  57         FMT_MOD_w,
  58         FMT_MOD_I,
  59         FMT_MOD_I32,
  60         FMT_MOD_I64
  61 } format_length_modifier_t;
  62
  63 typedef struct format_spec_t {
  64         const char    *name;     /**< name of the function */
  65         format_kind_t  fmt_kind; /**< kind */
  66         unsigned       fmt_idx;  /**< index of the format string */
  67         unsigned       arg_idx;  /**< index of the first argument */
  68 } format_spec_t;
  69
  70 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  71 {
  72         static const char* const names[] = {
  73                 [FMT_MOD_NONE] = "",
  74                 [FMT_MOD_L]    = "L",
  75                 [FMT_MOD_hh]   = "hh",
  76                 [FMT_MOD_h]    = "h",
  77                 [FMT_MOD_l]    = "l",
  78                 [FMT_MOD_ll]   = "ll",
  79                 [FMT_MOD_j]    = "j",
  80                 [FMT_MOD_t]    = "t",
  81                 [FMT_MOD_z]    = "z",
  82                 [FMT_MOD_q]    = "q",
  83                 /* only in microsoft mode */
  84                 [FMT_MOD_w]    = "w",
  85                 [FMT_MOD_I]    = "I",
  86                 [FMT_MOD_I32]  = "I32",
  87                 [FMT_MOD_I64]  = "I64"
  88         };
  89         assert(mod < sizeof(names) / sizeof(*names));
  90         return names[mod];
  91 }
  92
  93 static void warn_invalid_length_modifier(const source_position_t *pos,
  94                                          const format_length_modifier_t mod,
  95                                          const wchar_rep_t conversion)
  96 {
  97         warningf(pos,
  98                 "invalid length modifier '%s' for conversion specifier '%%%c'",
  99                 get_length_modifier_name(mod), conversion
 100         );
 101 }
 102
 103 typedef struct vchar_t vchar_t;
 104 struct vchar_t {
 105         const void *string;   /**< the string */
 106         size_t     position;  /**< current position */
 107         size_t     size;      /**< size of the string */
 108
 109         /** return the first character of the string and setthe position to 0. */
 110         unsigned (*first)(vchar_t *self);
 111         /** return the next character of the string */
 112         unsigned (*next)(vchar_t *self);
 113         /** return non_zero if the given character is a digit */
 114         int (*is_digit)(unsigned vchar);
 115 };
 116
 117 static unsigned string_first(vchar_t *self) {
 118         self->position = 0;
 119         const string_t *string = self->string;
 120         return string->begin[0];
 121 }
 122
 123 static unsigned string_next(vchar_t *self) {
 124         ++self->position;
 125         const string_t *string = self->string;
 126         return string->begin[self->position];
 127 }
 128
 129 static int string_isdigit(unsigned vchar) {
 130         return isdigit(vchar);
 131 }
 132
 133 static unsigned wstring_first(vchar_t *self) {
 134         self->position = 0;
 135         const wide_string_t *wstring = self->string;
 136         return wstring->begin[0];
 137 }
 138
 139 static unsigned wstring_next(vchar_t *self) {
 140         ++self->position;
 141         const wide_string_t *wstring = self->string;
 142         return wstring->begin[self->position];
 143 }
 144
 145 static int wstring_isdigit(unsigned vchar) {
 146         return iswdigit(vchar);
 147 }
 148
 149 static bool atend(vchar_t *self) {
 150         return self->position + 1 == self->size;
 151 }
 152
 153 /**
 154  * Check printf-style format.
