xref: /openbmc/linux/scripts/get_abi.pl (revision 997a5310) 
1#!/usr/bin/env perl
2# SPDX-License-Identifier: GPL-2.0
3
4BEGIN { $Pod::Usage::Formatter = 'Pod::Text::Termcap'; }
5
6use strict;
7use warnings;
8use utf8;
9use Pod::Usage qw(pod2usage);
10use Getopt::Long;
11use File::Find;
12use IO::Handle;
13use Fcntl ':mode';
14use Cwd 'abs_path';
15use Data::Dumper;
16
17my $help = 0;
18my $hint = 0;
19my $man = 0;
20my $debug = 0;
21my $enable_lineno = 0;
22my $show_warnings = 1;
23my $prefix="Documentation/ABI";
24my $sysfs_prefix="/sys";
25my $search_string;
26
27# Debug options
28my $dbg_what_parsing = 1;
29my $dbg_what_open = 2;
30my $dbg_dump_abi_structs = 4;
31my $dbg_undefined = 8;
32
33$Data::Dumper::Indent = 1;
34$Data::Dumper::Terse = 1;
35
36#
37# If true, assumes that the description is formatted with ReST
38#
39my $description_is_rst = 1;
40
41GetOptions(
42	"debug=i" => \$debug,
43	"enable-lineno" => \$enable_lineno,
44	"rst-source!" => \$description_is_rst,
45	"dir=s" => \$prefix,
46	'help|?' => \$help,
47	"show-hints" => \$hint,
48	"search-string=s" => \$search_string,
49	man => \$man
50) or pod2usage(2);
51
52pod2usage(1) if $help;
53pod2usage(-exitstatus => 0, -noperldoc, -verbose => 2) if $man;
54
55pod2usage(2) if (scalar @ARGV < 1 || @ARGV > 2);
56
57my ($cmd, $arg) = @ARGV;
58
59pod2usage(2) if ($cmd ne "search" && $cmd ne "rest" && $cmd ne "validate" && $cmd ne "undefined");
60pod2usage(2) if ($cmd eq "search" && !$arg);
61
62require Data::Dumper if ($debug & $dbg_dump_abi_structs);
63
64my %data;
65my %symbols;
66
67#
68# Displays an error message, printing file name and line
69#
70sub parse_error($$$$) {
71	my ($file, $ln, $msg, $data) = @_;
72
73	return if (!$show_warnings);
74
75	$data =~ s/\s+$/\n/;
76
77	print STDERR "Warning: file $file#$ln:\n\t$msg";
78
79	if ($data ne "") {
80		print STDERR ". Line\n\t\t$data";
81	} else {
82	    print STDERR "\n";
83	}
84}
85
86#
87# Parse an ABI file, storing its contents at %data
88#
89sub parse_abi {
90	my $file = $File::Find::name;
91
92	my $mode = (stat($file))[2];
93	return if ($mode & S_IFDIR);
94	return if ($file =~ m,/README,);
95	return if ($file =~ m,/\.,);
96
97	my $name = $file;
98	$name =~ s,.*/,,;
99
100	my $fn = $file;
101	$fn =~ s,.*Documentation/ABI/,,;
102
103	my $nametag = "File $fn";
104	$data{$nametag}->{what} = "File $name";
105	$data{$nametag}->{type} = "File";
106	$data{$nametag}->{file} = $name;
107	$data{$nametag}->{filepath} = $file;
108	$data{$nametag}->{is_file} = 1;
109	$data{$nametag}->{line_no} = 1;
110
111	my $type = $file;
112	$type =~ s,.*/(.*)/.*,$1,;
113
114	my $what;
115	my $new_what;
116	my $tag = "";
117	my $ln;
118	my $xrefs;
119	my $space;
120	my @labels;
121	my $label = "";
122
123	print STDERR "Opening $file\n" if ($debug & $dbg_what_open);
124	open IN, $file;
125	while(<IN>) {
126		$ln++;
127		if (m/^(\S+)(:\s*)(.*)/i) {
128			my $new_tag = lc($1);
129			my $sep = $2;
130			my $content = $3;
131
132			if (!($new_tag =~ m/(what|where|date|kernelversion|contact|description|users)/)) {
133				if ($tag eq "description") {
134					# New "tag" is actually part of
135					# description. Don't consider it a tag
136					$new_tag = "";
137				} elsif ($tag ne "") {
138					parse_error($file, $ln, "tag '$tag' is invalid", $_);
139				}
140			}
141
142			# Invalid, but it is a common mistake
143			if ($new_tag eq "where") {
144				parse_error($file, $ln, "tag 'Where' is invalid. Should be 'What:' instead", "");
145				$new_tag = "what";
146			}
147
148			if ($new_tag =~ m/what/) {
