Bump version to 0.2.2
[snf-image-creator] / image_creator / disk.py
index be728a2..e0ee867 100644 (file)
 # or implied, of GRNET S.A.
 
 from image_creator.util import get_command
-from image_creator.util import warn, progress, success, output, FatalError
+from image_creator.util import FatalError
+from image_creator.util import try_fail_repeat
+from image_creator.util import free_space
+from image_creator.gpt import GPTPartitionTable
+from image_creator.bundle_volume import BundleVolume
 
 import stat
 import os
@@ -41,19 +45,19 @@ import uuid
 import re
 import sys
 import guestfs
-import time
+import shutil
 from sendfile import sendfile
 
 
-class DiskError(Exception):
-    pass
-
 dd = get_command('dd')
 dmsetup = get_command('dmsetup')
 losetup = get_command('losetup')
 blockdev = get_command('blockdev')
 
 
+TMP_CANDIDATES = ['/var/tmp', os.path.expanduser('~'), '/mnt']
+
+
 class Disk(object):
     """This class represents a hard disk hosting an Operating System
 
@@ -62,12 +66,35 @@ class Disk(object):
     the Linux kernel.
     """
 
-    def __init__(self, source):
+    def __init__(self, source, output, tmp=None):
         """Create a new Disk instance out of a source media. The source
-        media can be an image file, a block device or a directory."""
+        media can be an image file, a block device or a directory.
+        """
         self._cleanup_jobs = []
         self._devices = []
         self.source = source
+        self.out = output
+        self.meta = {}
+        self.tmp = tempfile.mkdtemp(prefix='.snf_image_creator.',
+                                    dir=self._get_tmp_dir(tmp))
+
+        self._add_cleanup(shutil.rmtree, self.tmp)
+
+    def _get_tmp_dir(self, default=None):
+        if default is not None:
+            return default
+
+        space = map(free_space, TMP_CANDIDATES)
+
+        max_idx = 0
+        max_val = space[0]
+        for i, val in zip(range(len(space)), space):
+            if val > max_val:
+                max_val = val
+                max_idx = i
+
+        # Return the candidate path with more available space
+        return TMP_CANDIDATES[max_idx]
 
     def _add_cleanup(self, job, *args):
         self._cleanup_jobs.append((job, args))
@@ -75,73 +102,85 @@ class Disk(object):
     def _losetup(self, fname):
         loop = losetup('-f', '--show', fname)
         loop = loop.strip()  # remove the new-line char
-        self._add_cleanup(losetup, '-d', loop)
+        self._add_cleanup(try_fail_repeat, losetup, '-d', loop)
         return loop
 
     def _dir_to_disk(self):
-        raise NotImplementedError
+        if self.source == '/':
+            bundle = BundleVolume(self.out, self.meta)
+            image = '%s/%s.diskdump' % (self.tmp, uuid.uuid4().hex)
+
+            def check_unlink(path):
+                if os.path.exists(path):
+                    os.unlink(path)
+
+            self._add_cleanup(check_unlink, image)
+            bundle.create_image(image)
+            return self._losetup(image)
+        raise FatalError("Using a directory as media source is supported")
 
     def cleanup(self):
         """Cleanup internal data. This needs to be called before the
         program ends.
         """
-        while len(self._devices):
-            device = self._devices.pop()
-            device.destroy()
-
-        while len(self._cleanup_jobs):
-            job, args = self._cleanup_jobs.pop()
-            job(*args)
-
-    def get_device(self):
-        """Returns a newly created DiskDevice instance.
-
-        This instance is a snapshot of the original source media of
-        the Disk instance.
+        try:
+            while len(self._devices):
+                device = self._devices.pop()
+                device.destroy()
+        finally:
+            # Make sure those are executed even if one of the device.destroy
+            # methods throws exeptions.
+            while len(self._cleanup_jobs):
+                job, args = self._cleanup_jobs.pop()
+                job(*args)
+
+    def snapshot(self):
+        """Creates a snapshot of the original source media of the Disk
+        instance.
         """
 
