scripts/cstyle.pl

   1 #!/usr/bin/perl -w
   2 #
   3 # CDDL HEADER START
   4 #
   5 # The contents of this file are subject to the terms of the
   6 # Common Development and Distribution License (the "License").
   7 # You may not use this file except in compliance with the License.
   8 #
   9 # You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
  10 # or http://www.opensolaris.org/os/licensing.
  11 # See the License for the specific language governing permissions
  12 # and limitations under the License.
  13 #
  14 # When distributing Covered Code, include this CDDL HEADER in each
  15 # file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  16 # If applicable, add the following below this CDDL HEADER, with the
  17 # fields enclosed by brackets "[]" replaced with your own identifying
  18 # information: Portions Copyright [yyyy] [name of copyright owner]
  19 #
  20 # CDDL HEADER END
  21 #
  22 #
  23 # Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
  24 # Use is subject to license terms.
  25 #
  26 # @(#)cstyle 1.58 98/09/09 (from shannon)
  27 #ident  "%Z%%M% %I%     %E% SMI"
  28 #
  29 # cstyle - check for some common stylistic errors.
  30 #
  31 #       cstyle is a sort of "lint" for C coding style.
  32 #       It attempts to check for the style used in the
  33 #       kernel, sometimes known as "Bill Joy Normal Form".
  34 #
  35 #       There's a lot this can't check for, like proper indentation
  36 #       of code blocks.  There's also a lot more this could check for.
  37 #
  38 #       A note to the non perl literate:
  39 #
  40 #               perl regular expressions are pretty much like egrep
  41 #               regular expressions, with the following special symbols
  42 #
  43 #               \s      any space character
  44 #               \S      any non-space character
  45 #               \w      any "word" character [a-zA-Z0-9_]
  46 #               \W      any non-word character
  47 #               \d      a digit [0-9]
  48 #               \D      a non-digit
  49 #               \b      word boundary (between \w and \W)
  50 #               \B      non-word boundary
  51 #
  52
  53 require 5.0;
  54 use IO::File;
  55 use Getopt::Std;
  56 use strict;
  57
  58 my $usage =
  59 "usage: cstyle [-chpvCP] [-o constructs] file ...
  60         -c      check continuation indentation inside functions
  61         -h      perform heuristic checks that are sometimes wrong
  62         -p      perform some of the more picky checks
  63         -v      verbose
  64         -C      don't check anything in header block comments
  65         -P      check for use of non-POSIX types
  66         -o constructs
  67                 allow a comma-seperated list of optional constructs:
  68                     doxygen     allow doxygen-style block comments (/** /*!)
  69                     splint      allow splint-style lint comments (/*@ ... @*/)
  70 ";
  71
  72 my %opts;
  73
  74 if (!getopts("cho:pvCP", \%opts)) {
  75         print $usage;
  76         exit 2;
  77 }
  78
  79 my $check_continuation = $opts{'c'};
  80 my $heuristic = $opts{'h'};
  81 my $picky = $opts{'p'};
  82 my $verbose = $opts{'v'};
  83 my $ignore_hdr_comment = $opts{'C'};
  84 my $check_posix_types = $opts{'P'};
  85
  86 my $doxygen_comments = 0;
  87 my $splint_comments = 0;
  88
  89 if (defined($opts{'o'})) {
  90         for my $x (split /,/, $opts{'o'}) {
  91                 if ($x eq "doxygen") {
  92                         $doxygen_comments = 1;
  93                 } elsif ($x eq "splint") {
  94                         $splint_comments = 1;
  95                 } else {
  96                         print "cstyle: unrecognized construct \"$x\"\n";
  97                         print $usage;
  98                         exit 2;
  99                 }
 100         }
 101 }
 102
 103 my ($filename, $line, $prev);           # shared globals
 104
 105 my $fmt;
 106 my $hdr_comment_start;
 107
 108 if ($verbose) {
 109         $fmt = "%s: %d: %s\n%s\n";
 110 } else {
 111         $fmt = "%s: %d: %s\n";
 112 }
 113
 114 if ($doxygen_comments) {
 115         # doxygen comments look like "/*!" or "/**"; allow them.
 116         $hdr_comment_start = qr/^\s*\/\*[\!\*]?$/;
 117 } else {
 118         $hdr_comment_start = qr/^\s*\/\*$/;
 119 }
 120
 121 # Note, following must be in single quotes so that \s and \w work right.
 122 my $typename = '(int|char|short|long|unsigned|float|double' .
