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# Copyright 2015 Toomas Soome <tsoome@me.com> 23# 24# Copyright 2008 Sun Microsystems, Inc. All rights reserved. 25# Use is subject to license terms. 26# 27# Copyright (c) 2015 by Delphix. All rights reserved. 28# 29# @(#)cstyle 1.58 98/09/09 (from shannon) 30# 31# cstyle - check for some common stylistic errors. 32# 33# cstyle is a sort of "lint" for C coding style. 34# It attempts to check for the style used in the 35# kernel, sometimes known as "Bill Joy Normal Form". 36# 37# There's a lot this can't check for, like proper indentation 38# of code blocks. There's also a lot more this could check for. 39# 40# A note to the non perl literate: 41# 42# perl regular expressions are pretty much like egrep 43# regular expressions, with the following special symbols 44# 45# \s any space character 46# \S any non-space character 47# \w any "word" character [a-zA-Z0-9_] 48# \W any non-word character 49# \d a digit [0-9] 50# \D a non-digit 51# \b word boundary (between \w and \W) 52# \B non-word boundary 53# 54 55require 5.0; 56use IO::File; 57use Getopt::Std; 58use strict; 59 60my $usage = 61"usage: cstyle [-chpvCP] [-o constructs] file ... 62 -c check continuation indentation inside functions 63 -h perform heuristic checks that are sometimes wrong 64 -p perform some of the more picky checks 65 -v verbose 66 -C don't check anything in header block comments 67 -P check for use of non-POSIX types 68 -o constructs 69 allow a comma-seperated list of optional constructs: 70 doxygen allow doxygen-style block comments (/** /*!) 71 splint allow splint-style lint comments (/*@ ... @*/) 72"; 73 74my %opts; 75 76if (!getopts("cho:pvCP", \%opts)) { 77 print $usage; 78 exit 2; 79} 80 81my $check_continuation = $opts{'c'}; 82my $heuristic = $opts{'h'}; 83my $picky = $opts{'p'}; 84my $verbose = $opts{'v'}; 85my $ignore_hdr_comment = $opts{'C'}; 86my $check_posix_types = $opts{'P'}; 87 88my $doxygen_comments = 0; 89my $splint_comments = 0; 90 91if (defined($opts{'o'})) { 92 for my $x (split /,/, $opts{'o'}) { 93 if ($x eq "doxygen") { 94 $doxygen_comments = 1; 95 } elsif ($x eq "splint") { 96 $splint_comments = 1; 97 } else { 98 print "cstyle: unrecognized construct \"$x\"\n"; 99 print $usage; 100 exit 2; 101 } 102 } 103} 104 105my ($filename, $line, $prev); # shared globals 106 107my $fmt; 108my $hdr_comment_start; 109 110if ($verbose) { 111 $fmt = "%s: %d: %s\n%s\n"; 112} else { 113 $fmt = "%s: %d: %s\n"; 114} 115 116if ($doxygen_comments) { 117 # doxygen comments look like "/*!" or "/**"; allow them. 118 $hdr_comment_start = qr/^\s*\/\*[\!\*]?$/; 119} else { 120 $hdr_comment_start = qr/^\s*\/\*$/; 121} 122 123# Note, following must be in single quotes so that \s and \w work right. 124my $typename = '(int|char|short|long|unsigned|float|double' . 125 '|\w+_t|struct\s+\w+|union\s+\w+|FILE)'; 126 127# mapping of old types to POSIX compatible types 128my %old2posix = ( 129 'unchar' => 'uchar_t', 130 'ushort' => 'ushort_t', 131 'uint' => 'uint_t', 132 'ulong' => 'ulong_t', 133 'u_int' => 'uint_t', 134 'u_short' => 'ushort_t', 135 'u_long' => 'ulong_t', 136 'u_char' => 'uchar_t', 137 'quad' => 'quad_t' 138); 139 140my $lint_re = qr/\/\*(?: 141 ARGSUSED[0-9]*|NOTREACHED|LINTLIBRARY|VARARGS[0-9]*| 142 CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY| 143 FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*| 144 PROTOLIB[0-9]*|SCANFLIKE[0-9]*|CSTYLED.