1#!/usr/bin/env perl 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 2016 Nexenta Systems, Inc. 23# 24# Copyright 2008 Sun Microsystems, Inc. All rights reserved. 25# Use is subject to license terms. 26# 27# @(#)cstyle 1.58 98/09/09 (from shannon) 28#ident "%Z%%M% %I% %E% SMI" 29# 30# cstyle - check for some common stylistic errors. 31# 32# cstyle is a sort of "lint" for C coding style. 33# It attempts to check for the style used in the 34# kernel, sometimes known as "Bill Joy Normal Form". 35# 36# There's a lot this can't check for, like proper indentation 37# of code blocks. There's also a lot more this could check for. 38# 39# A note to the non perl literate: 40# 41# perl regular expressions are pretty much like egrep 42# regular expressions, with the following special symbols 43# 44# \s any space character 45# \S any non-space character 46# \w any "word" character [a-zA-Z0-9_] 47# \W any non-word character 48# \d a digit [0-9] 49# \D a non-digit 50# \b word boundary (between \w and \W) 51# \B non-word boundary 52# 53 54require 5.0; 55use warnings; 56use IO::File; 57use Getopt::Std; 58use strict; 59 60my $usage = 61"usage: cstyle [-cghpvCP] [-o constructs] file ... 62 -c check continuation indentation inside functions 63 -g print github actions' workflow commands 64 -h perform heuristic checks that are sometimes wrong 65 -p perform some of the more picky checks 66 -v verbose 67 -C don't check anything in header block comments 68 -P check for use of non-POSIX types 69 -o constructs 70 allow a comma-separated list of optional constructs: 71 doxygen allow doxygen-style block comments (/** /*!) 72 splint allow splint-style lint comments (/*@ ... @*/) 73"; 74 75my %opts; 76 77if (!getopts("cgho:pvCP", \%opts)) { 78 print $usage; 79 exit 2; 80} 81 82my $check_continuation = $opts{'c'}; 83my $github_workflow = $opts{'g'} || $ENV{'CI'}; 84my $heuristic = $opts{'h'}; 85my $picky = $opts{'p'}; 86my $verbose = $opts{'v'}; 87my $ignore_hdr_comment = $opts{'C'}; 88my $check_posix_types = $opts{'P'}; 89 90my $doxygen_comments = 0; 91my $splint_comments = 0; 92 93if (defined($opts{'o'})) { 94 for my $x (split /,/, $opts{'o'}) { 95 if ($x eq "doxygen") { 96 $doxygen_comments = 1; 97 } elsif ($x eq "splint") { 98 $splint_comments = 1; 99 } else { 100 print "cstyle: unrecognized construct \"$x\"\n"; 101 print $usage; 102 exit 2; 103 } 104 } 105} 106 107my ($filename, $line, $prev); # shared globals 108 109my $fmt; 110my $hdr_comment_start; 111 112if ($verbose) { 113 $fmt = "%s: %d: %s\n%s\n"; 114} else { 115 $fmt = "%s: %d: %s\n"; 116} 117 118if ($doxygen_comments) { 119 # doxygen comments look like "/*!" or "/**"; allow them. 120 $hdr_comment_start = qr/^\s*\/\*[\!\*]?$/; 121} else { 122 $hdr_comment_start = qr/^\s*\/\*$/; 123} 124 125# Note, following must be in single quotes so that \s and \w work right. 126my $typename = '(int|char|short|long|unsigned|float|double' . 127 '|\w+_t|struct\s+\w+|union\s+\w+|FILE)'; 128 129# mapping of old types to POSIX compatible types 130my %old2posix = ( 131 'unchar' => 'uchar_t', 132 'ushort' => 'ushort_t', 133 'uint' => 'uint_t', 134 'ulong' => 'ulong_t', 135 'u_int' => 'uint_t', 136 'u_short' => 'ushort_t', 137 'u_long' => 'ulong_t', 138 'u_char' => 'uchar_t', 139 'quad' => 'quad_t' 140); 141 142my $lint_re = qr/\/\*(?: 143 ARGSUSED[0-9]*|NOTREACHED|LINTLIBRARY|VARARGS[0-9]*| 144 CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY| 145 FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*| 146 PROTOLIB[0-9]*|SCANFLIKE[0-9]*|CSTYLED.