diskimage-builder/diskimage_builder/block_device/blockdevice.py

496 lines
18 KiB
Python
Raw Normal View History

# Copyright 2016-2017 Andreas Florath (andreas@florath.net)
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import codecs
import collections.abc
import json
import logging
import os
import pickle
import pprint
import shutil
import yaml
from diskimage_builder.block_device.config import config_tree_to_graph
from diskimage_builder.block_device.config import create_graph
from diskimage_builder.block_device.exception import \
BlockDeviceSetupException
from diskimage_builder.block_device.utils import exec_sudo
logger = logging.getLogger(__name__)
def _load_json(file_name):
"""Load file from .json file on disk, return None if not existing"""
if os.path.exists(file_name):
with codecs.open(file_name, encoding="utf-8", mode="r") as fd:
return json.load(fd)
return None
class BlockDeviceState(collections.abc.MutableMapping):
"""The global state singleton
An reference to an instance of this object is saved into nodes as
a global repository. It wraps a single dictionary "state" and
provides a few helper functions.
The state ends up used in two contexts:
- The node list (including this state) is pickled and dumped
between cmd_create() and later cmd_* calls that need to call
the nodes.
- Some other cmd_* calls, such as cmd_writefstab, only need
access to values inside the state and not the whole node list,
and load it from the json dump created after cmd_create()
"""
# XXX:
# - we could implement getters/setters such that if loaded from
# disk, the state is read-only? or make it append-only
# (i.e. you can't overwrite existing keys)
def __init__(self, filename=None):
"""Initialise state
:param filename: if :param:`filename` is passed and exists, it
will be loaded as the state. If it does not exist an
exception is raised. If :param:`filename` is not
passed, state will be initalised to a blank dictionary.
"""
if filename:
if not os.path.exists(filename):
raise BlockDeviceSetupException("State dump not found")
else:
self.state = _load_json(filename)
assert self.state is not None
else:
self.state = {}
def __delitem__(self, key):
del self.state[key]
def __getitem__(self, key):
return self.state[key]
def __setitem__(self, key, value):
self.state[key] = value
def __iter__(self):
return iter(self.state)
def __len__(self):
return len(self.state)
def save_state(self, filename):
"""Persist the state to disk
:param filename: The file to persist state to
"""
logger.debug("Writing state to: %s", filename)
self.debug_dump()
with open(filename, "w") as fd:
json.dump(self.state, fd)
def debug_dump(self):
"""Log state to debug"""
# This is pretty good for human consumption, but maybe a bit
# verbose.
nice_output = pprint.pformat(self.state, width=40)
for line in nice_output.split('\n'):
logger.debug('{0:{fill}{align}50}'.format(
line, fill=' ', align='<'))
class BlockDevice(object):
"""Handles block devices.
This class handles the complete setup and deletion of all aspects
of the block device level.
A typical call sequence:
cmd_init: initialize the block device level config. After this
call it is possible to e.g. query information from the (partially
automatic generated) internal state like root-label.
cmd_getval: retrieve information about the (internal) block device
state like the block image device (for bootloader) or the
root-label (for writing fstab).
cmd_create: creates all the different aspects of the block
device. When this call is successful, the complete block level
device is set up, filesystems are created and are mounted at
the correct position.
After this call it is possible to copy / install all the needed
files into the appropriate directories.
cmd_writefstab: creates the (complete) fstab for the system.
cmd_umount: unmount and detaches all directories and used many
resources. After this call the used (e.g.) images are still
available for further handling, e.g. converting from raw in
some other format.
cmd_cleanup: removes everything that was created with the
'cmd_create' call, i.e. all images files themselves and
internal temporary configuration.
cmd_delete: unmounts and removes everything that was created
during the 'cmd_create' all. This call should be used in error
conditions when there is the need to remove all allocated
resources immediately and as good as possible.
From the functional point of view this is mostly the same as a
call to 'cmd_umount' and 'cmd_cleanup' - but is typically more
error tolerance.
In a script this should be called in the following way:
dib-block-device init ...
# From that point the database can be queried, like
ROOT_LABEL=$(dib-block-device getval root-label)
Please note that currently the dib-block-device executable can
only be used outside the chroot.
dib-block-device create ...
trap "dib-block-device delete ..." EXIT
# copy / install files
dib-block-device umount ...
# convert image(s)
dib-block-device cleanup ...
trap - EXIT
"""
def _merge_rootfs_params(self):
"""Merge rootfs related parameters into configuration
To maintain compatability with some old block-device
environment variables from before we had a specific
block-device config, disk-image-create provides a "parameters"
file that translates the old bash-environment variables into a
YAML based configuration file (``self.params``).