149				$space = "";
150				$content =~ s/[,.;]$//;
151
152				push @{$symbols{$content}->{file}}, " $file:" . ($ln - 1);
153
154				if ($tag =~ m/what/) {
155					$what .= "\xac" . $content;
156				} else {
157					if ($what) {
158						parse_error($file, $ln, "What '$what' doesn't have a description", "") if (!$data{$what}->{description});
159
160						foreach my $w(split /\xac/, $what) {
161							$symbols{$w}->{xref} = $what;
162						};
163					}
164
165					$what = $content;
166					$label = $content;
167					$new_what = 1;
168				}
169				push @labels, [($content, $label)];
170				$tag = $new_tag;
171
172				push @{$data{$nametag}->{symbols}}, $content if ($data{$nametag}->{what});
173				next;
174			}
175
176			if ($tag ne "" && $new_tag) {
177				$tag = $new_tag;
178
179				if ($new_what) {
180					@{$data{$what}->{label_list}} = @labels if ($data{$nametag}->{what});
181					@labels = ();
182					$label = "";
183					$new_what = 0;
184
185					$data{$what}->{type} = $type;
186					if (!defined($data{$what}->{file})) {
187						$data{$what}->{file} = $name;
188						$data{$what}->{filepath} = $file;
189					} else {
190						$data{$what}->{description} .= "\n\n" if (defined($data{$what}->{description}));
191						if ($name ne $data{$what}->{file}) {
192							$data{$what}->{file} .= " " . $name;
193							$data{$what}->{filepath} .= " " . $file;
194						}
195					}
196					print STDERR "\twhat: $what\n" if ($debug & $dbg_what_parsing);
197					$data{$what}->{line_no} = $ln;
198				} else {
199					$data{$what}->{line_no} = $ln if (!defined($data{$what}->{line_no}));
200				}
201
202				if (!$what) {
203					parse_error($file, $ln, "'What:' should come first:", $_);
204					next;
205				}
206				if ($new_tag eq "description") {
207					$sep =~ s,:, ,;
208					$content = ' ' x length($new_tag) . $sep . $content;
209					while ($content =~ s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e) {}
210					if ($content =~ m/^(\s*)(\S.*)$/) {
211						# Preserve initial spaces for the first line
212						$space = $1;
213						$content = "$2\n";
214						$data{$what}->{$tag} .= $content;
215					} else {
216						undef($space);
217					}
218
219				} else {
220					$data{$what}->{$tag} = $content;
221				}
222				next;
223			}
224		}
225
226		# Store any contents before tags at the database
227		if (!$tag && $data{$nametag}->{what}) {
228			$data{$nametag}->{description} .= $_;
229			next;
230		}
231
232		if ($tag eq "description") {
233			my $content = $_;
234			while ($content =~ s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e) {}
235			if (m/^\s*\n/) {
236				$data{$what}->{$tag} .= "\n";
237				next;
238			}
239
240			if (!defined($space)) {
241				# Preserve initial spaces for the first line
242				if ($content =~ m/^(\s*)(\S.*)$/) {
243					$space = $1;
244					$content = "$2\n";
245				}
246			} else {
247				$space = "" if (!($content =~ s/^($space)//));
248			}
249			$data{$what}->{$tag} .= $content;
250
251			next;
252		}
253		if (m/^\s*(.*)/) {
254			$data{$what}->{$tag} .= "\n$1";
255			$data{$what}->{$tag} =~ s/\n+$//;
256			next;
257		}
258
259		# Everything else is error
260		parse_error($file, $ln, "Unexpected content", $_);
261	}
262	$data{$nametag}->{description} =~ s/^\n+// if ($data{$nametag}->{description});
263	if ($what) {
264		parse_error($file, $ln, "What '$what' doesn't have a description", "") if (!$data{$what}->{description});
265
266		foreach my $w(split /\xac/,$what) {
267			$symbols{$w}->{xref} = $what;
268		};
269	}
270	close IN;