 123     '|\w+_t|struct\s+\w+|union\s+\w+|FILE)';
 124
 125 # mapping of old types to POSIX compatible types
 126 my %old2posix = (
 127         'unchar' => 'uchar_t',
 128         'ushort' => 'ushort_t',
 129         'uint' => 'uint_t',
 130         'ulong' => 'ulong_t',
 131         'u_int' => 'uint_t',
 132         'u_short' => 'ushort_t',
 133         'u_long' => 'ulong_t',
 134         'u_char' => 'uchar_t',
 135         'quad' => 'quad_t'
 136 );
 137
 138 my $lint_re = qr/\/\*(?:
 139         ARGSUSED[0-9]*|NOTREACHED|LINTLIBRARY|VARARGS[0-9]*|
 140         CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY|
 141         FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*|
 142         PROTOLIB[0-9]*|SCANFLIKE[0-9]*|CSTYLED.*?
 143     )\*\//x;
 144
 145 my $splint_re = qr/\/\*@.*?@\*\//x;
 146
 147 my $warlock_re = qr/\/\*\s*(?:
 148         VARIABLES\ PROTECTED\ BY|
 149         MEMBERS\ PROTECTED\ BY|
 150         ALL\ MEMBERS\ PROTECTED\ BY|
 151         READ-ONLY\ VARIABLES:|
 152         READ-ONLY\ MEMBERS:|
 153         VARIABLES\ READABLE\ WITHOUT\ LOCK:|
 154         MEMBERS\ READABLE\ WITHOUT\ LOCK:|
 155         LOCKS\ COVERED\ BY|
 156         LOCK\ UNNEEDED\ BECAUSE|
 157         LOCK\ NEEDED:|
 158         LOCK\ HELD\ ON\ ENTRY:|
 159         READ\ LOCK\ HELD\ ON\ ENTRY:|
 160         WRITE\ LOCK\ HELD\ ON\ ENTRY:|
 161         LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:|
 162         READ\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:|
 163         WRITE\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:|
 164         LOCK\ RELEASED\ AS\ SIDE\ EFFECT:|
 165         LOCK\ UPGRADED\ AS\ SIDE\ EFFECT:|
 166         LOCK\ DOWNGRADED\ AS\ SIDE\ EFFECT:|
 167         FUNCTIONS\ CALLED\ THROUGH\ POINTER|
 168         FUNCTIONS\ CALLED\ THROUGH\ MEMBER|
 169         LOCK\ ORDER:
 170     )/x;
 171
 172 my $err_stat = 0;               # exit status
 173
 174 if ($#ARGV >= 0) {
 175         foreach my $arg (@ARGV) {
 176                 my $fh = new IO::File $arg, "r";
 177                 if (!defined($fh)) {
 178                         printf "%s: can not open\n", $arg;
 179                 } else {
 180                         &cstyle($arg, $fh);
 181                         close $fh;
 182                 }
 183         }
 184 } else {
 185         &cstyle("<stdin>", *STDIN);
 186 }
 187 exit $err_stat;
 188
 189 my $no_errs = 0;                # set for CSTYLED-protected lines
 190
 191 sub err($) {
 192         my ($error) = @_;
 193         unless ($no_errs) {
 194                 if ($verbose) {
 195                         printf $fmt, $filename, $., $error, $line;
 196                 } else {
 197                         printf $fmt, $filename, $., $error;
 198                 }
 199                 $err_stat = 1;
 200         }
 201 }
 202
 203 sub err_prefix($$) {
 204         my ($prevline, $error) = @_;
 205         my $out = $prevline."\n".$line;
 206         unless ($no_errs) {
 207                 if ($verbose) {
 208                         printf $fmt, $filename, $., $error, $out;
 209                 } else {
 210                         printf $fmt, $filename, $., $error;
 211                 }
 212                 $err_stat = 1;
 213         }
 214 }
 215
 216 sub err_prev($) {
 217         my ($error) = @_;
 218         unless ($no_errs) {
 219                 if ($verbose) {
 220                         printf $fmt, $filename, $. - 1, $error, $prev;
 221                 } else {
 222                         printf $fmt, $filename, $. - 1, $error;
 223                 }
 224                 $err_stat = 1;
 225         }
 226 }
 227
 228 sub cstyle($$) {
 229
 230 my ($fn, $filehandle) = @_;
 231 $filename = $fn;                        # share it globally
 232
 233 my $in_cpp = 0;
 234 my $next_in_cpp = 0;
 235
 236 my $in_comment = 0;
 237 my $in_header_comment = 0;
 238 my $comment_done = 0;
 239 my $in_warlock_comment = 0;
 240 my $in_function = 0;
 241 my $in_function_header = 0;
 242 my $in_declaration = 0;
 243 my $note_level = 0;
 244 my $nextok = 0;
 245 my $nocheck = 0;
 246
 247 my $in_string = 0;
 248
 249 my ($okmsg, $comment_prefix);
 250
 251 $line = '';
 252 $prev = '';
 253 reset_indent();
 254
 255 line: while (<$filehandle>) {
 256         s/\r?\n$//;     # strip return and newline
 257
 258         # save the original line, then remove all text from within
 259         # double or single quotes, we do not want to check such text.
