nsz Git - cparser/blob - format_check.c

   1 /*
   2  * This file is part of cparser.
   3  * Copyright (C) 2007-2009 Matthias Braun <matze@braunis.de>
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License
   7  * as published by the Free Software Foundation; either version 2
   8  * of the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  18  * 02111-1307, USA.
  19  */
  20 #include <config.h>
  21
  22 #include <ctype.h>
  23
  24 #include "adt/util.h"
  25 #include "format_check.h"
  26 #include "symbol_t.h"
  27 #include "ast_t.h"
  28 #include "entity_t.h"
  29 #include "diagnostic.h"
  30 #include "types.h"
  31 #include "type_t.h"
  32 #include "warning.h"
  33 #include "lang_features.h"
  34
  35 typedef enum format_flag_t {
  36         FMT_FLAG_NONE  = 0,
  37         FMT_FLAG_HASH  = 1U << 0,
  38         FMT_FLAG_ZERO  = 1U << 1,
  39         FMT_FLAG_MINUS = 1U << 2,
  40         FMT_FLAG_SPACE = 1U << 3,
  41         FMT_FLAG_PLUS  = 1U << 4,
  42         FMT_FLAG_TICK  = 1U << 5
  43 } format_flag_t;
  44
  45 typedef unsigned format_flags_t;
  46
  47 typedef enum format_length_modifier_t {
  48         FMT_MOD_NONE,
  49         FMT_MOD_L,
  50         FMT_MOD_hh,
  51         FMT_MOD_h,
  52         FMT_MOD_l,
  53         FMT_MOD_ll,
  54         FMT_MOD_j,
  55         FMT_MOD_t,
  56         FMT_MOD_z,
  57         FMT_MOD_q,
  58         /* only in microsoft mode */
  59         FMT_MOD_w,
  60         FMT_MOD_I,
  61         FMT_MOD_I32,
  62         FMT_MOD_I64
  63 } format_length_modifier_t;
  64
  65 typedef struct format_spec_t {
  66         const char    *name;     /**< name of the function */
  67         format_kind_t  fmt_kind; /**< kind */
  68         unsigned       fmt_idx;  /**< index of the format string */
  69         unsigned       arg_idx;  /**< index of the first argument */
  70 } format_spec_t;
  71
  72 static const char* get_length_modifier_name(const format_length_modifier_t mod)
  73 {
  74         static const char* const names[] = {
  75                 [FMT_MOD_NONE] = "",
  76                 [FMT_MOD_L]    = "L",
  77                 [FMT_MOD_hh]   = "hh",
  78                 [FMT_MOD_h]    = "h",
  79                 [FMT_MOD_l]    = "l",
  80                 [FMT_MOD_ll]   = "ll",
  81                 [FMT_MOD_j]    = "j",
  82                 [FMT_MOD_t]    = "t",
  83                 [FMT_MOD_z]    = "z",
  84                 [FMT_MOD_q]    = "q",
  85                 /* only in microsoft mode */
  86                 [FMT_MOD_w]    = "w",
  87                 [FMT_MOD_I]    = "I",
  88                 [FMT_MOD_I32]  = "I32",
  89                 [FMT_MOD_I64]  = "I64"
  90         };
  91         assert((size_t)mod < lengthof(names));
  92         return names[mod];
  93 }
  94
  95 static void warn_invalid_length_modifier(const source_position_t *pos,
  96                                          const format_length_modifier_t mod,
  97                                          const utf32 conversion)
  98 {
  99         char const *const lmod = get_length_modifier_name(mod);
 100         warningf(WARN_FORMAT, pos, "invalid length modifier '%s' for conversion specifier '%%%c'", lmod, conversion);
 101 }
 102
 103 /**
 104  * Check printf-style format. Returns number of expected arguments.
