1 |
#!/usr/bin/perl -w |
2 |
use strict; |
3 |
use threads; |
4 |
use threads::shared; |
5 |
|
6 |
use Fuse; |
7 |
use POSIX qw(ENOENT ENOSYS EEXIST EPERM O_RDONLY O_RDWR O_APPEND O_CREAT); |
8 |
use Fcntl qw(S_ISBLK S_ISCHR S_ISFIFO SEEK_SET); |
9 |
require 'syscall.ph'; # for SYS_mknod and SYS_lchown |
10 |
use PerlIO::gzip; |
11 |
use File::Path; |
12 |
use Data::Dump qw/dump/; |
13 |
use Carp qw/confess/; |
14 |
use IO::File; |
15 |
use Getopt::Long; |
16 |
|
17 |
my $debug = 0; |
18 |
my $fuse_debug = 0; |
19 |
my $stats = 1; |
20 |
|
21 |
GetOptions( |
22 |
'debug+' => \$debug, |
23 |
'fuse-debug+' => \$fuse_debug, |
24 |
'stats!' => \$stats, |
25 |
); |
26 |
|
27 |
my $mount = { |
28 |
from => '/tmp/comp', |
29 |
to => '/tmp/no-comp', |
30 |
tmp => '/dev/shm/comp', |
31 |
}; |
32 |
|
33 |
my $skip_extensions_regex = qr/gz|gz%|\.(?:sw[a-z]|gif|png|jpeg|jpg|avi|rar|zip|bz2|tgz|avi|mpeg|mpg|tmp|temp)$/i; |
34 |
|
35 |
# don't compress files smaller than this |
36 |
my $min_compress_size = 512; |
37 |
|
38 |
foreach my $dir ( keys %$mount ) { |
39 |
if ( ! -e $mount->{$dir} ) { |
40 |
warn "created $mount->{$dir}\n"; |
41 |
mkdir $mount->{$dir} || die "can't create $mount->{$dir}: $!"; |
42 |
} |
43 |
} |
44 |
|
45 |
my $pending; |
46 |
|
47 |
sub real_name { |
48 |
my ( $dir, $name ) = @_; |
49 |
if ( -e "$dir/${name}.gz" ) { |
50 |
confess "ASSERT: unexpected $dir/$name exists" if -e "$dir/$name"; |
51 |
return "${name}.gz"; |
52 |
} |
53 |
if ( $name =~ m/\.gz$/ ) { |
54 |
return $name . '%'; # protect (mingle) compressed files |
55 |
} else { |
56 |
return $name; |
57 |
} |
58 |
} |
59 |
|
60 |
sub fixup { |
61 |
my ( $path ) = @_; |
62 |
return $mount->{from} . '/' . real_name( $mount->{from}, $path ); |
63 |
} |
64 |
|
65 |
sub original_name { |
66 |
my $p = shift; |
67 |
$p =~ s/\.gz$//; |
68 |
$p =~ s/\.gz%$/.gz/; # demungle compressed .gz files |
69 |
return $p; |
70 |
}; |
71 |
|
72 |
sub gzip_original_size { |
73 |
my $file = shift; |
74 |
|
75 |
return unless -e $file; |
76 |
|
77 |
my $buff; |
78 |
|
79 |
open(my $fh, $file) || die "can't open $file: $!"; |
80 |
|
81 |
# read($fh, $buff, 10 ); |
82 |
# print dump( unpack("nccVcc", $buff ) ); |
83 |
|
84 |
seek($fh, -4, 2); |
85 |
read($fh, $buff, 4 ); |
86 |
close($fh); |
87 |
|
88 |
return unpack("L", $buff); |
89 |
} |
90 |
|
91 |
sub x_getattr { |
92 |
my ($file) = fixup(shift); |
93 |
my (@list) = lstat($file); |
94 |
$list[7] = gzip_original_size( $file ) if ( $list[7] && $file =~ m/\.gz$/ ); |
95 |
return -$! unless @list; |
96 |
return @list; |
97 |
} |
98 |
|
99 |
sub x_getdir { |
100 |
my ($dirname) = fixup(shift); |
101 |
unless(opendir(DIRHANDLE,$dirname)) { |
102 |
return -ENOENT(); |
103 |
} |
104 |
my @files = map { original_name( $_ ) } readdir(DIRHANDLE); |
105 |
closedir(DIRHANDLE); |
106 |
return (@files, 0); |