271}
272
273sub create_labels {
274	my %labels;
275
276	foreach my $what (keys %data) {
277		next if ($data{$what}->{file} eq "File");
278
279		foreach my $p (@{$data{$what}->{label_list}}) {
280			my ($content, $label) = @{$p};
281			$label = "abi_" . $label . " ";
282			$label =~ tr/A-Z/a-z/;
283
284			# Convert special chars to "_"
285			$label =~s/([\x00-\x2f\x3a-\x40\x5b-\x60\x7b-\xff])/_/g;
286			$label =~ s,_+,_,g;
287			$label =~ s,_$,,;
288
289			# Avoid duplicated labels
290			while (defined($labels{$label})) {
291			    my @chars = ("A".."Z", "a".."z");
292			    $label .= $chars[rand @chars];
293			}
294			$labels{$label} = 1;
295
296			$data{$what}->{label} = $label;
297
298			# only one label is enough
299			last;
300		}
301	}
302}
303
304#
305# Outputs the book on ReST format
306#
307
308# \b doesn't work well with paths. So, we need to define something else:
309# Boundaries are punct characters, spaces and end-of-line
310my $start = qr {(^|\s|\() }x;
311my $bondary = qr { ([,.:;\)\s]|\z) }x;
312my $xref_match = qr { $start(\/(sys|config|proc|dev|kvd)\/[^,.:;\)\s]+)$bondary }x;
313my $symbols = qr { ([\x01-\x08\x0e-\x1f\x21-\x2f\x3a-\x40\x7b-\xff]) }x;
314
315sub output_rest {
316	create_labels();
317
318	my $part = "";
319
320	foreach my $what (sort {
321				($data{$a}->{type} eq "File") cmp ($data{$b}->{type} eq "File") ||
322				$a cmp $b
323			       } keys %data) {
324		my $type = $data{$what}->{type};
325
326		my @file = split / /, $data{$what}->{file};
327		my @filepath = split / /, $data{$what}->{filepath};
328
329		if ($enable_lineno) {
330			printf ".. LINENO %s%s#%s\n\n",
331			       $prefix, $file[0],
332			       $data{$what}->{line_no};
333		}
334
335		my $w = $what;
336
337		if ($type ne "File") {
338			my $cur_part = $what;
339			if ($what =~ '/') {
340				if ($what =~ m#^(\/?(?:[\w\-]+\/?){1,2})#) {
341					$cur_part = "Symbols under $1";
342					$cur_part =~ s,/$,,;
343				}
344			}
345
346			if ($cur_part ne "" && $part ne $cur_part) {
347			    $part = $cur_part;
348			    my $bar = $part;
349			    $bar =~ s/./-/g;
350			    print "$part\n$bar\n\n";
351			}
352
353			printf ".. _%s:\n\n", $data{$what}->{label};
354
355			my @names = split /\xac/,$w;
356			my $len = 0;
357
358			foreach my $name (@names) {
359				$name =~ s/$symbols/\\$1/g;
360				$name = "**$name**";
361				$len = length($name) if (length($name) > $len);
362			}
363
364			print "+-" . "-" x $len . "-+\n";
365			foreach my $name (@names) {
366				printf "| %s", $name . " " x ($len - length($name)) . " |\n";
367				print "+-" . "-" x $len . "-+\n";
368			}
369
370			print "\n";
371		}
372
373		for (my $i = 0; $i < scalar(@filepath); $i++) {
374			my $path = $filepath[$i];
375			my $f = $file[$i];
376
377			$path =~ s,.*/(.*/.*),$1,;;
378			$path =~ s,[/\-],_,g;;
379			my $fileref = "abi_file_".$path;
380
381			if ($type eq "File") {
382				print ".. _$fileref:\n\n";
383			} else {
384				print "Defined on file :ref:`$f <$fileref>`\n\n";
385			}
386		}
387
388		if ($type eq "File") {
389			my $bar = $w;
390			$bar =~ s/./-/g;
391			print "$w\n$bar\n\n";
392		}
393
394		my $desc = "";
395		$desc = $data{$what}->{description} if (defined($data{$what}->{description}));
396		$desc =~ s/\s+$/\n/;
397
398		if (!($desc =~ /^\s*$/)) {
399			if ($description_is_rst) {
400				# Remove title markups from the description
401				# Having titles inside ABI files will only work if extra
402				# care would be taken in order to strictly follow the same
403				# level order for each markup.