*? 147 )\*\//x; 148 149my $splint_re = qr/\/\*@.*?@\*\//x; 150 151my $warlock_re = qr/\/\*\s*(?: 152 VARIABLES\ PROTECTED\ BY| 153 MEMBERS\ PROTECTED\ BY| 154 ALL\ MEMBERS\ PROTECTED\ BY| 155 READ-ONLY\ VARIABLES:| 156 READ-ONLY\ MEMBERS:| 157 VARIABLES\ READABLE\ WITHOUT\ LOCK:| 158 MEMBERS\ READABLE\ WITHOUT\ LOCK:| 159 LOCKS\ COVERED\ BY| 160 LOCK\ UNNEEDED\ BECAUSE| 161 LOCK\ NEEDED:| 162 LOCK\ HELD\ ON\ ENTRY:| 163 READ\ LOCK\ HELD\ ON\ ENTRY:| 164 WRITE\ LOCK\ HELD\ ON\ ENTRY:| 165 LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 166 READ\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 167 WRITE\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 168 LOCK\ RELEASED\ AS\ SIDE\ EFFECT:| 169 LOCK\ UPGRADED\ AS\ SIDE\ EFFECT:| 170 LOCK\ DOWNGRADED\ AS\ SIDE\ EFFECT:| 171 FUNCTIONS\ CALLED\ THROUGH\ POINTER| 172 FUNCTIONS\ CALLED\ THROUGH\ MEMBER| 173 LOCK\ ORDER: 174 )/x; 175 176my $err_stat = 0; # exit status 177 178if ($#ARGV >= 0) { 179 foreach my $arg (@ARGV) { 180 my $fh = new IO::File $arg, "r"; 181 if (!defined($fh)) { 182 printf "%s: can not open\n", $arg; 183 } else { 184 &cstyle($arg, $fh); 185 close $fh; 186 } 187 } 188} else { 189 &cstyle("<stdin>", *STDIN); 190} 191exit $err_stat; 192 193my $no_errs = 0; # set for CSTYLED-protected lines 194 195sub err($) { 196 my ($error) = @_; 197 unless ($no_errs) { 198 if ($verbose) { 199 printf $fmt, $filename, $., $error, $line; 200 } else { 201 printf $fmt, $filename, $., $error; 202 } 203 if ($github_workflow) { 204 printf "::error file=%s,line=%s::%s\n", $filename, $., $error; 205 } 206 $err_stat = 1; 207 } 208} 209 210sub err_prefix($$) { 211 my ($prevline, $error) = @_; 212 my $out = $prevline."\n".$line; 213 unless ($no_errs) { 214 if ($verbose) { 215 printf $fmt, $filename, $., $error, $out; 216 } else { 217 printf $fmt, $filename, $., $error; 218 } 219 $err_stat = 1; 220 } 221} 222 223sub err_prev($) { 224 my ($error) = @_; 225 unless ($no_errs) { 226 if ($verbose) { 227 printf $fmt, $filename, $. - 1, $error, $prev; 228 } else { 229 printf $fmt, $filename, $. - 1, $error; 230 } 231 $err_stat = 1; 232 } 233} 234 235sub cstyle($$) { 236 237my ($fn, $filehandle) = @_; 238$filename = $fn; # share it globally 239 240my $in_cpp = 0; 241my $next_in_cpp = 0; 242 243my $in_comment = 0; 244my $in_header_comment = 0; 245my $comment_done = 0; 246my $in_warlock_comment = 0; 247my $in_function = 0; 248my $in_function_header = 0; 249my $function_header_full_indent = 0; 250my $in_declaration = 0; 251my $note_level = 0; 252my $nextok = 0; 253my $nocheck = 0; 254 255my $in_string = 0; 256 257my ($okmsg, $comment_prefix); 258 259$line = ''; 260$prev = ''; 261reset_indent(); 262 263line: while (<$filehandle>) { 264 s/\r?\n$//; # strip return and newline 265 266 # save the original line, then remove all text from within 267 # double or single quotes, we do not want to check such text. 268 269 $line = $_; 270 271 # 272 # C allows strings to be continued with a backslash at the end of 273 # the line. We translate that into a quoted string on the previous 274 # line followed by an initial quote on the next line. 275 # 276 # (we assume that no-one will use backslash-continuation with character 277 # constants) 278 # 279 $_ = '"' . $_ if ($in_string && !$nocheck && !$in_comment); 280 281 # 282 # normal strings and characters 283 # 284 s/'([^\\']|\\[^xX0]|\\0[0-9]*|\\[xX][0-9a-fA-F]*)'/''/g; 285 s/"([^\\"]|\\.)