107 |
} |
108 |
|
109 |
sub file_copy { |
110 |
my ( $s_opt, $s_path, $d_opt, $d_path ) = @_; |
111 |
warn "## file_copy( $s_opt $s_path [",-s $s_path,"] $d_opt $d_path [",-e $d_path ? -s $d_path : 'new',"])\n" if $debug; |
112 |
open(my $s, $s_opt, $s_path ) || confess "can't open $s_path: $!\npending = ", dump( $pending ); |
113 |
open(my $d, $d_opt, $d_path ) || confess "can't open $d_path: $!"; |
114 |
my $buff; |
115 |
while( read( $s, $buff, 65535 ) ) { |
116 |
print $d $buff || confess "can't write into $d_path: $!"; |
117 |
warn ">> [", length($buff), "] offset ", tell($s), " -> ", tell($d), "\n" if $debug; |
118 |
} |
119 |
close($d) || warn "can't close $d_path: $!"; |
120 |
close($s) || warn "can't close $s_path: $!"; |
121 |
warn "-- $s_path [", -s $s_path, "] >>> $d_path [", -s $d_path, "]\n" if $debug; |
122 |
my ($mode,$uid,$gid,$atime,$mtime) = (stat $s_path)[2,4,5,8,9]; |
123 |
|
124 |
chmod $mode, $d_path || warn "chmod( $mode $d_path ) failed: $!\n"; |
125 |
chown $uid,$gid,$d_path || warn "chown( $uid $gid $d_path ) failed: $!\n"; |
126 |
utime $atime,$mtime,$d_path || warn "utime( $atime $mtime $d_path ) failed: $!\n"; |
127 |
|
128 |
undef $d; |
129 |
undef $s; |
130 |
} |
131 |
|
132 |
sub tmp_path { |
133 |
my $file = shift; |
134 |
|
135 |
my $path = fixup( $file ); |
136 |
|
137 |
my $op = 'UNKNOWN'; |
138 |
|
139 |
if (defined( $pending->{$file} )) { |
140 |
$path = $pending->{$file}->{path} || confess "no path for $file in ",dump( $pending ); |
141 |
$op = 'opened'; |
142 |
} else { |
143 |
my $tmp = $mount->{tmp} . '/' . $file; |
144 |
if ( -e $tmp ) { |
145 |
$path = $tmp; |
146 |
$op = 'existing'; |
147 |
} elsif ( $path =~ m/\.gz$/ ) { |
148 |
my $dest_path = $tmp; |
149 |
$dest_path =~ s!/[^/]+$!!; #!vim-fix |
150 |
mkpath $dest_path unless -e $dest_path; |
151 |
if ( -s $path ) { |
152 |
file_copy( '<:gzip', $path, '>', $tmp ) |
153 |
} else { |
154 |
confess "ASSERT: filesystem corruption, $path is zero size in ",dump( $pending ); |
155 |
} |
156 |
$path = $tmp; |
157 |
$op = 'created'; |
158 |
} |
159 |
confess "ASSERT: path shouldn't exist for $file in ", dump( $pending ) if defined( $pending->{$file}->{path} ); |
160 |
confess "ASSERT: open shouldn't exist for $file in ", dump( $pending ) if defined( $pending->{$file}->{open} ); |
161 |
$pending->{$file}->{path} = $path; |
162 |
$pending->{$file}->{open} = 0; # not really opened, just uncompressed |
163 |
warn "## tmp_file( $file ) $op $path [", -s $path, "]\n" if $debug; |
164 |
} |
165 |
return $path; |
166 |
} |
167 |
|
168 |
sub compress_file2path { |
169 |
my ( $file, $path ) = @_; |
170 |
|
171 |
my $dest = fixup( $file ); |
172 |
|
173 |
if ( defined($pending->{$file}) ) { |
174 |
my $pending_path = $pending->{$file}->{path} || confess "no path for $file in ",dump( $pending ); |