 155  */
 156 static void check_printf_format(const call_argument_t *arg, const format_spec_t *spec)
 157 {
 158         /* find format arg */
 159         unsigned idx = 0;
 160         for (; idx < spec->fmt_idx && arg != NULL; ++idx)
 161                 arg = arg->next;
 162
 163         if (arg == NULL)
 164                 return;
 165
 166         const expression_t *fmt_expr = arg->expression;
 167         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 168                 fmt_expr = fmt_expr->unary.value;
 169         }
 170
 171         vchar_t vchar;
 172         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 173                 vchar.string   = &fmt_expr->wide_string.value;
 174                 vchar.size     = fmt_expr->wide_string.value.size;
 175                 vchar.first    = wstring_first;
 176                 vchar.next     = wstring_next;
 177                 vchar.is_digit = wstring_isdigit;
 178         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 179                 vchar.string   = &fmt_expr->string.value;
 180                 vchar.size     = fmt_expr->string.value.size;
 181                 vchar.first    = string_first;
 182                 vchar.next     = string_next;
 183                 vchar.is_digit = string_isdigit;
 184         } else {
 185                 return;
 186         }
 187         /* find the real args */
 188         for(; idx < spec->arg_idx && arg != NULL; ++idx)
 189                 arg = arg->next;
 190
 191         const source_position_t *pos = &fmt_expr->base.source_position;
 192         unsigned fmt     = vchar.first(&vchar);
 193         unsigned num_fmt = 0;
 194         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 195                 if (fmt != '%')
 196                         continue;
 197                 fmt = vchar.next(&vchar);
 198
 199                 if (fmt == '%')
 200                         continue;
 201
 202                 ++num_fmt;
 203
 204                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 205                 if (fmt == '0') {
 206                         fmt = vchar.next(&vchar);
 207                         fmt_flags |= FMT_FLAG_ZERO;
 208                 }
 209
 210                 /* argument selector or minimum field width */
 211                 if (vchar.is_digit(fmt)) {
 212                         do {
 213                                 fmt = vchar.next(&vchar);
 214                         } while (vchar.is_digit(fmt));
 215
 216                         /* digit string was ... */
 217                         if (fmt == '$') {
 218                                 /* ... argument selector */
 219                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 220                                 /* TODO implement */
 221                                 return;
 222                         }
 223                         /* ... minimum field width */
 224                 } else {
 225                         /* flags */
 226                         for (;;) {
 227                                 format_flags_t flag;
 228                                 switch (fmt) {
 229                                         case '#':  flag = FMT_FLAG_HASH;  break;
 230                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 231                                         case '-':  flag = FMT_FLAG_MINUS; break;
 232                                         case '\'': flag = FMT_FLAG_TICK;  break;
 233
 234                                         case ' ':
 235                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 236                                                         warningf(pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 237                                                 }
 238                                                 flag = FMT_FLAG_SPACE;
 239                                                 break;
 240
 241                                         case '+':
 242                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 243                                                         warningf(pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 244                                                 }
 245                                                 flag = FMT_FLAG_PLUS;
 246                                                 break;
 247
 248                                         default: goto break_fmt_flags;
 249                                 }
 250                                 if (fmt_flags & flag) {
 251                                         warningf(pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 252                                 }
 253                                 fmt_flags |= flag;
 254                                 fmt = vchar.next(&vchar);
 255                         }
 256 break_fmt_flags:
 257
 258                         /* minimum field width */
 259                         if (fmt == '*') {
 260                                 fmt = vchar.next(&vchar);
 261                                 if (arg == NULL) {
 262                                         warningf(pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 263                                         return;
 264                                 }
 265                                 const type_t *const arg_type = arg->expression->base.type;
 266                                 if (arg_type != type_int) {
 267                                         warningf(pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 268                                 }
 269                                 arg = arg->next;
 270                         } else {
 271                                 while (vchar.is_digit(fmt)) {
 272                                         fmt = vchar.next(&vchar);
 273                                 }
 274                         }
 275                 }
 276
 277                 /* precision */
 278                 if (fmt == '.') {
 279                         fmt = vchar.next(&vchar);
 280                         if (fmt == '*') {
 281                                 fmt = vchar.next(&vchar);
 282                                 if (arg == NULL) {
 283                                         warningf(pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 284                                         return;
 285                                 }
 286                                 const type_t *const arg_type = arg->expression->base.type;
 287                                 if (arg_type != type_int) {
 288                                         warningf(pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 289                                 }
 290                                 arg = arg->next;
 291                         } else {
 292                                 /* digit string may be omitted */
 293                                 while (vchar.is_digit(fmt)) {
 294                                         fmt = vchar.next(&vchar);
 295                                 }
 296                         }
 297                 }
 298
 299                 /* length modifier */
 300                 format_length_modifier_t fmt_mod;
 301                 switch (fmt) {
 302                         case 'h':
 303                                 fmt = vchar.next(&vchar);
 304                                 if (fmt == 'h') {
 305                                         fmt = vchar.next(&vchar);
 306                                         fmt_mod = FMT_MOD_hh;
