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# Copyright 2016 Nexenta Systems, Inc. 24# Copyright 2020 Oxide Computer Company 25# 26# Copyright 2008 Sun Microsystems, Inc. All rights reserved. 27# Use is subject to license terms. 28# 29# Copyright (c) 2015 by Delphix. All rights reserved. 30# 31# @(#)cstyle 1.58 98/09/09 (from shannon) 32# 33# cstyle - check for some common stylistic errors. 34# 35# cstyle is a sort of "lint" for C coding style. 36# It attempts to check for the style used in the 37# kernel, sometimes known as "Bill Joy Normal Form". 38# 39# There's a lot this can't check for, like proper indentation 40# of code blocks. There's also a lot more this could check for. 41# 42# A note to the non perl literate: 43# 44# perl regular expressions are pretty much like egrep 45# regular expressions, with the following special symbols 46# 47# \s any space character 48# \S any non-space character 49# \w any "word" character [a-zA-Z0-9_] 50# \W any non-word character 51# \d a digit [0-9] 52# \D a non-digit 53# \b word boundary (between \w and \W) 54# \B non-word boundary 55# 56 57require 5.0; 58use IO::File; 59use Getopt::Std; 60use strict; 61 62my $usage = 63"usage: cstyle [-chpvCP] [-o constructs] file ... 64 -c check continuation indentation inside functions 65 -h perform heuristic checks that are sometimes wrong 66 -p perform some of the more picky checks 67 -v verbose 68 -C don't check anything in header block comments 69 -P check for use of non-POSIX types 70 -o constructs 71 allow a comma-seperated list of optional constructs: 72 doxygen allow doxygen-style block comments (/** /*!) 73 splint allow splint-style lint comments (/*@ ... @*/) 74"; 75 76my %opts; 77 78if (!getopts("cho:pvCP", \%opts)) { 79 print $usage; 80 exit 2; 81} 82 83my $check_continuation = $opts{'c'}; 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# FreeBSD uses comments styled as such for their license headers: 126# /*- 127# * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 128# * 129# ... 130# 131# In order to apply other cstyle checks to those files without stumbling over 132# the license header, tolerate such comment openings as well. 133my $fbsd_comment_start = qr/^\s*\/\*-$/; 134 135# Note, following must be in single quotes so that \s and \w work right. 136my $typename = '(int|char|short|long|unsigned|float|double' . 137 '|\w+_t|struct\s+\w+|union\s+\w+|FILE)'; 138 139# mapping of old types to POSIX compatible types 140my %old2posix = ( 141 'unchar' => 'uchar_t', 142 'ushort' => 'ushort_t', 143 'uint' => 'uint_t', 144 'ulong' => 'ulong_t', 145 'u_int' => 'uint_t', 146 'u_short' => 'ushort_t', 147 'u_long' => 'ulong_t', 148 'u_char' => 'uchar_t', 149 'u_int8_t' => 'uint8_t', 150 'u_int16_t' => 'uint16_t', 151 'u_int32_t' => 'uint32_t', 152 'u_int64_t' => 'uint64_t', 153 'u_quad_t' => 'uint64_t', 154 'quad' => 'quad_t' 155); 156 157my $lint_re = qr/\/\*(?: 158 ARGSUSED[0-9]*|NOTREACHED|LINTLIBRARY|VARARGS[0-9]*| 159 CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY| 160 FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*| 161 PROTOLIB[0-9]*|SCANFLIKE[0-9]*|CSTYLED.*? 162 )\*\//x; 163 164my $splint_re = qr/\/\*@.*?