175 |
|
176 |
if ( $pending->{$file}->{open} > 1 ) { |
177 |
warn "$file used ", $pending->{$file}->{open}, " times, delaying compression\n"; |
178 |
return; |
179 |
} elsif ( ! $path ) { |
180 |
$path = $pending_path; |
181 |
} elsif ( $pending_path ne $path ) { |
182 |
confess "ASSERT: compressing into $path instead of $pending_path"; |
183 |
} |
184 |
} |
185 |
|
186 |
confess "need path" unless $path; |
187 |
|
188 |
# cleanup old compressed copy |
189 |
if ( $dest =~ /\.gz$/ ) { |
190 |
warn "## remove old $dest\n"; |
191 |
unlink $dest || confess "can't remove $dest: $!"; |
192 |
$dest =~ s/\.gz$//; |
193 |
confess "ASSERT: uncompressed $dest shouldn't exist!" if -e $dest; |
194 |
} |
195 |
|
196 |
if ( $path =~ $skip_extensions_regex ) { |
197 |
warn "$path [",-s $path,"] skipped compression\n"; |
198 |
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
199 |
} elsif ( -s $path < $min_compress_size ) { |
200 |
warn "$path [",-s $path,"] uncompressed, too small\n"; |
201 |
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
202 |
} else { |
203 |
warn "$path [",-s $path,"] compressing\n"; |
204 |
|
205 |
my $comp = $dest . '.gz'; |
206 |
file_copy( '<', $path, '>:gzip', $comp ); |
207 |
|
208 |
my ( $size_path, $size_comp ) = ( -s $path, -s $comp ); |
209 |
|
210 |
if ( $size_path <= $size_comp ) { |
211 |
warn ">>> $size_path <= $size_comp leaving uncompressed $dest\n"; |
212 |
unlink $comp || confess "can't remove: $comp: $!"; |
213 |
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
214 |
} else { |
215 |
warn ">>> compressed $size_path -> $size_comp ",int(($size_comp * 100) / $size_path),"% $comp\n"; |
216 |
|
217 |
# FIXME add timeout to remove uncompressed version? |
218 |
unlink $path || confess "can't remove $path: $!"; |
219 |
|
220 |
if ( -e $dest ) { |
221 |
warn "## cleanup uncompressed $dest\n" if $debug; |
222 |
unlink $dest || confess "can't remove $dest: $!"; |
223 |
} |
224 |
} |
225 |
|
226 |
} |
227 |
} |
228 |
|
229 |
sub x_open { |
230 |
my ($file) = shift; |
231 |
my ($mode) = shift; |
232 |
|
233 |
if ( $file eq '/.debug' ) { |
234 |
my $path = $mount->{from} . '/.debug'; |
235 |
open( my $debug, '>', $path ) || die "can't open $path: $!"; |
236 |
my $dump = dump( $pending ); |
237 |
print $debug "pending = $dump\n"; |
238 |
close($debug); |
239 |
$pending->{'/.debug'}->{path} = $path; |
240 |
warn "## created dump $path $dump\n"; |
241 |
return 0; |
242 |
} |
243 |
|
244 |
my $mode_desc = { |
245 |
rdonly => $mode && O_RDONLY, |
246 |
rdwr => $mode && O_RDWR, |
247 |
append => $mode && O_APPEND, |
248 |
create => $mode && O_CREAT, |
249 |
trunc => $mode && O_TRUNC, |
250 |
}; |
251 |
|
252 |
my $path = tmp_path( $file ); |
253 |
|
254 |
warn "## open( $file, $mode ) pending: ", $pending->{$file}->{open}, " mode $mode: ", dump( $mode_desc )," $path [", -s $path, "]\n" if $debug; |