 260
 261         $line = $_;
 262
 263         #
 264         # C allows strings to be continued with a backslash at the end of
 265         # the line.  We translate that into a quoted string on the previous
 266         # line followed by an initial quote on the next line.
 267         #
 268         # (we assume that no-one will use backslash-continuation with character
 269         # constants)
 270         #
 271         $_ = '"' . $_           if ($in_string && !$nocheck && !$in_comment);
 272
 273         #
 274         # normal strings and characters
 275         #
 276         s/'([^\\']|\\[^xX0]|\\0[0-9]*|\\[xX][0-9a-fA-F]*)'/''/g;
 277         s/"([^\\"]|\\.)*"/\"\"/g;
 278
 279         #
 280         # detect string continuation
 281         #
 282         if ($nocheck || $in_comment) {
 283                 $in_string = 0;
 284         } else {
 285                 #
 286                 # Now that all full strings are replaced with "", we check
 287                 # for unfinished strings continuing onto the next line.
 288                 #
 289                 $in_string =
 290                     (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ ||
 291                     s/^("")*"([^\\"]|\\.)*\\$/""/);
 292         }
 293
 294         #
 295         # figure out if we are in a cpp directive
 296         #
 297         $in_cpp = $next_in_cpp || /^\s*#/;      # continued or started
 298         $next_in_cpp = $in_cpp && /\\$/;        # only if continued
 299
 300         # strip off trailing backslashes, which appear in long macros
 301         s/\s*\\$//;
 302
 303         # an /* END CSTYLED */ comment ends a no-check block.
 304         if ($nocheck) {
 305                 if (/\/\* *END *CSTYLED *\*\//) {
 306                         $nocheck = 0;
 307                 } else {
 308                         reset_indent();
 309                         next line;
 310                 }
 311         }
 312
 313         # a /*CSTYLED*/ comment indicates that the next line is ok.
 314         if ($nextok) {
 315                 if ($okmsg) {
 316                         err($okmsg);
 317                 }
 318                 $nextok = 0;
 319                 $okmsg = 0;
 320                 if (/\/\* *CSTYLED.*\*\//) {
 321                         /^.*\/\* *CSTYLED *(.*) *\*\/.*$/;
 322                         $okmsg = $1;
 323                         $nextok = 1;
 324                 }
 325                 $no_errs = 1;
 326         } elsif ($no_errs) {
 327                 $no_errs = 0;
 328         }
 329
 330         # check length of line.
 331         # first, a quick check to see if there is any chance of being too long.
 332         if (($line =~ tr/\t/\t/) * 7 + length($line) > 80) {
 333                 # yes, there is a chance.
 334                 # replace tabs with spaces and check again.
 335                 my $eline = $line;
 336                 1 while $eline =~
 337                     s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e;
 338                 if (length($eline) > 80) {
 339                         err("line > 80 characters");
 340                 }
 341         }
 342
 343         # ignore NOTE(...) annotations (assumes NOTE is on lines by itself).
 344         if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE
 345                 s/[^()]//g;                       # eliminate all non-parens
 346                 $note_level += s/\(//g - length;  # update paren nest level
 347                 next;
 348         }
 349
 350         # a /* BEGIN CSTYLED */ comment starts a no-check block.
 351         if (/\/\* *BEGIN *CSTYLED *\*\//) {
 352                 $nocheck = 1;
 353         }
 354
 355         # a /*CSTYLED*/ comment indicates that the next line is ok.
 356         if (/\/\* *CSTYLED.*\*\//) {
 357                 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/;
 358                 $okmsg = $1;
 359                 $nextok = 1;
 360         }
 361         if (/\/\/ *CSTYLED/) {
 362                 /^.*\/\/ *CSTYLED *(.*)$/;
 363                 $okmsg = $1;
 364                 $nextok = 1;
 365         }
 366
 367         # universal checks; apply to everything
 368         if (/\t +\t/) {
 369                 err("spaces between tabs");
 370         }
 371         if (/ \t+ /) {
 372                 err("tabs between spaces");
 373         }
 374         if (/\s$/) {
 375                 err("space or tab at end of line");
 376         }
 377         if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) {
 378                 err("comment preceded by non-blank");
 379         }
 380
 381         # is this the beginning or ending of a function?