*? 145 )\*\//x; 146 147my $splint_re = qr/\/\*@.*?@\*\//x; 148 149my $warlock_re = qr/\/\*\s*(?: 150 VARIABLES\ PROTECTED\ BY| 151 MEMBERS\ PROTECTED\ BY| 152 ALL\ MEMBERS\ PROTECTED\ BY| 153 READ-ONLY\ VARIABLES:| 154 READ-ONLY\ MEMBERS:| 155 VARIABLES\ READABLE\ WITHOUT\ LOCK:| 156 MEMBERS\ READABLE\ WITHOUT\ LOCK:| 157 LOCKS\ COVERED\ BY| 158 LOCK\ UNNEEDED\ BECAUSE| 159 LOCK\ NEEDED:| 160 LOCK\ HELD\ ON\ ENTRY:| 161 READ\ LOCK\ HELD\ ON\ ENTRY:| 162 WRITE\ LOCK\ HELD\ ON\ ENTRY:| 163 LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 164 READ\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 165 WRITE\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 166 LOCK\ RELEASED\ AS\ SIDE\ EFFECT:| 167 LOCK\ UPGRADED\ AS\ SIDE\ EFFECT:| 168 LOCK\ DOWNGRADED\ AS\ SIDE\ EFFECT:| 169 FUNCTIONS\ CALLED\ THROUGH\ POINTER| 170 FUNCTIONS\ CALLED\ THROUGH\ MEMBER| 171 LOCK\ ORDER: 172 )/x; 173 174my $err_stat = 0; # exit status 175 176if ($#ARGV >= 0) { 177 foreach my $arg (@ARGV) { 178 my $fh = new IO::File $arg, "r"; 179 if (!defined($fh)) { 180 printf "%s: can not open\n", $arg; 181 } else { 182 &cstyle($arg, $fh); 183 close $fh; 184 } 185 } 186} else { 187 &cstyle("<stdin>", *STDIN); 188} 189exit $err_stat; 190 191my $no_errs = 0; # set for CSTYLED-protected lines 192 193sub err($) { 194 my ($error) = @_; 195 unless ($no_errs) { 196 if ($verbose) { 197 printf $fmt, $filename, $., $error, $line; 198 } else { 199 printf $fmt, $filename, $., $error; 200 } 201 $err_stat = 1; 202 } 203} 204 205sub err_prefix($$) { 206 my ($prevline, $error) = @_; 207 my $out = $prevline."\n".$line; 208 unless ($no_errs) { 209 if ($verbose) { 210 printf $fmt, $filename, $., $error, $out; 211 } else { 212 printf $fmt, $filename, $., $error; 213 } 214 $err_stat = 1; 215 } 216} 217 218sub err_prev($) { 219 my ($error) = @_; 220 unless ($no_errs) { 221 if ($verbose) { 222 printf $fmt, $filename, $. - 1, $error, $prev; 223 } else { 224 printf $fmt, $filename, $. - 1, $error; 225 } 226 $err_stat = 1; 227 } 228} 229 230sub cstyle($$) { 231 232my ($fn, $filehandle) = @_; 233$filename = $fn; # share it globally 234 235my $in_cpp = 0; 236my $next_in_cpp = 0; 237 238my $in_comment = 0; 239my $in_header_comment = 0; 240my $comment_done = 0; 241my $in_warlock_comment = 0; 242my $in_function = 0; 243my $in_function_header = 0; 244my $function_header_full_indent = 0; 245my $in_declaration = 0; 246my $note_level = 0; 247my $nextok = 0; 248my $nocheck = 0; 249 250my $in_string = 0; 251 252my ($okmsg, $comment_prefix); 253 254$line = ''; 255$prev = ''; 256reset_indent(); 257 258line: while (<$filehandle>) { 259 s/\r?\n$//; # strip return and newline 260 261 # save the original line, then remove all text from within 262 # double or single quotes, we do not want to check such text. 263 264 $line = $_; 265 266 # 267 # C allows strings to be continued with a backslash at the end of 268 # the line. We translate that into a quoted string on the previous 269 # line followed by an initial quote on the next line. 270 # 271 # (we assume that no-one will use backslash-continuation with character 272 # constants) 273 # 274 $_ = '"' . $_ if ($in_string && !$nocheck && !$in_comment); 275 276 # 277 # normal strings and characters 278 # 279 s/'([^\\']|\\[^xX0]|\\0[0-9]*|\\[xX][0-9a-fA-F]*)'/''/g; 280 s/"([^\\"]|\\.)