@\*\//x; 165 166my $warlock_re = qr/\/\*\s*(?: 167 VARIABLES\ PROTECTED\ BY| 168 MEMBERS\ PROTECTED\ BY| 169 ALL\ MEMBERS\ PROTECTED\ BY| 170 READ-ONLY\ VARIABLES:| 171 READ-ONLY\ MEMBERS:| 172 VARIABLES\ READABLE\ WITHOUT\ LOCK:| 173 MEMBERS\ READABLE\ WITHOUT\ LOCK:| 174 LOCKS\ COVERED\ BY| 175 LOCK\ UNNEEDED\ BECAUSE| 176 LOCK\ NEEDED:| 177 LOCK\ HELD\ ON\ ENTRY:| 178 READ\ LOCK\ HELD\ ON\ ENTRY:| 179 WRITE\ LOCK\ HELD\ ON\ ENTRY:| 180 LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 181 READ\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 182 WRITE\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| 183 LOCK\ RELEASED\ AS\ SIDE\ EFFECT:| 184 LOCK\ UPGRADED\ AS\ SIDE\ EFFECT:| 185 LOCK\ DOWNGRADED\ AS\ SIDE\ EFFECT:| 186 FUNCTIONS\ CALLED\ THROUGH\ POINTER| 187 FUNCTIONS\ CALLED\ THROUGH\ MEMBER| 188 LOCK\ ORDER: 189 )/x; 190 191my $err_stat = 0; # exit status 192 193if ($#ARGV >= 0) { 194 foreach my $arg (@ARGV) { 195 my $fh = new IO::File $arg, "r"; 196 if (!defined($fh)) { 197 printf "%s: can not open\n", $arg; 198 } else { 199 &cstyle($arg, $fh); 200 close $fh; 201 } 202 } 203} else { 204 &cstyle("<stdin>", *STDIN); 205} 206exit $err_stat; 207 208my $no_errs = 0; # set for CSTYLED-protected lines 209 210sub err($) { 211 my ($error) = @_; 212 unless ($no_errs) { 213 if ($verbose) { 214 printf $fmt, $filename, $., $error, $line; 215 } else { 216 printf $fmt, $filename, $., $error; 217 } 218 $err_stat = 1; 219 } 220} 221 222sub err_prefix($$) { 223 my ($prevline, $error) = @_; 224 my $out = $prevline."\n".$line; 225 unless ($no_errs) { 226 if ($verbose) { 227 printf $fmt, $filename, $., $error, $out; 228 } else { 229 printf $fmt, $filename, $., $error; 230 } 231 $err_stat = 1; 232 } 233} 234 235sub err_prev($) { 236 my ($error) = @_; 237 unless ($no_errs) { 238 if ($verbose) { 239 printf $fmt, $filename, $. - 1, $error, $prev; 240 } else { 241 printf $fmt, $filename, $. - 1, $error; 242 } 243 $err_stat = 1; 244 } 245} 246 247sub cstyle($$) { 248 249my ($fn, $filehandle) = @_; 250$filename = $fn; # share it globally 251 252my $in_cpp = 0; 253my $next_in_cpp = 0; 254 255my $in_comment = 0; 256my $in_header_comment = 0; 257my $comment_done = 0; 258my $in_warlock_comment = 0; 259my $in_function = 0; 260my $in_function_header = 0; 261my $function_header_full_indent = 0; 262my $in_declaration = 0; 263my $note_level = 0; 264my $nextok = 0; 265my $nocheck = 0; 266 267my $in_string = 0; 268 269my ($okmsg, $comment_prefix); 270 271$line = ''; 272$prev = ''; 273reset_indent(); 274 275line: while (<$filehandle>) { 276 s/\r?\n$//; # strip return and newline 277 278 # save the original line, then remove all text from within 279 # double or single quotes, we do not want to check such text. 280 281 $line = $_; 282 283 # 284 # C allows strings to be continued with a backslash at the end of 285 # the line. We translate that into a quoted string on the previous 286 # line followed by an initial quote on the next line. 287 # 288 # (we assume that no-one will use backslash-continuation with character 289 # constants) 290 # 291 $_ = '"' . $_ if ($in_string && !$nocheck && !$in_comment); 292 293 # 294 # normal strings and characters 295 # 296 s/'([^\\']|\\[^xX0]|\\0[0-9]*|\\[xX][0-9a-fA-F]*)'/''/g; 297 s/"([^\\"]|\\.)*"/\"\"/g; 298 299 # 300 # detect string continuation 301 # 302 if ($nocheck || $in_comment) { 303 $in_string = 0; 304 } else { 305 # 306 # Now that all full strings are replaced with "", we check 307 # for unfinished strings continuing onto the next line. 308 # 309 $in_string = 310 (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ || 311 s/^("")*"([^\\"]|\\.)*\\$/""/); 312 } 313 314 # 315 # figure out if we are in a cpp directive 316 # 317 $in_cpp = $next_in_cpp || /^\s*#/; # continued or started 318 $next_in_cpp = $in_cpp && /\\$/; # only if continued 319 320 # strip off trailing backslashes, which appear in long macros 321 s/\s*\\$//; 322 323 # an /* END CSTYLED */ comment ends a no-check block. 