 382         # (not if "struct foo\n{\n")
 383         if (/^{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) {
 384                 $in_function = 1;
 385                 $in_declaration = 1;
 386                 $in_function_header = 0;
 387                 $prev = $line;
 388                 next line;
 389         }
 390         if (/^}\s*(\/\*.*\*\/\s*)*$/) {
 391                 if ($prev =~ /^\s*return\s*;/) {
 392                         err_prev("unneeded return at end of function");
 393                 }
 394                 $in_function = 0;
 395                 reset_indent();         # we don't check between functions
 396                 $prev = $line;
 397                 next line;
 398         }
 399         if (/^\w*\($/) {
 400                 $in_function_header = 1;
 401         }
 402
 403         if ($in_warlock_comment && /\*\//) {
 404                 $in_warlock_comment = 0;
 405                 $prev = $line;
 406                 next line;
 407         }
 408
 409         # a blank line terminates the declarations within a function.
 410         # XXX - but still a problem in sub-blocks.
 411         if ($in_declaration && /^$/) {
 412                 $in_declaration = 0;
 413         }
 414
 415         if ($comment_done) {
 416                 $in_comment = 0;
 417                 $in_header_comment = 0;
 418                 $comment_done = 0;
 419         }
 420         # does this looks like the start of a block comment?
 421         if (/$hdr_comment_start/) {
 422                 if (!/^\t*\/\*/) {
 423                         err("block comment not indented by tabs");
 424                 }
 425                 $in_comment = 1;
 426                 /^(\s*)\//;
 427                 $comment_prefix = $1;
 428                 if ($comment_prefix eq "") {
 429                         $in_header_comment = 1;
 430                 }
 431                 $prev = $line;
 432                 next line;
 433         }
 434         # are we still in the block comment?
 435         if ($in_comment) {
 436                 if (/^$comment_prefix \*\/$/) {
 437                         $comment_done = 1;
 438                 } elsif (/\*\//) {
 439                         $comment_done = 1;
 440                         err("improper block comment close")
 441                             unless ($ignore_hdr_comment && $in_header_comment);
 442                 } elsif (!/^$comment_prefix \*[ \t]/ &&
 443                     !/^$comment_prefix \*$/) {
 444                         err("improper block comment")
 445                             unless ($ignore_hdr_comment && $in_header_comment);
 446                 }
 447         }
 448
 449         if ($in_header_comment && $ignore_hdr_comment) {
 450                 $prev = $line;
 451                 next line;
 452         }
 453
 454         # check for errors that might occur in comments and in code.
 455
 456         # allow spaces to be used to draw pictures in all comments.
 457         if (/[^ ]     / && !/".*     .*"/ && !$in_comment) {
 458                 err("spaces instead of tabs");
 459         }
 460         if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ &&
 461             (!/^    \w/ || $in_function != 0)) {
 462                 err("indent by spaces instead of tabs");
 463         }
 464         if (/^\t+ [^ \t\*]/ || /^\t+  \S/ || /^\t+   \S/) {
 465                 err("continuation line not indented by 4 spaces");
 466         }
 467         if (/$warlock_re/ && !/\*\//) {
 468                 $in_warlock_comment = 1;
 469                 $prev = $line;
 470                 next line;
 471         }
 472         if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) {
 473                 err("improper first line of block comment");
 474         }
 475
 476         if ($in_comment) {      # still in comment, don't do further checks
 477                 $prev = $line;
 478                 next line;
 479         }
 480
 481         if ((/[^(]\/\*\S/ || /^\/\*\S/) &&
 482             !(/$lint_re/ || ($splint_comments && /$splint_re/))) {
 483                 err("missing blank after open comment");
 484         }
 485         if (/\S\*\/[^)]|\S\*\/$/ &&
 486             !(/$lint_re/ || ($splint_comments && /$splint_re/))) {
 487                 err("missing blank before close comment");
 488         }
 489         if (/\/\/\S/) {         # C++ comments
 490                 err("missing blank after start comment");
 491         }
 492         # check for unterminated single line comments, but allow them when
 493         # they are used to comment out the argument list of a function
 494         # declaration.
 495         if (/\S.*\/\*/ && !/\S.*\/\*.*\*\// && !/\(\/\*/) {
 496                 err("unterminated single line comment");
 497         }
 498
 499         if (/^(#else|#endif|#include)(.*)$/) {
 500                 $prev = $line;
 501                 if ($picky) {
 502                         my $directive = $1;
 503                         my $clause = $2;
 504                         # Enforce ANSI rules for #else and #endif: no noncomment
 505                         # identifiers are allowed after #endif or #else.  Allow
 506                         # C++ comments since they seem to be a fact of life.