*"/\"\"/g; 281 282 # 283 # detect string continuation 284 # 285 if ($nocheck || $in_comment) { 286 $in_string = 0; 287 } else { 288 # 289 # Now that all full strings are replaced with "", we check 290 # for unfinished strings continuing onto the next line. 291 # 292 $in_string = 293 (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ || 294 s/^("")*"([^\\"]|\\.)*\\$/""/); 295 } 296 297 # 298 # figure out if we are in a cpp directive 299 # 300 $in_cpp = $next_in_cpp || /^\s*#/; # continued or started 301 $next_in_cpp = $in_cpp && /\\$/; # only if continued 302 303 # strip off trailing backslashes, which appear in long macros 304 s/\s*\\$//; 305 306 # an /* END CSTYLED */ comment ends a no-check block. 307 if ($nocheck) { 308 if (/\/\* *END *CSTYLED *\*\//) { 309 $nocheck = 0; 310 } else { 311 reset_indent(); 312 next line; 313 } 314 } 315 316 # a /*CSTYLED*/ comment indicates that the next line is ok. 317 if ($nextok) { 318 if ($okmsg) { 319 err($okmsg); 320 } 321 $nextok = 0; 322 $okmsg = 0; 323 if (/\/\* *CSTYLED.*\*\//) { 324 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; 325 $okmsg = $1; 326 $nextok = 1; 327 } 328 $no_errs = 1; 329 } elsif ($no_errs) { 330 $no_errs = 0; 331 } 332 333 # check length of line. 334 # first, a quick check to see if there is any chance of being too long. 335 if (($line =~ tr/\t/\t/) * 7 + length($line) > 80) { 336 # yes, there is a chance. 337 # replace tabs with spaces and check again. 338 my $eline = $line; 339 1 while $eline =~ 340 s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e; 341 if (length($eline) > 80) { 342 err("line > 80 characters"); 343 } 344 } 345 346 # ignore NOTE(...) annotations (assumes NOTE is on lines by itself). 347 if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE 348 s/[^()]//g; # eliminate all non-parens 349 $note_level += s/\(//g - length; # update paren nest level 350 next; 351 } 352 353 # a /* BEGIN CSTYLED */ comment starts a no-check block. 354 if (/\/\* *BEGIN *CSTYLED *\*\//) { 355 $nocheck = 1; 356 } 357 358 # a /*CSTYLED*/ comment indicates that the next line is ok. 359 if (/\/\* *CSTYLED.*\*\//) { 360 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; 361 $okmsg = $1; 362 $nextok = 1; 363 } 364 if (/\/\/ *CSTYLED/) { 365 /^.*\/\/ *CSTYLED *(.*)$/; 366 $okmsg = $1; 367 $nextok = 1; 368 } 369 370 # universal checks; apply to everything 371 if (/\t +\t/) { 372 err("spaces between tabs"); 373 } 374 if (/ \t+ /) { 375 err("tabs between spaces"); 376 } 377 if (/\s$/) { 378 err("space or tab at end of line"); 379 } 380 if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) { 381 err("comment preceded by non-blank"); 382 } 383 384 # is this the beginning or ending of a function? 385 # (not if "struct foo\n{\n") 386 if (/^{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) { 387 $in_function = 1; 388 $in_declaration = 1; 389 $in_function_header = 0; 390 $function_header_full_indent = 0; 391 $prev = $line; 392 next line; 393 } 394 if (/^}\s*(\/\*.*\*\/\s*)*$/) { 395 if ($prev =~ /^\s*return\s*;/) { 396 err_prev("unneeded return at end of function"); 397 } 398 $in_function = 0; 399 reset_indent(); # we don't check between functions 400 $prev = $line; 401 next line; 402 } 403 if ($in_function_header && ! /^ \w/ ) { 404 if (/^{}$/) { 405 $in_function_header = 0; 406 $function_header_full_indent = 0; 407 } elsif ($picky && ! (/^\t/ && $function_header_full_indent != 0)) { 408 err("continuation line should be indented by 4 spaces"); 409 } 410 } 411 412 # 413 # If this matches something of form "foo(", it's probably a function 414 # definition, unless it ends with ") bar;", in which case it's a declaration 415 # that uses a macro to generate the type. 416 # 417 if (/^\w+\(/ && !