 105  */
 106 static int internal_check_printf_format(const expression_t *fmt_expr,
 107                                         const call_argument_t *arg,
 108                                         const format_spec_t *spec)
 109 {
 110         while (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 111                 fmt_expr = fmt_expr->unary.value;
 112         }
 113
 114         /*
 115          * gettext results in expressions like (X ? "format_string" : Y)
 116          * we assume the left part is the format string
 117          */
 118         if (fmt_expr->kind == EXPR_CONDITIONAL) {
 119                 conditional_expression_t const *const c = &fmt_expr->conditional;
 120                 expression_t             const *      t = c->true_expression;
 121                 if (t == NULL)
 122                         t = c->condition;
 123                 int const nt = internal_check_printf_format(t,                   arg, spec);
 124                 int const nf = internal_check_printf_format(c->false_expression, arg, spec);
 125                 return nt > nf ? nt : nf;
 126         }
 127
 128         if (fmt_expr->kind != EXPR_STRING_LITERAL
 129                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 130                 return -1;
 131
 132         const char *string = fmt_expr->literal.value.begin;
 133         size_t      size   = fmt_expr->literal.value.size;
 134         const char *c      = string;
 135
 136         const source_position_t *pos = &fmt_expr->base.source_position;
 137         unsigned num_fmt  = 0;
 138         unsigned num_args = 0;
 139         char     fmt;
 140         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 141                 if (fmt != '%')
 142                         continue;
 143                 fmt = *(++c);
 144
 145                 if (fmt == '\0') {
 146                         warningf(WARN_FORMAT, pos, "dangling %% in format string");
 147                         break;
 148                 }
 149                 if (fmt == '%')
 150                         continue;
 151
 152                 ++num_fmt;
 153                 ++num_args;
 154
 155                 format_flags_t fmt_flags = FMT_FLAG_NONE;
 156                 if (fmt == '0') {
 157                         fmt = *(++c);
 158                         fmt_flags |= FMT_FLAG_ZERO;
 159                 }
 160
 161                 /* argument selector or minimum field width */
 162                 if (isdigit(fmt)) {
 163                         do {
 164                                 fmt = *(++c);
 165                         } while (isdigit(fmt));
 166
 167                         /* digit string was ... */
 168                         if (fmt == '$') {
 169                                 /* ... argument selector */
 170                                 fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 171                                 /* TODO implement */
 172                                 return -1;
 173                         }
 174                         /* ... minimum field width */
 175                 } else {
 176                         /* flags */
 177                         for (;;) {
 178                                 format_flags_t flag;
 179                                 switch (fmt) {
 180                                         case '#':  flag = FMT_FLAG_HASH;  break;
 181                                         case '0':  flag = FMT_FLAG_ZERO;  break;
 182                                         case '-':  flag = FMT_FLAG_MINUS; break;
 183                                         case '\'': flag = FMT_FLAG_TICK;  break;
 184
 185                                         case ' ':
 186                                                 if (fmt_flags & FMT_FLAG_PLUS) {
 187                                                         warningf(WARN_FORMAT, pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 188                                                 }
 189                                                 flag = FMT_FLAG_SPACE;
 190                                                 break;
 191
 192                                         case '+':
 193                                                 if (fmt_flags & FMT_FLAG_SPACE) {
 194                                                         warningf(WARN_FORMAT, pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 195                                                 }
 196                                                 flag = FMT_FLAG_PLUS;
 197                                                 break;
 198
 199                                         default: goto break_fmt_flags;
 200                                 }
 201                                 if (fmt_flags & flag) {
 202                                         warningf(WARN_FORMAT, pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 203                                 }
 204                                 fmt_flags |= flag;
 205                                 fmt = *(++c);
 206                         }
 207 break_fmt_flags:
 208
 209                         /* minimum field width */
 210                         if (fmt == '*') {
 211                                 ++num_args;
 212                                 fmt = *(++c);
 213                                 if (arg == NULL) {
 214                                         warningf(WARN_FORMAT, pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
 215                                         return -1;
 216                                 }
 217                                 const type_t *const arg_type = arg->expression->base.type;
 218                                 if (arg_type != type_int) {
 219                                         warningf(WARN_FORMAT, pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 220                                 }
 221                                 arg = arg->next;
 222                         } else {
 223                                 while (isdigit(fmt)) {
 224                                         fmt = *(++c);
 225                                 }
 226                         }
 227                 }
 228
 229                 /* precision */
 230                 if (fmt == '.') {
 231                         if (fmt_flags & FMT_FLAG_ZERO) {