404				$desc =~ s/\n[\-\*\=\^\~]+\n/\n\n/g;
405
406				# Enrich text by creating cross-references
407
408				my $new_desc = "";
409				my $init_indent = -1;
410				my $literal_indent = -1;
411
412				open(my $fh, "+<", \$desc);
413				while (my $d = <$fh>) {
414					my $indent = $d =~ m/^(\s+)/;
415					my $spaces = length($indent);
416					$init_indent = $indent if ($init_indent < 0);
417					if ($literal_indent >= 0) {
418						if ($spaces > $literal_indent) {
419							$new_desc .= $d;
420							next;
421						} else {
422							$literal_indent = -1;
423						}
424					} else {
425						if ($d =~ /()::$/ && !($d =~ /^\s*\.\./)) {
426							$literal_indent = $spaces;
427						}
428					}
429
430					$d =~ s,Documentation/(?!devicetree)(\S+)\.rst,:doc:`/$1`,g;
431
432					my @matches = $d =~ m,Documentation/ABI/([\w\/\-]+),g;
433					foreach my $f (@matches) {
434						my $xref = $f;
435						my $path = $f;
436						$path =~ s,.*/(.*/.*),$1,;;
437						$path =~ s,[/\-],_,g;;
438						$xref .= " <abi_file_" . $path . ">";
439						$d =~ s,\bDocumentation/ABI/$f\b,:ref:`$xref`,g;
440					}
441
442					# Seek for cross reference symbols like /sys/...
443					@matches = $d =~ m/$xref_match/g;
444
445					foreach my $s (@matches) {
446						next if (!($s =~ m,/,));
447						if (defined($data{$s}) && defined($data{$s}->{label})) {
448							my $xref = $s;
449
450							$xref =~ s/$symbols/\\$1/g;
451							$xref = ":ref:`$xref <" . $data{$s}->{label} . ">`";
452
453							$d =~ s,$start$s$bondary,$1$xref$2,g;
454						}
455					}
456					$new_desc .= $d;
457				}
458				close $fh;
459
460
461				print "$new_desc\n\n";
462			} else {
463				$desc =~ s/^\s+//;
464
465				# Remove title markups from the description, as they won't work
466				$desc =~ s/\n[\-\*\=\^\~]+\n/\n\n/g;
467
468				if ($desc =~ m/\:\n/ || $desc =~ m/\n[\t ]+/  || $desc =~ m/[\x00-\x08\x0b-\x1f\x7b-\xff]/) {
469					# put everything inside a code block
470					$desc =~ s/\n/\n /g;
471
472					print "::\n\n";
473					print " $desc\n\n";
474				} else {
475					# Escape any special chars from description
476					$desc =~s/([\x00-\x08\x0b-\x1f\x21-\x2a\x2d\x2f\x3c-\x40\x5c\x5e-\x60\x7b-\xff])/\\$1/g;
477					print "$desc\n\n";
478				}
479			}
480		} else {
481			print "DESCRIPTION MISSING for $what\n\n" if (!$data{$what}->{is_file});
482		}
483
484		if ($data{$what}->{symbols}) {
485			printf "Has the following ABI:\n\n";
486
487			foreach my $content(@{$data{$what}->{symbols}}) {
488				my $label = $data{$symbols{$content}->{xref}}->{label};
489
490				# Escape special chars from content
491				$content =~s/([\x00-\x1f\x21-\x2f\x3a-\x40\x7b-\xff])/\\$1/g;
492
493				print "- :ref:`$content <$label>`\n\n";
494			}
495		}
496
497		if (defined($data{$what}->{users})) {
498			my $users = $data{$what}->{users};
499
500			$users =~ s/\n/\n\t/g;
501			printf "Users:\n\t%s\n\n", $users if ($users ne "");
502		}
503
504	}
505}
506
507#
508# Searches for ABI symbols
509#
510sub search_symbols {
511	foreach my $what (sort keys %data) {
512		next if (!($what =~ m/($arg)/));
513
514		my $type = $data{$what}->{type};
515		next if ($type eq "File");
516
517		my $file = $data{$what}->{filepath};
518
519		$what =~ s/\xac/, /g;
520		my $bar = $what;
521		$bar =~ s/./-/g;
522
523		print "\n$what\n$bar\n\n";
524
525		my $kernelversion = $data{$what}->{kernelversion} if (defined($data{$what}->{kernelversion}));
526		my $contact = $data{$what}->{contact} if (defined($data{$what}->{contact}));
527		my $users = $data{$what}->{users} if (defined($data{$what}->{users}));
528		my $date = $data{$what}->{date} if (defined($data{$what}->{date}));
529		my $desc = $data{$what}->{description} if (defined($data{$what}->{description}));
530
531		$kernelversion =~ s/^\s+// if ($kernelversion);
532		$contact =~ s/^\s+// if ($contact);
533		if ($users) {
534			$users =~ s/^\s+//;
535			$users =~ s/\n//g;
536		}
537		$date =~ s/^\s+// if ($date);
538		$desc =~ s/^\s+// if ($desc);