/\) \w+;$/) { 418 $in_function_header = 1; 419 if (/\($/) { 420 $function_header_full_indent = 1; 421 } 422 } 423 if ($in_function_header && /^{$/) { 424 $in_function_header = 0; 425 $function_header_full_indent = 0; 426 $in_function = 1; 427 } 428 if ($in_function_header && /\);$/) { 429 $in_function_header = 0; 430 $function_header_full_indent = 0; 431 } 432 if ($in_function_header && /{$/ ) { 433 if ($picky) { 434 err("opening brace on same line as function header"); 435 } 436 $in_function_header = 0; 437 $function_header_full_indent = 0; 438 $in_function = 1; 439 next line; 440 } 441 442 if ($in_warlock_comment && /\*\//) { 443 $in_warlock_comment = 0; 444 $prev = $line; 445 next line; 446 } 447 448 # a blank line terminates the declarations within a function. 449 # XXX - but still a problem in sub-blocks. 450 if ($in_declaration && /^$/) { 451 $in_declaration = 0; 452 } 453 454 if ($comment_done) { 455 $in_comment = 0; 456 $in_header_comment = 0; 457 $comment_done = 0; 458 } 459 # does this looks like the start of a block comment? 460 if (/$hdr_comment_start/) { 461 if (!/^\t*\/\*/) { 462 err("block comment not indented by tabs"); 463 } 464 $in_comment = 1; 465 /^(\s*)\//; 466 $comment_prefix = $1; 467 if ($comment_prefix eq "") { 468 $in_header_comment = 1; 469 } 470 $prev = $line; 471 next line; 472 } 473 # are we still in the block comment? 474 if ($in_comment) { 475 if (/^$comment_prefix \*\/$/) { 476 $comment_done = 1; 477 } elsif (/\*\//) { 478 $comment_done = 1; 479 err("improper block comment close") 480 unless ($ignore_hdr_comment && $in_header_comment); 481 } elsif (!/^$comment_prefix \*[ \t]/ && 482 !/^$comment_prefix \*$/) { 483 err("improper block comment") 484 unless ($ignore_hdr_comment && $in_header_comment); 485 } 486 } 487 488 if ($in_header_comment && $ignore_hdr_comment) { 489 $prev = $line; 490 next line; 491 } 492 493 # check for errors that might occur in comments and in code. 494 495 # allow spaces to be used to draw pictures in header comments. 496 if (/[^ ] / && !/".* .*"/ && !$in_header_comment) { 497 err("spaces instead of tabs"); 498 } 499 if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ && 500 (!/^ \w/ || $in_function != 0)) { 501 err("indent by spaces instead of tabs"); 502 } 503 if (/^\t+ [^ \t\*]/ || /^\t+ \S/ || /^\t+ \S/) { 504 err("continuation line not indented by 4 spaces"); 505 } 506 if (/$warlock_re/ && !/\*\//) { 507 $in_warlock_comment = 1; 508 $prev = $line; 509 next line; 510 } 511 if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) { 512 err("improper first line of block comment"); 513 } 514 515 if ($in_comment) { # still in comment, don't do further checks 516 $prev = $line; 517 next line; 518 } 519 520 if ((/[^(]\/\*\S/ || /^\/\*\S/) && 521 !(/$lint_re/ || ($splint_comments && /$splint_re/))) { 522 err("missing blank after open comment"); 523 } 524 if (/\S\*\/[^)]|\S\*\/$/ && 525 !(/$lint_re/ || ($splint_comments && /$splint_re/))) { 526 err("missing blank before close comment"); 527 } 528 if (/\/\/\S/) { # C++ comments 529 err("missing blank after start comment"); 530 } 531 # check for unterminated single line comments, but allow them when 532 # they are used to comment out the argument list of a function 533 # declaration. 534 if (/\S.*\/\*/ && !/\S.*\/\*.