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                         ++num_args;
 232                         fmt = *(++c);
 233                         if (fmt == '*') {
 234                                 fmt = *(++c);
 235                                 if (arg == NULL) {
 236                                         warningf(WARN_FORMAT, pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 237                                         return -1;
 238                                 }
 239                                 const type_t *const arg_type = arg->expression->base.type;
 240                                 if (arg_type != type_int) {
 241                                         warningf(WARN_FORMAT, pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 242                                 }
 243                                 arg = arg->next;
 244                         } else {
 245                                 /* digit string may be omitted */
 246                                 while (isdigit(fmt)) {
 247                                         fmt = *(++c);
 248                                 }
 249                         }
 250                 }
 251
 252                 /* length modifier */
 253                 format_length_modifier_t fmt_mod;
 254                 switch (fmt) {
 255                         case 'h':
 256                                 fmt = *(++c);
 257                                 if (fmt == 'h') {
 258                                         fmt = *(++c);
 259                                         fmt_mod = FMT_MOD_hh;
 260                                 } else {
 261                                         fmt_mod = FMT_MOD_h;
 262                                 }
 263                                 break;
 264
 265                         case 'l':
 266                                 fmt = *(++c);
 267                                 if (fmt == 'l') {
 268                                         fmt = *(++c);
 269                                         fmt_mod = FMT_MOD_ll;
 270                                 } else {
 271                                         fmt_mod = FMT_MOD_l;
 272                                 }
 273                                 break;
 274
 275                         case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
 276                         case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
 277                         case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
 278                         case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
 279                         case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 280                         /* microsoft mode */
 281                         case 'w':
 282                                 if (c_mode & _MS) {
 283                                         fmt = *(++c); fmt_mod = FMT_MOD_w;
 284                                 } else {
 285                                         fmt_mod = FMT_MOD_NONE;
 286                                 }
 287                                 break;
 288                         case 'I':
 289                                 if (c_mode & _MS) {
 290                                         fmt = *(++c); fmt_mod = FMT_MOD_I;
 291                                         if (fmt == '3') {
 292                                                 fmt = *(++c);
 293                                                 if (fmt == '2') {
 294                                                         fmt = *(++c);
 295                                                         fmt_mod = FMT_MOD_I32;
 296                                                 } else {
 297                                                         /* rewind */
 298                                                         fmt = *(--c);
 299                                                 }
 300                                         } else if (fmt == '6') {
 301                                                 fmt = *(++c);
 302                                                 if (fmt == '4') {
 303                                                         fmt = *(++c);
 304                                                         fmt_mod = FMT_MOD_I64;
 305                                                 } else {
 306                                                         /* rewind */
 307                                                         fmt = *(--c);
 308                                                 }
 309                                         }
 310                                 } else {
 311                                         fmt_mod = FMT_MOD_NONE;
 312                                 }
 313                                 break;
 314                         default:
 315                                 fmt_mod = FMT_MOD_NONE;
 316                                 break;
 317                 }
 318
 319
 320                 type_t            *expected_type;
 321                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 322                 format_flags_t     allowed_flags;
 323                 switch (fmt) {
 324                         case 'd':
 325                         case 'i':
 326                                 switch (fmt_mod) {
 327                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 328                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 329                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 330                                         case FMT_MOD_l:    expected_type = type_long;      break;
 331                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 332                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 333                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 334                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 335                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 336                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 337                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 338
 339                                         default:
 340                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 341                                                 goto next_arg;
 342                                 }
 343                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 344                                 break;
 345
 346                         case 'o':
 347                         case 'X':
 348                         case 'x':
 349                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 350                                 goto eval_fmt_mod_unsigned;
 351
 352                         case 'u':
 353                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 354 eval_fmt_mod_unsigned:
 355                                 switch (fmt_mod) {
 356                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 357                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 358                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 359                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 360                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 361                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 362                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 363                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 364                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 365                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 366                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 367
 368                                         default:
 369                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 370                                                 goto next_arg;
 371                                 }
 372                                 break;
 373
 374                         case 'A':
 375                         case 'a':
 376                         case 'E':