*"/\"\"/g; 286 287 # 288 # detect string continuation 289 # 290 if ($nocheck || $in_comment) { 291 $in_string = 0; 292 } else { 293 # 294 # Now that all full strings are replaced with "", we check 295 # for unfinished strings continuing onto the next line. 296 # 297 $in_string = 298 (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ || 299 s/^("")*"([^\\"]|\\.)*\\$/""/); 300 } 301 302 # 303 # figure out if we are in a cpp directive 304 # 305 $in_cpp = $next_in_cpp || /^\s*#/; # continued or started 306 $next_in_cpp = $in_cpp && /\\$/; # only if continued 307 308 # strip off trailing backslashes, which appear in long macros 309 s/\s*\\$//; 310 311 # an /* END CSTYLED */ comment ends a no-check block. 312 if ($nocheck) { 313 if (/\/\* *END *CSTYLED *\*\//) { 314 $nocheck = 0; 315 } else { 316 reset_indent(); 317 next line; 318 } 319 } 320 321 # a /*CSTYLED*/ comment indicates that the next line is ok. 322 if ($nextok) { 323 if ($okmsg) { 324 err($okmsg); 325 } 326 $nextok = 0; 327 $okmsg = 0; 328 if (/\/\* *CSTYLED.*\*\//) { 329 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; 330 $okmsg = $1; 331 $nextok = 1; 332 } 333 $no_errs = 1; 334 } elsif ($no_errs) { 335 $no_errs = 0; 336 } 337 338 # check length of line. 339 # first, a quick check to see if there is any chance of being too long. 340 if (($line =~ tr/\t/\t/) * 7 + length($line) > 80) { 341 # yes, there is a chance. 342 # replace tabs with spaces and check again. 343 my $eline = $line; 344 1 while $eline =~ 345 s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e; 346 if (length($eline) > 80) { 347 err("line > 80 characters"); 348 } 349 } 350 351 # ignore NOTE(...) annotations (assumes NOTE is on lines by itself). 352 if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE 353 s/[^()]//g; # eliminate all non-parens 354 $note_level += s/\(//g - length; # update paren nest level 355 next; 356 } 357 358 # a /* BEGIN CSTYLED */ comment starts a no-check block. 359 if (/\/\* *BEGIN *CSTYLED *\*\//) { 360 $nocheck = 1; 361 } 362 363 # a /*CSTYLED*/ comment indicates that the next line is ok. 364 if (/\/\* *CSTYLED.*\*\//) { 365 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; 366 $okmsg = $1; 367 $nextok = 1; 368 } 369 if (/\/\/ *CSTYLED/) { 370 /^.*\/\/ *CSTYLED *(.*)$/; 371 $okmsg = $1; 372 $nextok = 1; 373 } 374 375 # universal checks; apply to everything 376 if (/\t +\t/) { 377 err("spaces between tabs"); 378 } 379 if (/ \t+ /) { 380 err("tabs between spaces"); 381 } 382 if (/\s$/) { 383 err("space or tab at end of line"); 384 } 385 if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) { 386 err("comment preceded by non-blank"); 387 } 388 389 # is this the beginning or ending of a function? 390 # (not if "struct foo\n{\n") 391 if (/^\{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) { 392 $in_function = 1; 393 $in_declaration = 1; 394 $in_function_header = 0; 395 $function_header_full_indent = 0; 396 $prev = $line; 397 next line; 398 } 399 if (/^\}\s*(\/\*.*\*\/\s*)*$/) { 400 if ($prev =~ /^\s*return\s*;/) { 401 err_prev("unneeded return at end of function"); 402 } 403 $in_function = 0; 404 reset_indent(); # we don't check between functions 405 $prev = $line; 406 next line; 407 } 408 if ($in_function_header && ! /^ (\w|\.)/ ) { 409 if (/^\{\}$/ # empty functions 410 || /;/ #run function with multiline arguments 411 || /#/ #preprocessor commands 412 || /^[^\s\\]*\(.