539
540		printf "Kernel version:\t\t%s\n", $kernelversion if ($kernelversion);
541		printf "Date:\t\t\t%s\n", $date if ($date);
542		printf "Contact:\t\t%s\n", $contact if ($contact);
543		printf "Users:\t\t\t%s\n", $users if ($users);
544		print "Defined on file(s):\t$file\n\n";
545		print "Description:\n\n$desc";
546	}
547}
548
549# Exclude /sys/kernel/debug and /sys/kernel/tracing from the search path
550sub dont_parse_special_attributes {
551	if (($File::Find::dir =~ m,^/sys/kernel,)) {
552		return grep {!/(debug|tracing)/ } @_;
553	}
554
555	if (($File::Find::dir =~ m,^/sys/fs,)) {
556		return grep {!/(pstore|bpf|fuse)/ } @_;
557	}
558
559	return @_
560}
561
562my %leaf;
563my %aliases;
564my @files;
565my %root;
566
567sub graph_add_file {
568	my $file = shift;
569	my $type = shift;
570
571	my $dir = $file;
572	$dir =~ s,^(.*/).*,$1,;
573	$file =~ s,.*/,,;
574
575	my $name;
576	my $file_ref = \%root;
577	foreach my $edge(split "/", $dir) {
578		$name .= "$edge/";
579		if (!defined ${$file_ref}{$edge}) {
580			${$file_ref}{$edge} = { };
581		}
582		$file_ref = \%{$$file_ref{$edge}};
583		${$file_ref}{"__name"} = [ $name ];
584	}
585	$name .= "$file";
586	${$file_ref}{$file} = {
587		"__name" => [ $name ]
588	};
589
590	return \%{$$file_ref{$file}};
591}
592
593sub graph_add_link {
594	my $file = shift;
595	my $link = shift;
596
597	# Traverse graph to find the reference
598	my $file_ref = \%root;
599	foreach my $edge(split "/", $file) {
600		$file_ref = \%{$$file_ref{$edge}} || die "Missing node!";
601	}
602
603	# do a BFS
604
605	my @queue;
606	my %seen;
607	my $st;
608
609	push @queue, $file_ref;
610	$seen{$start}++;
611
612	while (@queue) {
613		my $v = shift @queue;
614		my @child = keys(%{$v});
615
616		foreach my $c(@child) {
617			next if $seen{$$v{$c}};
618			next if ($c eq "__name");
619
620			if (!defined($$v{$c}{"__name"})) {
621				printf STDERR "Error: Couldn't find a non-empty name on a children of $file/.*: ";
622				print STDERR Dumper(%{$v});
623				exit;
624			}
625
626			# Add new name
627			my $name = @{$$v{$c}{"__name"}}[0];
628			if ($name =~ s#^$file/#$link/#) {
629				push @{$$v{$c}{"__name"}}, $name;
630			}
631			# Add child to the queue and mark as seen
632			push @queue, $$v{$c};
633			$seen{$c}++;
634		}
635	}
636}
637
638my $escape_symbols = qr { ([\x01-\x08\x0e-\x1f\x21-\x29\x2b-\x2d\x3a-\x40\x7b-\xfe]) }x;
639sub parse_existing_sysfs {
640	my $file = $File::Find::name;
641
642	my $mode = (lstat($file))[2];
643	my $abs_file = abs_path($file);
644
645	my @tmp;
646	push @tmp, $file;
647	push @tmp, $abs_file if ($abs_file ne $file);
648
649	foreach my $f(@tmp) {
650		# Ignore cgroup, as this is big and has zero docs under ABI
651		return if ($f =~ m#^/sys/fs/cgroup/#);
652
653		# Ignore firmware as it is documented elsewhere
654		# Either ACPI or under Documentation/devicetree/bindings/
655		return if ($f =~ m#^/sys/firmware/#);
656
657		# Ignore some sysfs nodes that aren't actually part of ABI
658		return if ($f =~ m#/sections|notes/#);
659
660		# Would need to check at
661		# Documentation/admin-guide/kernel-parameters.txt, but this
662		# is not easily parseable.
663		return if ($f =~ m#/parameters/#);
664	}
665
666	if (S_ISLNK($mode)) {
667		$aliases{$file} = $abs_file;
668		return;
669	}
670
671	return if (S_ISDIR($mode));
672
673	# Trivial: file is defined exactly the same way at ABI What:
674	return if (defined($data{$file}));
675	return if (defined($data{$abs_file}));
676
677	push @files, graph_add_file($abs_file, "file");
678}
679
680sub get_leave($)
681{
682	my $what = shift;
683	my $leave;
684
685	my $l = $what;
686	my $stop = 1;
687
688	$leave = $l;
689	$leave =~ s,/$,,;
690	$leave =~ s,.*/,,;
691	$leave =~ s/[\(\)]//g;
692
693	# $leave is used to improve search performance at
694	# check_undefined_symbols, as the algorithm there can seek
695	# for a small number of "what". It also allows giving a
696	# hint about a leave with the same name somewhere else.