 377                         case 'e':
 378                         case 'F':
 379                         case 'f':
 380                         case 'G':
 381                         case 'g':
 382                                 switch (fmt_mod) {
 383                                         case FMT_MOD_l:    /* l modifier is ignored */
 384                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 385                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 386
 387                                         default:
 388                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 389                                                 goto next_arg;
 390                                 }
 391                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 392                                 break;
 393
 394                         case 'C':
 395                                 if (fmt_mod != FMT_MOD_NONE) {
 396                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 397                                         goto next_arg;
 398                                 }
 399                                 expected_type = type_wchar_t;
 400                                 allowed_flags = FMT_FLAG_NONE;
 401                                 break;
 402
 403                         case 'c':
 404                                 expected_type = type_int;
 405                                 switch (fmt_mod) {
 406                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 407                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 408                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 409
 410                                         default:
 411                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 412                                                 goto next_arg;
 413                                 }
 414                                 allowed_flags = FMT_FLAG_NONE;
 415                                 break;
 416
 417                         case 'S':
 418                                 if (fmt_mod != FMT_MOD_NONE) {
 419                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 420                                         goto next_arg;
 421                                 }
 422                                 expected_type = type_wchar_t_ptr;
 423                                 expected_qual = TYPE_QUALIFIER_CONST;
 424                                 allowed_flags = FMT_FLAG_MINUS;
 425                                 break;
 426
 427                         case 's':
 428                                 switch (fmt_mod) {
 429                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 430                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 431                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 432
 433                                         default:
 434                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 435                                                 goto next_arg;
 436                                 }
 437                                 expected_qual = TYPE_QUALIFIER_CONST;
 438                                 allowed_flags = FMT_FLAG_MINUS;
 439                                 break;
 440
 441                         case 'p':
 442                                 if (fmt_mod != FMT_MOD_NONE) {
 443                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 444                                         goto next_arg;
 445                                 }
 446                                 expected_type = type_void_ptr;
 447                                 allowed_flags = FMT_FLAG_NONE;
 448                                 break;
 449
 450                         case 'n':
 451                                 switch (fmt_mod) {
 452                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 453                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 454                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 455                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 456                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 457                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 458                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 459                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   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                         default:
 469                                 warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at position %u", fmt, num_fmt);
 470                                 if (arg == NULL) {
 471                                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 472                                         return -1;
 473                                 }
 474                                 goto next_arg;
 475                 }
 476
 477                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 478                 if (wrong_flags != 0) {
 479                         char  wrong[8];
 480                         char *p = wrong;
 481                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 482                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 483                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 484                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 485                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 486                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 487                         *p = '\0';
 488
 489                         warningf(WARN_FORMAT, pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 490                 }
 491
 492                 if (arg == NULL) {
 493                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 494                         return -1;
 495                 }
 496
 497                 { /* create a scope here to prevent warning about the jump to next_arg */
 498                         type_t *const arg_type           = arg->expression->base.type;
 499                         type_t *const arg_skip           = skip_typeref(arg_type);
 500                         type_t *const expected_type_skip = skip_typeref(expected_type);
 501
 502                         if (fmt == 'p') {
 503                                 /* allow any pointer type for %p, not just void */
 504                                 if (is_type_pointer(arg_skip))
 505                                         goto next_arg;
 506                         }
 507
 508                         if (is_type_pointer(expected_type_skip)) {
 509                                 if (is_type_pointer(arg_skip)) {
 510                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 511                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 512                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 513                                                 get_unqualified_type(arg_to) == exp_to) {
 514                                                 goto next_arg;
 515                                         }
 516                                 }
 517                         } else if (get_unqualified_type(arg_skip) == expected_type_skip) {
 518                                 goto next_arg;
 519                         }
 520                         if (is_type_valid(arg_skip)) {
 521                                 source_position_t const *const apos = &arg->expression->base.source_position;
 522                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 523                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 524                         }
 525                 }
 526 next_arg:
 527                 arg = arg->next;
 528         }
 529         assert(fmt == '\0');
 530         if (c+1 < string + size) {
 531                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 532         }
 533         return num_args;
 534 }
 535
 536 /**
 537  * Check printf-style format.