Here we merge the values in this parameters file that relate
to the root file-system into the final configuration. We look
for the ``mkfs_root`` node in the new config, and pull the
relevant settings from the parameters into it.
"""
for entry in self.config:
for k, v in entry.items():
if k == 'mkfs':
if 'name' not in v:
continue
if v['name'] != 'mkfs_root':
continue
if 'type' not in v \
and 'root-fs-type' in self.params:
v['type'] = self.params['root-fs-type']
if 'opts' not in v \
and 'root-fs-opts' in self.params:
v['opts'] = self.params['root-fs-opts']
if 'label' not in v \
and 'root-label' in self.params:
if self.params['root-label'] is not None:
v['label'] = self.params['root-label']
else:
# The default label is "cloudimg-rootfs"
# for historical reasons (upstream
# images/EC2 defaults/cloud-init etc). We
# want to remain backwards compatible, but
# unfortunately that's too long for XFS so
# we've decided on 'img-rootfs' in that
# case. Note there's separate checks if
# the label is specified explicitly.
if v.get('type') == 'xfs':
v['label'] = 'img-rootfs'
else:
v['label'] = 'cloudimg-rootfs'
def __init__(self, params):
"""Create BlockDevice object
Arguments:
:param params: YAML file from --params
"""
logger.debug("Creating BlockDevice object")
self.params = params
logger.debug("Params [%s]", self.params)
self.state_dir = os.path.join(
self.params['build-dir'], "states/block-device")
self.state_json_file_name \
= os.path.join(self.state_dir, "state.json")
self.config_json_file_name \
= os.path.join(self.state_dir, "config.json")
self.node_pickle_file_name \
= os.path.join(self.state_dir, "nodes.pickle")
self.config = _load_json(self.config_json_file_name)
# This needs to exists for the state and config files
try:
os.makedirs(self.state_dir)
except OSError:
pass
def cmd_init(self):
"""Initialize block device setup
This initializes the block device setup layer. One major task
is to parse and check the configuration, write it down for
later examiniation and execution.
"""
with open(self.params['config'], "rt") as config_fd:
self.config = yaml.safe_load(config_fd)
logger.debug("Config before merge [%s]", self.config)
self.config = config_tree_to_graph(self.config)
logger.debug("Config before merge [%s]", self.config)
self._merge_rootfs_params()
logger.debug("Final config [%s]", self.config)
# Write the final config
with open(self.config_json_file_name, "wt") as fd:
json.dump(self.config, fd)
logger.info("Wrote final block device config to [%s]",
self.config_json_file_name)
def _config_get_mount(self, path):
for entry in self.config:
for k, v in entry.items():
if k == 'mount' and v['mount_point'] == path:
return v
assert False
def _config_get_all_mount_points(self):
rvec = []
for entry in self.config:
for k, v in entry.items():
if k == 'mount':
rvec.append(v['mount_point'])
return rvec
def _config_get_mkfs(self, name):
for entry in self.config:
for k, v in entry.items():
if k == 'mkfs' and v['name'] == name:
return v
assert False
def cmd_getval(self, symbol):
"""Retrieve value from block device level
The value of SYMBOL is printed to stdout. This is intended to
be captured into bash-variables for backward compatibility
(non python) access to internal configuration.