-        output("Examining source media `%s'..." % self.source, False)
+        self.out.output("Examining source media `%s' ..." % self.source, False)
         sourcedev = self.source
         mode = os.stat(self.source).st_mode
         if stat.S_ISDIR(mode):
-            success('looks like a directory')
-            return self._losetup(self._dir_to_disk())
+            self.out.success('looks like a directory')
+            return self._dir_to_disk()
         elif stat.S_ISREG(mode):
-            success('looks like an image file')
+            self.out.success('looks like an image file')
             sourcedev = self._losetup(self.source)
         elif not stat.S_ISBLK(mode):
             raise ValueError("Invalid media source. Only block devices, "
-                            "regular files and directories are supported.")
+                             "regular files and directories are supported.")
         else:
-            success('looks like a block device')
+            self.out.success('looks like a block device')
 
         # Take a snapshot and return it to the user
-        output("Snapshotting media source...", False)
-        size = blockdev('--getsize', sourcedev)
-        cowfd, cow = tempfile.mkstemp()
+        self.out.output("Snapshotting media source...", False)
+        size = blockdev('--getsz', sourcedev)
+        cowfd, cow = tempfile.mkstemp(dir=self.tmp)
         os.close(cowfd)
         self._add_cleanup(os.unlink, cow)
-        # Create 1G cow sparse file
-        dd('if=/dev/null', 'of=%s' % cow, 'bs=1k', \
-                                        'seek=%d' % (1024 * 1024))
+        # Create cow sparse file
+        dd('if=/dev/null', 'of=%s' % cow, 'bs=512', 'seek=%d' % int(size))
         cowdev = self._losetup(cow)
 
         snapshot = uuid.uuid4().hex
         tablefd, table = tempfile.mkstemp()
         try:
-            os.write(tablefd, "0 %d snapshot %s %s n 8" % \
-                                        (int(size), sourcedev, cowdev))
+            os.write(tablefd, "0 %d snapshot %s %s n 8" %
+                              (int(size), sourcedev, cowdev))
             dmsetup('create', snapshot, table)
-            self._add_cleanup(dmsetup, 'remove', snapshot)
-            # Sometimes dmsetup remove fails with Device or resource busy,
-            # although everything is cleaned up and the snapshot is not
-            # used by anyone. Add a 2 seconds delay to be on the safe side.
-            self._add_cleanup(time.sleep, 2)
+            self._add_cleanup(try_fail_repeat, dmsetup, 'remove', snapshot)
 
         finally:
             os.unlink(table)
-        success('done')
-        new_device = DiskDevice("/dev/mapper/%s" % snapshot)
+        self.out.success('done')
+        return "/dev/mapper/%s" % snapshot
+
+    def get_device(self, media):
+        """Returns a newly created DiskDevice instance."""
+
+        new_device = DiskDevice(media, self.out)
         self._devices.append(new_device)
         new_device.enable()
         return new_device
@@ -159,15 +198,33 @@ class DiskDevice(object):
     as created by the device-mapper.
     """
 
-    def __init__(self, device, bootable=True):
+    def __init__(self, device, output, bootable=True, meta={}):
         """Create a new DiskDevice."""
 
-        self.device = device
+        self.real_device = device
+        self.out = output
         self.bootable = bootable
+        self.meta = meta
         self.progress_bar = None
+        self.guestfs_device = None
+        self.size = 0
 
         self.g = guestfs.GuestFS()
-        self.g.add_drive_opts(self.device, readonly=0)
+        self.g.add_drive_opts(self.real_device, readonly=0)
+
+        # Before version 1.17.14 the recovery process, which is a fork of the
+        # original process that called libguestfs, did not close its inherited
+        # file descriptors. This can cause problems especially if the parent
+        # process has opened pipes. Since the recovery process is an optional
+        # feature of libguestfs, it's better to disable it.
+        self.g.set_recovery_proc(0)
+        version = self.g.version()
+        if version['major'] > 1 or \
+            (version['major'] == 1 and (version['minor'] >= 18 or
+                                        (version['minor'] == 17 and
+                                         version['release'] >= 14))):
+            self.g.set_recovery_proc(1)
+            self.out.output("Enabling recovery proc")
 