 307                                 } else {
 308                                         fmt_mod = FMT_MOD_h;
 309                                 }
 310                                 break;
 311
 312                         case 'l':
 313                                 fmt = vchar.next(&vchar);
 314                                 if (fmt == 'l') {
 315                                         fmt = vchar.next(&vchar);
 316                                         fmt_mod = FMT_MOD_ll;
 317                                 } else {
 318                                         fmt_mod = FMT_MOD_l;
 319                                 }
 320                                 break;
 321
 322                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 323                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 324                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 325                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 326                         case 'q': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_q;    break;
 327                         /* microsoft mode */
 328                         case 'w':
 329                                 if (c_mode & _MS) {
 330                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 331                                 } else {
 332                                         fmt_mod = FMT_MOD_NONE;
 333                                 }
 334                                 break;
 335                         case 'I':
 336                                 if (c_mode & _MS) {
 337                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 338                                         if (fmt == '3') {
 339                                                 fmt = vchar.next(&vchar);
 340                                                 if (fmt == '2') {
 341                                                         fmt = vchar.next(&vchar);
 342                                                         fmt_mod = FMT_MOD_I32;
 343                                                 } else {
 344                                                         /* rewind */
 345                                                         --vchar.position;
 346                                                 }
 347                                         } else if (fmt == '6') {
 348                                                 fmt = vchar.next(&vchar);
 349                                                 if (fmt == '4') {
 350                                                         fmt = vchar.next(&vchar);
 351                                                         fmt_mod = FMT_MOD_I64;
 352                                                 } else {
 353                                                         /* rewind */
 354                                                         --vchar.position;
 355                                                 }
 356                                         }
 357                                 } else {
 358                                         fmt_mod = FMT_MOD_NONE;
 359                                 }
 360                                 break;
 361                         default:
 362                                 fmt_mod = FMT_MOD_NONE;
 363                                 break;
 364                 }
 365
 366                 if (fmt == '\0') {
 367                         warningf(pos, "dangling %% in format string");
 368                         break;
 369                 }
 370
 371                 type_t            *expected_type;
 372                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 373                 format_flags_t     allowed_flags;
 374                 switch (fmt) {
 375                         case 'd':
 376                         case 'i':
 377                                 switch (fmt_mod) {
 378                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 379                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 380                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 381                                         case FMT_MOD_l:    expected_type = type_long;      break;
 382                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 383                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 384                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 385                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 386                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 387                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 388                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 389
 390                                         default:
 391                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 392                                                 goto next_arg;
 393                                 }
 394                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 395                                 break;
 396
 397                         case 'o':
 398                         case 'X':
 399                         case 'x':
 400                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 401                                 goto eval_fmt_mod_unsigned;
 402
 403                         case 'u':
 404                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 405 eval_fmt_mod_unsigned:
 406                                 switch (fmt_mod) {
 407                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 408                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 409                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 410                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 411                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 412                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 413                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 414                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 415                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 416                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 417                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 418
 419                                         default:
 420                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 421                                                 goto next_arg;
 422                                 }
 423                                 break;
 424
 425                         case 'A':
 426                         case 'a':
 427                         case 'E':
 428                         case 'e':
 429                         case 'F':
 430                         case 'f':
 431                         case 'G':
 432                         case 'g':
 433                                 switch (fmt_mod) {
 434                                         case FMT_MOD_l:    /* l modifier is ignored */
 435                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 436                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 437
 438                                         default:
 439                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 440                                                 goto next_arg;
 441                                 }
 442                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 443                                 break;
