| 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 $skip_extensions_regex = qr/\.(?:sw[a-z]|gif|png|jpeg|jpg|avi|rar|zip|bz2|gz|tgz|avi|mpeg|mpg|tmp|temp)$/i; |
| 34 |
|
| 35 |
# don't compress files smaller than this |
| 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 |
return $name; |
| 54 |
} |
| 55 |
|
| 56 |
sub fixup { |
| 57 |
my ( $path ) = @_; |
| 58 |
return $mount->{from} . '/' . real_name( $mount->{from}, $path ); |
| 59 |
} |
| 60 |
|
| 61 |
sub original_name { |
| 62 |
my $p = shift; |
| 63 |
$p =~ s/\.gz$//; |
| 64 |
return $p; |
| 65 |
}; |
| 66 |
|
| 67 |
sub gzip_original_size { |
| 68 |
my $file = shift; |
| 69 |
|
| 70 |
return unless -e $file; |
| 71 |
|
| 72 |
my $buff; |
| 73 |
|
| 74 |
open(my $fh, $file) || die "can't open $file: $!"; |
| 75 |
|
| 76 |
# read($fh, $buff, 10 ); |
| 77 |
# print dump( unpack("nccVcc", $buff ) ); |
| 78 |
|
| 79 |
seek($fh, -4, 2); |
| 80 |
read($fh, $buff, 4 ); |
| 81 |
close($fh); |
| 82 |
|
| 83 |
return unpack("L", $buff); |
| 84 |
} |
| 85 |
|
| 86 |
sub x_getattr { |
| 87 |
my ($file) = fixup(shift); |
| 88 |
my (@list) = lstat($file); |
| 89 |
$list[7] = gzip_original_size( $file ) if ( $list[7] && $file =~ m/\.gz$/ ); |
| 90 |
return -$! unless @list; |
| 91 |
return @list; |
| 92 |
} |
| 93 |
|
| 94 |
sub x_getdir { |
| 95 |
my ($dirname) = fixup(shift); |
| 96 |
unless(opendir(DIRHANDLE,$dirname)) { |
| 97 |
return -ENOENT(); |
| 98 |
} |
| 99 |
my @files = map { original_name( $_ ) } readdir(DIRHANDLE); |
| 100 |
closedir(DIRHANDLE); |
| 101 |
return (@files, 0); |
| 102 |
} |
| 103 |
|
| 104 |
sub file_copy { |
| 105 |
my ( $s_opt, $s_path, $d_opt, $d_path ) = @_; |
| 106 |
warn "## file_copy( $s_opt $s_path [",-s $s_path,"] $d_opt $d_path [",-e $d_path ? -s $d_path : 'new',"])\n" if $debug; |
| 107 |
open(my $s, $s_opt, $s_path ) || confess "can't open $s_path: $!\npending = ", dump( $pending ); |
| 108 |
open(my $d, $d_opt, $d_path ) || confess "can't open $d_path: $!"; |
| 109 |
my $buff; |
| 110 |
while( read( $s, $buff, 65535 ) ) { |
| 111 |
print $d $buff || confess "can't write into $d_path: $!"; |
| 112 |
warn ">> [", length($buff), "] offset ", tell($s), " -> ", tell($d), "\n" if $debug; |
| 113 |
} |
| 114 |
close($d) || warn "can't close $d_path: $!"; |
| 115 |
close($s) || warn "can't close $s_path: $!"; |
| 116 |
warn "-- $s_path [", -s $s_path, "] >>> $d_path [", -s $d_path, "]\n" if $debug; |
| 117 |
my ($mode,$uid,$gid,$atime,$mtime) = (stat $s_path)[2,4,5,8,9]; |
| 118 |
|
| 119 |
chmod $mode, $d_path || warn "chmod( $mode $d_path ) failed: $!\n"; |
| 120 |
chown $uid,$gid,$d_path || warn "chown( $uid $gid $d_path ) failed: $!\n"; |
| 121 |
utime $atime,$mtime,$d_path || warn "utime( $atime $mtime $d_path ) failed: $!\n"; |
| 122 |
|
| 123 |
undef $d; |
| 124 |
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 |
|
| 211 |
# FIXME add timeout to remove uncompressed version? |
| 212 |
unlink $path || confess "can't remove $path: $!"; |
| 213 |
|
| 214 |
if ( -e $dest ) { |
| 215 |
warn "## cleanup uncompressed $dest\n" if $debug; |
| 216 |
unlink $dest || confess "can't remove $dest: $!"; |