*\)$/ #functions without ; at the end 413 || /^$/ #function declaration can't have empty line 414 ) { 415 $in_function_header = 0; 416 $function_header_full_indent = 0; 417 } elsif ($prev =~ /^__attribute__/) { #__attribute__((*)) 418 $in_function_header = 0; 419 $function_header_full_indent = 0; 420 $prev = $line; 421 next line; 422 } elsif ($picky && ! (/^\t/ && $function_header_full_indent != 0)) { 423 424 err("continuation line should be indented by 4 spaces"); 425 } 426 } 427 428 # 429 # If this matches something of form "foo(", it's probably a function 430 # definition, unless it ends with ") bar;", in which case it's a declaration 431 # that uses a macro to generate the type. 432 # 433 if (/^\w+\(/ && !/\) \w+;/) { 434 $in_function_header = 1; 435 if (/\($/) { 436 $function_header_full_indent = 1; 437 } 438 } 439 if ($in_function_header && /^\{$/) { 440 $in_function_header = 0; 441 $function_header_full_indent = 0; 442 $in_function = 1; 443 } 444 if ($in_function_header && /\);$/) { 445 $in_function_header = 0; 446 $function_header_full_indent = 0; 447 } 448 if ($in_function_header && /\{$/ ) { 449 if ($picky) { 450 err("opening brace on same line as function header"); 451 } 452 $in_function_header = 0; 453 $function_header_full_indent = 0; 454 $in_function = 1; 455 next line; 456 } 457 458 if ($in_warlock_comment && /\*\//) { 459 $in_warlock_comment = 0; 460 $prev = $line; 461 next line; 462 } 463 464 # a blank line terminates the declarations within a function. 465 # XXX - but still a problem in sub-blocks. 466 if ($in_declaration && /^$/) { 467 $in_declaration = 0; 468 } 469 470 if ($comment_done) { 471 $in_comment = 0; 472 $in_header_comment = 0; 473 $comment_done = 0; 474 } 475 # does this looks like the start of a block comment? 476 if (/$hdr_comment_start/) { 477 if (!/^\t*\/\*/) { 478 err("block comment not indented by tabs"); 479 } 480 $in_comment = 1; 481 /^(\s*)\//; 482 $comment_prefix = $1; 483 if ($comment_prefix eq "") { 484 $in_header_comment = 1; 485 } 486 $prev = $line; 487 next line; 488 } 489 # are we still in the block comment? 490 if ($in_comment) { 491 if (/^$comment_prefix \*\/$/) { 492 $comment_done = 1; 493 } elsif (/\*\//) { 494 $comment_done = 1; 495 err("improper block comment close") 496 unless ($ignore_hdr_comment && $in_header_comment); 497 } elsif (!/^$comment_prefix \*[ \t]/ && 498 !/^$comment_prefix \*$/) { 499 err("improper block comment") 500 unless ($ignore_hdr_comment && $in_header_comment); 501 } 502 } 503 504 if ($in_header_comment && $ignore_hdr_comment) { 505 $prev = $line; 506 next line; 507 } 508 509 # check for errors that might occur in comments and in code. 510 511 # allow spaces to be used to draw pictures in all comments. 512 if (/[^ ] / && !/".* .*"/ && !$in_comment) { 513 err("spaces instead of tabs"); 514 } 515 if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ && 516 (!/^ (\w|\.)/ || $in_function != 0)) { 517 err("indent by spaces instead of tabs"); 518 } 519 if (/^\t+ [^ \t\*]/ || /^\t+ \S/ || /^\t+ \S/) { 520 err("continuation line not indented by 4 spaces"); 521 } 522 if (/$warlock_re/ && !/\*\//) { 523 $in_warlock_comment = 1; 524 $prev = $line; 525 next line; 526 } 527 if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) { 528 err("improper first line of block comment"); 529 } 530 531 if ($in_comment) { # still in comment, don't do further checks 532 $prev = $line; 533 next line; 534 } 535 536 if ((/[^(]\/\*\S/ || /^\/\*\S/) && 537 !(/$lint_re/ || ($splint_comments && /$splint_re/))) { 538 err("missing blank after open comment"); 539 } 540 if (/\S\*\/[^)]|\S\*\/$/ && 541 !