 232                                 warningf(WARN_FORMAT, pos, "'0' flag ignored with precision in conversion specification %u", num_fmt);
 233                         }
 234
 235                         ++num_args;
 236                         fmt = *(++c);
 237                         if (fmt == '*') {
 238                                 fmt = *(++c);
 239                                 if (arg == NULL) {
 240                                         warningf(WARN_FORMAT, pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
 241                                         return -1;
 242                                 }
 243                                 const type_t *const arg_type = arg->expression->base.type;
 244                                 if (arg_type != type_int) {
 245                                         warningf(WARN_FORMAT, pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 246                                 }
 247                                 arg = arg->next;
 248                         } else {
 249                                 /* digit string may be omitted */
 250                                 while (isdigit(fmt)) {
 251                                         fmt = *(++c);
 252                                 }
 253                         }
 254                 }
 255
 256                 /* length modifier */
 257                 format_length_modifier_t fmt_mod;
 258                 switch (fmt) {
 259                         case 'h':
 260                                 fmt = *(++c);
 261                                 if (fmt == 'h') {
 262                                         fmt = *(++c);
 263                                         fmt_mod = FMT_MOD_hh;
 264                                 } else {
 265                                         fmt_mod = FMT_MOD_h;
 266                                 }
 267                                 break;
 268
 269                         case 'l':
 270                                 fmt = *(++c);
 271                                 if (fmt == 'l') {
 272                                         fmt = *(++c);
 273                                         fmt_mod = FMT_MOD_ll;
 274                                 } else {
 275                                         fmt_mod = FMT_MOD_l;
 276                                 }
 277                                 break;
 278
 279                         case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
 280                         case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
 281                         case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
 282                         case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
 283                         case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 284                         /* microsoft mode */
 285                         case 'w':
 286                                 if (c_mode & _MS) {
 287                                         fmt = *(++c); fmt_mod = FMT_MOD_w;
 288                                 } else {
 289                                         fmt_mod = FMT_MOD_NONE;
 290                                 }
 291                                 break;
 292                         case 'I':
 293                                 if (c_mode & _MS) {
 294                                         fmt = *(++c); fmt_mod = FMT_MOD_I;
 295                                         if (fmt == '3') {
 296                                                 fmt = *(++c);
 297                                                 if (fmt == '2') {
 298                                                         fmt = *(++c);
 299                                                         fmt_mod = FMT_MOD_I32;
 300                                                 } else {
 301                                                         /* rewind */
 302                                                         fmt = *(--c);
 303                                                 }
 304                                         } else if (fmt == '6') {
 305                                                 fmt = *(++c);
 306                                                 if (fmt == '4') {
 307                                                         fmt = *(++c);
 308                                                         fmt_mod = FMT_MOD_I64;
 309                                                 } else {
 310                                                         /* rewind */
 311                                                         fmt = *(--c);
 312                                                 }
 313                                         }
 314                                 } else {
 315                                         fmt_mod = FMT_MOD_NONE;
 316                                 }
 317                                 break;
 318                         default:
 319                                 fmt_mod = FMT_MOD_NONE;
 320                                 break;
 321                 }
 322
 323
 324                 type_t            *expected_type;
 325                 type_qualifiers_t  expected_qual = TYPE_QUALIFIER_NONE;
 326                 format_flags_t     allowed_flags;
 327                 switch (fmt) {
 328                         case 'd':
 329                         case 'i':
 330                                 switch (fmt_mod) {
 331                                         case FMT_MOD_NONE: expected_type = type_int;       break;
 332                                         case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
 333                                         case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
 334                                         case FMT_MOD_l:    expected_type = type_long;      break;
 335                                         case FMT_MOD_ll:   expected_type = type_long_long; break;
 336                                         case FMT_MOD_j:    expected_type = type_intmax_t;  break;
 337                                         case FMT_MOD_z:    expected_type = type_ssize_t;   break;
 338                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
 339                                         case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
 340                                         case FMT_MOD_I32:  expected_type = type_int32;     break;