 507                         if ((($1 eq "#endif") || ($1 eq "#else")) &&
 508                             ($clause ne "") &&
 509                             (!($clause =~ /^\s+\/\*.*\*\/$/)) &&
 510                             (!($clause =~ /^\s+\/\/.*$/))) {
 511                                 err("non-comment text following " .
 512                                     "$directive (or malformed $directive " .
 513                                     "directive)");
 514                         }
 515                 }
 516                 next line;
 517         }
 518
 519         #
 520         # delete any comments and check everything else.  Note that
 521         # ".*?" is a non-greedy match, so that we don't get confused by
 522         # multiple comments on the same line.
 523         #
 524         s/\/\*.*?\*\//\ 1/g;
 525         s/\/\/.*$/\ 1/;           # C++ comments
 526
 527         # delete any trailing whitespace; we have already checked for that.
 528         s/\s*$//;
 529
 530         # following checks do not apply to text in comments.
 531
 532         if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=[^\s,]/ ||
 533             (/[^->]>[^,=>\s]/ && !/[^->]>$/) ||
 534             (/[^<]<[^,=<\s]/ && !/[^<]<$/) ||
 535             /[^<\s]<[^<]/ || /[^->\s]>[^>]/) {
 536                 err("missing space around relational operator");
 537         }
 538         if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ ||
 539             (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) ||
 540             (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) {
 541                 # XXX - should only check this for C++ code
 542                 # XXX - there are probably other forms that should be allowed
 543                 if (!/\soperator=/) {
 544                         err("missing space around assignment operator");
 545                 }
 546         }
 547         if (/[,;]\S/ && !/\bfor \(;;\)/) {
 548                 err("comma or semicolon followed by non-blank");
 549         }
 550         # allow "for" statements to have empty "while" clauses
 551         if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) {
 552                 err("comma or semicolon preceded by blank");
 553         }
 554         if (/^\s*(&&|\|\|)/) {
 555                 err("improper boolean continuation");
 556         }
 557         if (/\S   *(&&|\|\|)/ || /(&&|\|\|)   *\S/) {
 558                 err("more than one space around boolean operator");
 559         }
 560         if (/\b(for|if|while|switch|sizeof|return|case)\(/) {
 561                 err("missing space between keyword and paren");
 562         }
 563         if (/(\b(for|if|while|switch|return)\b.*){2,}/ && !/^#define/) {
 564                 # multiple "case" and "sizeof" allowed
 565                 err("more than one keyword on line");
 566         }
 567         if (/\b(for|if|while|switch|sizeof|return|case)\s\s+\(/ &&
 568             !/^#if\s+\(/) {
 569                 err("extra space between keyword and paren");
 570         }
 571         # try to detect "func (x)" but not "if (x)" or
 572         # "#define foo (x)" or "int (*func)();"
 573         if (/\w\s\(/) {
 574                 my $s = $_;
 575                 # strip off all keywords on the line
 576                 s/\b(for|if|while|switch|return|case|sizeof)\s\(/XXX(/g;
 577                 s/#elif\s\(/XXX(/g;
 578                 s/^#define\s+\w+\s+\(/XXX(/;
 579                 # do not match things like "void (*f)();"
 580                 # or "typedef void (func_t)();"
 581                 s/\w\s\(+\*/XXX(*/g;
 582                 s/\b($typename|void)\s+\(+/XXX(/og;
 583                 if (/\w\s\(/) {
 584                         err("extra space between function name and left paren");
 585                 }
 586                 $_ = $s;
 587         }
 588         # try to detect "int foo(x)", but not "extern int foo(x);"
 589         # XXX - this still trips over too many legitimate things,
 590         # like "int foo(x,\n\ty);"
 591 #               if (/^(\w+(\s|\*)+)+\w+\(/ && !/\)[;,](\s|\ 1)*$/ &&
 592 #                   !/^(extern|static)\b/) {
 593 #                       err("return type of function not on separate line");
 594 #               }
 595         # this is a close approximation
 596         if (/^(\w+(\s|\*)+)+\w+\(.*\)(\s|\ 1)*$/ &&
 597             !/^(extern|static)\b/) {
 598                 err("return type of function not on separate line");
 599         }
 600         if (/^#define /) {
 601                 err("#define followed by space instead of tab");
 602         }
 603         if (/^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) {
 604                 err("unparenthesized return expression");
 605         }
 606         if (/\bsizeof\b/ && !/\bsizeof\s*\(.*\)/) {
 607                 err("unparenthesized sizeof expression");
 608         }
 609         if (/\(\s/) {
 610                 err("whitespace after left paren");
 611         }
 612         # Allow "for" statements to have empty "continue" clauses.