255 |
|
256 |
my $fh; |
257 |
my $rv = 0; |
258 |
|
259 |
if ( ! -w $path ) { |
260 |
my $old_mode = (stat $path)[2]; |
261 |
my $new_mode = $old_mode | 0600; |
262 |
chmod $new_mode, $path || confess "can't chmod $new_mode $path"; |
263 |
warn "### modify mode $old_mode -> $new_mode for $path\n"; |
264 |
$pending->{$file}->{mode} = $old_mode; |
265 |
} |
266 |
|
267 |
if ( sysopen($fh , $path, $mode) ) { |
268 |
close($fh) || confess "can't close $path: $!"; |
269 |
warn "<<< sysopen $path [", -e $path ? -s $path : 'new' , "]\n"; |
270 |
$pending->{$file}->{open}++; |
271 |
} else { |
272 |
warn "ERROR: can't open $path -- $!"; |
273 |
$rv = -$!; |
274 |
} |
275 |
|
276 |
return $rv; |
277 |
|
278 |
} |
279 |
|
280 |
sub x_read { |
281 |
my ($file,$bufsize,$off) = @_; |
282 |
my ($rv) = -ENOSYS(); |
283 |
my $path = fixup( $file ); |
284 |
|
285 |
confess "no pending file $file ", dump( $pending ) unless defined( $pending->{$file} ); |
286 |
|
287 |
return -ENOENT() unless -e $path; |
288 |
|
289 |
my $fh = new IO::File; |
290 |
return -ENOSYS() unless open($fh,$pending->{$file}->{path}); |
291 |
|
292 |
if(seek($fh,$off,SEEK_SET)) { |
293 |
read($fh,$rv,$bufsize); |
294 |
$pending->{$file}->{read} += length($rv) if $stats; |
295 |
} |
296 |
|
297 |
return $rv; |
298 |
} |
299 |
|
300 |
sub x_write { |
301 |
my ($file,$buf,$off) = @_; |
302 |
|
303 |
my $rv; |
304 |
my $path = fixup($file); |
305 |
|
306 |
confess "no pending file $file ", dump( $pending ) unless defined( $pending->{$file} ); |
307 |
|
308 |
return -ENOENT() unless -e $path; |
309 |
|
310 |
$path = $pending->{$file}->{path} || confess "no path for $file in ", dump( $pending ); |
311 |
confess "write into non-existant $path for $file" unless -e $path; |
312 |
|
313 |
my $fh = new IO::File; |
314 |
return -ENOSYS() unless open($fh,'+<',$path); |
315 |
if($rv = seek( $fh ,$off,SEEK_SET)) { |
316 |
$rv = print( $fh $buf ); |
317 |
my $size = length($buf); |
318 |
warn "## write $path offset $off [$size]\n" if $debug; |
319 |
$pending->{$file}->{write} += $size; |
320 |
} |
321 |
$rv = -ENOSYS() unless $rv; |
322 |
close($fh) || warn "can't close $path: $!"; |
323 |
return length($buf); |
324 |
} |
325 |
|
326 |
sub err { return (-shift || -$!) } |
327 |
|
328 |
sub x_readlink { return readlink(fixup(shift)); } |
329 |
|
330 |
sub x_unlink { |
331 |
my $file = shift; |
332 |
my $path = fixup( $file ); |
333 |
|
334 |
if ( $file =~ m#\Q/.fuse_hidden\E# ) { |
335 |
return unlink $path ? 0 : -$1; |
336 |
} |
337 |
|
338 |
warn "# unlink( $file )\n"; |
339 |
|
340 |
unlink $path || return 0; |
341 |
|
342 |
my $tmp = $mount->{tmp} . '/' . $file; |
343 |
unlink $tmp if ( -e $tmp ); |
344 |
|
345 |
delete( $pending->{$file} ); |
346 |
return 0; |
347 |
} |
348 |
|
349 |
sub x_symlink { |
350 |