(/$lint_re/ || ($splint_comments && /$splint_re/))) { 542 err("missing blank before close comment"); 543 } 544 if (/\/\/\S/) { # C++ comments 545 err("missing blank after start comment"); 546 } 547 # check for unterminated single line comments, but allow them when 548 # they are used to comment out the argument list of a function 549 # declaration. 550 if (/\S.*\/\*/ && !/\S.*\/\*.*\*\// && !/\(\/\*/) { 551 err("unterminated single line comment"); 552 } 553 554 if (/^(#else|#endif|#include)(.*)$/) { 555 $prev = $line; 556 if ($picky) { 557 my $directive = $1; 558 my $clause = $2; 559 # Enforce ANSI rules for #else and #endif: no noncomment 560 # identifiers are allowed after #endif or #else. Allow 561 # C++ comments since they seem to be a fact of life. 562 if ((($1 eq "#endif") || ($1 eq "#else")) && 563 ($clause ne "") && 564 (!($clause =~ /^\s+\/\*.*\*\/$/)) && 565 (!($clause =~ /^\s+\/\/.*$/))) { 566 err("non-comment text following " . 567 "$directive (or malformed $directive " . 568 "directive)"); 569 } 570 } 571 next line; 572 } 573 574 # 575 # delete any comments and check everything else. Note that 576 # ".*?" is a non-greedy match, so that we don't get confused by 577 # multiple comments on the same line. 578 # 579 s/\/\*.*?\*\///g; 580 s/\/\/.*$//; # C++ comments 581 582 # delete any trailing whitespace; we have already checked for that. 583 s/\s*$//; 584 585 # following checks do not apply to text in comments. 586 587 if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=[^\s,]/ || 588 (/[^->]>[^,=>\s]/ && !/[^->]>$/) || 589 (/[^<]<[^,=<\s]/ && !/[^<]<$/) || 590 /[^<\s]<[^<]/ || /[^->\s]>[^>]/) { 591 err("missing space around relational operator"); 592 } 593 if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ || 594 (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) || 595 (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) { 596 # XXX - should only check this for C++ code 597 # XXX - there are probably other forms that should be allowed 598 if (!/\soperator=/) { 599 err("missing space around assignment operator"); 600 } 601 } 602 if (/[,;]\S/ && !/\bfor \(;;\)/) { 603 err("comma or semicolon followed by non-blank"); 604 } 605 # allow "for" statements to have empty "while" clauses 606 if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) { 607 err("comma or semicolon preceded by blank"); 608 } 609 if (/^\s*(&&|\|\|)/) { 610 err("improper boolean continuation"); 611 } 612 if (/\S *(&&|\|\|)/ || /(&&|\|\|) *\S/) { 613 err("more than one space around boolean operator"); 614 } 615 if (/\b(for|if|while|switch|sizeof|return|case)\(/) { 616 err("missing space between keyword and paren"); 617 } 618 if (/(\b(for|if|while|switch|return)\b.*){2,}/ && !/^#define/) { 619 # multiple "case" and "sizeof" allowed 620 err("more than one keyword on line"); 621 } 622 if (/\b(for|if|while|switch|sizeof|return|case)\s\s+\(/ && 623 !/^#if\s+\(/) { 624 err("extra space between keyword and paren"); 625 } 626 # try to detect "func (x)" but not "if (x)" or 627 # "#define foo (x)" or "int (*func)();" 628 if (/\w\s\(/) { 629 my $s = $_; 630 # strip off all keywords on the line 631 s/\b(for|if|while|switch|return|case|sizeof)\s\(/XXX(/g; 632 s/#elif\s\(/XXX(/g; 633 s/^#define\s+\w+\s+\(/XXX(/; 634 # do not match things like "void (*f)();" 635 # or "typedef void (func_t)();" 636 s/\w\s\(+\*/XXX(*/g; 637 s/\b($typename|void)\s+\(+/XXX(/og; 638 if (/\w\s\(/) { 639 err("extra space between function name and left paren"); 640 } 641 $_ = $s; 642 } 643 # try to detect "int foo(x)", but not "extern int foo(x);" 644 # XXX - this still trips over too many legitimate things, 645 # like "int foo(x,\n\ty);" 646# if (/^(\w+(\s|\*)+)+\w+\(/ && !