 538  */
 539 static void check_printf_format(call_argument_t const *arg,
 540                                 format_spec_t const *const spec)
 541 {
 542         /* find format arg */
 543         size_t idx = 0;
 544         for (; idx < spec->fmt_idx; ++idx) {
 545                 if (arg == NULL)
 546                         return;
 547                 arg = arg->next;
 548         }
 549
 550         expression_t const *const fmt_expr = arg->expression;
 551
 552         /* find the real args */
 553         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 554                 arg = arg->next;
 555
 556         int const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
 557         if (num_fmt < 0)
 558                 return;
 559
 560         size_t num_args = 0;
 561         for (; arg != NULL; arg = arg->next)
 562                 ++num_args;
 563         if (num_args > (size_t)num_fmt) {
 564                 source_position_t const *const pos = &fmt_expr->base.source_position;
 565                 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" : "");
 566         }
 567 }
 568
 569 /**
 570  * Check scanf-style format.
 571  */
 572 static void check_scanf_format(const call_argument_t *arg,
 573                                const format_spec_t *spec)
 574 {
 575         /* find format arg */
 576         unsigned idx = 0;
 577         for (; idx < spec->fmt_idx; ++idx) {
 578                 if (arg == NULL)
 579                         return;
 580                 arg = arg->next;
 581         }
 582
 583         const expression_t *fmt_expr = arg->expression;
 584         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 585                 fmt_expr = fmt_expr->unary.value;
 586         }
 587
 588         if (fmt_expr->kind != EXPR_STRING_LITERAL
 589                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 590                 return;
 591
 592         const char *string = fmt_expr->literal.value.begin;
 593         size_t      size   = fmt_expr->literal.value.size;
 594         const char *c      = string;
 595
 596         /* find the real args */
 597         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 598                 arg = arg->next;
 599
 600         const source_position_t *pos = &fmt_expr->base.source_position;
 601         unsigned num_fmt = 0;
 602         char     fmt;
 603         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 604                 if (fmt != '%')
 605                         continue;
 606                 fmt = *(++c);
 607                 if (fmt == '\0') {
 608                         warningf(WARN_FORMAT, pos, "dangling '%%' in format string");
 609                         break;
 610                 }
 611                 if (fmt == '%')
 612                         continue;
 613
 614                 ++num_fmt;
 615
 616                 /* look for length modifiers */
 617                 format_length_modifier_t fmt_mod = FMT_MOD_NONE;
 618                 switch (fmt) {
 619                 case 'h':
 620                         fmt = *(++c);
 621                         if (fmt == 'h') {
 622                                 fmt = *(++c);
 623                                 fmt_mod = FMT_MOD_hh;
 624                         } else {
 625                                 fmt_mod = FMT_MOD_h;
 626                         }
 627                         break;
 628
 629                 case 'l':
 630                         fmt = *(++c);
 631                         if (fmt == 'l') {
 632                                 fmt = *(++c);
 633                                 fmt_mod = FMT_MOD_ll;
 634                         } else {
 635                                 fmt_mod = FMT_MOD_l;
 636                         }
 637                         break;
 638
 639                 case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
 640                 case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
 641                 case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
 642                 case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
 643                 /* microsoft mode */
 644                 case 'w':
 645                         if (c_mode & _MS) {
 646                                 fmt = *(++c);
 647                                 fmt_mod = FMT_MOD_w;
 648                         }
 649                         break;
 650                 case 'I':
 651                         if (c_mode & _MS) {
 652                                 fmt = *(++c);
 653                                 fmt_mod = FMT_MOD_I;
 654                                 if (fmt == '3') {
 655                                         fmt = *(++c);
 656                                         if (fmt == '2') {
 657                                                 fmt = *(++c);
 658                                                 fmt_mod = FMT_MOD_I32;
 659                                         } else {
 660                                                 /* rewind */
 661                                                 fmt = *(--c);
 662                                         }
 663                                 } else if (fmt == '6') {
 664                                         fmt = *(++c);
 665                                         if (fmt == '4') {
 666                                                 fmt = *(++c);
 667                                                 fmt_mod = FMT_MOD_I64;
 668                                         } else {
 669                                                 /* rewind */
 670                                                 fmt = *(--c);
 671                                         }
 672                                 }
 673                         }
 674                         break;
 675                 }
 676
 677                 if (fmt == '\0') {
 678                         warningf(WARN_FORMAT, pos, "dangling %% with conversion specififer in format string");
 679                         break;
 680                 }
 681
 682                 type_t *expected_type;
 683                 switch (fmt) {
 684                 case 'd':
 685                 case 'i':
 686                         switch (fmt_mod) {
 687                         case FMT_MOD_NONE: expected_type = type_int;         break;
 688                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 689                         case FMT_MOD_h:    expected_type = type_short;       break;
 690                         case FMT_MOD_l:    expected_type = type_long;        break;