my ($from,$to) = @_; |
351 |
|
352 |
my $from_path = $from; #fixup( $from ); |
353 |
my $to_path = fixup( $to ); |
354 |
|
355 |
my $rv = symlink( $from_path, $to_path ) ? 0 : -$!; |
356 |
warn "# symlink( $from_path -> $to_path ) = $rv\n" if $debug; |
357 |
|
358 |
my $tmp = $mount->{tmp} . '/' . $from; |
359 |
if ( -e $tmp ) { |
360 |
my $tmp_to = $mount->{$tmp} . '/' . $to; |
361 |
symlink( $tmp, $tmp_to ) || confess "can't symlink $tmp -> $tmp_to: $!"; |
362 |
} |
363 |
return $rv; |
364 |
} |
365 |
|
366 |
sub x_link { |
367 |
my ($from,$to) = @_; |
368 |
|
369 |
my $from_path = fixup($from); |
370 |
my $to_path = fixup($to); |
371 |
$to_path .= '.gz' if ( $from_path =~ m/\.gz$/ && $to_path !~ m/\.gz$/ ); |
372 |
|
373 |
my $rv = link( $from_path, $to_path ) ? 0 : -$!; |
374 |
|
375 |
warn "# link( $from_path -> $to_path ) = $rv\n" if $debug; |
376 |
|
377 |
return $rv; |
378 |
} |
379 |
|
380 |
sub x_rename { |
381 |
my ($old,$new) = @_; |
382 |
my $old_path = fixup($old); |
383 |
my $new_path = fixup($new); |
384 |
$new_path .= '.gz' if ( $old_path =~ m/\.gz$/ && $new_path !~ m/\.gz$/ ); |
385 |
|
386 |
my $err = rename($old_path,$new_path) ? 0 : -ENOENT(); |
387 |
warn "## rename( $old_path => $new_path ) = $err\n"; |
388 |
|
389 |
my $tmp = $mount->{tmp} . '/' . $old; |
390 |
if ( -e $tmp ) { |
391 |
if ( $new =~ m#\Q/.fuse_hidden\E# ) { |
392 |
unlink $tmp || confess "can't unlink $tmp for $new"; |
393 |
} else { |
394 |
my $new_tmp = $mount->{tmp} . '/' . $new; |
395 |
rename $tmp, $new_tmp || confess "can't rename $tmp -> $new_tmp : $!"; |
396 |
} |
397 |
} |
398 |
|
399 |
if (defined( $pending->{$old} )) { |
400 |
$pending->{$new} = $pending->{$old}; |
401 |
|
402 |
my $path = $pending->{$old}->{path}; |
403 |
$path =~ s/\Q$old\E/$new/; |
404 |
$pending->{$new}->{path} = $path; |
405 |
|
406 |
delete( $pending->{$old} ); |
407 |
warn "## tweaking pending to ", dump( $pending ) if $debug; |
408 |
} |
409 |
|
410 |
return $err; |
411 |
} |
412 |
|
413 |
sub x_chown { |
414 |
my ($file,$uid,$gid) = @_; |
415 |
my $path = fixup($file); |
416 |
print "nonexistent $path\n" unless -e $path; |
417 |
# perl's chown() does not chown symlinks, it chowns the symlink's |
418 |
# target. it fails when the link's target doesn't exist, because |
419 |
# the stat64() syscall fails. |
420 |
# this causes error messages when unpacking symlinks in tarballs. |
421 |
my ($err) = syscall(&SYS_lchown,$path,$uid,$gid,$path) ? -$! : 0; |
422 |
|
423 |
my $tmp = $mount->{tmp} . '/' . $file; |
424 |
syscall(&SYS_lchown,$file,$uid,$gid,$path) if -e $tmp; |
425 |
|
426 |
return $err; |
427 |
} |
428 |
|
429 |
sub x_chmod { |
430 |
my ($path) = fixup(shift); |
431 |
my ($mode) = shift; |
432 |
my ($err) = chmod($mode,$path) ? 0 : -$!; |
433 |
return $err; |
434 |
} |
435 |
|
436 |
sub x_truncate { |
