16 |
|
|
17 |
my $debug = 0; |
my $debug = 0; |
18 |
my $fuse_debug = 0; |
my $fuse_debug = 0; |
19 |
|
my $stats = 1; |
20 |
|
|
21 |
GetOptions( |
GetOptions( |
22 |
'debug+' => \$debug, |
'debug+' => \$debug, |
23 |
'fuse-debug+' => \$fuse_debug, |
'fuse-debug+' => \$fuse_debug, |
24 |
|
'stats!' => \$stats, |
25 |
); |
); |
26 |
|
|
27 |
my $mount = { |
my $mount = { |
30 |
tmp => '/dev/shm/comp', |
tmp => '/dev/shm/comp', |
31 |
}; |
}; |
32 |
|
|
|
my $debug = shift @ARGV; |
|
|
|
|
33 |
my $skip_extensions_regex = qr/\.(?:sw[a-z]|gif|png|jpeg|jpg|avi|rar|zip|bz2|gz|tgz|avi|mpeg|mpg|tmp|temp)$/i; |
my $skip_extensions_regex = qr/\.(?:sw[a-z]|gif|png|jpeg|jpg|avi|rar|zip|bz2|gz|tgz|avi|mpeg|mpg|tmp|temp)$/i; |
34 |
|
|
35 |
# don't compress files smaller than this |
# don't compress files smaller than this |
44 |
|
|
45 |
my $pending; |
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 |
|
return $name; |
54 |
|
} |
55 |
|
|
56 |
sub fixup { |
sub fixup { |
57 |
my ( $path ) = @_; |
my ( $path ) = @_; |
58 |
my $full = $mount->{from} . '/' . $path; |
return $mount->{from} . '/' . real_name( $mount->{from}, $path ); |
|
if ( -e $full . '.gz' ) { |
|
|
return $full . '.gz'; |
|
|
} |
|
|
return $full; |
|
59 |
} |
} |
60 |
|
|
61 |
sub original_name { |
sub original_name { |
83 |
return unpack("L", $buff); |
return unpack("L", $buff); |
84 |
} |
} |
85 |
|
|
|
sub unlink_all { |
|
|
my $file = shift; |
|
|
warn "# unlink_all( $file )\n"; |
|
|
|
|
|
my $path = fixup( $file ); |
|
|
unlink $path || return 0; |
|
|
|
|
|
my $tmp = $mount->{tmp} . '/' . $file; |
|
|
unlink $tmp if ( -e $tmp ); |
|
|
|
|
|
delete( $pending->{$file} ); |
|
|
return 1; |
|
|
} |
|
|
|
|
86 |
sub x_getattr { |
sub x_getattr { |
87 |
my ($file) = fixup(shift); |
my ($file) = fixup(shift); |
88 |
my (@list) = lstat($file); |
my (@list) = lstat($file); |
124 |
undef $s; |
undef $s; |
125 |
} |
} |
126 |
|
|
127 |
|
sub tmp_path { |
128 |
|
my $file = shift; |
129 |
|
|
130 |
|
my $path = fixup( $file ); |
131 |
|
|
132 |
|
my $op = 'UNKNOWN'; |
133 |
|
|
134 |
|
if (defined( $pending->{$file} )) { |
135 |
|
$path = $pending->{$file}->{path} || confess "no path for $file in ",dump( $pending ); |
136 |
|
$op = 'opened'; |
137 |
|
} else { |
138 |
|
my $tmp = $mount->{tmp} . '/' . $file; |
139 |
|
if ( -e $tmp ) { |
140 |
|
$path = $tmp; |
141 |
|
$op = 'existing'; |
142 |
|
} elsif ( $path =~ m/\.gz$/ ) { |
143 |
|
my $dest_path = $tmp; |
144 |
|
$dest_path =~ s!/[^/]+$!!; #!vim-fix |
145 |
|
mkpath $dest_path unless -e $dest_path; |
146 |
|
if ( -s $path ) { |
147 |
|
file_copy( '<:gzip', $path, '>', $tmp ) |
148 |
|
} else { |
149 |
|
confess "ASSERT: filesystem corruption, $path is zero size\n"; |
150 |
|
} |
151 |
|
$path = $tmp; |
152 |
|
$op = 'created'; |
153 |
|
} |
154 |
|