697	# However, there are a few occurences where the leave is
698	# either a wildcard or a number. Just group such cases
699	# altogether.
700	if ($leave =~ m/\.\*/ || $leave eq "" || $leave =~ /\\d/) {
701		$leave = "others";
702	}
703
704	return $leave;
705}
706
707my @not_found;
708
709sub check_file($$)
710{
711	my $file_ref = shift;
712	my $names_ref = shift;
713	my @names = @{$names_ref};
714	my $file = $names[0];
715
716	my $found_string;
717
718	my $leave = get_leave($file);
719	if (!defined($leaf{$leave})) {
720		$leave = "others";
721	}
722	my @expr = @{$leaf{$leave}->{expr}};
723	die ("\rmissing rules for $leave") if (!defined($leaf{$leave}));
724
725	my $path = $file;
726	$path =~ s,(.*/).*,$1,;
727
728	if ($search_string) {
729		return if (!($file =~ m#$search_string#));
730		$found_string = 1;
731	}
732
733	for (my $i = 0; $i < @names; $i++) {
734		if ($found_string && $hint) {
735			if (!$i) {
736				print STDERR "--> $names[$i]\n";
737			} else {
738				print STDERR "    $names[$i]\n";
739			}
740		}
741		foreach my $re (@expr) {
742			print STDERR "$names[$i] =~ /^$re\$/\n" if ($debug && $dbg_undefined);
743			if ($names[$i] =~ $re) {
744				return;
745			}
746		}
747	}
748
749	if ($leave ne "others") {
750		my @expr = @{$leaf{"others"}->{expr}};
751		for (my $i = 0; $i < @names; $i++) {
752			foreach my $re (@expr) {
753				print STDERR "$names[$i] =~ /^$re\$/\n" if ($debug && $dbg_undefined);
754				if ($names[$i] =~ $re) {
755					return;
756				}
757			}
758		}
759	}
760
761	push @not_found, $file if (!$search_string || $found_string);
762
763	if ($hint && (!$search_string || $found_string)) {
764		my $what = $leaf{$leave}->{what};
765		$what =~ s/\xac/\n\t/g;
766		if ($leave ne "others") {
767			print STDERR "\r    more likely regexes:\n\t$what\n";
768		} else {
769			print STDERR "\r    tested regexes:\n\t$what\n";
770		}
771	}
772}
773
774sub check_undefined_symbols {
775	my $num_files = scalar @files;
776	my $next_i = 0;
777	my $start_time = times;
778
779	@files = sort @files;
780
781	my $last_time = $start_time;
782
783	# When either debug or hint is enabled, there's no sense showing
784	# progress, as the progress will be overriden.
785	if ($hint || ($debug && $dbg_undefined)) {
786		$next_i = $num_files;
787	}
788
789	my $is_console;
790	$is_console = 1 if (-t STDERR);
791
792	for (my $i = 0; $i < $num_files; $i++) {
793		my $file_ref = $files[$i];
794		my @names = @{$$file_ref{"__name"}};
795
796		check_file($file_ref, \@names);
797
798		my $cur_time = times;
799
800		if ($i == $next_i || $cur_time > $last_time + 1) {
801			my $percent = $i * 100 / $num_files;
802
803			my $tm = $cur_time - $start_time;
804			my $time = sprintf "%d:%02d", int($tm), 60 * ($tm - int($tm));
805
806			printf STDERR "\33[2K\r", if ($is_console);
807			printf STDERR "%s: processing sysfs files... %i%%: $names[0]", $time, $percent;
808			printf STDERR "\n", if (!$is_console);
809			STDERR->flush();
810
811			$next_i = int (($percent + 1) * $num_files / 100);
812			$last_time = $cur_time;
813		}
814	}
815
816	my $cur_time = times;
817	my $tm = $cur_time - $start_time;
818	my $time = sprintf "%d:%02d", int($tm), 60 * ($tm - int($tm));
819
820	printf STDERR "\33[2K\r", if ($is_console);
821	printf STDERR "%s: processing sysfs files... done\n", $time;
822
823	foreach my $file (@not_found) {
824		print "$file not found.\n";
825	}
826}
827
828sub undefined_symbols {
829	print STDERR "Reading $sysfs_prefix directory contents...";
830	find({
831		wanted =>\&parse_existing_sysfs,
832		preprocess =>\&dont_parse_special_attributes,
833		no_chdir => 1
834	     }, $sysfs_prefix);
835	print STDERR "done.\n";
836
837	$leaf{"others"}->{what} = "";
838
839	print STDERR "Converting ABI What fields into regexes...";
840	foreach my $w (sort keys %data) {
