#!/usr/bin/perl -w # virt-make-fs # Copyright (C) 2010 Red Hat Inc. # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. use warnings; use strict; use Sys::Guestfs; use Sys::Guestfs::Lib qw(feature_available); use Pod::Usage; use Getopt::Long; use File::Temp qw(tempdir); use POSIX qw(mkfifo floor); use Data::Dumper; use String::ShellQuote qw(shell_quote); use Locale::TextDomain 'libguestfs'; =encoding utf8 =head1 NAME virt-make-fs - Make a filesystem from a tar archive or files =head1 SYNOPSIS virt-make-fs [--options] input.tar output.img virt-make-fs [--options] input.tar.gz output.img virt-make-fs [--options] directory output.img =head1 DESCRIPTION Virt-make-fs is a command line tool for creating a filesystem from a tar archive or some files in a directory. It is similar to tools like L, L and L. Unlike those tools, it can create common filesystem types like ext2/3 or NTFS, which can be useful if you want to attach these filesystems to existing virtual machines (eg. to import large amounts of read-only data to a VM). Basic usage is: virt-make-fs input output where C is either a directory containing files that you want to add, or a tar archive (either uncompressed tar or gzip-compressed tar); and C is a disk image. The input type is detected automatically. The output disk image defaults to a raw ext2 image unless you specify extra flags (see L below). =head2 EXTRA SPACE Unlike formats such as tar and squashfs, a filesystem does not "just fit" the files that it contains, but might have extra space. Depending on how you are going to use the output, you might think this extra space is wasted and want to minimize it, or you might want to leave space so that more files can be added later. Virt-make-fs defaults to minimizing the extra space, but you can use the C<--size> flag to leave space in the filesystem if you want it. An alternative way to leave extra space but not make the output image any bigger is to use an alternative disk image format (instead of the default "raw" format). Using C<--format=qcow2> will use the native QEmu/KVM qcow2 image format (check your hypervisor supports this before using it). This allows you to choose a large C<--size> but the extra space won't actually be allocated in the image until you try to store something in it. Don't forget that you can also use local commands including L and L to resize existing filesystems, or rerun virt-make-resize to build another image from scratch. =head3 EXAMPLE virt-make-fs --format=qcow2 --size=+200M input output.img =head2 FILESYSTEM TYPE The default filesystem type is C. Just about any filesystem type that libguestfs supports can be used (but I read-only formats like ISO9660). Here are some of the more common choices: =over 4 =item I Note that ext3 filesystems contain a journal, typically 1-32 MB in size. If you are not going to use the filesystem in a way that requires the journal, then this is just wasted overhead. =item I or I Useful if exporting data to a Windows guest. I: The tar archive or local directory must only contain files which are owned by root (ie. UID:GID = 0:0). The reason is that the tar program running within libguestfs is unable to change the ownership of non-root files, since vfat itself does not support this. =item I Lower overhead than C, but certain limitations on filename length and total filesystem size. =back =head3 EXAMPLE virt-make-fs --type=minix input minixfs.img =head2 TO PARTITION OR NOT TO PARTITION Optionally virt-make-fs can add a partition table to the output disk. Adding a partition can make the disk image more compatible with certain virtualized operating systems which don't expect to see a filesystem directly located on a block device (Linux doesn't care and will happily handle both types). On the other hand, if you have a partition table then the output image is no longer a straight filesystem. For example you cannot run L directly on a partitioned disk image. (However libguestfs tools such as L and L can still be used). =head3 EXAMPLE Add an MBR partition: virt-make-fs --partition -- input disk.img If the output disk image could be terabyte-sized or larger, it's better to use an EFI/GPT-compatible partition table: virt-make-fs --partition=gpt --size=+4T --format=qcow2 input disk.img =head1 OPTIONS =over 4 =cut my $help; =item B<--help> Display brief help. =cut my $version; =item B<--version> Display