 444
 445                         case 'C':
 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_wchar_t;
 451                                 allowed_flags = FMT_FLAG_NONE;
 452                                 break;
 453
 454                         case 'c':
 455                                 expected_type = type_int;
 456                                 switch (fmt_mod) {
 457                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 458                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 459                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 460
 461                                         default:
 462                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 463                                                 goto next_arg;
 464                                 }
 465                                 allowed_flags = FMT_FLAG_NONE;
 466                                 break;
 467
 468                         case 'S':
 469                                 if (fmt_mod != FMT_MOD_NONE) {
 470                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 471                                         goto next_arg;
 472                                 }
 473                                 expected_type = type_wchar_t_ptr;
 474                                 expected_qual = TYPE_QUALIFIER_CONST;
 475                                 allowed_flags = FMT_FLAG_MINUS;
 476                                 break;
 477
 478                         case 's':
 479                                 switch (fmt_mod) {
 480                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 481                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 482                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 483
 484                                         default:
 485                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 486                                                 goto next_arg;
 487                                 }
 488                                 expected_qual = TYPE_QUALIFIER_CONST;
 489                                 allowed_flags = FMT_FLAG_MINUS;
 490                                 break;
 491
 492                         case 'p':
 493                                 if (fmt_mod != FMT_MOD_NONE) {
 494                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 495                                         goto next_arg;
 496                                 }
 497                                 expected_type = type_void_ptr;
 498                                 allowed_flags = FMT_FLAG_NONE;
 499                                 break;
 500
 501                         case 'n':
 502                                 switch (fmt_mod) {
 503                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 504                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 505                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 506                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 507                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 508                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 509                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 510                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 511
 512                                         default:
 513                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 514                                                 goto next_arg;
 515                                 }
 516                                 allowed_flags = FMT_FLAG_NONE;
 517                                 break;
 518
 519                         default:
 520                                 warningf(pos, "encountered unknown conversion specifier '%%%C' at position %u", (wint_t)fmt, num_fmt);
 521                                 goto next_arg;
 522                 }
 523
 524                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 525                 if (wrong_flags != 0) {
 526                         char  wrong[8];
 527                         char *p = wrong;
 528                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 529                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 530                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 531                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 532                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 533                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 534                         *p = '\0';
 535
 536                         warningf(pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 537                 }
 538
 539                 if (arg == NULL) {
 540                         warningf(pos, "too few arguments for format string");
 541                         return;
 542                 }
 543
 544                 {       /* create a scope here to prevent warning about the jump to next_arg */
 545                         type_t *const arg_type           = arg->expression->base.type;
 546                         type_t *const arg_skip           = skip_typeref(arg_type);
 547                         type_t *const expected_type_skip = skip_typeref(expected_type);
 548
 549                         if (fmt == 'p') {
 550                                 /* allow any pointer type for %p, not just void */
 551                                 if (is_type_pointer(arg_skip))
 552                                         goto next_arg;
 553                         }
 554
 555                         if (is_type_pointer(expected_type_skip)) {
 556                                 if (is_type_pointer(arg_skip)) {
 557                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 558                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 559                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 560                                                 get_unqualified_type(arg_to) == exp_to) {
 561                                                 goto next_arg;
 562                                         }
 563                                 }
 564                         } else {
 565                                 if (get_unqualified_type(arg_skip) == expected_type_skip) {
 566                                         goto next_arg;
 567                                 }
 568                         }
 569                         if (is_type_valid(arg_skip)) {
 570                                 warningf(pos,
 571                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 572                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 573                         }
 574                 }
 575 next_arg:
 576                 arg = arg->next;
 577         }
 578         if (!atend(&vchar)) {
 579                 warningf(pos, "format string contains NUL");
 580         }
 581         if (arg != NULL) {
 582                 unsigned num_args = num_fmt;
 583                 while (arg != NULL) {
 584                         ++num_args;
 585                         arg = arg->next;
 586                 }
 587                 warningf(pos, "%u argument%s but only %u format specifier%s",
 588                         num_args, num_args != 1 ? "s" : "",
 589                         num_fmt, num_fmt != 1 ? "s" : "");
 590         }
 591 }
 592
 593 /**
 594  * Check scanf-style format.