| 217 |
} |
| 218 |
} |
| 219 |
|
| 220 |
} |
| 221 |
} |
| 222 |
|
| 223 |
sub x_open { |
| 224 |
my ($file) = shift; |
| 225 |
my ($mode) = shift; |
| 226 |
|
| 227 |
if ( $file eq '/.debug' ) { |
| 228 |
my $path = $mount->{from} . '/.debug'; |
| 229 |
open( my $debug, '>', $path ) || die "can't open $path: $!"; |
| 230 |
my $dump = dump( $pending ); |
| 231 |
print $debug "pending = $dump\n"; |
| 232 |
close($debug); |
| 233 |
$pending->{'/.debug'}->{path} = $path; |
| 234 |
warn "## created dump $path $dump\n"; |
| 235 |
return 0; |
| 236 |
} |
| 237 |
|
| 238 |
my $mode_desc = { |
| 239 |
rdonly => $mode && O_RDONLY, |
| 240 |
rdwr => $mode && O_RDWR, |
| 241 |
append => $mode && O_APPEND, |
| 242 |
create => $mode && O_CREAT, |
| 243 |
trunc => $mode && O_TRUNC, |
| 244 |
}; |
| 245 |
|
| 246 |
my $path = tmp_path( $file ); |
| 247 |
|
| 248 |
warn "## open( $file, $mode ) pending: ", $pending->{$file}->{open}, " mode $mode: ", dump( $mode_desc )," $path [", -s $path, "]\n" if $debug; |
| 249 |
|
| 250 |
my $fh; |
| 251 |
|
| 252 |
if ( sysopen($fh , $path, $mode) ) { |
| 253 |
close($fh) || confess "can't close $path: $!"; |
| 254 |
warn "<<< sysopen $path [", -e $path ? -s $path : 'new' , "]\n"; |
| 255 |
$pending->{$file}->{open}++; |
| 256 |
return 0; |
| 257 |
} else { |
| 258 |
warn "ERROR: can't open $path -- $!"; |
| 259 |
return -$!; |
| 260 |
} |
| 261 |
|
| 262 |
} |
| 263 |
|
| 264 |
sub x_read { |
| 265 |
my ($file,$bufsize,$off) = @_; |
| 266 |
my ($rv) = -ENOSYS(); |
| 267 |
my $path = fixup( $file ); |
| 268 |
|
| 269 |
confess "no pending file $file ", dump( $pending ) unless defined( $pending->{$file} ); |
| 270 |
|
| 271 |
return -ENOENT() unless -e $path; |
| 272 |
|
| 273 |
my $fh = new IO::File; |
| 274 |
return -ENOSYS() unless open($fh,$pending->{$file}->{path}); |
| 275 |
|
| 276 |
if(seek($fh,$off,SEEK_SET)) { |
| 277 |
read($fh,$rv,$bufsize); |
| 278 |
$pending->{$file}->{read} += length($rv) if $stats; |
| 279 |
} |
| 280 |
|
| 281 |
return $rv; |
| 282 |
} |
| 283 |
|
| 284 |
sub x_write { |
| 285 |
my ($file,$buf,$off) = @_; |
| 286 |
|
| 287 |
my $rv; |
| 288 |
my $path = fixup($file); |
| 289 |
|
| 290 |
confess "no pending file $file ", dump( $pending ) unless defined( $pending->{$file} ); |
| 291 |
|
| 292 |
return -ENOENT() unless -e $path; |
| 293 |
|
| 294 |
$path = $pending->{$file}->{path} || confess "no path for $file in ", dump( $pending ); |
| 295 |
confess "write into non-existant $path for $file" unless -e $path; |
| 296 |
|
| 297 |
my $fh = new IO::File; |
| 298 |
return -ENOSYS() unless open($fh,'+<',$path); |
| 299 |
if($rv = seek( $fh ,$off,SEEK_SET)) { |
| 300 |
$rv = print( $fh $buf ); |
| 301 |
my $size = length($buf); |
| 302 |
warn "## write $path offset $off [$size]\n" if $debug; |
| 303 |
$pending->{$file}->{write} += $size; |
| 304 |
} |
| 305 |
$rv = -ENOSYS() unless $rv; |
| 306 |
close($fh) || warn "can't close $path: $!"; |
| 307 |
return length($buf); |
| 308 |
} |
| 309 |
|
| 310 |
sub err { return (-shift || -$!) } |
| 311 |
|
| 312 |
sub x_readlink { return readlink(fixup(shift)); } |
| 313 |
|
| 314 |