         #self.g.set_trace(1)
         #self.g.set_verbose(1)
@@ -176,52 +233,60 @@ class DiskDevice(object):
 
     def enable(self):
         """Enable a newly created DiskDevice"""
-        self.progressbar = progress("Launching helper VM: ")
-        self.progressbar.next()
-        eh = self.g.set_event_callback(self.progress_callback,
-                                                    guestfs.EVENT_PROGRESS)
+
+        self.out.output('Launching helper VM (may take a while) ...', False)
+        # self.progressbar = self.out.Progress(100, "Launching helper VM",
+        #                                     "percent")
+        # eh = self.g.set_event_callback(self.progress_callback,
+        #                               guestfs.EVENT_PROGRESS)
         self.g.launch()
         self.guestfs_enabled = True
-        self.g.delete_event_callback(eh)
-        if self.progressbar is not None:
-            self.progressbar.send(100)
-            self.progressbar = None
+        # self.g.delete_event_callback(eh)
+        # self.progressbar.success('done')
+        # self.progressbar = None
+        self.out.success('done')
 
-        output('Inspecting Operating System...', False)
+        self.out.output('Inspecting Operating System ...', False)
         roots = self.g.inspect_os()
         if len(roots) == 0:
             raise FatalError("No operating system found")
         if len(roots) > 1:
             raise FatalError("Multiple operating systems found."
-                            "We only support images with one filesystem.")
+                             "We only support images with one OS.")
         self.root = roots[0]
+        self.guestfs_device = self.g.part_to_dev(self.root)
+        self.size = self.g.blockdev_getsize64(self.guestfs_device)
+        self.meta['PARTITION_TABLE'] = \
+            self.g.part_get_parttype(self.guestfs_device)
+
         self.ostype = self.g.inspect_get_type(self.root)
         self.distro = self.g.inspect_get_distro(self.root)
-        success('found a %s system' % self.distro)
+        self.out.success('found a(n) %s system' % self.distro)
 
     def destroy(self):
         """Destroy this DiskDevice instance."""
 
-        if self.guestfs_enabled:
-            self.g.umount_all()
-            self.g.sync()
-
-        # Close the guestfs handler if open
-        self.g.close()
-
-    def progress_callback(self, ev, eh, buf, array):
-        position = array[2]
-        total = array[3]
-
-        self.progressbar.send((position * 100) // total)
+        # In new guestfs versions, there is a handy shutdown method for this
+        try:
+            if self.guestfs_enabled:
+                self.g.umount_all()
+                self.g.sync()
+        finally:
+            # Close the guestfs handler if open
+            self.g.close()
 
-        if position == total:
-            self.progressbar = None
+#    def progress_callback(self, ev, eh, buf, array):
+#        position = array[2]
+#        total = array[3]
+#
+#        self.progressbar.goto((position * 100) // total)
 
-    def mount(self):
+    def mount(self, readonly=False):
         """Mount all disk partitions in a correct order."""
 
-        output("Mounting image...", False)
+        mount = self.g.mount_ro if readonly else self.g.mount
+        msg = " read-only" if readonly else ""
+        self.out.output("Mounting the media%s ..." % msg, False)
         mps = self.g.inspect_get_mountpoints(self.root)
 
         # Sort the keys to mount the fs in a correct order.
@@ -236,74 +301,166 @@ class DiskDevice(object):
         mps.sort(compare)
         for mp, dev in mps:
             try:
-                self.g.mount(dev, mp)
+                mount(dev, mp)
             except RuntimeError as msg:
-                warn("%s (ignored)" % msg)
-        success("done")
+                self.out.warn("%s (ignored)" % msg)
+        self.out.success("done")
 
     def umount(self):
         """Umount all mounted filesystems."""
         self.g.umount_all()
 
+    def _last_partition(self):
+        if self.meta['PARTITION_TABLE'] not in 'msdos' 'gpt':
+            msg = "Unsupported partition table: %s. Only msdos and gpt " \
+                "partition tables are supported" % self.meta['PARTITION_TABLE']
+            raise FatalError(msg)
+
+        is_extended = lambda p: \
+            self.g.part_get_mbr_id(self.guestfs_device, p['part_num']) \
+            in (0x5, 0xf)
+        is_logical = lambda p: \
+            self.meta['PARTITION_TABLE'] == 'msdos' and p['part_num'] > 4
+
+        partitions = self.g.part_list(self.guestfs_device)
+        last_partition = partitions[-1]
+
+        if is_logical(last_partition):
+            # The disk contains extended and logical partitions....
+            extended = filter(is_extended, partitions)[0]
+            last_primary = [p for p in partitions if p['part_num'] <= 4][-1]
+
+            # check if extended is the last primary partition
+            if last_primary['part_num'] > extended['part_num']:
+                last_partition = last_primary
+
+        return last_partition
+
     def shrink(self):
         """Shrink the disk.
 