*\*\// && !/\(\/\*/) { 535 err("unterminated single line comment"); 536 } 537 538 if (/^(#else|#endif|#include)(.*)$/) { 539 $prev = $line; 540 if ($picky) { 541 my $directive = $1; 542 my $clause = $2; 543 # Enforce ANSI rules for #else and #endif: no noncomment 544 # identifiers are allowed after #endif or #else. Allow 545 # C++ comments since they seem to be a fact of life. 546 if ((($1 eq "#endif") || ($1 eq "#else")) && 547 ($clause ne "") && 548 (!($clause =~ /^\s+\/\*.*\*\/$/)) && 549 (!($clause =~ /^\s+\/\/.*$/))) { 550 err("non-comment text following " . 551 "$directive (or malformed $directive " . 552 "directive)"); 553 } 554 } 555 next line; 556 } 557 558 # 559 # delete any comments and check everything else. Note that 560 # ".*?" is a non-greedy match, so that we don't get confused by 561 # multiple comments on the same line. 562 # 563 s/\/\*.*?\*\///g; 564 s/\/\/.*$//; # C++ comments 565 566 # delete any trailing whitespace; we have already checked for that. 567 s/\s*$//; 568 569 # following checks do not apply to text in comments. 570 571 if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=[^\s,]/ || 572 (/[^->]>[^,=>\s]/ && !/[^->]>$/) || 573 (/[^<]<[^,=<\s]/ && !/[^<]<$/) || 574 /[^<\s]<[^<]/ || /[^->\s]>[^>]/) { 575 err("missing space around relational operator"); 576 } 577 if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ || 578 (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) || 579 (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) { 580 # XXX - should only check this for C++ code 581 # XXX - there are probably other forms that should be allowed 582 if (!/\soperator=/) { 583 err("missing space around assignment operator"); 584 } 585 } 586 if (/[,;]\S/ && !/\bfor \(;;\)/) { 587 err("comma or semicolon followed by non-blank"); 588 } 589 # allow "for" statements to have empty "while" clauses 590 if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) { 591 err("comma or semicolon preceded by blank"); 592 } 593 if (/^\s*(&&|\|\|)/) { 594 err("improper boolean continuation"); 595 } 596 if (/\S *(&&|\|\|)/ || /(&&|\|\|) *\S/) { 597 err("more than one space around boolean operator"); 598 } 599 if (/\b(for|if|while|switch|sizeof|return|case)\(/) { 600 err("missing space between keyword and paren"); 601 } 602 if (/(\b(for|if|while|switch|return)\b.*){2,}/ && !/^#define/) { 603 # multiple "case" and "sizeof" allowed 604 err("more than one keyword on line"); 605 } 606 if (/\b(for|if|while|switch|sizeof|return|case)\s\s+\(/ && 607 !/^#if\s+\(/) { 608 err("extra space between keyword and paren"); 609 } 610 # try to detect "func (x)" but not "if (x)" or 611 # "#define foo (x)" or "int (*func)();" 612 if (/\w\s\(/) { 613 my $s = $_; 614 # strip off all keywords on the line 615 s/\b(for|if|while|switch|return|case|sizeof)\s\(/XXX(/g; 616 s/#elif\s\(/XXX(/g; 617 s/^#define\s+\w+\s+\(/XXX(/; 618 # do not match things like "void (*f)();" 619 # or "typedef void (func_t)();" 620 s/\w\s\(+\*/XXX(*/g; 621 s/\b($typename|void)\s+\(+/XXX(/og; 622 if (/\w\s\(/) { 623 err("extra space between function name and left paren"); 624 } 625 $_ = $s; 626 } 627 # try to detect "int foo(x)", but not "extern int foo(x);" 628 # XXX - this still trips over too many legitimate things, 629 # like "int foo(x,\n\ty);" 630# if (/^(\w+(\s|\*)+)+\w+\(/ && !/\)[;,](\s|)*$/ && 631# !/^(extern|static)\b/) { 632# err("return type of function not on separate line"); 633# } 634 # this is a close approximation 635 if (/^(\w+(\s|\*)+)+\w+\(.