 341                                         case FMT_MOD_I64:  expected_type = type_int64;     break;
 342
 343                                         default:
 344                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 345                                                 goto next_arg;
 346                                 }
 347                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_ZERO;
 348                                 break;
 349
 350                         case 'o':
 351                         case 'X':
 352                         case 'x':
 353                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 354                                 goto eval_fmt_mod_unsigned;
 355
 356                         case 'u':
 357                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 358 eval_fmt_mod_unsigned:
 359                                 switch (fmt_mod) {
 360                                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 361                                         case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
 362                                         case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
 363                                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 364                                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 365                                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 366                                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 367                                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 368                                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 369                                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 370                                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 371
 372                                         default:
 373                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 374                                                 goto next_arg;
 375                                 }
 376                                 break;
 377
 378                         case 'A':
 379                         case 'a':
 380                         case 'E':
 381                         case 'e':
 382                         case 'F':
 383                         case 'f':
 384                         case 'G':
 385                         case 'g':
 386                                 switch (fmt_mod) {
 387                                         case FMT_MOD_l:    /* l modifier is ignored */
 388                                         case FMT_MOD_NONE: expected_type = type_double;      break;
 389                                         case FMT_MOD_L:    expected_type = type_long_double; break;
 390
 391                                         default:
 392                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 393                                                 goto next_arg;
 394                                 }
 395                                 allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_SPACE | FMT_FLAG_PLUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 396                                 break;
 397
 398                         case 'C':
 399                                 if (fmt_mod != FMT_MOD_NONE) {
 400                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 401                                         goto next_arg;
 402                                 }
 403                                 expected_type = type_wchar_t;
 404                                 allowed_flags = FMT_FLAG_NONE;
 405                                 break;
 406
 407                         case 'c':
 408                                 expected_type = type_int;
 409                                 switch (fmt_mod) {
 410                                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 411                                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 412                                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 413
 414                                         default:
 415                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 416                                                 goto next_arg;
 417                                 }
 418                                 allowed_flags = FMT_FLAG_NONE;
 419                                 break;
 420
 421                         case 'S':
 422                                 if (fmt_mod != FMT_MOD_NONE) {
 423                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 424                                         goto next_arg;
 425                                 }
 426                                 expected_type = type_wchar_t_ptr;
 427                                 expected_qual = TYPE_QUALIFIER_CONST;
 428                                 allowed_flags = FMT_FLAG_MINUS;
 429                                 break;
 430
 431                         case 's':
 432                                 switch (fmt_mod) {
 433                                         case FMT_MOD_NONE: expected_type = type_char_ptr;    break;
 434                                         case FMT_MOD_l:    expected_type = type_wchar_t_ptr; break;
 435                                         case FMT_MOD_w:    expected_type = type_wchar_t_ptr; break;
 436
 437                                         default:
 438                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 439                                                 goto next_arg;
 440                                 }
 441                                 expected_qual = TYPE_QUALIFIER_CONST;
 442                                 allowed_flags = FMT_FLAG_MINUS;
 443                                 break;
 444
 445                         case 'p':
 446                                 if (fmt_mod != FMT_MOD_NONE) {
 447                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 448                                         goto next_arg;
 449                                 }
 450                                 expected_type = type_void_ptr;
 451                                 allowed_flags = FMT_FLAG_NONE;
 452                                 break;
 453