841		foreach my $what (split /\xac/,$w) {
842			next if (!($what =~ m/^$sysfs_prefix/));
843
844			# Convert what into regular expressions
845
846			# Escape dot characters
847			$what =~ s/\./\xf6/g;
848
849			# Temporarily change [0-9]+ type of patterns
850			$what =~ s/\[0\-9\]\+/\xff/g;
851
852			# Temporarily change [\d+-\d+] type of patterns
853			$what =~ s/\[0\-\d+\]/\xff/g;
854			$what =~ s/\[(\d+)\]/\xf4$1\xf5/g;
855
856			# Temporarily change [0-9] type of patterns
857			$what =~ s/\[(\d)\-(\d)\]/\xf4$1-$2\xf5/g;
858
859			# Handle multiple option patterns
860			$what =~ s/[\{\<\[]([\w_]+)(?:[,|]+([\w_]+)){1,}[\}\>\]]/($1|$2)/g;
861
862			# Handle wildcards
863			$what =~ s,\*,.*,g;
864			$what =~ s,/\xf6..,/.*,g;
865			$what =~ s/\<[^\>]+\>/.*/g;
866			$what =~ s/\{[^\}]+\}/.*/g;
867			$what =~ s/\[[^\]]+\]/.*/g;
868
869			$what =~ s/[XYZ]/.*/g;
870
871			# Recover [0-9] type of patterns
872			$what =~ s/\xf4/[/g;
873			$what =~ s/\xf5/]/g;
874
875			# Remove duplicated spaces
876			$what =~ s/\s+/ /g;
877
878			# Special case: this ABI has a parenthesis on it
879			$what =~ s/sqrt\(x^2\+y^2\+z^2\)/sqrt\(x^2\+y^2\+z^2\)/;
880
881			# Special case: drop comparition as in:
882			#	What: foo = <something>
883			# (this happens on a few IIO definitions)
884			$what =~ s,\s*\=.*$,,;
885
886			# Escape all other symbols
887			$what =~ s/$escape_symbols/\\$1/g;
888			$what =~ s/\\\\/\\/g;
889			$what =~ s/\\([\[\]\(\)\|])/$1/g;
890			$what =~ s/(\d+)\\(-\d+)/$1$2/g;
891
892			$what =~ s/\xff/\\d+/g;
893
894			# Special case: IIO ABI which a parenthesis.
895			$what =~ s/sqrt(.*)/sqrt\(.*\)/;
896
897			# Simplify regexes with multiple .*
898			$what =~ s#(?:\.\*){2,}##g;
899#			$what =~ s#\.\*/\.\*#.*#g;
900
901			# Recover dot characters
902			$what =~ s/\xf6/\./g;
903
904			my $leave = get_leave($what);
905
906			my $added = 0;
907			foreach my $l (split /\|/, $leave) {
908				if (defined($leaf{$l})) {
909					next if ($leaf{$l}->{what} =~ m/\b$what\b/);
910					$leaf{$l}->{what} .= "\xac" . $what;
911					$added = 1;
912				} else {
913					$leaf{$l}->{what} = $what;
914					$added = 1;
915				}
916			}
917			if ($search_string && $added) {
918				print STDERR "What: $what\n" if ($what =~ m#$search_string#);
919			}
920
921		}
922	}
923	# Compile regexes
924	foreach my $l (sort keys %leaf) {
925		my @expr;
926		foreach my $w(sort split /\xac/, $leaf{$l}->{what}) {
927			push @expr, qr /^$w$/;
928		}
929		$leaf{$l}->{expr} = \@expr;
930	}
931
932	# Take links into account
933	foreach my $link (sort keys %aliases) {
934		my $abs_file = $aliases{$link};
935		graph_add_link($abs_file, $link);
936	}
937	print STDERR "done.\n";
938
939	check_undefined_symbols;
940}
941
942# Ensure that the prefix will always end with a slash
943# While this is not needed for find, it makes the patch nicer
944# with --enable-lineno
945$prefix =~ s,/?$,/,;
946
947if ($cmd eq "undefined" || $cmd eq "search") {
948	$show_warnings = 0;
949}
950#
951# Parses all ABI files located at $prefix dir
952#
953find({wanted =>\&parse_abi, no_chdir => 1}, $prefix);
954
955print STDERR Data::Dumper->Dump([\%data], [qw(*data)]) if ($debug & $dbg_dump_abi_structs);
956
957#
958# Handles the command
959#
960if ($cmd eq "undefined") {
961	undefined_symbols;
962} elsif ($cmd eq "search") {
963	search_symbols;
964} else {
965	if ($cmd eq "rest") {
966		output_rest;
967	}
968
969	# Warn about duplicated ABI entries
970	foreach my $what(sort keys %symbols) {
971		my @files = @{$symbols{$what}->{file}};
972
973		next if (scalar(@files) == 1);
974
975		printf STDERR "Warning: $what is defined %d times: @files\n",
976		    scalar(@files);
977	}
978}
979
980__END__
981
982=head1 NAME
983
984get_abi.pl - parse the Linux ABI files and produce a ReST book.