*\)(\s|)*$/ && 636 !/^(extern|static)\b/) { 637 err("return type of function not on separate line"); 638 } 639 if (/^#define /) { 640 err("#define followed by space instead of tab"); 641 } 642 if (/^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) { 643 err("unparenthesized return expression"); 644 } 645 if (/\bsizeof\b/ && !/\bsizeof\s*\(.*\)/) { 646 err("unparenthesized sizeof expression"); 647 } 648 if (/\(\s/) { 649 err("whitespace after left paren"); 650 } 651 # allow "for" statements to have empty "continue" clauses 652 if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/) { 653 err("whitespace before right paren"); 654 } 655 if (/^\s*\(void\)[^ ]/) { 656 err("missing space after (void) cast"); 657 } 658 if (/\S\{/ && !/\{\{/) { 659 err("missing space before left brace"); 660 } 661 if ($in_function && /^\s+{/ && 662 ($prev =~ /\)\s*$/ || $prev =~ /\bstruct\s+\w+$/)) { 663 err("left brace starting a line"); 664 } 665 if (/}(else|while)/) { 666 err("missing space after right brace"); 667 } 668 if (/}\s\s+(else|while)/) { 669 err("extra space after right brace"); 670 } 671 if (/\b_VOID\b|\bVOID\b|\bSTATIC\b/) { 672 err("obsolete use of VOID or STATIC"); 673 } 674 if (/\b$typename\*/o) { 675 err("missing space between type name and *"); 676 } 677 if (/^\s+#/) { 678 err("preprocessor statement not in column 1"); 679 } 680 if (/^#\s/) { 681 err("blank after preprocessor #"); 682 } 683 if (/!\s*(strcmp|strncmp|bcmp)\s*\(/) { 684 err("don't use boolean ! with comparison functions"); 685 } 686 687 # 688 # We completely ignore, for purposes of indentation: 689 # * lines outside of functions 690 # * preprocessor lines 691 # 692 if ($check_continuation && $in_function && !$in_cpp) { 693 process_indent($_); 694 } 695 if ($picky) { 696 # try to detect spaces after casts, but allow (e.g.) 697 # "sizeof (int) + 1", "void (*funcptr)(int) = foo;", and 698 # "int foo(int) __NORETURN;" 699 if ((/^\($typename( \*+)?\)\s/o || 700 /\W\($typename( \*+)?\)\s/o) && 701 !/sizeof\s*\($typename( \*)?\)\s/o && 702 !/\($typename( \*+)?\)\s+=[^=]/o) { 703 err("space after cast"); 704 } 705 if (/\b$typename\s*\*\s/o && 706 !/\b$typename\s*\*\s+const\b/o) { 707 err("unary * followed by space"); 708 } 709 } 710 if ($check_posix_types) { 711 # try to detect old non-POSIX types. 712 # POSIX requires all non-standard typedefs to end in _t, 713 # but historically these have been used. 714 if (/\b(unchar|ushort|uint|ulong|u_int|u_short|u_long|u_char|quad)\b/) { 715 err("non-POSIX typedef $1 used: use $old2posix{$1} instead"); 716 } 717 } 718 if ($heuristic) { 719 # cannot check this everywhere due to "struct {\n...\n} foo;" 720 if ($in_function && !$in_declaration && 721 /}./ && !/}\s+=/ && !/{.*}[;,]$/ && !/}(\s|)*$/ && 722 !/} (else|while)/ && !/}}/) { 723 err("possible bad text following right brace"); 724 } 725 # cannot check this because sub-blocks in 726 # the middle of code are ok 727 if ($in_function && /^\s+{/) { 728 err("possible left brace starting a line"); 729 } 730 } 731 if (/^\s*else\W/) { 732 if ($prev =~ /^\s*}$/) { 733 err_prefix($prev, 734 "else and right brace should be on same line"); 735 } 736 } 737 $prev = $line; 738} 739 740if ($prev eq "") { 741 err("last line in file is blank"); 742} 743 744} 745 746# 747# Continuation-line checking 748# 749# The rest of this file contains the code for the continuation checking 750# engine. It's a pretty simple state machine which tracks the expression 751# depth (unmatched '('s and '['s). 