 595  */
 596 static void check_scanf_format(const call_argument_t *arg, const format_spec_t *spec)
 597 {
 598         /* find format arg */
 599         unsigned idx = 0;
 600         for (; idx < spec->fmt_idx; ++idx)
 601                 arg = arg->next;
 602
 603         const expression_t *fmt_expr = arg->expression;
 604         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 605                 fmt_expr = fmt_expr->unary.value;
 606         }
 607
 608         vchar_t vchar;
 609         if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
 610                 vchar.string   = &fmt_expr->wide_string.value;
 611                 vchar.size     = fmt_expr->wide_string.value.size;
 612                 vchar.first    = wstring_first;
 613                 vchar.next     = wstring_next;
 614                 vchar.is_digit = wstring_isdigit;
 615         } else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
 616                 vchar.string   = &fmt_expr->string.value;
 617                 vchar.size     = fmt_expr->string.value.size;
 618                 vchar.first    = string_first;
 619                 vchar.next     = string_next;
 620                 vchar.is_digit = string_isdigit;
 621         } else {
 622                 return;
 623         }
 624         /* find the real args */
 625         for (; idx < spec->arg_idx; ++idx)
 626                 arg = arg->next;
 627
 628         const source_position_t *pos = &fmt_expr->base.source_position;
 629         unsigned fmt     = vchar.first(&vchar);
 630         unsigned num_fmt = 0;
 631         for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
 632                 if (fmt != '%')
 633                         continue;
 634                 fmt = vchar.next(&vchar);
 635
 636                 if (fmt == '%')
 637                         continue;
 638
 639                 ++num_fmt;
 640
 641                 /* length modifier */
 642                 format_length_modifier_t fmt_mod;
 643                 switch (fmt) {
 644                         case 'h':
 645                                 fmt = vchar.next(&vchar);
 646                                 if (fmt == 'h') {
 647                                         fmt = vchar.next(&vchar);
 648                                         fmt_mod = FMT_MOD_hh;
 649                                 } else {
 650                                         fmt_mod = FMT_MOD_h;
 651                                 }
 652                                 break;
 653
 654                         case 'l':
 655                                 fmt = vchar.next(&vchar);
 656                                 if (fmt == 'l') {
 657                                         fmt = vchar.next(&vchar);
 658                                         fmt_mod = FMT_MOD_ll;
 659                                 } else {
 660                                         fmt_mod = FMT_MOD_l;
 661                                 }
 662                                 break;
 663
 664                         case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
 665                         case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
 666                         case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
 667                         case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
 668                         /* microsoft mode */
 669                         case 'w':
 670                                 if (c_mode & _MS) {
 671                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
 672                                 } else {
 673                                         fmt_mod = FMT_MOD_NONE;
 674                                 }
 675                                 break;
 676                         case 'I':
 677                                 if (c_mode & _MS) {
 678                                         fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
 679                                         if (fmt == '3') {
 680                                                 fmt = vchar.next(&vchar);
 681                                                 if (fmt == '2') {
 682                                                         fmt = vchar.next(&vchar);
 683                                                         fmt_mod = FMT_MOD_I32;
 684                                                 } else {
 685                                                         /* rewind */
 686                                                         --vchar.position;
 687                                                 }
 688                                         } else if (fmt == '6') {
 689                                                 fmt = vchar.next(&vchar);
 690                                                 if (fmt == '4') {
 691                                                         fmt = vchar.next(&vchar);
 692                                                         fmt_mod = FMT_MOD_I64;
 693                                                 } else {
 694                                                         /* rewind */
 695                                                         --vchar.position;
 696                                                 }
 697                                         }
 698                                 } else {
 699                                         fmt_mod = FMT_MOD_NONE;
 700                                 }
 701                                 break;
 702                         default:
 703                                 fmt_mod = FMT_MOD_NONE;
 704                                 break;
 705                 }
 706
 707                 if (fmt == '\0') {
 708                         warningf(pos, "dangling %% in format string");
 709                         break;
 710                 }
 711
 712                 type_t            *expected_type;
 713                 switch (fmt) {
 714                         case 'd':
 715                         case 'i':
 716                                 switch (fmt_mod) {
 717                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 718                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 719                                         case FMT_MOD_h:    expected_type = type_short;       break;
 720                                         case FMT_MOD_l:    expected_type = type_long;        break;
 721                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 722                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 723                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 724                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 725                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 726                                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 727                                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 728
 729                                         default:
 730                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 731                                                 goto next_arg;
 732                                 }
 733                                 break;
 734
 735                         case 'o':
 736                         case 'X':
 737                         case 'x':
 738                                 goto eval_fmt_mod_unsigned;
 739
 740                         case 'u':
 741 eval_fmt_mod_unsigned:
 742                                 switch (fmt_mod) {
 743                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 744                                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 745                                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 746                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 747                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 748                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 749                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 750                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 751                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 752                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 753                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 754
 755                                         default:
 756                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 757                                                 goto next_arg;
 758                                 }
 759                                 break;
 760
 761                         case 'A':
 762                         case 'a':
 763                         case 'E':
 764                         case 'e':
 765                         case 'F':
 766                         case 'f':
 767                         case 'G':
 768                         case 'g':
 769                                 switch (fmt_mod) {
 770                                         case FMT_MOD_l:    /* l modifier is ignored */
 771                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 772                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 773
 774                                         default:
 775                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 776                                                 goto next_arg;
 777                                 }
 778                                 break;
 779
 780                         case 'C':
 781                                 if (fmt_mod != FMT_MOD_NONE) {
 782                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 783                                         goto next_arg;
 784                                 }
 785                                 expected_type = type_wchar_t;
 786                                 break;
 787
 788                         case 'c':
 789                                 expected_type = type_int;
 790                                 switch (fmt_mod) {
 791                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 792                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 793                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 794
 795                                         default:
 796                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 797                                                 goto next_arg;
 798                                 }
 799                                 break;
 800
 801                         case 'S':
 802                                 if (fmt_mod != FMT_MOD_NONE) {
 803                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 804                                         goto next_arg;
 805                                 }
 806                                 expected_type = type_wchar_t;
 807                                 break;
 808
 809                         case 's':
 810                         case '[':
 811                                 switch (fmt_mod) {
 812                                         case FMT_MOD_NONE: expected_type = type_char;    break;
 813                                         case FMT_MOD_l:    expected_type = type_wchar_t; break;
 814                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 815
 816                                         default:
 817                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 818                                                 goto next_arg;
 819                                 }
 820                                 break;
 821
 822                         case 'p':
 823                                 if (fmt_mod != FMT_MOD_NONE) {
 824                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 825                                         goto next_arg;
 826                                 }
 827                                 expected_type = type_void_ptr;
 828                                 break;
 829
 830                         case 'n':
 831                                 switch (fmt_mod) {
 832                                         case FMT_MOD_NONE: expected_type = type_int;         break;
 833                                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 834                                         case FMT_MOD_h:    expected_type = type_short;       break;
 835                                         case FMT_MOD_l:    expected_type = type_long;        break;
 836                                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 837                                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 838                                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 839                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 840
 841                                         default:
 842                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 843                                                 goto next_arg;
 844                                 }
 845                                 break;
 846
 847                         default:
 848                                 warningf(pos, "encountered unknown conversion specifier '%%%C' at position %u", (wint_t)fmt, num_fmt);
 849                                 goto next_arg;
 850                 }
 851
 852                 if (arg == NULL) {
 853                         warningf(pos, "too few arguments for format string");
 854                         return;
 855                 }
 856
 857                 {       /* create a scope here to prevent warning about the jump to next_arg */
 858                         type_t *const arg_type           = arg->expression->base.type;
 859                         type_t *const arg_skip           = skip_typeref(arg_type);
 860                         type_t *const expected_type_skip = skip_typeref(expected_type);