 454                         case 'n':
 455                                 switch (fmt_mod) {
 456                                         case FMT_MOD_NONE: expected_type = type_int_ptr;         break;
 457                                         case FMT_MOD_hh:   expected_type = type_signed_char_ptr; break;
 458                                         case FMT_MOD_h:    expected_type = type_short_ptr;       break;
 459                                         case FMT_MOD_l:    expected_type = type_long_ptr;        break;
 460                                         case FMT_MOD_ll:   expected_type = type_long_long_ptr;   break;
 461                                         case FMT_MOD_j:    expected_type = type_intmax_t_ptr;    break;
 462                                         case FMT_MOD_z:    expected_type = type_ssize_t_ptr;     break;
 463                                         case FMT_MOD_t:    expected_type = type_ptrdiff_t_ptr;   break;
 464
 465                                         default:
 466                                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 467                                                 goto next_arg;
 468                                 }
 469                                 allowed_flags = FMT_FLAG_NONE;
 470                                 break;
 471
 472                         default:
 473                                 warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at position %u", fmt, num_fmt);
 474                                 if (arg == NULL) {
 475                                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 476                                         return -1;
 477                                 }
 478                                 goto next_arg;
 479                 }
 480
 481                 format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
 482                 if (wrong_flags != 0) {
 483                         char  wrong[8];
 484                         char *p = wrong;
 485                         if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
 486                         if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
 487                         if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
 488                         if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
 489                         if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
 490                         if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
 491                         *p = '\0';
 492
 493                         warningf(WARN_FORMAT, pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 494                 }
 495
 496                 if (arg == NULL) {
 497                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 498                         return -1;
 499                 }
 500
 501                 { /* create a scope here to prevent warning about the jump to next_arg */
 502                         type_t *const arg_type           = arg->expression->base.type;
 503                         type_t *const arg_skip           = skip_typeref(arg_type);
 504                         type_t *const expected_type_skip = skip_typeref(expected_type);
 505
 506                         if (fmt == 'p') {
 507                                 /* allow any pointer type for %p, not just void */
 508                                 if (is_type_pointer(arg_skip))
 509                                         goto next_arg;
 510                         }
 511
 512                         if (is_type_pointer(expected_type_skip)) {
 513                                 if (is_type_pointer(arg_skip)) {
 514                                         type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 515                                         type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
 516                                         if ((arg_to->base.qualifiers & ~expected_qual) == 0 &&
 517                                                 get_unqualified_type(arg_to) == exp_to) {
 518                                                 goto next_arg;
 519                                         }
 520                                 }
 521                         } else if (get_unqualified_type(arg_skip) == expected_type_skip) {
 522                                 goto next_arg;
 523                         }
 524                         if (is_type_valid(arg_skip)) {
 525                                 source_position_t const *const apos = &arg->expression->base.source_position;
 526                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 527                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 528                         }
 529                 }
 530 next_arg:
 531                 arg = arg->next;
 532         }
 533         assert(fmt == '\0');
 534         if (c+1 < string + size) {
 535                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 536         }
 537         return num_args;
 538 }
 539
 540 /**
 541  * Check printf-style format.
 542  */
 543 static void check_printf_format(call_argument_t const *arg,
 544                                 format_spec_t const *const spec)
 545 {
 546         /* find format arg */
 547         size_t idx = 0;
 548         for (; idx < spec->fmt_idx; ++idx) {
 549                 if (arg == NULL)
 550                         return;
 551                 arg = arg->next;
 552         }
 553
 554         expression_t const *const fmt_expr = arg->expression;
 555
 556         /* find the real args */
 557         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 558                 arg = arg->next;
 559
 560         int const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
 561         if (num_fmt < 0)
 562                 return;
 563
 564         size_t num_args = 0;
 565         for (; arg != NULL; arg = arg->next)
 566                 ++num_args;
 567         if (num_args > (size_t)num_fmt) {
 568                 source_position_t const *const pos = &fmt_expr->base.source_position;
 569                 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" : "");
 570         }
 571 }
 572
 573 /**
 574  * Check scanf-style format.
 575  */
 576 static void check_scanf_format(const call_argument_t *arg,
 577                                const format_spec_t *spec)