Arguments:
:param symbol: the symbol to get
"""
logger.info("Getting value for [%s]", symbol)
if symbol == "root-label":
root_mount = self._config_get_mount("/")
root_fs = self._config_get_mkfs(root_mount['base'])
logger.debug("root-label [%s]", root_fs['label'])
print("%s" % root_fs['label'])
return 0
if symbol == "root-fstype":
root_mount = self._config_get_mount("/")
root_fs = self._config_get_mkfs(root_mount['base'])
logger.debug("root-fstype [%s]", root_fs['type'])
print("%s" % root_fs['type'])
return 0
Correct boot path to cover FIPS usage cases When your booting a Linux system using dracut, i.e. with any redhat style distribution, dracut's internal code looks to validate the kernel hmac signature in before proceeding to userspace. It does this by looking at the /boot/ folder file for the kernel hmac file. And it normally does this with the root filesystem. Except if the kernel is not on the root filesystem and is instead on a /boot filesystem, this breaks horribly. This is compounded because DIB enables the operator to restructure the OS image/layout to fit their needs. In order for this to be navigated, as dracut is written, we need to pass a "boot=" argument to the kernel. So now we attempt to purge any prior boot entry in the disk image content, which is good because any filesystem operations invalidate it, and then we attempt to identify the boot filesystem, and save a boot kernel command line parameter so the resulting image can boot properly if FIPS was enabled in the prior image. Regex developed with https://sed.js.org utilizing stdin: VAR="quiet boot=UUID=173c759f-1302-48a3-9d51-a17784c21e03 text" VAR="quiet boot=PARTUUID=173c759f-1302-48a3-9d51-a17784c21e03" VAR="quiet boot=PARTUUID=173c759f-1302-48a3-9d51-a17784c21e03 reboot=meow" VAR="quiet boot=UUID=/dev/sda1 text" VAR="quiet boot=/dev/sda1" VAR="quiet boot=/dev/sda1 reboot=meow" VAR="quiet after_boot=1 reboot=meow boot=/dev/sda1" VAR="quiet after_boot=1 reboot=meow" Which resulted in stdout: VAR="quiet text" VAR="quiet" VAR="quiet reboot=meow" VAR="quiet text" VAR="quiet" VAR="quiet reboot=meow" VAR="quiet after_boot=1 reboot=meow" VAR="quiet after_boot=1 reboot=meow" Change-Id: I9034c21e84deda2ba2c0ec0d1d6d6595ed10bed4
2023-03-02 16:43:50 +00:00
if symbol == "boot-label":
try:
boot_mount = self._config_get_mount("/boot")
boot_fs = self._config_get_mkfs(boot_mount['base'])
# If not explicitly defined, we appear to fallback
# to name for a label, which we can only get from the
# resulting filesystem config.
boot_label = boot_fs.get('label', boot_fs.get('name', ''))
except AssertionError:
boot_label = ''
logger.debug("boot-label [%s]", boot_label)
print("%s" % boot_label)
return 0
if symbol == 'mount-points':
mount_points = self._config_get_all_mount_points()
# we return the mountpoints joined by a pipe, because it is not
# a valid char in directories, so it is a safe separator for the
# mountpoints list
print("%s" % "|".join(mount_points))
return 0
# the following symbols all come from the global state
# dictionary. They can only be accessed after the state has
# been dumped; i.e. after cmd_create() called.
state = BlockDeviceState(self.state_json_file_name)
Pass all blockdevices to bootloader Currently we only export "image-block-device" which is the loopback device (/dev/loopX) for the underlying image. This is the device we install grub to (from inside the chroot ...) This is ok for x86, but is insufficient for some platforms like PPC which have a separate boot partition. They do not want to install to the loop device, but do things like dd special ELF files into special boot partitions. The first problem seems to be that in level1/partitioning.py we have a whole bunch of different paths that either call partprobe on the loop device, or kpartx. We have _all_part_devices_exist() that gates the kpartx for unknown reasons. We have detach_loopback() that does not seem to remove losetup created devices. I don't think this does cleanup if it uses kpartx correctly. It is extremley unclear what's going to be mapped where. This moves to us *only* using kpartx to map the partitions of the loop device. We will *not* call partprobe and create the /dev/loopXpN devices and will only have the devicemapper nodes kpartx creates. This seems to be best. Cleanup happens inside partitioning.py. practice. Deeper thinking about this, and more cleanup of the variables will be welcome. This adds "image-block-devices" (note the extra "s") which exports all the block devices with name and path. This is in a string format that can be eval'd to an array (you can't export arrays). This is then used in a follow-on (I0918e8df8797d6dbabf7af618989ab7f79ee9580) to pick the right partition on PPC. Change-Id: If8e33106b4104da2d56d7941ce96ffcb014907bc
2017-06-06 02:09:24 +00:00
# The path to the .raw file for conversion
if symbol == 'image-path':
print("%s" % state['blockdev']['image0']['image'])
return 0