/\)[;,](\s|)*$/ && 647# !/^(extern|static)\b/) { 648# err("return type of function not on separate line"); 649# } 650 # this is a close approximation 651 if (/^(\w+(\s|\*)+)+\w+\(.*\)(\s|)*$/ && 652 !/^(extern|static)\b/) { 653 err("return type of function not on separate line"); 654 } 655 if (/^#define /) { 656 err("#define followed by space instead of tab"); 657 } 658 if (/^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) { 659 err("unparenthesized return expression"); 660 } 661 if (/\bsizeof\b/ && !/\bsizeof\s*\(.*\)/) { 662 err("unparenthesized sizeof expression"); 663 } 664 if (/\(\s/) { 665 err("whitespace after left paren"); 666 } 667 # Allow "for" statements to have empty "continue" clauses. 668 # Allow right paren on its own line unless we're being picky (-p). 669 if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/ && ($picky || !/^\s*\)/)) { 670 err("whitespace before right paren"); 671 } 672 if (/^\s*\(void\)[^ ]/) { 673 err("missing space after (void) cast"); 674 } 675 if (/\S\{/ && !/\{\{/) { 676 err("missing space before left brace"); 677 } 678 if ($in_function && /^\s+\{/ && 679 ($prev =~ /\)\s*$/ || $prev =~ /\bstruct\s+\w+$/)) { 680 err("left brace starting a line"); 681 } 682 if (/\}(else|while)/) { 683 err("missing space after right brace"); 684 } 685 if (/\}\s\s+(else|while)/) { 686 err("extra space after right brace"); 687 } 688 if (/\b_VOID\b|\bVOID\b|\bSTATIC\b/) { 689 err("obsolete use of VOID or STATIC"); 690 } 691 if (/\b$typename\*/o) { 692 err("missing space between type name and *"); 693 } 694 if (/^\s+#/) { 695 err("preprocessor statement not in column 1"); 696 } 697 if (/^#\s/) { 698 err("blank after preprocessor #"); 699 } 700 if (/!\s*(strcmp|strncmp|bcmp)\s*\(/) { 701 err("don't use boolean ! with comparison functions"); 702 } 703 704 # 705 # We completely ignore, for purposes of indentation: 706 # * lines outside of functions 707 # * preprocessor lines 708 # 709 if ($check_continuation && $in_function && !$in_cpp) { 710 process_indent($_); 711 } 712 if ($picky) { 713 # try to detect spaces after casts, but allow (e.g.) 714 # "sizeof (int) + 1", "void (*funcptr)(int) = foo;", and 715 # "int foo(int) __NORETURN;" 716 if ((/^\($typename( \*+)?\)\s/o || 717 /\W\($typename( \*+)?\)\s/o) && 718 !/sizeof\s*\($typename( \*)?\)\s/o && 719 !/\($typename( \*+)?\)\s+=[^=]/o) { 720 err("space after cast"); 721 } 722 if (/\b$typename\s*\*\s/o && 723 !/\b$typename\s*\*\s+const\b/o) { 724 err("unary * followed by space"); 725 } 726 } 727 if ($check_posix_types) { 728 # try to detect old non-POSIX types. 729 # POSIX requires all non-standard typedefs to end in _t, 730 # but historically these have been used. 731 if (/\b(unchar|ushort|uint|ulong|u_int|u_short|u_long|u_char|quad)\b/) { 732 err("non-POSIX typedef $1 used: use $old2posix{$1} instead"); 733 } 734 } 735 if ($heuristic) { 736 # cannot check this everywhere due to "struct {\n...\n} foo;" 737 if ($in_function && !$in_declaration && 738 /\}./ && !/\}\s+=/ && !/\{.*\}[;,]$/ && !/\}(\s|)*$/ && 739 !/\} (else|while)/ && !/\}\}/) { 740 err("possible bad text following right brace"); 741 } 742 # cannot check this because sub-blocks in 743 # the middle of code are ok 744 if ($in_function && /^\s+\{/) { 745 err("possible left brace starting a line"); 746 } 747 } 748 if (/^\s*else\W/) { 749 if ($prev =~ /^\s*\}$/) { 750 err_prefix($prev, 751 "else and right brace should be on same line"); 752 } 753 } 754 $prev = $line; 755} 756 757if ($prev eq "") { 758 err("last line in file is blank"); 759} 760 761} 762 763# 764# Continuation-line checking 765# 766# The rest of this file contains the code for the continuation checking 767# engine. It's a pretty simple state machine which tracks the expression 768# depth (unmatched '('s and '['s). 