 578 {
 579         /* find format arg */
 580         unsigned idx = 0;
 581         for (; idx < spec->fmt_idx; ++idx) {
 582                 if (arg == NULL)
 583                         return;
 584                 arg = arg->next;
 585         }
 586
 587         const expression_t *fmt_expr = arg->expression;
 588         if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
 589                 fmt_expr = fmt_expr->unary.value;
 590         }
 591
 592         if (fmt_expr->kind != EXPR_STRING_LITERAL
 593                         && fmt_expr->kind != EXPR_WIDE_STRING_LITERAL)
 594                 return;
 595
 596         const char *string = fmt_expr->literal.value.begin;
 597         size_t      size   = fmt_expr->literal.value.size;
 598         const char *c      = string;
 599
 600         /* find the real args */
 601         for (; idx < spec->arg_idx && arg != NULL; ++idx)
 602                 arg = arg->next;
 603
 604         const source_position_t *pos = &fmt_expr->base.source_position;
 605         unsigned num_fmt = 0;
 606         char     fmt;
 607         for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 608                 if (fmt != '%')
 609                         continue;
 610                 fmt = *(++c);
 611                 if (fmt == '\0') {
 612                         warningf(WARN_FORMAT, pos, "dangling '%%' in format string");
 613                         break;
 614                 }
 615                 if (fmt == '%')
 616                         continue;
 617
 618                 ++num_fmt;
 619
 620                 /* look for length modifiers */
 621                 format_length_modifier_t fmt_mod = FMT_MOD_NONE;
 622                 switch (fmt) {
 623                 case 'h':
 624                         fmt = *(++c);
 625                         if (fmt == 'h') {
 626                                 fmt = *(++c);
 627                                 fmt_mod = FMT_MOD_hh;
 628                         } else {
 629                                 fmt_mod = FMT_MOD_h;
 630                         }
 631                         break;
 632
 633                 case 'l':
 634                         fmt = *(++c);
 635                         if (fmt == 'l') {
 636                                 fmt = *(++c);
 637                                 fmt_mod = FMT_MOD_ll;
 638                         } else {
 639                                 fmt_mod = FMT_MOD_l;
 640                         }
 641                         break;
 642
 643                 case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
 644                 case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
 645                 case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
 646                 case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
 647                 /* microsoft mode */
 648                 case 'w':
 649                         if (c_mode & _MS) {
 650                                 fmt = *(++c);
 651                                 fmt_mod = FMT_MOD_w;
 652                         }
 653                         break;
 654                 case 'I':
 655                         if (c_mode & _MS) {
 656                                 fmt = *(++c);
 657                                 fmt_mod = FMT_MOD_I;
 658                                 if (fmt == '3') {
 659                                         fmt = *(++c);
 660                                         if (fmt == '2') {
 661                                                 fmt = *(++c);
 662                                                 fmt_mod = FMT_MOD_I32;
 663                                         } else {
 664                                                 /* rewind */
 665                                                 fmt = *(--c);
 666                                         }
 667                                 } else if (fmt == '6') {
 668                                         fmt = *(++c);
 669                                         if (fmt == '4') {
 670                                                 fmt = *(++c);
 671                                                 fmt_mod = FMT_MOD_I64;
 672                                         } else {
 673                                                 /* rewind */
 674                                                 fmt = *(--c);
 675                                         }
 676                                 }
 677                         }
 678                         break;
 679                 }
 680
 681                 if (fmt == '\0') {
 682                         warningf(WARN_FORMAT, pos, "dangling %% with conversion specififer in format string");
 683                         break;
 684                 }
 685
 686                 type_t *expected_type;
 687                 switch (fmt) {
 688                 case 'd':
 689                 case 'i':
 690                         switch (fmt_mod) {
 691                         case FMT_MOD_NONE: expected_type = type_int;         break;
 692                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 693                         case FMT_MOD_h:    expected_type = type_short;       break;
 694                         case FMT_MOD_l:    expected_type = type_long;        break;
 695                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 696                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 697                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 698                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 699                         case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
 700                         case FMT_MOD_I32:  expected_type = type_int32;       break;
 701                         case FMT_MOD_I64:  expected_type = type_int64;       break;
 702
 703                         default:
 704                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 705                                 goto next_arg;
 706                         }
 707                         break;
 708
 709                 case 'o':
 710                 case 'X':
 711                 case 'x':
 712                 case 'u':
 713                         switch (fmt_mod) {
 714                         case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