confess "ASSERT: path shouldn't exist for $file in ", dump( $pending ) if defined( $pending->{$file}->{path} ); |
155 |
|
confess "ASSERT: open shouldn't exist for $file in ", dump( $pending ) if defined( $pending->{$file}->{open} ); |
156 |
|
$pending->{$file}->{path} = $path; |
157 |
|
$pending->{$file}->{open} = 0; # not really opened, just uncompressed |
158 |
|
warn "## tmp_file( $file ) $op $path [", -s $path, "]\n"; |
159 |
|
} |
160 |
|
return $path; |
161 |
|
} |
162 |
|
|
163 |
|
sub compress_file2path { |
164 |
|
my ( $file, $path ) = @_; |
165 |
|
|
166 |
|
my $dest = fixup( $file ); |
167 |
|
|
168 |
|
if ( defined($pending->{$file}) ) { |
169 |
|
my $pending_path = $pending->{$file}->{path} || confess "no path for $file in ",dump( $pending ); |
170 |
|
|
171 |
|
if ( $pending->{$file}->{open} > 1 ) { |
172 |
|
warn "$file used ", $pending->{$file}->{open}, " times, delaying compression\n"; |
173 |
|
return; |
174 |
|
} elsif ( ! $path ) { |
175 |
|
$path = $pending_path; |
176 |
|
} elsif ( $pending_path ne $path ) { |
177 |
|
confess "ASSERT: compressing into $path instead of $pending_path\n"; |
178 |
|
} |
179 |
|
} |
180 |
|
|
181 |
|
confess "need path" unless $path; |
182 |
|
|
183 |
|
# cleanup old compressed copy |
184 |
|
if ( $dest =~ /\.gz$/ ) { |
185 |
|
warn "## remove old $dest\n"; |
186 |
|
unlink $dest || confess "can't remove $dest: $!"; |
187 |
|
$dest =~ s/\.gz$//; |
188 |
|
confess "ASSERT: uncompressed $dest shouldn't exist!" if -e $dest; |
189 |
|
} |
190 |
|
|
191 |
|
if ( $path =~ $skip_extensions_regex ) { |
192 |
|
warn "$path [",-s $path,"] skipped compression\n"; |
193 |
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
194 |
|
} elsif ( -s $path < $min_compress_size ) { |
195 |
|
warn "$path [",-s $path,"] uncompressed, too small\n"; |
196 |
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
197 |
|
} else { |
198 |
|
warn "$path [",-s $path,"] compressing\n"; |
199 |
|
|
200 |
|
my $comp = $dest . '.gz'; |
201 |
|
file_copy( '<', $path, '>:gzip', $comp ); |
202 |
|
|
203 |
|
my ( $size_path, $size_comp ) = ( -s $path, -s $comp ); |
204 |
|
|
205 |
|
if ( $size_path <= $size_comp ) { |
206 |
|
warn ">>> $size_path <= $size_comp leaving uncompressed\n"; |
207 |
|
unlink $comp || confess "can't remove: $comp: $!"; |
208 |
|
} else { |
209 |
|
warn ">>> compressed $size_path -> $size_comp ",int(($size_comp * 100) / $size_path),"% $comp\n"; |
210 |
|
# FIXME add timeout to remove uncompressed version? |
211 |
|
unlink $path || confess "can't remove $path: $!"; |
212 |
|
} |
213 |
|
} |
214 |
|
} |
215 |
|
|
216 |
sub x_open { |
sub x_open { |
217 |
my ($file) = shift; |
my ($file) = shift; |
218 |
my ($mode) = shift; |
my ($mode) = shift; |
228 |
return 0; |
return 0; |
229 |
} |
} |
230 |
|
|
|
$pending->{$file}->{open}++; |
|
|
|
|
231 |
my $mode_desc = { |
my $mode_desc = { |
232 |