769# 770# Keep in mind that the argument to process_indent() has already been heavily 771# processed; all comments have been replaced by control-A, and the contents of 772# strings and character constants have been elided. 773# 774 775my $cont_in; # currently inside of a continuation 776my $cont_off; # skipping an initializer or definition 777my $cont_noerr; # suppress cascading errors 778my $cont_start; # the line being continued 779my $cont_base; # the base indentation 780my $cont_first; # this is the first line of a statement 781my $cont_multiseg; # this continuation has multiple segments 782 783my $cont_special; # this is a C statement (if, for, etc.) 784my $cont_macro; # this is a macro 785my $cont_case; # this is a multi-line case 786 787my @cont_paren; # the stack of unmatched ( and [s we've seen 788 789sub 790reset_indent() 791{ 792 $cont_in = 0; 793 $cont_off = 0; 794} 795 796sub 797delabel($) 798{ 799 # 800 # replace labels with tabs. Note that there may be multiple 801 # labels on a line. 802 # 803 local $_ = $_[0]; 804 805 while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) { 806 my ($pre_tabs, $label, $rest) = ($1, $2, $3); 807 $_ = $pre_tabs; 808 while ($label =~ s/^([^\t]*)(\t+)//) { 809 $_ .= "\t" x (length($2) + length($1) / 8); 810 } 811 $_ .= ("\t" x (length($label) / 8)).$rest; 812 } 813 814 return ($_); 815} 816 817sub 818process_indent($) 819{ 820 require strict; 821 local $_ = $_[0]; # preserve the global $_ 822 823 s///g; # No comments 824 s/\s+$//; # Strip trailing whitespace 825 826 return if (/^$/); # skip empty lines 827 828 # regexps used below; keywords taking (), macros, and continued cases 829 my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b'; 830 my $macro = '[A-Z_][A-Z_0-9]*\('; 831 my $case = 'case\b[^:]*$'; 832 833 # skip over enumerations, array definitions, initializers, etc. 834 if ($cont_off <= 0 && !/^\s*$special/ && 835 (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*))\{/ || 836 (/^\s*\{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) { 837 $cont_in = 0; 838 $cont_off = tr/{/{/ - tr/}/}/; 839 return; 840 } 841 if ($cont_off) { 842 $cont_off += tr/{/{/ - tr/}/}/; 843 return; 844 } 845 846 if (!$cont_in) { 847 $cont_start = $line; 848 849 if (/^\t* /) { 850 err("non-continuation indented 4 spaces"); 851 $cont_noerr = 1; # stop reporting 852 } 853 $_ = delabel($_); # replace labels with tabs 854 855 # check if the statement is complete 856 return if (/^\s*\}?$/); 857 return if (/^\s*\}?\s*else\s*\{?$/); 858 return if (/^\s*do\s*\{?$/); 859 return if (/\{$/); 860 return if (/\}[,;]?$/); 861 862 # Allow macros on their own lines 863 return if (/^\s*[A-Z_][A-Z_0-9]*$/); 864 865 # cases we don't deal with, generally non-kosher 866 if (/\{/) { 867 err("stuff after {"); 868 return; 869 } 870 871 # Get the base line, and set up the state machine 872 /^(\t*)/; 873 $cont_base = $1; 874 $cont_in = 1; 875 @cont_paren = (); 876 $cont_first = 1; 877 $cont_multiseg = 0; 878 879 # certain things need special processing 880 $cont_special = /^\s*$special/? 1 : 0; 881 $cont_macro = /^\s*$macro/? 1 : 0; 882 $cont_case = /^\s*$case/? 1 : 0; 883 } else { 884 $cont_first = 0; 885 886 # Strings may be pulled back to an earlier (half-)tabstop 887 unless ($cont_noerr || /^$cont_base / || 888 (/^\t*(?: )?