sub x_unlink { |
| 315 |
my $file = shift; |
| 316 |
my $path = fixup( $file ); |
| 317 |
|
| 318 |
if ( $file =~ m#\Q/.fuse_hidden\E# ) { |
| 319 |
return unlink $path ? 0 : -$1; |
| 320 |
} |
| 321 |
|
| 322 |
warn "# unlink( $file )\n"; |
| 323 |
|
| 324 |
unlink $path || return 0; |
| 325 |
|
| 326 |
my $tmp = $mount->{tmp} . '/' . $file; |
| 327 |
unlink $tmp if ( -e $tmp ); |
| 328 |
|
| 329 |
delete( $pending->{$file} ); |
| 330 |
return 0; |
| 331 |
} |
| 332 |
|
| 333 |
sub x_symlink { |
| 334 |
my ($from,$to) = @_; |
| 335 |
|
| 336 |
my $from_path = $from; #fixup( $from ); |
| 337 |
my $to_path = fixup( $to ); |
| 338 |
|
| 339 |
my $rv = symlink( $from_path, $to_path ) ? 0 : -$!; |
| 340 |
warn "# symlink( $from_path -> $to_path ) = $rv\n" if $debug; |
| 341 |
|
| 342 |
my $tmp = $mount->{tmp} . '/' . $from; |
| 343 |
if ( -e $tmp ) { |
| 344 |
my $tmp_to = $mount->{$tmp} . '/' . $to; |
| 345 |
symlink( $tmp, $tmp_to ) || confess "can't symlink $tmp -> $tmp_to: $!"; |
| 346 |
} |
| 347 |
return $rv; |
| 348 |
} |
| 349 |
|
| 350 |
sub x_link { |
| 351 |
my ($from,$to) = @_; |
| 352 |
|
| 353 |
my $from_path = fixup($from); |
| 354 |
my $to_path = fixup($to); |
| 355 |
$to_path .= '.gz' if ( $from_path =~ m/\.gz$/ && $to_path !~ m/\.gz$/ ); |
| 356 |
|
| 357 |
my $rv = link( $from_path, $to_path ) ? 0 : -$!; |
| 358 |
|
| 359 |
warn "# link( $from_path -> $to_path ) = $rv\n" if $debug; |
| 360 |
|
| 361 |
return $rv; |
| 362 |
} |
| 363 |
|
| 364 |
sub x_rename { |
| 365 |
my ($old,$new) = @_; |
| 366 |
my $old_path = fixup($old); |
| 367 |
my $new_path = fixup($new); |
| 368 |
$new_path .= '.gz' if ( $old_path =~ m/\.gz$/ && $new_path !~ m/\.gz$/ ); |
| 369 |
|
| 370 |
my $err = rename($old_path,$new_path) ? 0 : -ENOENT(); |
| 371 |
warn "## rename( $old_path => $new_path ) = $err\n"; |
| 372 |
|
| 373 |
my $tmp = $mount->{tmp} . '/' . $old; |
| 374 |
if ( -e $tmp ) { |
| 375 |
if ( $new =~ m#\Q/.fuse_hidden\E# ) { |
| 376 |
unlink $tmp || confess "can't unlink $tmp for $new\n"; |
| 377 |
} else { |
| 378 |
my $new_tmp = $mount->{tmp} . '/' . $new; |
| 379 |
rename $tmp, $new_tmp || confess "can't rename $tmp -> $new_tmp : $!"; |
| 380 |
} |
| 381 |
} |
| 382 |
|
| 383 |
if (defined( $pending->{$old} )) { |
| 384 |
$pending->{$new} = $pending->{$old}; |
| 385 |
|
| 386 |
my $path = $pending->{$old}->{path}; |
| 387 |
$path =~ s/\Q$old\E/$new/; |
| 388 |
$pending->{$new}->{path} = $path; |
| 389 |
|
| 390 |
delete( $pending->{$old} ); |
| 391 |
warn "## tweaking pending to ", dump( $pending ) if $debug; |
| 392 |
} |
| 393 |
|
| 394 |
return $err; |
| 395 |
} |
| 396 |
|
| 397 |
sub x_chown { |
| 398 |
my ($file,$uid,$gid) = @_; |
| 399 |
my $path = fixup($file); |
| 400 |
print "nonexistent $path\n" unless -e $path; |
| 401 |
# perl's chown() does not chown symlinks, it chowns the symlink's |
| 402 |
# target. it fails when the link's target doesn't exist, because |
| 403 |
# the stat64() syscall fails. |
| 404 |
# this causes error messages when unpacking symlinks in tarballs. |
| 405 |
my ($err) = syscall(&SYS_lchown,$path,$uid,$gid,$path) ? -$! : 0; |
| 406 |
|
| 407 |