324 if ($nocheck) { 325 if (/\/\* *END *CSTYLED *\*\//) { 326 $nocheck = 0; 327 } else { 328 reset_indent(); 329 next line; 330 } 331 } 332 333 # a /*CSTYLED*/ comment indicates that the next line is ok. 334 if ($nextok) { 335 if ($okmsg) { 336 err($okmsg); 337 } 338 $nextok = 0; 339 $okmsg = 0; 340 if (/\/\* *CSTYLED.*\*\//) { 341 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; 342 $okmsg = $1; 343 $nextok = 1; 344 } 345 $no_errs = 1; 346 } elsif ($no_errs) { 347 $no_errs = 0; 348 } 349 350 # check length of line. 351 # first, a quick check to see if there is any chance of being too long. 352 if (($line =~ tr/\t/\t/) * 7 + length($line) > 80) { 353 # yes, there is a chance. 354 # replace tabs with spaces and check again. 355 my $eline = $line; 356 1 while $eline =~ 357 s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e; 358 if (length($eline) > 80) { 359 err("line > 80 characters"); 360 } 361 } 362 363 # ignore NOTE(...) annotations (assumes NOTE is on lines by itself). 364 if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE 365 s/[^()]//g; # eliminate all non-parens 366 $note_level += s/\(//g - length; # update paren nest level 367 next; 368 } 369 370 # a /* BEGIN CSTYLED */ comment starts a no-check block. 371 if (/\/\* *BEGIN *CSTYLED *\*\//) { 372 $nocheck = 1; 373 } 374 375 # a /*CSTYLED*/ comment indicates that the next line is ok. 376 if (/\/\* *CSTYLED.*\*\//) { 377 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; 378 $okmsg = $1; 379 $nextok = 1; 380 } 381 if (/\/\/ *CSTYLED/) { 382 /^.*\/\/ *CSTYLED *(.*)$/; 383 $okmsg = $1; 384 $nextok = 1; 385 } 386 387 # universal checks; apply to everything 388 if (/\t +\t/) { 389 err("spaces between tabs"); 390 } 391 if (/ \t+ /) { 392 err("tabs between spaces"); 393 } 394 if (/\s$/) { 395 err("space or tab at end of line"); 396 } 397 if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) { 398 err("comment preceded by non-blank"); 399 } 400 401 # is this the beginning or ending of a function? 402 # (not if "struct foo\n{\n") 403 if (/^\{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) { 404 $in_function = 1; 405 $in_declaration = 1; 406 $in_function_header = 0; 407 $function_header_full_indent = 0; 408 $prev = $line; 409 next line; 410 } 411 if (/^\}\s*(\/\*.*\*\/\s*)*$/) { 412 if ($prev =~ /^\s*return\s*;/) { 413 err_prev("unneeded return at end of function"); 414 } 415 $in_function = 0; 416 reset_indent(); # we don't check between functions 417 $prev = $line; 418 next line; 419 } 420 if ($in_function_header && ! /^ (\w|\.)/ ) { 421 if (/^\{\}$/) { 422 $in_function_header = 0; 423 $function_header_full_indent = 0; 424 } elsif ($picky && ! (/^\t/ && $function_header_full_indent != 0)) { 425 err("continuation line should be indented by 4 spaces"); 426 } 427 } 428 429 # 430 # If this matches something of form "foo(", it's probably a function 431 # definition, unless it ends with ") bar;", in which case it's a declaration 432 # that uses a macro to generate the type. 433 # 434 if (/^\w+\(/ && !/\) \w+;$/) { 435 $in_function_header = 1; 436 if (/\($/) { 437 $function_header_full_indent = 1; 438 } 439 } 440 if ($in_function_header && /^\{$/) { 441 $in_function_header = 0; 442 $function_header_full_indent = 0; 443 $in_function = 1; 444 } 445 if ($in_function_header && /\);$/) { 446 $in_function_header = 0; 447 $function_header_full_indent = 0; 448 } 449 if ($in_function_header && /\{$/ ) { 450 if ($picky) { 451 err("opening brace on same line as function header"); 452 } 453 $in_function_header = 0; 454 $function_header_full_indent = 0; 455 $in_function = 1; 456 next line; 457 } 458 459 if ($in_warlock_comment && /\*\//) { 460 $in_warlock_comment = 0; 461 $prev = $line; 462 next line; 463 } 464 465 # a blank line terminates the declarations within a function. 