985
986=head1 SYNOPSIS
987
988B<get_abi.pl> [--debug <level>] [--enable-lineno] [--man] [--help]
989	       [--(no-)rst-source] [--dir=<dir>] [--show-hints]
990	       [--search-string <regex>]
991	       <COMMAND> [<ARGUMENT>]
992
993Where B<COMMAND> can be:
994
995=over 8
996
997B<search> I<SEARCH_REGEX> - search for I<SEARCH_REGEX> inside ABI
998
999B<rest>                   - output the ABI in ReST markup language
1000
1001B<validate>               - validate the ABI contents
1002
1003B<undefined>              - existing symbols at the system that aren't
1004                            defined at Documentation/ABI
1005
1006=back
1007
1008=head1 OPTIONS
1009
1010=over 8
1011
1012=item B<--dir>
1013
1014Changes the location of the ABI search. By default, it uses
1015the Documentation/ABI directory.
1016
1017=item B<--rst-source> and B<--no-rst-source>
1018
1019The input file may be using ReST syntax or not. Those two options allow
1020selecting between a rst-compliant source ABI (B<--rst-source>), or a
1021plain text that may be violating ReST spec, so it requres some escaping
1022logic (B<--no-rst-source>).
1023
1024=item B<--enable-lineno>
1025
1026Enable output of .. LINENO lines.
1027
1028=item B<--debug> I<debug level>
1029
1030Print debug information according with the level, which is given by the
1031following bitmask:
1032
1033    -  1: Debug parsing What entries from ABI files;
1034    -  2: Shows what files are opened from ABI files;
1035    -  4: Dump the structs used to store the contents of the ABI files.
1036
1037=item B<--show-hints>
1038
1039Show hints about possible definitions for the missing ABI symbols.
1040Used only when B<undefined>.
1041
1042=item B<--search-string> I<regex string>
1043
1044Show only occurences that match a search string.
1045Used only when B<undefined>.
1046
1047=item B<--help>
1048
1049Prints a brief help message and exits.
1050
1051=item B<--man>
1052
1053Prints the manual page and exits.
1054
1055=back
1056
1057=head1 DESCRIPTION
1058
1059Parse the Linux ABI files from ABI DIR (usually located at Documentation/ABI),
1060allowing to search for ABI symbols or to produce a ReST book containing
1061the Linux ABI documentation.
1062
1063=head1 EXAMPLES
1064
1065Search for all stable symbols with the word "usb":
1066
1067=over 8
1068
1069$ scripts/get_abi.pl search usb --dir Documentation/ABI/stable
1070
1071=back
1072
1073Search for all symbols that match the regex expression "usb.*cap":
1074
1075=over 8
1076
1077$ scripts/get_abi.pl search usb.*cap
1078
1079=back
1080
1081Output all obsoleted symbols in ReST format
1082
1083=over 8
1084
1085$ scripts/get_abi.pl rest --dir Documentation/ABI/obsolete
1086
1087=back
1088
1089=head1 BUGS
1090
1091Report bugs to Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
1092
1093=head1 COPYRIGHT
1094
1095Copyright (c) 2016-2021 by Mauro Carvalho Chehab <mchehab+huawei@kernel.org>.
1096
1097License GPLv2: GNU GPL version 2 <http://gnu.org/licenses/gpl.html>.
1098
1099This is free software: you are free to change and redistribute it.
1100There is NO WARRANTY, to the extent permitted by law.
1101
1102=cut
1103