 613         # Allow right paren on its own line unless we're being picky (-p).
 614         if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/ && ($picky || !/^\s*\)/)) {
 615                 err("whitespace before right paren");
 616         }
 617         if (/^\s*\(void\)[^ ]/) {
 618                 err("missing space after (void) cast");
 619         }
 620         if (/\S\{/ && !/\{\{/) {
 621                 err("missing space before left brace");
 622         }
 623         if ($in_function && /^\s+{/ &&
 624             ($prev =~ /\)\s*$/ || $prev =~ /\bstruct\s+\w+$/)) {
 625                 err("left brace starting a line");
 626         }
 627         if (/}(else|while)/) {
 628                 err("missing space after right brace");
 629         }
 630         if (/}\s\s+(else|while)/) {
 631                 err("extra space after right brace");
 632         }
 633         if (/\b_VOID\b|\bVOID\b|\bSTATIC\b/) {
 634                 err("obsolete use of VOID or STATIC");
 635         }
 636         if (/\b$typename\*/o) {
 637                 err("missing space between type name and *");
 638         }
 639         if (/^\s+#/) {
 640                 err("preprocessor statement not in column 1");
 641         }
 642         if (/^#\s/) {
 643                 err("blank after preprocessor #");
 644         }
 645         if (/!\s*(strcmp|strncmp|bcmp)\s*\(/) {
 646                 err("don't use boolean ! with comparison functions");
 647         }
 648
 649         #
 650         # We completely ignore, for purposes of indentation:
 651         #  * lines outside of functions
 652         #  * preprocessor lines
 653         #
 654         if ($check_continuation && $in_function && !$in_cpp) {
 655                 process_indent($_);
 656         }
 657         if ($picky) {
 658                 # try to detect spaces after casts, but allow (e.g.)
 659                 # "sizeof (int) + 1", "void (*funcptr)(int) = foo;", and
 660                 # "int foo(int) __NORETURN;"
 661                 if ((/^\($typename( \*+)?\)\s/o ||
 662                     /\W\($typename( \*+)?\)\s/o) &&
 663                     !/sizeof\s*\($typename( \*)?\)\s/o &&
 664                     !/\($typename( \*+)?\)\s+=[^=]/o) {
 665                         err("space after cast");
 666                 }
 667                 if (/\b$typename\s*\*\s/o &&
 668                     !/\b$typename\s*\*\s+const\b/o) {
 669                         err("unary * followed by space");
 670                 }
 671         }
 672         if ($check_posix_types) {
 673                 # try to detect old non-POSIX types.
 674                 # POSIX requires all non-standard typedefs to end in _t,
 675                 # but historically these have been used.
 676                 if (/\b(unchar|ushort|uint|ulong|u_int|u_short|u_long|u_char|quad)\b/) {
 677                         err("non-POSIX typedef $1 used: use $old2posix{$1} instead");
 678                 }
 679         }
 680         if ($heuristic) {
 681                 # cannot check this everywhere due to "struct {\n...\n} foo;"
 682                 if ($in_function && !$in_declaration &&
 683                     /}./ && !/}\s+=/ && !/{.*}[;,]$/ && !/}(\s|\ 1)*$/ &&
 684                     !/} (else|while)/ && !/}}/) {
 685                         err("possible bad text following right brace");
 686                 }
 687                 # cannot check this because sub-blocks in
 688                 # the middle of code are ok
 689                 if ($in_function && /^\s+{/) {
 690                         err("possible left brace starting a line");
 691                 }
 692         }
 693         if (/^\s*else\W/) {
 694                 if ($prev =~ /^\s*}$/) {
 695                         err_prefix($prev,
 696                             "else and right brace should be on same line");
 697                 }
 698         }
 699         $prev = $line;
 700 }
 701
 702 if ($prev eq "") {
 703         err("last line in file is blank");
 704 }
 705
 706 }
 707
 708 #
 709 # Continuation-line checking
 710 #
 711 # The rest of this file contains the code for the continuation checking
 712 # engine.  It's a pretty simple state machine which tracks the expression
 713 # depth (unmatched '('s and '['s).
 714 #
 715 # Keep in mind that the argument to process_indent() has already been heavily
 716 # processed; all comments have been replaced by control-A, and the contents of
 717 # strings and character constants have been elided.