466 # XXX - but still a problem in sub-blocks. 467 if ($in_declaration && /^$/) { 468 $in_declaration = 0; 469 } 470 471 if ($comment_done) { 472 $in_comment = 0; 473 $in_header_comment = 0; 474 $comment_done = 0; 475 } 476 # does this looks like the start of a block comment? 477 if (/$hdr_comment_start/ || /$fbsd_comment_start/) { 478 if (!/^\t*\/\*/) { 479 err("block comment not indented by tabs"); 480 } 481 $in_comment = 1; 482 /^(\s*)\//; 483 $comment_prefix = $1; 484 if ($comment_prefix eq "") { 485 $in_header_comment = 1; 486 } 487 $prev = $line; 488 next line; 489 } 490 # are we still in the block comment? 491 if ($in_comment) { 492 if (/^$comment_prefix \*\/$/) { 493 $comment_done = 1; 494 } elsif (/\*\//) { 495 $comment_done = 1; 496 err("improper block comment close") 497 unless ($ignore_hdr_comment && $in_header_comment); 498 } elsif (!/^$comment_prefix \*[ \t]/ && 499 !/^$comment_prefix \*$/) { 500 err("improper block comment") 501 unless ($ignore_hdr_comment && $in_header_comment); 502 } 503 } 504 505 if ($in_header_comment && $ignore_hdr_comment) { 506 $prev = $line; 507 next line; 508 } 509 510 # check for errors that might occur in comments and in code. 511 512 # allow spaces to be used to draw pictures in header comments. 513 if (/[^ ] / && !/".* .*"/ && !$in_header_comment) { 514 err("spaces instead of tabs"); 515 } 516 if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ && 517 (!/^ (\w|\.)/ || $in_function != 0)) { 518 err("indent by spaces instead of tabs"); 519 } 520 if (/^\t+ [^ \t\*]/ || /^\t+ \S/ || /^\t+ \S/) { 521 err("continuation line not indented by 4 spaces"); 522 } 523 if (/$warlock_re/ && !/\*\//) { 524 $in_warlock_comment = 1; 525 $prev = $line; 526 next line; 527 } 528 if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) { 529 err("improper first line of block comment"); 530 } 531 532 if ($in_comment) { # still in comment, don't do further checks 533 $prev = $line; 534 next line; 535 } 536 537 if ((/[^(]\/\*\S/ || /^\/\*\S/) && 538 !(/$lint_re/ || ($splint_comments && /$splint_re/))) { 539 err("missing blank after open comment"); 540 } 541 if (/\S\*\/[^)]|\S\*\/$/ && 542 !(/$lint_re/ || ($splint_comments && /$splint_re/))) { 543 err("missing blank before close comment"); 544 } 545 if (/\/\/\S/) { # C++ comments 546 err("missing blank after start comment"); 547 } 548 # check for unterminated single line comments, but allow them when 549 # they are used to comment out the argument list of a function 550 # declaration. 551 if (/\S.*\/\*/ && !/\S.*\/\*.*\*\// && !/\(\/\*/) { 552 err("unterminated single line comment"); 553 } 554 555 if (/^(#else|#endif|#include)(.*)$/) { 556 $prev = $line; 557 if ($picky) { 558 my $directive = $1; 559 my $clause = $2; 560 # Enforce ANSI rules for #else and #endif: no noncomment 561 # identifiers are allowed after #endif or #else. Allow 562 # C++ comments since they seem to be a fact of life. 563 if ((($1 eq "#endif") || ($1 eq "#else")) && 564 ($clause ne "") && 565 (!($clause =~ /^\s+\/\*.*\*\/$/)) && 566 (!($clause =~ /^\s+\/\/.*$/))) { 567 err("non-comment text following " . 568 "$directive (or malformed $directive " . 569 "directive)"); 570 } 571 } 572 next line; 573 } 574 575 # 576 # delete any comments and check everything else. Note that 577 # ".*?" is a non-greedy match, so that we don't get confused by 578 # multiple comments on the same line. 579 # 580 s/\/\*.*?\*\///g; 581 s/\/\/.*$//; # C++ comments 582 583 # delete any trailing whitespace; we have already checked for that. 584 s/\s*$//; 585 586 # following checks do not apply to text in comments. 