 861
 862                         if (! is_type_pointer(arg_skip))
 863                                 goto error_arg_type;
 864                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 865
 866                         if (fmt == 'p') {
 867                                 /* allow any pointer type for %p, not just void */
 868                                 if (is_type_pointer(ptr_skip))
 869                                         goto next_arg;
 870                         }
 871
 872                         /* do NOT allow const or restrict, all other should be ok */
 873                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 874                                 goto error_arg_type;
 875                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 876                         if (unqual_ptr == expected_type_skip) {
 877                                 goto next_arg;
 878                         } else if (expected_type_skip == type_char) {
 879                                 /* char matches with unsigned char AND signed char */
 880                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 881                                         goto next_arg;
 882                         }
 883 error_arg_type:
 884                         if (is_type_valid(arg_skip)) {
 885                                 warningf(pos,
 886                                         "argument type '%T' does not match conversion specifier '%%%s%c' at position %u",
 887                                         arg_type, get_length_modifier_name(fmt_mod), (char)fmt, num_fmt);
 888                         }
 889                 }
 890 next_arg:
 891                 arg = arg->next;
 892         }
 893         if (!atend(&vchar)) {
 894                 warningf(pos, "format string contains NUL");
 895         }
 896         if (arg != NULL) {
 897                 unsigned num_args = num_fmt;
 898                 while (arg != NULL) {
 899                         ++num_args;
 900                         arg = arg->next;
 901                 }
 902                 warningf(pos, "%u argument%s but only %u format string%s",
 903                         num_args, num_args != 1 ? "s" : "",
 904                         num_fmt, num_fmt != 1 ? "s" : "");
 905         }
 906 }
 907
 908 static const format_spec_t builtin_table[] = {
 909         { "printf",        FORMAT_PRINTF,   0, 1 },
 910         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 911         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 912         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 913         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 914         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 915         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 916         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 917         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 918         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 919
 920         { "scanf",         FORMAT_SCANF,    0, 1 },
 921         { "wscanf",        FORMAT_SCANF,    0, 1 },
 922         { "sscanf",        FORMAT_SCANF,    1, 2 },
 923         { "swscanf",       FORMAT_SCANF,    1, 2 },
 924         { "fscanf",        FORMAT_SCANF,    1, 2 },
 925         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 926
 927         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 928         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 929
 930         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 931
 932         /* MS extensions */
 933         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 934         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 935         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 936         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 937         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 938         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 939         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 940         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 941         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 942         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 943         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 944         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 945         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 946         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 947         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 948         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 949         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 950         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 951         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 952         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 953         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 954         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 955 };
 956
 957 void check_format(const call_expression_t *const call)
 958 {
 959         if (!warning.format)
 960                 return;
 961
 962         const expression_t *const func_expr = call->function;
 963         if (func_expr->kind != EXPR_REFERENCE)
 964                 return;
 965
 966         const entity_t        *const entity = func_expr->reference.entity;
 967         const call_argument_t *      arg    = call->arguments;
 968
 969         if (false) {
 970                 /* the declaration has a GNU format attribute, check it */
 971         } else {
 972                 /*
 973                  * For some functions we always check the format, even if it was not specified.
 974                  * This allows to check format even in MS mode or without header included.
 975                  */
 976                 const char *const name = entity->base.symbol->string;
 977                 for (size_t i = 0; i < sizeof(builtin_table) / sizeof(builtin_table[0]); ++i) {
 978                         if (strcmp(name, builtin_table[i].name) == 0) {
 979                                 switch (builtin_table[i].fmt_kind) {
 980                                 case FORMAT_PRINTF:
 981                                         check_printf_format(arg, &builtin_table[i]);
 982                                         break;
 983                                 case FORMAT_SCANF:
 984                                         check_scanf_format(arg, &builtin_table[i]);
 985                                         break;
 986                                 case FORMAT_STRFTIME:
 987                                 case FORMAT_STRFMON:
 988                                         /* TODO: implement other cases */
 989                                         break;
 990                                 }
 991                                 break;
 992                         }
 993                 }
 994         }
 995 }