752# 753# Keep in mind that the argument to process_indent() has already been heavily 754# processed; all comments have been replaced by control-A, and the contents of 755# strings and character constants have been elided. 756# 757 758my $cont_in; # currently inside of a continuation 759my $cont_off; # skipping an initializer or definition 760my $cont_noerr; # suppress cascading errors 761my $cont_start; # the line being continued 762my $cont_base; # the base indentation 763my $cont_first; # this is the first line of a statement 764my $cont_multiseg; # this continuation has multiple segments 765 766my $cont_special; # this is a C statement (if, for, etc.) 767my $cont_macro; # this is a macro 768my $cont_case; # this is a multi-line case 769 770my @cont_paren; # the stack of unmatched ( and [s we've seen 771 772sub 773reset_indent() 774{ 775 $cont_in = 0; 776 $cont_off = 0; 777} 778 779sub 780delabel($) 781{ 782 # 783 # replace labels with tabs. Note that there may be multiple 784 # labels on a line. 785 # 786 local $_ = $_[0]; 787 788 while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) { 789 my ($pre_tabs, $label, $rest) = ($1, $2, $3); 790 $_ = $pre_tabs; 791 while ($label =~ s/^([^\t]*)(\t+)//) { 792 $_ .= "\t" x (length($2) + length($1) / 8); 793 } 794 $_ .= ("\t" x (length($label) / 8)).$rest; 795 } 796 797 return ($_); 798} 799 800sub 801process_indent($) 802{ 803 require strict; 804 local $_ = $_[0]; # preserve the global $_ 805 806 s///g; # No comments 807 s/\s+$//; # Strip trailing whitespace 808 809 return if (/^$/); # skip empty lines 810 811 # regexps used below; keywords taking (), macros, and continued cases 812 my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b'; 813 my $macro = '[A-Z_][A-Z_0-9]*\('; 814 my $case = 'case\b[^:]*$'; 815 816 # skip over enumerations, array definitions, initializers, etc. 817 if ($cont_off <= 0 && !/^\s*$special/ && 818 (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*)){/ || 819 (/^\s*{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) { 820 $cont_in = 0; 821 $cont_off = tr/{/{/ - tr/}/}/; 822 return; 823 } 824 if ($cont_off) { 825 $cont_off += tr/{/{/ - tr/}/}/; 826 return; 827 } 828 829 if (!$cont_in) { 830 $cont_start = $line; 831 832 if (/^\t* /) { 833 err("non-continuation indented 4 spaces"); 834 $cont_noerr = 1; # stop reporting 835 } 836 $_ = delabel($_); # replace labels with tabs 837 838 # check if the statement is complete 839 return if (/^\s*\}?$/); 840 return if (/^\s*\}?\s*else\s*\{?$/); 841 return if (/^\s*do\s*\{?$/); 842 return if (/{$/); 843 return if (/}[,;]?$/); 844 845 # Allow macros on their own lines 846 return if (/^\s*[A-Z_][A-Z_0-9]*$/); 847 848 # cases we don't deal with, generally non-kosher 849 if (/{/) { 850 err("stuff after {"); 851 return; 852 } 853 854 # Get the base line, and set up the state machine 855 /^(\t*)/; 856 $cont_base = $1; 857 $cont_in = 1; 858 @cont_paren = (); 859 $cont_first = 1; 860 $cont_multiseg = 0; 861 862 # certain things need special processing 863 $cont_special = /^\s*$special/? 1 : 0; 864 $cont_macro = /^\s*$macro/? 1 : 0; 865 $cont_case = /^\s*$case/? 1 : 0; 866 } else { 867 $cont_first = 0; 868 869 # Strings may be pulled back to an earlier (half-)tabstop 870 unless ($cont_noerr || /^$cont_base / || 871 (/^\t*(?: )?(?:gettext\()?\"/ && !