 718 #
 719
 720 my $cont_in;            # currently inside of a continuation
 721 my $cont_off;           # skipping an initializer or definition
 722 my $cont_noerr;         # suppress cascading errors
 723 my $cont_start;         # the line being continued
 724 my $cont_base;          # the base indentation
 725 my $cont_first;         # this is the first line of a statement
 726 my $cont_multiseg;      # this continuation has multiple segments
 727
 728 my $cont_special;       # this is a C statement (if, for, etc.)
 729 my $cont_macro;         # this is a macro
 730 my $cont_case;          # this is a multi-line case
 731
 732 my @cont_paren;         # the stack of unmatched ( and [s we've seen
 733
 734 sub
 735 reset_indent()
 736 {
 737         $cont_in = 0;
 738         $cont_off = 0;
 739 }
 740
 741 sub
 742 delabel($)
 743 {
 744         #
 745         # replace labels with tabs.  Note that there may be multiple
 746         # labels on a line.
 747         #
 748         local $_ = $_[0];
 749
 750         while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) {
 751                 my ($pre_tabs, $label, $rest) = ($1, $2, $3);
 752                 $_ = $pre_tabs;
 753                 while ($label =~ s/^([^\t]*)(\t+)//) {
 754                         $_ .= "\t" x (length($2) + length($1) / 8);
 755                 }
 756                 $_ .= ("\t" x (length($label) / 8)).$rest;
 757         }
 758
 759         return ($_);
 760 }
 761
 762 sub
 763 process_indent($)
 764 {
 765         require strict;
 766         local $_ = $_[0];                       # preserve the global $_
 767
 768         s/\ 1//g; # No comments
 769         s/\s+$//;       # Strip trailing whitespace
 770
 771         return                  if (/^$/);      # skip empty lines
 772
 773         # regexps used below; keywords taking (), macros, and continued cases
 774         my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b';
 775         my $macro = '[A-Z_][A-Z_0-9]*\(';
 776         my $case = 'case\b[^:]*$';
 777
 778         # skip over enumerations, array definitions, initializers, etc.
 779         if ($cont_off <= 0 && !/^\s*$special/ &&
 780             (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*)){/ ||
 781             (/^\s*{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) {
 782                 $cont_in = 0;
 783                 $cont_off = tr/{/{/ - tr/}/}/;
 784                 return;
 785         }
 786         if ($cont_off) {
 787                 $cont_off += tr/{/{/ - tr/}/}/;
 788                 return;
 789         }
 790
 791         if (!$cont_in) {
 792                 $cont_start = $line;
 793
 794                 if (/^\t* /) {
 795                         err("non-continuation indented 4 spaces");
 796                         $cont_noerr = 1;                # stop reporting
 797                 }
 798                 $_ = delabel($_);       # replace labels with tabs
 799
 800                 # check if the statement is complete
 801                 return          if (/^\s*\}?$/);
 802                 return          if (/^\s*\}?\s*else\s*\{?$/);
 803                 return          if (/^\s*do\s*\{?$/);
 804                 return          if (/{$/);
 805                 return          if (/}[,;]?$/);
 806
 807                 # Allow macros on their own lines
 808                 return          if (/^\s*[A-Z_][A-Z_0-9]*$/);
 809
 810                 # cases we don't deal with, generally non-kosher
 811                 if (/{/) {
 812                         err("stuff after {");
 813                         return;
 814                 }
 815
 816                 # Get the base line, and set up the state machine
 817                 /^(\t*)/;
 818                 $cont_base = $1;
 819                 $cont_in = 1;
 820                 @cont_paren = ();
 821                 $cont_first = 1;
 822                 $cont_multiseg = 0;
 823
 824                 # certain things need special processing
 825                 $cont_special = /^\s*$special/? 1 : 0;
 826                 $cont_macro = /^\s*$macro/? 1 : 0;
 827                 $cont_case = /^\s*$case/? 1 : 0;
 828         } else {
 829                 $cont_first = 0;
 830
 831                 # Strings may be pulled back to an earlier (half-)tabstop
 832                 unless ($cont_noerr || /^$cont_base    / ||
 833                     (/^\t*(?:    )?(?:gettext\()?\"/ && !/^$cont_base\t/)) {
 834                         err_prefix($cont_start,
 835                             "continuation should be indented 4 spaces");
 836                 }
 837         }
 838
 839         my $rest = $_;                  # keeps the remainder of the line
 840
 841         #
 842         # The split matches 0 characters, so that each 'special' character
 843         # is processed separately.  Parens and brackets are pushed and
 844         # popped off the @cont_paren stack.  For normal processing, we wait
 845         # until a ; or { terminates the statement.  "special" processing
 846         # (if/for/while/switch) is allowed to stop when the stack empties,
 847         # as is macro processing.  Case statements are terminated with a :
 848         # and an empty paren stack.