rdonly => $mode && O_RDONLY, |
rdonly => $mode && O_RDONLY, |
233 |
rdwr => $mode && O_RDWR, |
rdwr => $mode && O_RDWR, |
235 |
create => $mode && O_CREAT, |
create => $mode && O_CREAT, |
236 |
trunc => $mode && O_TRUNC, |
trunc => $mode && O_TRUNC, |
237 |
}; |
}; |
238 |
my $path = fixup($file); |
|
239 |
|
my $path = tmp_path( $file ); |
240 |
|
|
241 |
warn "## open( $file, $mode ) pending: ", $pending->{$file}->{open}, " mode $mode: ", dump( $mode_desc )," $path [", -s $path, "]\n" if $debug; |
warn "## open( $file, $mode ) pending: ", $pending->{$file}->{open}, " mode $mode: ", dump( $mode_desc )," $path [", -s $path, "]\n" if $debug; |
|
my $fh; |
|
242 |
|
|
243 |
my $tmp = $mount->{tmp} . '/' . $file; |
my $fh; |
|
if ( -e $tmp ) { |
|
|
$path = $tmp; |
|
|
} elsif ( $path =~ m/\.gz$/ ) { |
|
|
my $dest_path = $tmp; |
|
|
$dest_path =~ s!/[^/]+$!!; #!vim-fix |
|
|
mkpath $dest_path unless -e $dest_path; |
|
|
if ( -s $path ) { |
|
|
file_copy( '<:gzip', $path, '>', $tmp ) |
|
|
} else { |
|
|
warn "ERROR: filesystem corruption, $path is zero size\n"; |
|
|
} |
|
|
$path = $tmp; |
|
|
} |
|
244 |
|
|
245 |
if ( sysopen($fh , $path, $mode) ) { |
if ( sysopen($fh , $path, $mode) ) { |
246 |
close($fh) || confess "can't close $path: $!"; |
close($fh) || confess "can't close $path: $!"; |
247 |
warn "<<< open $path [", -e $path ? -s $path : 'new' , "]\n"; |
warn "<<< sysopen $path [", -e $path ? -s $path : 'new' , "]\n"; |
248 |
$pending->{$file}->{path} = $path; |
$pending->{$file}->{open}++; |
249 |
return 0; |
return 0; |
250 |
} else { |
} else { |
251 |
warn "ERROR: can't open $path : $!"; |
warn "ERROR: can't open $path -- $!"; |
252 |
return -$!; |
return -$!; |
253 |
} |
} |
254 |
|
|
268 |
|
|
269 |
if(seek($fh,$off,SEEK_SET)) { |
if(seek($fh,$off,SEEK_SET)) { |
270 |
read($fh,$rv,$bufsize); |
read($fh,$rv,$bufsize); |
271 |
|
$pending->{$file}->{read} += length($rv) if $stats; |
272 |
} |
} |
273 |
|
|
274 |
return $rv; |
return $rv; |
276 |
|
|
277 |
sub x_write { |
sub x_write { |
278 |
my ($file,$buf,$off) = @_; |
my ($file,$buf,$off) = @_; |
279 |
$pending->{$file}->{write}++; |
|
280 |
my $rv; |
my $rv; |
281 |
my $path = fixup($file); |
my $path = fixup($file); |
282 |
|
|
284 |
|
|
285 |
return -ENOENT() unless -e $path; |
return -ENOENT() unless -e $path; |
286 |
|
|
287 |
|
$path = $pending->{$file}->{path} || confess "no path for $file in ", dump( $pending ); |
288 |
|
confess "write into non-existant $path for $file" unless -e $path; |
289 |
|
|
290 |
my $fh = new IO::File; |
my $fh = new IO::File; |
291 |
return -ENOSYS() unless open($fh,'+<',$pending->{$file}->{path}); |
return -ENOSYS() unless open($fh,'+<',$path); |
292 |
if($rv = seek( $fh ,$off,SEEK_SET)) { |
if($rv = seek( $fh ,$off,SEEK_SET)) { |
293 |
$rv = print( $fh $buf ); |
$rv = print( $fh $buf ); |
294 |