version number and exit. =cut my $debug; =item B<--debug> Enable debugging information. =cut my $size; =item B<--size=ENE> =item B<--size=+ENE> =item B<-s ENE> =item B<-s +ENE> Use the C<--size> (or C<-s>) option to choose the size of the output image. If this option is I given, then the output image will be just large enough to contain all the files, with not much wasted space. To choose a fixed size output disk, specify an absolute number followed by b/K/M/G/T/P/E to mean bytes, Kilobytes, Megabytes, Gigabytes, Terabytes, Petabytes or Exabytes. This must be large enough to contain all the input files, else you will get an error. To leave extra space, specify C<+> (plus sign) and a number followed by b/K/M/G/T/P/E to mean bytes, Kilobytes, Megabytes, Gigabytes, Terabytes, Petabytes or Exabytes. For example: C<--size=+200M> means enough space for the input files, and (approximately) an extra 200 MB free space. Note that virt-make-fs estimates free space, and therefore will not produce filesystems containing precisely the free space requested. (It is much more expensive and time-consuming to produce a filesystem which has precisely the desired free space). =cut my $format = "raw"; =item B<--format=EfmtE> =item B<-F EfmtE> Choose the output disk image format. The default is C (raw disk image). For other choices, see the L manpage. The only other choice that would really make sense here is C. =cut my $type = "ext2"; =item B<--type=EfsE> =item B<-t EfsE> Choose the output filesystem type. The default is C. Any filesystem which is supported read-write by libguestfs can be used here. =cut my $partition; =item B<--partition> =item B<--partition=EparttypeE> If specified, this flag adds an MBR partition table to the output disk image. You can change the partition table type, eg. C<--partition=gpt> for large disks. Note that if you just use a lonesome C<--partition>, the Perl option parser might consider the next parameter to be the partition type. For example: virt-make-fs --partition input.tar ... would cause virt-make-fs to think you wanted to use a partition type of C which is completely wrong. To avoid this, use C<--> (a double dash) between options and the input file argument: virt-make-fs --partition -- input.tar ... =back =cut GetOptions ("help|?" => \$help, "version" => \$version, "debug" => \$debug, "s|size=s" => \$size, "F|format=s" => \$format, "t|type=s" => \$type, "partition:s" => \$partition, ) or pod2usage (2); pod2usage (1) if $help; if ($version) { my $g = Sys::Guestfs->new (); my %h = $g->version (); print "$h{major}.$h{minor}.$h{release}$h{extra}\n"; exit } die __"virt-make-fs input output\n" if @ARGV != 2; my $input = $ARGV[0]; my $output = $ARGV[1]; # Input. What is it? Estimate how much space it will need. # # Estimation is a Hard Problem. Some factors which make it hard: # # - Superblocks, block free bitmaps, FAT and other fixed overhead # - Indirect blocks (ext2, ext3), and extents # - Journal size # - Internal fragmentation of files # # What we could also do is try shrinking the filesystem after creating # and populating it, but that is complex given partitions. my $estimate; # Estimated size required (in bytes). my $ifmt; # Input format. if (-d $input) { $ifmt = "directory"; my @cmd = ("du", "--apparent-size", "-b", "-s", $input); open PIPE, "-|", @cmd or die "du $input: $!"; $_ = ; if (/^(\d+)/) { $estimate = $1; } else { die __"unexpected output from 'du' command"; } } else { local $ENV{LANG} = "C"; my @cmd = ("file", "-bsLz", $input); open PIPE, "-|", @cmd or die "file $input: $!"; $ifmt = ; chomp $ifmt; close PIPE; if ($ifmt !~ /tar archive/) { die __x("{f}: unknown input format: {fmt}\n", f => $input, fmt => $ifmt); } if ($ifmt =~ /compress.d/) { if ($ifmt =~ /compress'd/) { @cmd = ("uncompress", "-c", $input); } elsif ($ifmt =~ /gzip compressed/) { @cmd = ("gzip", "-cd", $input); } elsif ($ifmt =~ /bzip2 compressed/) { @cmd = ("bzip2", "-cd", $input); } elsif ($ifmt =~ /xz compressed/) { @cmd = ("xz", "-cd", $input); } else { die __x("{f}: unknown input format: {fmt}\n", f => $input, fmt => $ifmt); } open PIPE, "-|", @cmd or die "uncompress $input: $!"; $estimate = 0; $estimate += length while ; close PIPE or die "close: $!"; } else { # Plain tar file, just get the size directly. Tar files have # a 512 byte block size (compared with typically 1K or 4K for # filesystems) so this isn't very accurate. $estimate = -s $input; } } if ($debug) { printf STDERR "input format = %s\n", $ifmt; printf STDERR "estimate = %s bytes (%s 1K blocks, %s 4K blocks)\n", $estimate, $estimate / 1024, $estimate / 4096; } $estimate += 256 * 1024; # For superblocks &c. if ($type =~ /^ext[3-9]/) { $estimate += 1024 * 1024; # For ext3/4, add some more for the journal. } if ($type =~ /^ntfs/) { $estimate += 4 * 1024 * 1024; # NTFS journal. } $estimate *= 1.10; # Add 10%, see above. # Calculate the output size. if (!defined $size) { $size = $estimate; } else { if ($size =~ /^\+([.