 715                         case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
 716                         case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 717                         case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 718                         case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 719                         case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
 720                         case FMT_MOD_z:    expected_type = type_size_t;             break;
 721                         case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
 722                         case FMT_MOD_I:    expected_type = type_size_t;             break;
 723                         case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
 724                         case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
 725
 726                         default:
 727                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 728                                 goto next_arg;
 729                         }
 730                         break;
 731
 732                 case 'A':
 733                 case 'a':
 734                 case 'E':
 735                 case 'e':
 736                 case 'F':
 737                 case 'f':
 738                 case 'G':
 739                 case 'g':
 740                         switch (fmt_mod) {
 741                         case FMT_MOD_l:    expected_type = type_double;      break;
 742                         case FMT_MOD_NONE: expected_type = type_float;       break;
 743                         case FMT_MOD_L:    expected_type = type_long_double; break;
 744
 745                         default:
 746                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 747                                 goto next_arg;
 748                         }
 749                         break;
 750
 751                 case 'C':
 752                         if (fmt_mod != FMT_MOD_NONE) {
 753                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 754                                 goto next_arg;
 755                         }
 756                         expected_type = type_wchar_t;
 757                         break;
 758
 759                 case 'c':
 760                         expected_type = type_int;
 761                         switch (fmt_mod) {
 762                         case FMT_MOD_NONE: expected_type = type_int;     break; /* TODO promoted char */
 763                         case FMT_MOD_l:    expected_type = type_wint_t;  break;
 764                         case FMT_MOD_w:    expected_type = type_wchar_t; break;
 765
 766                         default:
 767                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 768                                 goto next_arg;
 769                         }
 770                         break;
 771
 772                 case 'S':
 773                         if (fmt_mod != FMT_MOD_NONE) {
 774                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 775                                 goto next_arg;
 776                         }
 777                         expected_type = type_wchar_t;
 778                         break;
 779
 780                 case 's':
 781                 case '[':
 782                         switch (fmt_mod) {
 783                                 case FMT_MOD_NONE: expected_type = type_char;    break;
 784                                 case FMT_MOD_l:    expected_type = type_wchar_t; break;
 785                                 case FMT_MOD_w:    expected_type = type_wchar_t; break;
 786
 787                                 default:
 788                                         warn_invalid_length_modifier(pos, fmt_mod, fmt);
 789                                         goto next_arg;
 790                         }
 791                         break;
 792
 793                 case 'p':
 794                         if (fmt_mod != FMT_MOD_NONE) {
 795                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 796                                 goto next_arg;
 797                         }
 798                         expected_type = type_void_ptr;
 799                         break;
 800
 801                 case 'n':
 802                         switch (fmt_mod) {
 803                         case FMT_MOD_NONE: expected_type = type_int;         break;
 804                         case FMT_MOD_hh:   expected_type = type_signed_char; break;
 805                         case FMT_MOD_h:    expected_type = type_short;       break;
 806                         case FMT_MOD_l:    expected_type = type_long;        break;
 807                         case FMT_MOD_ll:   expected_type = type_long_long;   break;
 808                         case FMT_MOD_j:    expected_type = type_intmax_t;    break;
 809                         case FMT_MOD_z:    expected_type = type_ssize_t;     break;
 810                         case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
 811
 812                         default:
 813                                 warn_invalid_length_modifier(pos, fmt_mod, fmt);
 814                                 goto next_arg;
 815                         }
 816                         break;
 817
 818                 default:
 819                         warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at format %u", fmt, num_fmt);
 820                         if (arg == NULL) {
 821                                 warningf(WARN_FORMAT, pos, "too few arguments for format string");
 822                                 return;
 823                         }
 824                         goto next_arg;
 825                 }
 826
 827                 if (arg == NULL) {
 828                         warningf(WARN_FORMAT, pos, "too few arguments for format string");
 829                         return;
 830                 }
 831
 832                 { /* create a scope here to prevent warning about the jump to next_arg */
 833                         type_t *const arg_type           = arg->expression->base.type;
 834                         type_t *const arg_skip           = skip_typeref(arg_type);
 835                         type_t *const expected_type_skip = skip_typeref(expected_type);
 836
 837                         if (! is_type_pointer(arg_skip))
 838                                 goto error_arg_type;