warn "## write ", $pending->{$file}->{path}, " $off ",length( $buf ), "\n" if $debug; |
my $size = length($buf); |
295 |
|
warn "## write $path offset $off [$size]\n" if $debug; |
296 |
|
$pending->{$file}->{write} += $size; |
297 |
} |
} |
298 |
$rv = -ENOSYS() unless $rv; |
$rv = -ENOSYS() unless $rv; |
299 |
close($fh); |
close($fh) || warn "can't close $path: $!"; |
300 |
return length($buf); |
return length($buf); |
301 |
} |
} |
302 |
|
|
303 |
sub err { return (-shift || -$!) } |
sub err { return (-shift || -$!) } |
304 |
|
|
305 |
sub x_readlink { return readlink(fixup(shift)); } |
sub x_readlink { return readlink(fixup(shift)); } |
306 |
sub x_unlink { return unlink_all( shift ) ? 0 : -$! } |
|
307 |
|
sub x_unlink { |
308 |
|
my $file = shift; |
309 |
|
my $path = fixup( $file ); |
310 |
|
|
311 |
|
if ( $file =~ m#\Q/.fuse_hidden\E# ) { |
312 |
|
return unlink $path ? 0 : -$1; |
313 |
|
} |
314 |
|
|
315 |
|
warn "# unlink( $file )\n"; |
316 |
|
|
317 |
|
unlink $path || return 0; |
318 |
|
|
319 |
|
my $tmp = $mount->{tmp} . '/' . $file; |
320 |
|
unlink $tmp if ( -e $tmp ); |
321 |
|
|
322 |
|
delete( $pending->{$file} ); |
323 |
|
return 0; |
324 |
|
} |
325 |
|
|
326 |
|
sub x_symlink { |
327 |
|
my ($from,$to) = @_; |
328 |
|
|
329 |
|
my $from_path = $from; #fixup( $from ); |
330 |
|
my $to_path = fixup( $to ); |
331 |
|
|
332 |
|
my $rv = symlink( $from_path, $to_path ) ? 0 : -$!; |
333 |
|
warn "# symlink( $from_path -> $to_path ) = $rv\n" if $debug; |
334 |
|
|
335 |
|
my $tmp = $mount->{tmp} . '/' . $from; |
336 |
|
if ( -e $tmp ) { |
337 |
|
my $tmp_to = $mount->{$tmp} . '/' . $to; |
338 |
|
symlink( $tmp, $tmp_to ) || confess "can't symlink $tmp -> $tmp_to: $!"; |
339 |
|
} |
340 |
|
return $rv; |
341 |
|
} |
342 |
|
|
343 |
|
sub x_link { |
344 |
|
my ($from,$to) = @_; |
345 |
|
|
346 |
|
my $from_path = fixup($from); |
347 |
|
my $to_path = fixup($to); |
348 |
|
$to_path .= '.gz' if ( $from_path =~ m/\.gz$/ && $to_path !~ m/\.gz$/ ); |
349 |
|
|
350 |
|
my $rv = link( $from_path, $to_path ) ? 0 : -$!; |
351 |
|
|
352 |
sub x_symlink { return symlink(shift,fixup(shift)) ? 0 : -$!; } |
warn "# link( $from_path -> $to_path ) = $rv\n" if $debug; |
353 |
|
|
354 |
|
return $rv; |
355 |
|
} |
356 |
|
|
357 |
sub x_rename { |
sub x_rename { |
358 |
my ($old,$new) = @_; |
my ($old,$new) = @_; |
365 |
|
|
366 |
my $tmp = $mount->{tmp} . '/' . $old; |
my $tmp = $mount->{tmp} . '/' . $old; |
367 |
if ( -e $tmp ) { |
if ( -e $tmp ) { |
368 |
my $new_tmp = $mount->{tmp} . '/' . $new; |
if ( $new =~ m#\Q/.fuse_hidden\E# ) { |
369 |
rename $tmp, $new_tmp || confess "can't rename $tmp -> $new_tmp : $!"; |
unlink $tmp || confess "can't unlink $tmp for $new\n"; |
370 |
|
} else { |
371 |
|
my $new_tmp = $mount->{tmp} . '/' . $new; |
372 |
|
rename $tmp, $new_tmp || confess "can't rename $tmp -> $new_tmp : $!"; |