/^$cont_base\t/)) { 872 err_prefix($cont_start, 873 "continuation should be indented 4 spaces"); 874 } 875 } 876 877 my $rest = $_; # keeps the remainder of the line 878 879 # 880 # The split matches 0 characters, so that each 'special' character 881 # is processed separately. Parens and brackets are pushed and 882 # popped off the @cont_paren stack. For normal processing, we wait 883 # until a ; or { terminates the statement. "special" processing 884 # (if/for/while/switch) is allowed to stop when the stack empties, 885 # as is macro processing. Case statements are terminated with a : 886 # and an empty paren stack. 887 # 888 foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) { 889 next if (length($_) == 0); 890 891 # rest contains the remainder of the line 892 my $rxp = "[^\Q$_\E]*\Q$_\E"; 893 $rest =~ s/^$rxp//; 894 895 if (/\(/ || /\[/) { 896 push @cont_paren, $_; 897 } elsif (/\)/ || /\]/) { 898 my $cur = $_; 899 tr/\)\]/\(\[/; 900 901 my $old = (pop @cont_paren); 902 if (!defined($old)) { 903 err("unexpected '$cur'"); 904 $cont_in = 0; 905 last; 906 } elsif ($old ne $_) { 907 err("'$cur' mismatched with '$old'"); 908 $cont_in = 0; 909 last; 910 } 911 912 # 913 # If the stack is now empty, do special processing 914 # for if/for/while/switch and macro statements. 915 # 916 next if (@cont_paren != 0); 917 if ($cont_special) { 918 if ($rest =~ /^\s*{?$/) { 919 $cont_in = 0; 920 last; 921 } 922 if ($rest =~ /^\s*;$/) { 923 err("empty if/for/while body ". 924 "not on its own line"); 925 $cont_in = 0; 926 last; 927 } 928 if (!$cont_first && $cont_multiseg == 1) { 929 err_prefix($cont_start, 930 "multiple statements continued ". 931 "over multiple lines"); 932 $cont_multiseg = 2; 933 } elsif ($cont_multiseg == 0) { 934 $cont_multiseg = 1; 935 } 936 # We've finished this section, start 937 # processing the next. 938 goto section_ended; 939 } 940 if ($cont_macro) { 941 if ($rest =~ /^$/) { 942 $cont_in = 0; 943 last; 944 } 945 } 946 } elsif (/\;/) { 947 if ($cont_case) { 948 err("unexpected ;"); 949 } elsif (!$cont_special) { 950 err("unexpected ;") if (@cont_paren != 0); 951 if (!$cont_first && $cont_multiseg == 1) { 952 err_prefix($cont_start, 953 "multiple statements continued ". 954 "over multiple lines"); 955 $cont_multiseg = 2; 956 } elsif ($cont_multiseg == 0) { 957 $cont_multiseg = 1; 958 } 959 if ($rest =~ /^$/) { 960 $cont_in = 0; 961 last; 962 } 963 if ($rest =~ /^\s*special/) { 964 err("if/for/while/switch not started ". 965 "on its own line"); 966 } 967 goto section_ended; 968 } 969 } elsif (/\{/) { 970 err("{ while in parens/brackets") if (@cont_paren != 0); 971 err("stuff after {") if ($rest =~ /[^\s}]/); 972 $cont_in = 0; 973 last; 974 } elsif (/\}/) { 975 err("} while in parens/brackets") if (@cont_paren != 0); 976 if (!$cont_special && $rest !~ /^\s*(while|else)\b/) { 977 if ($rest =~ /^$/) { 978 err("unexpected }"); 979 } else { 980 err("stuff after }"); 981 } 982 $cont_in = 0; 983 last; 984 } 985 } elsif (/\:/ && $cont_case && @cont_paren == 0) { 986 err("stuff after multi-line case") if ($rest !~ /$^/); 987 $cont_in = 0; 988 last; 989 } 990 next; 991section_ended: 992 # End of a statement or if/while/for loop. Reset 993 # cont_special and cont_macro based on the rest of the 994 # line. 995 $cont_special = ($rest =~ /^\s*$special/)? 1 : 0; 996 $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0; 997 $cont_case = 0; 998 next; 999 } 1000 $cont_noerr = 0 if (!$cont_in); 1001} 1002