 839                         type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
 840
 841                         if (fmt == 'p') {
 842                                 /* allow any pointer type for %p, not just void */
 843                                 if (is_type_pointer(ptr_skip))
 844                                         goto next_arg;
 845                         }
 846
 847                         /* do NOT allow const or restrict, all other should be ok */
 848                         if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
 849                                 goto error_arg_type;
 850                         type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
 851                         if (unqual_ptr == expected_type_skip) {
 852                                 goto next_arg;
 853                         } else if (expected_type_skip == type_char) {
 854                                 /* char matches with unsigned char AND signed char */
 855                                 if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
 856                                         goto next_arg;
 857                         }
 858 error_arg_type:
 859                         if (is_type_valid(arg_skip)) {
 860                                 source_position_t const *const apos = &arg->expression->base.source_position;
 861                                 char              const *const mod  = get_length_modifier_name(fmt_mod);
 862                                 warningf(WARN_FORMAT, apos, "argument type '%T' does not match conversion specifier '%%%s%c' at position %u", arg_type, mod, (char)fmt, num_fmt);
 863                         }
 864                 }
 865 next_arg:
 866                 arg = arg->next;
 867         }
 868         assert(fmt == '\0');
 869         if (c+1 < string + size) {
 870                 warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 871         }
 872         if (arg != NULL) {
 873                 unsigned num_args = num_fmt;
 874                 while (arg != NULL) {
 875                         ++num_args;
 876                         arg = arg->next;
 877                 }
 878                 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" : "");
 879         }
 880 }
 881
 882 static const format_spec_t builtin_table[] = {
 883         { "printf",        FORMAT_PRINTF,   0, 1 },
 884         { "wprintf",       FORMAT_PRINTF,   0, 1 },
 885         { "sprintf",       FORMAT_PRINTF,   1, 2 },
 886         { "swprintf",      FORMAT_PRINTF,   1, 2 },
 887         { "snprintf",      FORMAT_PRINTF,   2, 3 },
 888         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 889         { "fprintf",       FORMAT_PRINTF,   1, 2 },
 890         { "fwprintf",      FORMAT_PRINTF,   1, 2 },
 891         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 892         { "snwprintf",     FORMAT_PRINTF,   2, 3 },
 893
 894         { "scanf",         FORMAT_SCANF,    0, 1 },
 895         { "wscanf",        FORMAT_SCANF,    0, 1 },
 896         { "sscanf",        FORMAT_SCANF,    1, 2 },
 897         { "swscanf",       FORMAT_SCANF,    1, 2 },
 898         { "fscanf",        FORMAT_SCANF,    1, 2 },
 899         { "fwscanf",       FORMAT_SCANF,    1, 2 },
 900
 901         { "strftime",      FORMAT_STRFTIME, 3, 4 },
 902         { "wcstrftime",    FORMAT_STRFTIME, 3, 4 },
 903
 904         { "strfmon",       FORMAT_STRFMON,  3, 4 },
 905
 906         /* MS extensions */
 907         { "_snprintf",     FORMAT_PRINTF,   2, 3 },
 908         { "_snwprintf",    FORMAT_PRINTF,   2, 3 },
 909         { "_scrintf",      FORMAT_PRINTF,   0, 1 },
 910         { "_scwprintf",    FORMAT_PRINTF,   0, 1 },
 911         { "printf_s",      FORMAT_PRINTF,   0, 1 },
 912         { "wprintf_s",     FORMAT_PRINTF,   0, 1 },
 913         { "sprintf_s",     FORMAT_PRINTF,   3, 4 },
 914         { "swprintf_s",    FORMAT_PRINTF,   3, 4 },
 915         { "fprintf_s",     FORMAT_PRINTF,   1, 2 },
 916         { "fwprintf_s",    FORMAT_PRINTF,   1, 2 },
 917         { "_sprintf_l",    FORMAT_PRINTF,   1, 3 },
 918         { "_swprintf_l",   FORMAT_PRINTF,   1, 3 },
 919         { "_printf_l",     FORMAT_PRINTF,   0, 2 },
 920         { "_wprintf_l",    FORMAT_PRINTF,   0, 2 },
 921         { "_fprintf_l",    FORMAT_PRINTF,   1, 3 },
 922         { "_fwprintf_l",   FORMAT_PRINTF,   1, 3 },
 923         { "_printf_s_l",   FORMAT_PRINTF,   0, 2 },
 924         { "_wprintf_s_l",  FORMAT_PRINTF,   0, 2 },
 925         { "_sprintf_s_l",  FORMAT_PRINTF,   3, 5 },
 926         { "_swprintf_s_l", FORMAT_PRINTF,   3, 5 },
 927         { "_fprintf_s_l",  FORMAT_PRINTF,   1, 3 },
 928         { "_fwprintf_s_l", FORMAT_PRINTF,   1, 3 },
 929 };
 930
 931 void check_format(const call_expression_t *const call)
 932 {
 933         if (!is_warn_on(WARN_FORMAT))
 934                 return;
 935
 936         const expression_t *const func_expr = call->function;
 937         if (func_expr->kind != EXPR_REFERENCE)
 938                 return;
 939
 940         const entity_t        *const entity = func_expr->reference.entity;
 941         const call_argument_t *      arg    = call->arguments;
 942
 943         /*
 944          * For some functions we always check the format, even if it was not
 945          * specified. This allows to check format even in MS mode or without
 946          * header included.
 947          */
 948         const char *const name = entity->base.symbol->string;
 949         for (size_t i = 0; i < lengthof(builtin_table); ++i) {
 950                 if (strcmp(name, builtin_table[i].name) == 0) {
 951                         switch (builtin_table[i].fmt_kind) {
 952                         case FORMAT_PRINTF:
 953                                 check_printf_format(arg, &builtin_table[i]);
 954                                 break;
 955                         case FORMAT_SCANF:
 956                                 check_scanf_format(arg, &builtin_table[i]);
 957                                 break;
 958                         case FORMAT_STRFTIME:
 959                         case FORMAT_STRFMON:
 960                                 /* TODO: implement other cases */
 961                                 break;
 962                         }
 963                         break;
 964                 }
 965         }
 966 }