373 |
|
} |
374 |
} |
} |
375 |
|
|
376 |
if (defined( $pending->{$old} )) { |
if (defined( $pending->{$old} )) { |
379 |
my $path = $pending->{$old}->{path}; |
my $path = $pending->{$old}->{path}; |
380 |
$path =~ s/\Q$old\E/$new/; |
$path =~ s/\Q$old\E/$new/; |
381 |
$pending->{$new}->{path} = $path; |
$pending->{$new}->{path} = $path; |
|
$pending->{$old}->{path} = $path; |
|
382 |
|
|
383 |
#delete( $pending->{$old} ); |
delete( $pending->{$old} ); |
384 |
|
warn "## tweaking pending to ", dump( $pending ) if $debug; |
385 |
} |
} |
386 |
|
|
387 |
return $err; |
return $err; |
388 |
} |
} |
|
sub x_link { return link(fixup(shift),fixup(shift)) ? 0 : -$! } |
|
389 |
|
|
390 |
sub x_chown { |
sub x_chown { |
391 |
my ($path) = fixup(shift); |
my ($file,$uid,$gid) = @_; |
392 |
|
my $path = fixup($file); |
393 |
print "nonexistent $path\n" unless -e $path; |
print "nonexistent $path\n" unless -e $path; |
|
my ($uid,$gid) = @_; |
|
394 |
# perl's chown() does not chown symlinks, it chowns the symlink's |
# perl's chown() does not chown symlinks, it chowns the symlink's |
395 |
# target. it fails when the link's target doesn't exist, because |
# target. it fails when the link's target doesn't exist, because |
396 |
# the stat64() syscall fails. |
# the stat64() syscall fails. |
397 |
# this causes error messages when unpacking symlinks in tarballs. |
# this causes error messages when unpacking symlinks in tarballs. |
398 |
my ($err) = syscall(&SYS_lchown,$path,$uid,$gid,$path) ? -$! : 0; |
my ($err) = syscall(&SYS_lchown,$path,$uid,$gid,$path) ? -$! : 0; |
399 |
|
|
400 |
|
my $tmp = $mount->{tmp} . '/' . $file; |
401 |
|
syscall(&SYS_lchown,$file,$uid,$gid,$path) if -e $tmp; |
402 |
|
|
403 |
return $err; |
return $err; |
404 |
} |
} |
405 |
|
|
412 |
|
|
413 |
sub x_truncate { |
sub x_truncate { |
414 |
my ( $file,$size ) = @_; |
my ( $file,$size ) = @_; |
415 |
my $path = fixup($file); |
|
416 |
|
#confess "no pending file $file to truncate in ", dump( $pending ) unless defined( $pending->{$file} ); |
417 |
|
|
418 |
|
my $path = tmp_path( $file ); |
419 |
my $rv = truncate( $path, $size ) ? 0 : -$! ; |
my $rv = truncate( $path, $size ) ? 0 : -$! ; |
|
if ( $path =~ m/\.gz$/ ) { |
|
|
my $no_gz = $path; |
|
|
$no_gz =~ s/\.gz$//; |
|
|
rename $path, $no_gz || confess "can't rename $path -> $no_gz: $!"; |
|
|
} |
|
420 |
warn "## truncate( $file $size ) $path [", -s $path, "] = $rv\n" if $debug; |
warn "## truncate( $file $size ) $path [", -s $path, "] = $rv\n" if $debug; |
421 |
$pending->{$file}->{write}++; |
compress_file2path( $file, $path ); |
422 |
|
|
423 |
return $rv; |
return $rv; |
424 |
} |
} |
425 |
|
|
426 |
sub x_utime { return utime($_[1],$_[2],fixup($_[0])) ? 0:-$!; } |
sub x_utime { return utime($_[1],$_[2],fixup($_[0])) ? 0:-$!; } |
427 |
|
|
428 |
