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