587 588 if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=[^\s,]/ || 589 (/[^->]>[^,=>\s]/ && !/[^->]>$/) || 590 (/[^<]<[^,=<\s]/ && !/[^<]<$/) || 591 /[^<\s]<[^<]/ || /[^->\s]>[^>]/) { 592 err("missing space around relational operator"); 593 } 594 if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ || 595 (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) || 596 (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) { 597 # XXX - should only check this for C++ code 598 # XXX - there are probably other forms that should be allowed 599 if (!/\soperator=/) { 600 err("missing space around assignment operator"); 601 } 602 } 603 if (/[,;]\S/ && !/\bfor \(;;\)/) { 604 err("comma or semicolon followed by non-blank"); 605 } 606 # allow "for" statements to have empty "while" clauses 607 if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) { 608 err("comma or semicolon preceded by blank"); 609 } 610 if (/^\s*(&&|\|\|)/) { 611 err("improper boolean continuation"); 612 } 613 if (/\S *(&&|\|\|)/ || /(&&|\|\|) *\S/) { 614 err("more than one space around boolean operator"); 615 } 616 if (/\b(for|if|while|switch|sizeof|return|case)\(/) { 617 err("missing space between keyword and paren"); 618 } 619 if (/(\b(for|if|while|switch|return)\b.*){2,}/ && !/^#define/) { 620 # multiple "case" and "sizeof" allowed 621 err("more than one keyword on line"); 622 } 623 if (/\b(for|if|while|switch|sizeof|return|case)\s\s+\(/ && 624 !/^#if\s+\(/) { 625 err("extra space between keyword and paren"); 626 } 627 # try to detect "func (x)" but not "if (x)" or 628 # "#define foo (x)" or "int (*func)();" 629 if (/\w\s\(/) { 630 my $s = $_; 631 # strip off all keywords on the line 632 s/\b(for|if|while|switch|return|case|sizeof)\s\(/XXX(/g; 633 s/#elif\s\(/XXX(/g; 634 s/^#define\s+\w+\s+\(/XXX(/; 635 # do not match things like "void (*f)();" 636 # or "typedef void (func_t)();" 637 s/\w\s\(+\*/XXX(*/g; 638 s/\b($typename|void)\s+\(+/XXX(/og; 639 if (/\w\s\(/) { 640 err("extra space between function name and left paren"); 641 } 642 $_ = $s; 643 } 644 # try to detect "int foo(x)", but not "extern int foo(x);" 645 # XXX - this still trips over too many legitimate things, 646 # like "int foo(x,\n\ty);" 647# if (/^(\w+(\s|\*)+)+\w+\(/ && !/\)[;,](\s|)*$/ && 648# !/^(extern|static)\b/) { 649# err("return type of function not on separate line"); 650# } 651 # this is a close approximation 652 if (/^(\w+(\s|\*)+)+\w+\(.*\)(\s|)*$/ && 653 !/^(extern|static)\b/) { 654 err("return type of function not on separate line"); 655 } 656 if (/^#define /) { 657 err("#define followed by space instead of tab"); 658 } 659 if (/^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) { 660 err("unparenthesized return expression"); 661 } 662 if (/\bsizeof\b/ && !/\bsizeof\s*\(.*\)/) { 663 err("unparenthesized sizeof expression"); 664 } 665 if (/\(\s/) { 666 err("whitespace after left paren"); 667 } 668 # allow "for" statements to have empty "continue" clauses 669 if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/) { 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 my $types = join '|', keys %old2posix; 732 if (/\b($types)\b/) { 733 err("non-POSIX typedef $1 used: use $old2posix{$1} instead"); 734 } 735 } 736 if ($heuristic) { 737 # cannot check this everywhere due to "struct {\n...\n} foo;" 738 if ($in_function && !$in_declaration && 739 /\}./ && !/\}\s+=/ && !/\{.*\}[;,]$/ && !/\}(\s|)*$/ && 740 !/\} (else|while)/ && !