sub x_mkdir { my ($name, $perm) = @_; return 0 if mkdir(fixup($name),$perm); return -$!; } |
sub x_mkdir { my ($name, $perm) = @_; return 0 if mkdir(fixup($name),$perm); return -$!; } |
440 |
|
|
441 |
sub x_release { |
sub x_release { |
442 |
my ( $file, $mode ) = @_; |
my ( $file, $mode ) = @_; |
443 |
|
|
444 |
if ( ! defined( $pending->{$file} ) ) { |
if ( ! defined( $pending->{$file} ) ) { |
445 |
warn "release $file, NO PENDING DATA\n"; |
warn "release $file, NO PENDING DATA\n"; |
446 |
return 0; |
return 0; |
447 |
} elsif ( ! defined( $pending->{$file}->{write} ) ) { |
} elsif ( ! defined( $pending->{$file}->{write} ) ) { |
448 |
warn "release $file, not written into\n"; |
warn "release $file, not written into\n"; |
449 |
} elsif ( defined( $pending->{$file}->{open} ) && $pending->{$file}->{open} == 1 ) { |
} elsif ( $file =~ m#\Q/.fuse_hidden\E# ) { |
450 |
my $path = $pending->{$file}->{path} || confess "no path for $file ? ", dump( $pending ); |
warn "release internal $file\n" if $debug; |
|
my $dest = fixup( $file ); |
|
|
|
|
|
# cleanup old compressed copy |
|
|
if ( $dest =~ /\.gz$/ ) { |
|
|
warn "## remove old $dest\n"; |
|
|
unlink $dest || confess "can't remove $dest: $!"; |
|
|
$dest =~ s/\.gz$//; |
|
|
} |
|
|
|
|
|
if ( $file =~ $skip_extensions_regex ) { |
|
|
warn "release $file [",-s $path,"] skipped compression\n"; |
|
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
|
|
} elsif ( -s $path < $min_compress_size ) { |
|
|
warn "release $file [",-s $path,"] uncompressed, too small\n"; |
|
|
file_copy( '<', $path, '>', $dest ) if ( $path ne $dest ); |
|
|
} else { |
|
|
warn "release $file [",-s $path,"] compressing\n"; |
|
|
|
|
|
my $comp = $dest . '.gz'; |
|
|
file_copy( '<', $path, '>:gzip', $comp ); |
|
|
|
|
|
my ( $size_path, $size_comp ) = ( -s $path, -s $comp ); |
|
|
|
|
|
if ( $size_path <= $size_comp ) { |
|
|
warn ">>> $size_path <= $size_comp leaving uncompressed\n"; |
|
|
unlink $comp || warn "can't reamove: $comp: $!"; |
|
|
} else { |
|
|
warn ">>> compressed $size_path -> $size_comp ",int(($size_comp * 100) / $size_path),"%\n"; |
|
|
# FIXME add timeout to remove uncompressed version? |
|
|
unlink $path || warn "can't remove $path: $!"; |
|
|
} |
|
|
} |
|
451 |
} else { |
} else { |
452 |
warn "release $file, but still used ", $pending->{$file}->{open} , " times, delaying compression\n"; |
compress_file2path( $file ); |
453 |
} |
} |
454 |
|
|
455 |
$pending->{$file}->{open}--; |
$pending->{$file}->{open}--; |
456 |
if ( $pending->{$file}->{open} == 0 ) { |
if ( $pending->{$file}->{open} == 0 ) { |
457 |
warn "## cleanup pending $file [", -s fixup($file), "]\n" if $debug; |
warn "## cleanup pending $file [", -s fixup($file), "]\n" if $debug; |
458 |
delete( $pending->{$file} ); |
delete( $pending->{$file} ); |
459 |
} |
} |
460 |
|
|
461 |
return 0; |
return 0; |
462 |
} |
} |
463 |
|
|