437 |
my ( $file,$size ) = @_; |
438 |
|
439 |
#confess "no pending file $file to truncate in ", dump( $pending ) unless defined( $pending->{$file} ); |
440 |
|
441 |
my $path = tmp_path( $file ); |
442 |
my $rv = truncate( $path, $size ) ? 0 : -$! ; |
443 |
warn "## truncate( $file $size ) $path [", -s $path, "] = $rv\n" if $debug; |
444 |
compress_file2path( $file, $path ); |
445 |
|
446 |
return $rv; |
447 |
} |
448 |
|
449 |
sub x_utime { return utime($_[1],$_[2],fixup($_[0])) ? 0:-$!; } |
450 |
|
451 |
sub x_mkdir { my ($name, $perm) = @_; return 0 if mkdir(fixup($name),$perm); return -$!; } |
452 |
sub x_rmdir { return 0 if rmdir fixup(shift); return -$!; } |
453 |
|
454 |
sub x_mknod { |
455 |
# since this is called for ALL files, not just devices, I'll do some checks |
456 |
# and possibly run the real mknod command. |
457 |
my ($file, $modes, $dev) = @_; |
458 |
$file = fixup($file); |
459 |
$! = 0; |
460 |
syscall(&SYS_mknod,$file,$modes,$dev); |
461 |
return -$!; |
462 |
} |
463 |
|
464 |
sub x_release { |
465 |
my ( $file, $mode ) = @_; |
466 |
|
467 |
if ( ! defined( $pending->{$file} ) ) { |
468 |
warn "release $file, NO PENDING DATA\n"; |
469 |
return 0; |
470 |
} elsif ( ! defined( $pending->{$file}->{write} ) ) { |
471 |
warn "release $file, not written into\n"; |
472 |
} elsif ( $file =~ m#\Q/.fuse_hidden\E# ) { |
473 |
warn "release internal $file\n" if $debug; |
474 |
} else { |
475 |
compress_file2path( $file ); |
476 |
} |
477 |
|
478 |
$pending->{$file}->{open}--; |
479 |
if ( $pending->{$file}->{open} == 0 ) { |
480 |
|
481 |
my $path = fixup( $file ); |
482 |
|
483 |
if ( my $old_mode = $pending->{$file}->{mode} ) { |
484 |
chmod $old_mode, $path || confess "can't chmod $old_mode $path"; |
485 |
warn "### restored mode $old_mode $path\n"; |
486 |
|
487 |
} |
488 |
|
489 |
warn "## cleanup pending $file [", -s $path, "]\n" if $debug; |
490 |
delete( $pending->{$file} ); |
491 |
} |
492 |
|
493 |
return 0; |
494 |
} |
495 |
|
496 |
# kludge |
497 |
sub x_statfs {return 255,1000000,500000,1000000,500000,4096} |
498 |
Fuse::main( |
499 |
mountpoint=>$mount->{to}, |
500 |
getattr =>"main::x_getattr", |
501 |
readlink=>"main::x_readlink", |
502 |
getdir =>"main::x_getdir", |
503 |
mknod =>"main::x_mknod", |
504 |
mkdir =>"main::x_mkdir", |
505 |
unlink =>"main::x_unlink", |
506 |
rmdir =>"main::x_rmdir", |
507 |
symlink =>"main::x_symlink", |
508 |
rename =>"main::x_rename", |
509 |
link =>"main::x_link", |
510 |
chmod =>"main::x_chmod", |
511 |
chown =>"main::x_chown", |
512 |
truncate=>"main::x_truncate", |
513 |
utime =>"main::x_utime", |
514 |
open =>"main::x_open", |
515 |
read =>"main::x_read", |
516 |
write =>"main::x_write", |
517 |
statfs =>"main::x_statfs", |
518 |
release =>"main::x_release", |
519 |
# threaded=>1, |
520 |
debug => $fuse_debug, |
521 |
); |