 691                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 692                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 693                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 694                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 695                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 696                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 697                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 698
 699                         default:
 700                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 701                                 goto next_arg;
 702                         }
 703                         break;
 704
 705                 case 'o':
 706                 case 'X':
 707                 case 'x':
 708                 case 'u':
 709                         switch (fmt_mod) {
 710                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 711                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 712                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 713                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 714                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 715                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 716                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 717                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 718                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 719                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 720                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 721
 722                         default:
 723                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 724                                 goto next_arg;
 725                         }
 726                         break;
 727
 728                 case 'A':
 729                 case 'a':
 730                 case 'E':
 731                 case 'e':
 732                 case 'F':
 733                 case 'f':
 734                 case 'G':
 735                 case 'g':
 736                         switch (fmt_mod) {
 737                         case FMT_MOD_l:    expected_type = type_double;      break;
 738                         case FMT_MOD_NONE: expected_type = type_float;       break;
 739                         case FMT_MOD_L:    expected_type = type_long_double; break;
 740
 741                         default:
 742                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 743                                 goto next_arg;
 744                         }
 745                         break;
 746
 747                 case 'C':
 748                         if (fmt_mod != FMT_MOD_NONE) {
 749                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 750                                 goto next_arg;
 751                         }
 752                         expected_type = type_wchar_t;
 753                         break;
 754
 755                 case 'c':
 756                         expected_type = type_int;
 757                         switch (fmt_mod) {
 758                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 759                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 760                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 761
 762                         default:
 763                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 764                                 goto next_arg;
 765                         }
 766                         break;
 767
 768                 case 'S':
 769                         if (fmt_mod != FMT_MOD_NONE) {
 770                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 771                                 goto next_arg;
 772                         }
 773                         expected_type = type_wchar_t;
 774                         break;
 775
 776                 case 's':
 777                 case '[':
 778                         switch (fmt_mod) {
 779                                 case FMT_MOD_NONE: expected_type = type_char;    break;
 780                                 case FMT_MOD_l:    expected_type = type_wchar_t; break;
 781                                 case FMT_MOD_w:    expected_type = type_wchar_t; break;
 782
 783                                 default:
 784                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 785                                         goto next_arg;
 786                         }
 787                         break;
 788
 789                 case 'p':
 790                         if (fmt_mod != FMT_MOD_NONE) {
 791                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 792                                 goto next_arg;
 793                         }
 794                         expected_type = type_void_ptr;
 795                         break;
 796
 797                 case 'n':
 798                         switch (fmt_mod) {
 799                         case FMT_MOD_NONE: expected_type = type_int;         break;
 800                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 801                         case FMT_MOD_h:    expected_type = type_short;       break;
 802                         case FMT_MOD_l:    expected_type = type_long;        break;
 803                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 804                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 805                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 806                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 807
 808                         default:
 809                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 810                                 goto next_arg;
 811                         }
 812                         break;
 813
 814                 default:
 815                         warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at format %u", fmt, num_fmt);
 816                         if (arg == NULL) {
 817                                 warningf(WARN_FORMAT, pos, "too few arguments for format string");
 818                                 return;
 819                         }
 820                         goto next_arg;
 821                 }
 822
 823                 if (arg == NULL) {
 824                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 825                         return;
 826                 }
 827
 828                 { /* create a scope here to prevent warning about the jump to next_arg */
 829                         type_t *const arg_type           = arg->expression->base.type;
 830                         type_t *const arg_skip           = skip_typeref(arg_type);