 849         #
 850         foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) {
 851                 next            if (length($_) == 0);
 852
 853                 # rest contains the remainder of the line
 854                 my $rxp = "[^\Q$_\E]*\Q$_\E";
 855                 $rest =~ s/^$rxp//;
 856
 857                 if (/\(/ || /\[/) {
 858                         push @cont_paren, $_;
 859                 } elsif (/\)/ || /\]/) {
 860                         my $cur = $_;
 861                         tr/\)\]/\(\[/;
 862
 863                         my $old = (pop @cont_paren);
 864                         if (!defined($old)) {
 865                                 err("unexpected '$cur'");
 866                                 $cont_in = 0;
 867                                 last;
 868                         } elsif ($old ne $_) {
 869                                 err("'$cur' mismatched with '$old'");
 870                                 $cont_in = 0;
 871                                 last;
 872                         }
 873
 874                         #
 875                         # If the stack is now empty, do special processing
 876                         # for if/for/while/switch and macro statements.
 877                         #
 878                         next            if (@cont_paren != 0);
 879                         if ($cont_special) {
 880                                 if ($rest =~ /^\s*{?$/) {
 881                                         $cont_in = 0;
 882                                         last;
 883                                 }
 884                                 if ($rest =~ /^\s*;$/) {
 885                                         err("empty if/for/while body ".
 886                                             "not on its own line");
 887                                         $cont_in = 0;
 888                                         last;
 889                                 }
 890                                 if (!$cont_first && $cont_multiseg == 1) {
 891                                         err_prefix($cont_start,
 892                                             "multiple statements continued ".
 893                                             "over multiple lines");
 894                                         $cont_multiseg = 2;
 895                                 } elsif ($cont_multiseg == 0) {
 896                                         $cont_multiseg = 1;
 897                                 }
 898                                 # We've finished this section, start
 899                                 # processing the next.
 900                                 goto section_ended;
 901                         }
 902                         if ($cont_macro) {
 903                                 if ($rest =~ /^$/) {
 904                                         $cont_in = 0;
 905                                         last;
 906                                 }
 907                         }
 908                 } elsif (/\;/) {
 909                         if ($cont_case) {
 910                                 err("unexpected ;");
 911                         } elsif (!$cont_special) {
 912                                 err("unexpected ;")     if (@cont_paren != 0);
 913                                 if (!$cont_first && $cont_multiseg == 1) {
 914                                         err_prefix($cont_start,
 915                                             "multiple statements continued ".
 916                                             "over multiple lines");
 917                                         $cont_multiseg = 2;
 918                                 } elsif ($cont_multiseg == 0) {
 919                                         $cont_multiseg = 1;
 920                                 }
 921                                 if ($rest =~ /^$/) {
 922                                         $cont_in = 0;
 923                                         last;
 924                                 }
 925                                 if ($rest =~ /^\s*special/) {
 926                                         err("if/for/while/switch not started ".
 927                                             "on its own line");
 928                                 }
 929                                 goto section_ended;
 930                         }
 931                 } elsif (/\{/) {
 932                         err("{ while in parens/brackets") if (@cont_paren != 0);
 933                         err("stuff after {")            if ($rest =~ /[^\s}]/);
 934                         $cont_in = 0;
 935                         last;
 936                 } elsif (/\}/) {
 937                         err("} while in parens/brackets") if (@cont_paren != 0);
 938                         if (!$cont_special && $rest !~ /^\s*(while|else)\b/) {
 939                                 if ($rest =~ /^$/) {
 940                                         err("unexpected }");
 941                                 } else {
 942                                         err("stuff after }");
 943                                 }
 944                                 $cont_in = 0;
 945                                 last;
 946                         }
 947                 } elsif (/\:/ && $cont_case && @cont_paren == 0) {
 948                         err("stuff after multi-line case") if ($rest !~ /$^/);
 949                         $cont_in = 0;
 950                         last;
 951                 }
 952                 next;
 953 section_ended:
 954                 # End of a statement or if/while/for loop.  Reset
 955                 # cont_special and cont_macro based on the rest of the
 956                 # line.
 957                 $cont_special = ($rest =~ /^\s*$special/)? 1 : 0;
 958                 $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0;
 959                 $cont_case = 0;
 960                 next;
 961         }
 962         $cont_noerr = 0                 if (!$cont_in);
 963 }