Pass all blockdevices to bootloader Currently we only export "image-block-device" which is the loopback device (/dev/loopX) for the underlying image. This is the device we install grub to (from inside the chroot ...) This is ok for x86, but is insufficient for some platforms like PPC which have a separate boot partition. They do not want to install to the loop device, but do things like dd special ELF files into special boot partitions. The first problem seems to be that in level1/partitioning.py we have a whole bunch of different paths that either call partprobe on the loop device, or kpartx. We have _all_part_devices_exist() that gates the kpartx for unknown reasons. We have detach_loopback() that does not seem to remove losetup created devices. I don't think this does cleanup if it uses kpartx correctly. It is extremley unclear what's going to be mapped where. This moves to us *only* using kpartx to map the partitions of the loop device. We will *not* call partprobe and create the /dev/loopXpN devices and will only have the devicemapper nodes kpartx creates. This seems to be best. Cleanup happens inside partitioning.py. practice. Deeper thinking about this, and more cleanup of the variables will be welcome. This adds "image-block-devices" (note the extra "s") which exports all the block devices with name and path. This is in a string format that can be eval'd to an array (you can't export arrays). This is then used in a follow-on (I0918e8df8797d6dbabf7af618989ab7f79ee9580) to pick the right partition on PPC. Change-Id: If8e33106b4104da2d56d7941ce96ffcb014907bc
2017-06-06 02:09:24 +00:00
# This is the loopback device where the above image is setup
if symbol == 'image-block-device':
print("%s" % state['blockdev']['image0']['device'])
return 0
# Full list of created devices by name. Some bootloaders, for
# example, want to be able to see their boot partitions to
# copy things in. Intended to be read into a bash array
if symbol == 'image-block-devices':
out = ""
for k, v in state['blockdev'].items():
out += " [%s]=%s " % (k, v['device'])
print(out)
return 0
logger.error("Invalid symbol [%s] for getval", symbol)
return 1
def cmd_writefstab(self):
"""Creates the fstab"""
logger.info("Creating fstab")
# State should have been created by prior calls; we only need
# the dict
state = BlockDeviceState(self.state_json_file_name)
tmp_fstab = os.path.join(self.state_dir, "fstab")
with open(tmp_fstab, "wt") as fstab_fd:
# This gives the order in which this must be mounted
for mp in state['mount_order']:
logger.debug("Writing fstab entry for [%s]", mp)
fs_base = state['mount'][mp]['base']
fs_name = state['mount'][mp]['name']
fs_val = state['filesys'][fs_base]
if 'label' in fs_val:
diskid = "LABEL=%s" % fs_val['label']
else:
diskid = "UUID=%s" % fs_val['uuid']
# If there is no fstab entry - do not write anything
if 'fstab' not in state:
continue
if fs_name not in state['fstab']:
continue
options = state['fstab'][fs_name]['options']
dump_freq = state['fstab'][fs_name]['dump-freq']
fsck_passno = state['fstab'][fs_name]['fsck-passno']
fstab_fd.write("%s %s %s %s %s %s\n"
% (diskid, mp, fs_val['fstype'],
options, dump_freq, fsck_passno))
target_etc_dir = os.path.join(self.params['build-dir'], 'built', 'etc')
exec_sudo(['mkdir', '-p', target_etc_dir])
exec_sudo(['cp', tmp_fstab, os.path.join(target_etc_dir, "fstab")])
return 0
def cmd_create(self):
"""Creates the block device"""
logger.info("create() called")
logger.debug("Using config [%s]", self.config)
# Create a new, empty state
state = BlockDeviceState()
try:
dg, call_order = create_graph(self.config, self.params, state)
for node in call_order:
node.create()
except Exception:
logger.exception("Create failed; rollback initiated")
reverse_order = reversed(call_order)
for node in reverse_order:
node.rollback()
# save the state for debugging
state.save_state(self.state_json_file_name)
logger.error("Rollback complete, exiting")
raise
# dump state and nodes, in order
# XXX: we only dump the call_order (i.e. nodes) not the whole
# graph here, because later calls do not need the graph
# at this stage. might they?
state.save_state(self.state_json_file_name)
pickle.dump(call_order, open(self.node_pickle_file_name, 'wb'))
logger.info("create() finished")
return 0
def cmd_umount(self):
"""Unmounts the blockdevice and cleanup resources"""
# If the state is not here, cmd_cleanup removed it? Nothing
# more to do?
# XXX: better understand this...
if not os.path.exists(self.node_pickle_file_name):
logger.info("State already cleaned - no way to do anything here")
return 0
call_order = pickle.load(open(self.node_pickle_file_name, 'rb'))
reverse_order = reversed(call_order)
for node in reverse_order:
node.umount()
return 0
def cmd_cleanup(self):
"""Cleanup all remaining relicts - in good case"""
# Cleanup must be done in reverse order
try:
call_order = pickle.load(open(self.node_pickle_file_name, 'rb'))
except IOError:
raise BlockDeviceSetupException("Pickle file not found")
reverse_order = reversed(call_order)
for node in reverse_order:
node.cleanup()
logger.info("Removing temporary state dir [%s]", self.state_dir)
shutil.rmtree(self.state_dir)
return 0
def cmd_delete(self):
"""Cleanup all remaining relicts - in case of an error"""
# Deleting must be done in reverse order
try:
call_order = pickle.load(open(self.node_pickle_file_name, 'rb'))
except IOError:
raise BlockDeviceSetupException("Pickle file not found")
reverse_order = reversed(call_order)
for node in reverse_order:
node.delete()
logger.info("Removing temporary state dir [%s]", self.state_dir)
shutil.rmtree(self.state_dir)
return 0