/\}\}/) { 741 err("possible bad text following right brace"); 742 } 743 # cannot check this because sub-blocks in 744 # the middle of code are ok 745 if ($in_function && /^\s+\{/) { 746 err("possible left brace starting a line"); 747 } 748 } 749 if (/^\s*else\W/) { 750 if ($prev =~ /^\s*\}$/) { 751 err_prefix($prev, 752 "else and right brace should be on same line"); 753 } 754 } 755 $prev = $line; 756} 757 758if ($prev eq "") { 759 err("last line in file is blank"); 760} 761 762} 763 764# 765# Continuation-line checking 766# 767# The rest of this file contains the code for the continuation checking 768# engine. It's a pretty simple state machine which tracks the expression 769# depth (unmatched '('s and '['s). 770# 771# Keep in mind that the argument to process_indent() has already been heavily 772# processed; all comments have been replaced by control-A, and the contents of 773# strings and character constants have been elided. 774# 775 776my $cont_in; # currently inside of a continuation 777my $cont_off; # skipping an initializer or definition 778my $cont_noerr; # suppress cascading errors 779my $cont_start; # the line being continued 780my $cont_base; # the base indentation 781my $cont_first; # this is the first line of a statement 782my $cont_multiseg; # this continuation has multiple segments 783 784my $cont_special; # this is a C statement (if, for, etc.) 785my $cont_macro; # this is a macro 786my $cont_case; # this is a multi-line case 787 788my @cont_paren; # the stack of unmatched ( and [s we've seen 789 790sub 791reset_indent() 792{ 793 $cont_in = 0; 794 $cont_off = 0; 795} 796 797sub 798delabel($) 799{ 800 # 801 # replace labels with tabs. Note that there may be multiple 802 # labels on a line. 803 # 804 local $_ = $_[0]; 805 806 while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) { 807 my ($pre_tabs, $label, $rest) = ($1, $2, $3); 808 $_ = $pre_tabs; 809 while ($label =~ s/^([^\t]*)(\t+)//) { 810 $_ .= "\t" x (length($2) + length($1) / 8); 811 } 812 $_ .= ("\t" x (length($label) / 8)).$rest; 813 } 814 815 return ($_); 816} 817 818sub 819process_indent($) 820{ 821 require strict; 822 local $_ = $_[0]; # preserve the global $_ 823 824 s///g; # No comments 825 s/\s+$//; # Strip trailing whitespace 826 827 return if (/^$/); # skip empty lines 828 829 # regexps used below; keywords taking (), macros, and continued cases 830 my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b'; 831 my $macro = '[A-Z_][A-Z_0-9]*\('; 832 my $case = 'case\b[^:]*$'; 833 834 # skip over enumerations, array definitions, initializers, etc. 835 if ($cont_off <= 0 && !/^\s*$special/ && 836 (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*))\{/ || 837 (/^\s*\{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) { 838 $cont_in = 0; 839 $cont_off = tr/{/{/ - tr/}/}/; 840 return; 841 } 842 if ($cont_off) { 843 $cont_off += tr/{/{/ - tr/}/}/; 844 return; 845 } 846 847 if (!$cont_in) { 848 $cont_start = $line; 849 850 if (/^\t* /) { 851 err("non-continuation indented 4 spaces"); 852 $cont_noerr = 1; # stop reporting 853 } 854 $_ = delabel($_); # replace labels with tabs 855 856 # check if the statement is complete 857 return if (/^\s*\}?$/); 858 return if (/^\s*\}?\s*else\s*\{?$/); 859 return if (/^\s*do\s*\{?$/); 860 return if (/\{$/); 861 return if (/\}[,;]?$/); 862 863 # Allow macros on their own lines 864 return if (/^\s*[A-Z_][A-Z_0-9]*$/); 865 866 # cases we don't deal with, generally non-kosher 867 if (/\{/) { 868 err("stuff after {"); 869 return; 870 } 871 872 # Get the base line, and set up the state machine 873 /^(\t*)/; 874 $cont_base = $1; 875 $cont_in = 1; 876 @cont_paren = (); 877 $cont_first = 1; 878 $cont_multiseg = 0; 879 880 # certain things need special processing 881 $cont_special = /^\s*$special/? 1 : 0; 882 $cont_macro = /^\s*$macro/? 1 : 0; 883 $cont_case = /^\s*$case/? 