(?:gettext\()?\"/ && !/^$cont_base\t/)) { 889 err_prefix($cont_start, 890 "continuation should be indented 4 spaces"); 891 } 892 } 893 894 my $rest = $_; # keeps the remainder of the line 895 896 # 897 # The split matches 0 characters, so that each 'special' character 898 # is processed separately. Parens and brackets are pushed and 899 # popped off the @cont_paren stack. For normal processing, we wait 900 # until a ; or { terminates the statement. "special" processing 901 # (if/for/while/switch) is allowed to stop when the stack empties, 902 # as is macro processing. Case statements are terminated with a : 903 # and an empty paren stack. 904 # 905 foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) { 906 next if (length($_) == 0); 907 908 # rest contains the remainder of the line 909 my $rxp = "[^\Q$_\E]*\Q$_\E"; 910 $rest =~ s/^$rxp//; 911 912 if (/\(/ || /\[/) { 913 push @cont_paren, $_; 914 } elsif (/\)/ || /\]/) { 915 my $cur = $_; 916 tr/\)\]/\(\[/; 917 918 my $old = (pop @cont_paren); 919 if (!defined($old)) { 920 err("unexpected '$cur'"); 921 $cont_in = 0; 922 last; 923 } elsif ($old ne $_) { 924 err("'$cur' mismatched with '$old'"); 925 $cont_in = 0; 926 last; 927 } 928 929 # 930 # If the stack is now empty, do special processing 931 # for if/for/while/switch and macro statements. 932 # 933 next if (@cont_paren != 0); 934 if ($cont_special) { 935 if ($rest =~ /^\s*\{?$/) { 936 $cont_in = 0; 937 last; 938 } 939 if ($rest =~ /^\s*;$/) { 940 err("empty if/for/while body ". 941 "not on its own line"); 942 $cont_in = 0; 943 last; 944 } 945 if (!$cont_first && $cont_multiseg == 1) { 946 err_prefix($cont_start, 947 "multiple statements continued ". 948 "over multiple lines"); 949 $cont_multiseg = 2; 950 } elsif ($cont_multiseg == 0) { 951 $cont_multiseg = 1; 952 } 953 # We've finished this section, start 954 # processing the next. 955 goto section_ended; 956 } 957 if ($cont_macro) { 958 if ($rest =~ /^$/) { 959 $cont_in = 0; 960 last; 961 } 962 } 963 } elsif (/\;/) { 964 if ($cont_case) { 965 err("unexpected ;"); 966 } elsif (!$cont_special) { 967 err("unexpected ;") if (@cont_paren != 0); 968 if (!$cont_first && $cont_multiseg == 1) { 969 err_prefix($cont_start, 970 "multiple statements continued ". 971 "over multiple lines"); 972 $cont_multiseg = 2; 973 } elsif ($cont_multiseg == 0) { 974 $cont_multiseg = 1; 975 } 976 if ($rest =~ /^$/) { 977 $cont_in = 0; 978 last; 979 } 980 if ($rest =~ /^\s*special/) { 981 err("if/for/while/switch not started ". 982 "on its own line"); 983 } 984 goto section_ended; 985 } 986 } elsif (/\{/) { 987 err("{ while in parens/brackets") if (@cont_paren != 0); 988 err("stuff after {") if ($rest =~ /[^\s}]/); 989 $cont_in = 0; 990 last; 991 } elsif (/\}/) { 992 err("} while in parens/brackets") if (@cont_paren != 0); 993 if (!$cont_special && $rest !~ /^\s*(while|else)\b/) { 994 if ($rest =~ /^$/) { 995 err("unexpected }"); 996 } else { 997 err("stuff after }"); 998 } 999 $cont_in = 0; 1000 last; 1001 } 1002 } elsif (/\:/ && $cont_case && @cont_paren == 0) { 1003 err("stuff after multi-line case") if ($rest !~ /$^/); 1004 $cont_in = 0; 1005 last; 1006 } 1007 next; 1008section_ended: 1009 # End of a statement or if/while/for loop. Reset 1010 # cont_special and cont_macro based on the rest of the 1011 # line. 1012 $cont_special = ($rest =~ /^\s*$special/)? 1 : 0; 1013 $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0; 1014 $cont_case = 0; 1015 next; 1016 } 1017 $cont_noerr = 0 if (!$cont_in); 1018} 1019