Merge tag 'kvm-x86-mmu-6.4-2' of https://github.com/kvm-x86/linux into HEAD
[platform/kernel/linux-rpi.git] / scripts / get_abi.pl
1 #!/usr/bin/env perl
2 # SPDX-License-Identifier: GPL-2.0
3
4 BEGIN { $Pod::Usage::Formatter = 'Pod::Text::Termcap'; }
5
6 use strict;
7 use warnings;
8 use utf8;
9 use Pod::Usage qw(pod2usage);
10 use Getopt::Long;
11 use File::Find;
12 use IO::Handle;
13 use Fcntl ':mode';
14 use Cwd 'abs_path';
15 use Data::Dumper;
16
17 my $help = 0;
18 my $hint = 0;
19 my $man = 0;
20 my $debug = 0;
21 my $enable_lineno = 0;
22 my $show_warnings = 1;
23 my $prefix="Documentation/ABI";
24 my $sysfs_prefix="/sys";
25 my $search_string;
26
27 # Debug options
28 my $dbg_what_parsing = 1;
29 my $dbg_what_open = 2;
30 my $dbg_dump_abi_structs = 4;
31 my $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 #
39 my $description_is_rst = 1;
40
41 GetOptions(
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
52 pod2usage(1) if $help;
53 pod2usage(-exitstatus => 0, -noperldoc, -verbose => 2) if $man;
54
55 pod2usage(2) if (scalar @ARGV < 1 || @ARGV > 2);
56
57 my ($cmd, $arg) = @ARGV;
58
59 pod2usage(2) if ($cmd ne "search" && $cmd ne "rest" && $cmd ne "validate" && $cmd ne "undefined");
60 pod2usage(2) if ($cmd eq "search" && !$arg);
61
62 require Data::Dumper if ($debug & $dbg_dump_abi_structs);
63
64 my %data;
65 my %symbols;
66
67 #
68 # Displays an error message, printing file name and line
69 #
70 sub 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 #
89 sub 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
273 sub 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
310 my $start = qr {(^|\s|\() }x;
311 my $bondary = qr { ([,.:;\)\s]|\z) }x;
312 my $xref_match = qr { $start(\/(sys|config|proc|dev|kvd)\/[^,.:;\)\s]+)$bondary }x;
313 my $symbols = qr { ([\x01-\x08\x0e-\x1f\x21-\x2f\x3a-\x40\x7b-\xff]) }x;
314
315 sub 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 #
510 sub 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
550 sub 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
562 my %leaf;
563 my %aliases;
564 my @files;
565 my %root;
566
567 sub 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
593 sub 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
638 my $escape_symbols = qr { ([\x01-\x08\x0e-\x1f\x21-\x29\x2b-\x2d\x3a-\x40\x7b-\xfe]) }x;
639 sub 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($abs_file, "file");
678 }
679
680 sub get_leave($)
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
707 my @not_found;
708
709 sub 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
774 sub 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
828 sub 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
947 if ($cmd eq "undefined" || $cmd eq "search") {
948         $show_warnings = 0;
949 }
950 #
951 # Parses all ABI files located at $prefix dir
952 #
953 find({wanted =>\&parse_abi, no_chdir => 1}, $prefix);
954
955 print STDERR Data::Dumper->Dump([\%data], [qw(*data)]) if ($debug & $dbg_dump_abi_structs);
956
957 #
958 # Handles the command
959 #
960 if ($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
984 get_abi.pl - parse the Linux ABI files and produce a ReST book.
985
986 =head1 SYNOPSIS
987
988 B<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
993 Where B<COMMAND> can be:
994
995 =over 8
996
997 B<search> I<SEARCH_REGEX> - search for I<SEARCH_REGEX> inside ABI
998
999 B<rest>                   - output the ABI in ReST markup language
1000
1001 B<validate>               - validate the ABI contents
1002
1003 B<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
1014 Changes the location of the ABI search. By default, it uses
1015 the Documentation/ABI directory.
1016
1017 =item B<--rst-source> and B<--no-rst-source>
1018
1019 The input file may be using ReST syntax or not. Those two options allow
1020 selecting between a rst-compliant source ABI (B<--rst-source>), or a
1021 plain text that may be violating ReST spec, so it requres some escaping
1022 logic (B<--no-rst-source>).
1023
1024 =item B<--enable-lineno>
1025
1026 Enable output of .. LINENO lines.
1027
1028 =item B<--debug> I<debug level>
1029
1030 Print debug information according with the level, which is given by the
1031 following 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
1039 Show hints about possible definitions for the missing ABI symbols.
1040 Used only when B<undefined>.
1041
1042 =item B<--search-string> I<regex string>
1043
1044 Show only occurences that match a search string.
1045 Used only when B<undefined>.
1046
1047 =item B<--help>
1048
1049 Prints a brief help message and exits.
1050
1051 =item B<--man>
1052
1053 Prints the manual page and exits.
1054
1055 =back
1056
1057 =head1 DESCRIPTION
1058
1059 Parse the Linux ABI files from ABI DIR (usually located at Documentation/ABI),
1060 allowing to search for ABI symbols or to produce a ReST book containing
1061 the Linux ABI documentation.
1062
1063 =head1 EXAMPLES
1064
1065 Search 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
1073 Search 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
1081 Output 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
1091 Report bugs to Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
1092
1093 =head1 COPYRIGHT
1094
1095 Copyright (c) 2016-2021 by Mauro Carvalho Chehab <mchehab+huawei@kernel.org>.
1096
1097 License GPLv2: GNU GPL version 2 <http://gnu.org/licenses/gpl.html>.
1098
1099 This is free software: you are free to change and redistribute it.
1100 There is NO WARRANTY, to the extent permitted by law.
1101
1102 =cut