 831                         type_t *const expected_type_skip = skip_typeref(expected_type);
 832
 833                         if (! is_type_pointer(arg_skip))
 834                                 goto error_arg_type;
 835                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 836
 837                         if (fmt == 'p') {
 838                                 /* allow any pointer type for %p, not just void */
 839                                 if (is_type_pointer(ptr_skip))
 840                                         goto next_arg;
 841                         }
 842
 843                         /* do NOT allow const or restrict, all other should be ok */
 844                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 845                                 goto error_arg_type;
 846                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 847                         if (unqual_ptr == expected_type_skip) {
 848                                 goto next_arg;
 849                         } else if (expected_type_skip == type_char) {
 850                                 /* char matches with unsigned char AND signed char */
 851                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 852                                         goto next_arg;
 853                         }
 854 error_arg_type:
 855                         if (is_type_valid(arg_skip)) {
 856                                 source_position_t const *const apos = &arg->expression->base.source_position;
 857                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 858                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 859                         }
 860                 }
 861 next_arg:
 862                 arg = arg->next;
 863         }
 864         assert(fmt == '\0');
 865         if (c+1 < string + size) {
 866                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 867         }
 868         if (arg != NULL) {
 869                 unsigned num_args = num_fmt;
 870                 while (arg != NULL) {
 871                         ++num_args;
 872                         arg = arg->next;
 873                 }
 874                 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" : "");
 875         }
 876 }
 877
 878 static const format_spec_t builtin_table[] = {
 879         { "printf",        FORMAT_PRINTF,   0, 1 },
 880         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 881         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 882         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 883         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 884         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 885         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 886         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 887         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 888         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 889
 890         { "scanf",         FORMAT_SCANF,    0, 1 },
 891         { "wscanf",        FORMAT_SCANF,    0, 1 },
 892         { "sscanf",        FORMAT_SCANF,    1, 2 },
 893         { "swscanf",       FORMAT_SCANF,    1, 2 },
 894         { "fscanf",        FORMAT_SCANF,    1, 2 },
 895         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 896
 897         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 898         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 899
 900         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 901
 902         /* MS extensions */
 903         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 904         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 905         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 906         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 907         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 908         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 909         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 910         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 911         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 912         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 913         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 914         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 915         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 916         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 917         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 918         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 919         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 920         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 921         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 922         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 923         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 924         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 925 };
 926
 927 void check_format(const call_expression_t *const call)
 928 {
 929         if (!is_warn_on(WARN_FORMAT))
 930                 return;
 931
 932         const expression_t *const func_expr = call->function;
 933         if (func_expr->kind != EXPR_REFERENCE)
 934                 return;
 935
 936         const entity_t        *const entity = func_expr->reference.entity;
 937         const call_argument_t *      arg    = call->arguments;
 938
 939         /*
 940          * For some functions we always check the format, even if it was not
 941          * specified. This allows to check format even in MS mode or without
 942          * header included.
 943          */
 944         const char *const name = entity->base.symbol->string;
 945         for (size_t i = 0; i < lengthof(builtin_table); ++i) {
 946                 if (strcmp(name, builtin_table[i].name) == 0) {
 947                         switch (builtin_table[i].fmt_kind) {
 948                         case FORMAT_PRINTF:
 949                                 check_printf_format(arg, &builtin_table[i]);
 950                                 break;
 951                         case FORMAT_SCANF:
 952                                 check_scanf_format(arg, &builtin_table[i]);
 953                                 break;
 954                         case FORMAT_STRFTIME:
 955                         case FORMAT_STRFMON:
 956                                 /* TODO: implement other cases */
 957                                 break;
 958                         }
 959                         break;
 960                 }
 961         }
 962 }