my $tmp = $mount->{tmp} . '/' . $file; |
| 408 |
syscall(&SYS_lchown,$file,$uid,$gid,$path) if -e $tmp; |
| 409 |
|
| 410 |
return $err; |
| 411 |
} |
| 412 |
|
| 413 |
sub x_chmod { |
| 414 |
my ($path) = fixup(shift); |
| 415 |
my ($mode) = shift; |
| 416 |
my ($err) = chmod($mode,$path) ? 0 : -$!; |
| 417 |
return $err; |
| 418 |
} |
| 419 |
|
| 420 |
sub x_truncate { |
| 421 |
my ( $file,$size ) = @_; |
| 422 |
|
| 423 |
#confess "no pending file $file to truncate in ", dump( $pending ) unless defined( $pending->{$file} ); |
| 424 |
|
| 425 |
my $path = tmp_path( $file ); |
| 426 |
my $rv = truncate( $path, $size ) ? 0 : -$! ; |
| 427 |
warn "## truncate( $file $size ) $path [", -s $path, "] = $rv\n" if $debug; |
| 428 |
compress_file2path( $file, $path ); |
| 429 |
|
| 430 |
return $rv; |
| 431 |
} |
| 432 |
|
| 433 |
sub x_utime { return utime($_[1],$_[2],fixup($_[0])) ? 0:-$!; } |
| 434 |
|
| 435 |
sub x_mkdir { my ($name, $perm) = @_; return 0 if mkdir(fixup($name),$perm); return -$!; } |
| 436 |
sub x_rmdir { return 0 if rmdir fixup(shift); return -$!; } |
| 437 |
|
| 438 |
sub x_mknod { |
| 439 |
# since this is called for ALL files, not just devices, I'll do some checks |
| 440 |
# and possibly run the real mknod command. |
| 441 |
my ($file, $modes, $dev) = @_; |
| 442 |
$file = fixup($file); |
| 443 |
$! = 0; |
| 444 |
syscall(&SYS_mknod,$file,$modes,$dev); |
| 445 |
return -$!; |
| 446 |
} |
| 447 |
|
| 448 |
sub x_release { |
| 449 |
my ( $file, $mode ) = @_; |
| 450 |
|
| 451 |
if ( ! defined( $pending->{$file} ) ) { |
| 452 |
warn "release $file, NO PENDING DATA\n"; |
| 453 |
return 0; |
| 454 |
} elsif ( ! defined( $pending->{$file}->{write} ) ) { |
| 455 |
warn "release $file, not written into\n"; |
| 456 |
} elsif ( $file =~ m#\Q/.fuse_hidden\E# ) { |
| 457 |
warn "release internal $file\n" if $debug; |
| 458 |
} else { |
| 459 |
compress_file2path( $file ); |
| 460 |
} |
| 461 |
|
| 462 |
$pending->{$file}->{open}--; |
| 463 |
if ( $pending->{$file}->{open} == 0 ) { |
| 464 |
warn "## cleanup pending $file [", -s fixup($file), "]\n" if $debug; |
| 465 |
delete( $pending->{$file} ); |
| 466 |
} |
| 467 |
|
| 468 |
return 0; |
| 469 |
} |
| 470 |
|
| 471 |
# kludge |
| 472 |
sub x_statfs {return 255,1000000,500000,1000000,500000,4096} |
| 473 |
Fuse::main( |
| 474 |
mountpoint=>$mount->{to}, |
| 475 |
getattr =>"main::x_getattr", |
| 476 |
readlink=>"main::x_readlink", |
| 477 |
getdir =>"main::x_getdir", |
| 478 |
mknod =>"main::x_mknod", |
| 479 |
mkdir =>"main::x_mkdir", |
| 480 |
unlink =>"main::x_unlink", |
| 481 |
rmdir =>"main::x_rmdir", |
| 482 |
symlink =>"main::x_symlink", |
| 483 |
rename =>"main::x_rename", |
| 484 |
link =>"main::x_link", |
| 485 |
chmod =>"main::x_chmod", |
| 486 |
chown =>"main::x_chown", |
| 487 |
truncate=>"main::x_truncate", |
| 488 |
utime =>"main::x_utime", |
| 489 |
open =>"main::x_open", |
| 490 |
read =>"main::x_read", |
| 491 |
write =>"main::x_write", |
| 492 |
statfs =>"main::x_statfs", |
| 493 |
release =>"main::x_release", |
| 494 |
# threaded=>1, |
| 495 |
debug => $fuse_debug, |
| 496 |
); |
Parent Directory
| 
Revision Log