         This is accomplished by shrinking the last filesystem in the
         disk and then updating the partition table. The new disk size
         (in bytes) is returned.
-        """
-        output("Shrinking image (this may take a while)...", False)
-
-        dev = self.g.part_to_dev(self.root)
-        parttype = self.g.part_get_parttype(dev)
-        if parttype != 'msdos':
-            raise FatalError("You have a %s partition table. "
-                "Only msdos partitions are supported" % parttype)
-
-        last_partition = self.g.part_list(dev)[-1]
-
-        if last_partition['part_num'] > 4:
-            raise FatalError("This disk contains logical partitions. "
-                "Only primary partitions are supported.")
-
-        part_dev = "%s%d" % (dev, last_partition['part_num'])
-        fs_type = self.g.vfs_type(part_dev)
-        if not re.match("ext[234]", fs_type):
-            warn("Don't know how to resize %s partitions." % vfs_type)
-            return
 
+        ATTENTION: make sure unmount is called before shrink
+        """
+        get_fstype = lambda p: \
+            self.g.vfs_type("%s%d" % (self.guestfs_device, p['part_num']))
+        is_logical = lambda p: \
+            self.meta['PARTITION_TABLE'] == 'msdos' and p['part_num'] > 4
+        is_extended = lambda p: \
+            self.meta['PARTITION_TABLE'] == 'msdos' and \
+            self.g.part_get_mbr_id(self.guestfs_device, p['part_num']) \
+            in (0x5, 0xf)
+
+        part_add = lambda ptype, start, stop: \
+            self.g.part_add(self.guestfs_device, ptype, start, stop)
+        part_del = lambda p: self.g.part_del(self.guestfs_device, p)
+        part_get_id = lambda p: self.g.part_get_mbr_id(self.guestfs_device, p)
+        part_set_id = lambda p, id: \
+            self.g.part_set_mbr_id(self.guestfs_device, p, id)
+        part_get_bootable = lambda p: \
+            self.g.part_get_bootable(self.guestfs_device, p)
+        part_set_bootable = lambda p, bootable: \
+            self.g.part_set_bootable(self.guestfs_device, p, bootable)
+
+        MB = 2 ** 20
+
+        self.out.output("Shrinking image (this may take a while) ...", False)
+
+        sector_size = self.g.blockdev_getss(self.guestfs_device)
+
+        last_part = None
+        fstype = None
+        while True:
+            last_part = self._last_partition()
+            fstype = get_fstype(last_part)
+
+            if fstype == 'swap':
+                self.meta['SWAP'] = "%d:%s" % \
+                    (last_part['part_num'],
+                     (last_part['part_size'] + MB - 1) // MB)
+                part_del(last_part['part_num'])
+                continue
+            elif is_extended(last_part):
+                part_del(last_part['part_num'])
+                continue
+
+            # Most disk manipulation programs leave 2048 sectors after the last
+            # partition
+            new_size = last_part['part_end'] + 1 + 2048 * sector_size
+            self.size = min(self.size, new_size)
+            break
+
+        if not re.match("ext[234]", fstype):
+            self.out.warn("Don't know how to resize %s partitions." % fstype)
+            return self.size
+
+        part_dev = "%s%d" % (self.guestfs_device, last_part['part_num'])
         self.g.e2fsck_f(part_dev)
         self.g.resize2fs_M(part_dev)
 
-        output = self.g.tune2fs_l(part_dev)
-        block_size = int(
-            filter(lambda x: x[0] == 'Block size', output)[0][1])
-        block_cnt = int(
-            filter(lambda x: x[0] == 'Block count', output)[0][1])
-
-        sector_size = self.g.blockdev_getss(dev)
+        out = self.g.tune2fs_l(part_dev)
+        block_size = int(filter(lambda x: x[0] == 'Block size', out)[0][1])
+        block_cnt = int(filter(lambda x: x[0] == 'Block count', out)[0][1])
 
-        start = last_partition['part_start'] / sector_size
+        start = last_part['part_start'] / sector_size
         end = start + (block_size * block_cnt) / sector_size - 1
 