1 : 0; 884 } else { 885 $cont_first = 0; 886 887 # Strings may be pulled back to an earlier (half-)tabstop 888 unless ($cont_noerr || /^$cont_base / || 889 (/^\t*(?: )?(?:gettext\()?\"/ && !/^$cont_base\t/)) { 890 err_prefix($cont_start, 891 "continuation should be indented 4 spaces"); 892 } 893 } 894 895 my $rest = $_; # keeps the remainder of the line 896 897 # 898 # The split matches 0 characters, so that each 'special' character 899 # is processed separately. Parens and brackets are pushed and 900 # popped off the @cont_paren stack. For normal processing, we wait 901 # until a ; or { terminates the statement. "special" processing 902 # (if/for/while/switch) is allowed to stop when the stack empties, 903 # as is macro processing. Case statements are terminated with a : 904 # and an empty paren stack. 905 # 906 foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) { 907 next if (length($_) == 0); 908 909 # rest contains the remainder of the line 910 my $rxp = "[^\Q$_\E]*\Q$_\E"; 911 $rest =~ s/^$rxp//; 912 913 if (/\(/ || /\[/) { 914 push @cont_paren, $_; 915 } elsif (/\)/ || /\]/) { 916 my $cur = $_; 917 tr/\)\]/\(\[/; 918 919 my $old = (pop @cont_paren); 920 if (!defined($old)) { 921 err("unexpected '$cur'"); 922 $cont_in = 0; 923 last; 924 } elsif ($old ne $_) { 925 err("'$cur' mismatched with '$old'"); 926 $cont_in = 0; 927 last; 928 } 929 930 # 931 # If the stack is now empty, do special processing 932 # for if/for/while/switch and macro statements. 933 # 934 next if (@cont_paren != 0); 935 if ($cont_special) { 936 if ($rest =~ /^\s*\{?$/) { 937 $cont_in = 0; 938 last; 939 } 940 if ($rest =~ /^\s*;$/) { 941 err("empty if/for/while body ". 942 "not on its own line"); 943 $cont_in = 0; 944 last; 945 } 946 if (!$cont_first && $cont_multiseg == 1) { 947 err_prefix($cont_start, 948 "multiple statements continued ". 949 "over multiple lines"); 950 $cont_multiseg = 2; 951 } elsif ($cont_multiseg == 0) { 952 $cont_multiseg = 1; 953 } 954 # We've finished this section, start 955 # processing the next. 956 goto section_ended; 957 } 958 if ($cont_macro) { 959 if ($rest =~ /^$/) { 960 $cont_in = 0; 961 last; 962 } 963 } 964 } elsif (/\;/) { 965 if ($cont_case) { 966 err("unexpected ;"); 967 } elsif (!$cont_special) { 968 err("unexpected ;") if (@cont_paren != 0); 969 if (!$cont_first && $cont_multiseg == 1) { 970 err_prefix($cont_start, 971 "multiple statements continued ". 972 "over multiple lines"); 973 $cont_multiseg = 2; 974 } elsif ($cont_multiseg == 0) { 975 $cont_multiseg = 1; 976 } 977 if ($rest =~ /^$/) { 978 $cont_in = 0; 979 last; 980 } 981 if ($rest =~ /^\s*special/) { 982 err("if/for/while/switch not started ". 983 "on its own line"); 984 } 985 goto section_ended; 986 } 987 } elsif (/\{/) { 988 err("{ while in parens/brackets") if (@cont_paren != 0); 989 err("stuff after {") if ($rest =~ /[^\s}]/); 990 $cont_in = 0; 991 last; 992 } elsif (/\}/) { 993 err("} while in parens/brackets") if (@cont_paren != 0); 994 if (!$cont_special && $rest !~ /^\s*(while|else)\b/) { 995 if ($rest =~ /^$/) { 996 err("unexpected }"); 997 } else { 998 err("stuff after }"); 999 } 1000 $cont_in = 0; 1001 last; 1002 } 1003 } elsif (/\:/ && $cont_case && @cont_paren == 0) { 1004 err("stuff after multi-line case") if ($rest !~ /$^/); 1005 $cont_in = 0; 1006 last; 1007 } 1008 next; 1009section_ended: 1010 # End of a statement or if/while/for loop. Reset 1011 # cont_special and cont_macro based on the rest of the 1012 # line. 1013 $cont_special = ($rest =~ /^\s*$special/)? 1 : 0; 1014 $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0; 1015 $cont_case = 0; 1016 next; 1017 } 1018 $cont_noerr = 0 if (!$cont_in); 1019} 1020