\d]+)([bKMGTPE])$/) { $size = $estimate + sizebytes ($1, $2); } elsif ($size =~ /^([.\d]+)([bKMGTPE])$/) { $size = sizebytes ($1, $2); } else { die __x("virt-make-fs: cannot parse size parameter: {sz}\n", sz => $size); } } # Create the output disk. # Take the unusual step of invoking qemu-img here. my @cmd = ("qemu-img", "create", "-f", $format, $output, $size); system (@cmd) == 0 or die __"qemu-img create: failed to create disk image, see earlier error messages\n"; eval { print STDERR "starting libguestfs ...\n" if $debug; # Run libguestfs. my $g = Sys::Guestfs->new (); $g->add_drive ($output); $g->launch (); if ($type eq "ntfs" && !feature_available ($g, "ntfs3g", "ntfsprogs")) { die __"virt-make-fs: NTFS support was disabled when libguestfs was compiled\n" } # Partition the disk. my $dev = "/dev/sda"; if (defined $partition) { $partition = "mbr" if $partition eq ""; $g->part_disk ($dev, $partition); $dev = "/dev/sda1"; } print STDERR "creating $type filesystem on $dev ...\n" if $debug; # Create the filesystem. $g->mkfs ($type, $dev); $g->mount_options ("", $dev, "/"); # Copy the data in. my $ifile; if ($ifmt eq "directory") { my $pfile = create_pipe (); my $cmd = sprintf ("tar -C %s -cf - . > $pfile &", shell_quote ($input)); print STDERR "command: $cmd\n" if $debug; system ($cmd) == 0 or die __"tar: failed, see earlier messages\n"; $ifile = $pfile; } else { if ($ifmt =~ /compress.d/) { my $pfile = create_pipe (); my $cmd; if ($ifmt =~ /compress'd/) { $cmd = sprintf ("uncompress -c %s > $pfile", shell_quote ($input)); } elsif ($ifmt =~ /gzip compressed/) { $cmd = sprintf ("gzip -cd %s", shell_quote ($input)); } elsif ($ifmt =~ /bzip2 compressed/) { $cmd = sprintf ("bzip2 -cd %s", shell_quote ($input)); } elsif ($ifmt =~ /xz compressed/) { $cmd = sprintf ("xz -cd %s", shell_quote ($input)); } else { die __x("{f}: unknown input format: {fmt}\n", f => $input, fmt => $ifmt); } $cmd .= " > $pfile &"; print STDERR "command: $cmd\n" if $debug; system ($cmd) == 0 or die __"uncompress command failed, see earlier messages\n"; $ifile = $pfile; } else { print STDERR "reading directly from $input\n" if $debug; $ifile = $input; } } if ($debug) { # For debugging, print statvfs before and after doing # the tar-in. my %stat = $g->statvfs ("/"); print STDERR "Before uploading ...\n"; print STDERR Dumper(\%stat); } print STDERR "Uploading from $ifile to / ...\n" if $debug; $g->tar_in ($ifile, "/"); if ($debug) { my %stat = $g->statvfs ("/"); print STDERR "After uploading ...\n"; print STDERR Dumper(\%stat); } print STDERR "finishing off\n" if $debug; $g->umount_all (); $g->sync (); undef $g; }; if ($@) { # Error: delete the output before exiting. my $err = $@; unlink $output; if ($err =~ /tar_in/) { print STDERR __"virt-make-fs: error copying contents into filesystem\nAn error here usually means that the program did not estimate the\nfilesystem size correctly. Please read the BUGS section of the manpage.\n"; } print STDERR $err; exit 1; } exit 0; sub sizebytes { local $_ = shift; my $unit = shift; $_ *= 1024 if $unit =~ /[KMGTPE]/; $_ *= 1024 if $unit =~ /[MGTPE]/; $_ *= 1024 if $unit =~ /[GTPE]/; $_ *= 1024 if $unit =~ /[TPE]/; $_ *= 1024 if $unit =~ /[PE]/; $_ *= 1024 if $unit =~ /[E]/; return floor($_); } sub create_pipe { local $_; my $dir = tempdir (CLEANUP => 1); my $pipe = "$dir/pipe"; mkfifo ($pipe, 0600) or die "mkfifo: $pipe: $!"; return $pipe; } =head1 SEE ALSO L, L, L, L, L, L, L, L, L, L, L. =head1 BUGS When reporting bugs, please enable debugging and capture the I output: export LIBGUESTFS_DEBUG=1 virt-make-fs --debug [...] > /tmp/virt-make-fs.log 2>&1 Attach /tmp/virt-make-fs.log to a new bug report at L =head1 AUTHOR Richard W.M. Jones L =head1 COPYRIGHT Copyright (C) 2010 Red Hat Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.