-        self.g.part_del(dev, last_partition['part_num'])
-        self.g.part_add(dev, 'p', start, end)
+        if is_logical(last_part):
+            partitions = self.g.part_list(self.guestfs_device)
+
+            logical = []  # logical partitions
+            for partition in partitions:
+                if partition['part_num'] < 4:
+                    continue
+                logical.append({
+                    'num': partition['part_num'],
+                    'start': partition['part_start'] / sector_size,
+                    'end': partition['part_end'] / sector_size,
+                    'id': part_get_id(partition['part_num']),
+                    'bootable': part_get_bootable(partition['part_num'])
+                })
+
+            logical[-1]['end'] = end  # new end after resize
+
+            # Recreate the extended partition
+            extended = filter(is_extended, partitions)[0]
+            part_del(extended['part_num'])
+            part_add('e', extended['part_start'] / sector_size, end)
+
+            # Create all the logical partitions back
+            for l in logical:
+                part_add('l', l['start'], l['end'])
+                part_set_id(l['num'], l['id'])
+                part_set_bootable(l['num'], l['bootable'])
+        else:
+            # Recreate the last partition
+            if self.meta['PARTITION_TABLE'] == 'msdos':
+                last_part['id'] = part_get_id(last_part['part_num'])
+
+            last_part['bootable'] = part_get_bootable(last_part['part_num'])
+            part_del(last_part['part_num'])
+            part_add('p', start, end)
+            part_set_bootable(last_part['part_num'], last_part['bootable'])
+
+            if self.meta['PARTITION_TABLE'] == 'msdos':
+                part_set_id(last_part['part_num'], last_part['id'])
 
         new_size = (end + 1) * sector_size
-        success("new image size is %dMB" %
-                            ((new_size + 2 ** 20 - 1) // 2 ** 20))
-        return new_size
 
-    def size(self):
-        """Returns the "payload" size of the device.
+        assert (new_size <= self.size)
 
-        The size returned by this method is the size of the space occupied by
-        the partitions (including the space before the first partition).
-        """
-        dev = self.g.part_to_dev(self.root)
-        last = self.g.part_list(dev)[-1]
+        if self.meta['PARTITION_TABLE'] == 'gpt':
+            ptable = GPTPartitionTable(self.real_device)
+            self.size = ptable.shrink(new_size, self.size)
+        else:
+            self.size = min(new_size + 2048 * sector_size, self.size)
+
+        self.out.success("new size is %dMB" % ((self.size + MB - 1) // MB))
 
-        return last['part_end'] + 1
+        return self.size
 
     def dump(self, outfile):
         """Dumps the content of device into a file.
@@ -311,31 +468,31 @@ class DiskDevice(object):
         This method will only dump the actual payload, found by reading the
         partition table. Empty space in the end of the device will be ignored.
         """
-        blocksize = 2 ** 22  # 4MB
-        size = self.size()
-        progress_size = (size + 2 ** 20 - 1) // 2 ** 20  # in MB
-        progressbar = progress("Dumping image file: ", progress_size)
-
-        source = open(self.device, "r")
-        try:
-            dest = open(outfile, "w")
-            try:
+        MB = 2 ** 20
+        blocksize = 4 * MB  # 4MB
+        size = self.size
+        progr_size = (size + MB - 1) // MB  # in MB
+        progressbar = self.out.Progress(progr_size, "Dumping image file", 'mb')
+
+        with open(self.real_device, 'r') as src:
+            with open(outfile, "w") as dst:
                 left = size
                 offset = 0
                 progressbar.next()
                 while left > 0:
                     length = min(left, blocksize)
-                    sent = sendfile(dest.fileno(), source.fileno(), offset,
-                                                                        length)
+                    sent = sendfile(dst.fileno(), src.fileno(), offset, length)
+
+                    # Workaround for python-sendfile API change. In
+                    # python-sendfile 1.2.x (py-sendfile) the returning value
+                    # of sendfile is a tuple, where in version 2.x (pysendfile)
+                    # it is just a sigle integer.
+                    if isinstance(sent, tuple):
+                        sent = sent[1]
+
                     offset += sent
                     left -= sent
-                    for i in range((length + 2 ** 20 - 1) // 2 ** 20):
-                        progressbar.next()
-            finally:
-                dest.close()
-        finally:
-            source.close()
-
-        success('Image file %s was successfully created' % outfile)
+                    progressbar.goto((size - left) // MB)
+        progressbar.success('image file %s was successfully created' % outfile)
 
 # vim: set sta sts=4 shiftwidth=4 sw=4 et ai :