diff --git a/.lgtm.yml b/.lgtm.yml index 5948d8c2bc..51ec50e3b1 100644 --- a/.lgtm.yml +++ b/.lgtm.yml @@ -5,6 +5,9 @@ extraction: - python3-pip - python3-setuptools - python3-wheel + - libfdisk-dev + - libp11-kit-dev + - libssl-dev after_prepare: - pip3 install meson - export PATH="$HOME/.local/bin/:$PATH" diff --git a/.mkosi/mkosi.fedora b/.mkosi/mkosi.fedora index 911908cb77..dbd9b4df99 100644 --- a/.mkosi/mkosi.fedora +++ b/.mkosi/mkosi.fedora @@ -8,7 +8,7 @@ Distribution=fedora Release=31 [Output] -Format=raw_btrfs +Format=gpt_ext4 Bootable=yes KernelCommandLine=printk.devkmsg=on @@ -38,19 +38,20 @@ BuildPackages= libblkid-devel libcap-devel libcurl-devel + libfdisk-devel libgcrypt-devel libidn2-devel libmicrohttpd-devel libmount-devel libseccomp-devel libselinux-devel - libtool libxkbcommon-devel libxslt lz4 lz4-devel m4 meson + openssl-devel pam-devel pcre2-devel pkgconfig @@ -58,10 +59,17 @@ BuildPackages= python3-lxml qrencode-devel tree + valgrind-devel xz-devel Packages= + coreutils + cryptsetup-libs + kmod-libs libidn2 + libseccomp + procps-ng + util-linux BuildDirectory=mkosi.builddir Cache=mkosi.cache diff --git a/TODO b/TODO index a66e009202..d77f5ffb87 100644 --- a/TODO +++ b/TODO @@ -19,6 +19,11 @@ Janitorial Clean-ups: Features: +* bootctl: + - teach it to prepare an ESP wholesale, i.e. with mkfs.vfat invocation + - teach it to copy in unified kernel images and maybe type #1 boot loader spec entries from host + - make it operate on loopback files, dissecting enough to find ESP to operate on + * when dissecting images, warn about unrecognized partition flags * honour specifiers in unit files that resolve to some very basic @@ -27,8 +32,53 @@ Features: * socket units: allow creating a udev monitor socket with ListenDevices= or so, with matches, then actviate app thorugh that passing socket oveer +* unify on openssl: + - port sd_id128_get_machine_app_specific() over from khash + - port resolved over from libgcrypt (DNSSEC code) + - port journald + fsprg over from libgcrypt + - port importd over from libgcrypt + - when that's done: kill khash.c + - when that's done: kill gnutls support in resolved + * kill zenata, all hail weblate? +* when we resize disks (homed?) always round up to 4K sectors, not 512K + +* add growvol and makevol options for /etc/crypttab, similar to + x-systemd.growfs and x-systemd-makefs. + +* hook up the TPM to /etc/crypttab, with a new option that is similar to the + new PKCS#11 option in crypttab, and allows unlocking a LUKS volume via a key + unsealed from the TPM. Optionally, if TPM is not available fall back to + TPM-less mode, and set up linear DM mapping instead (inspired by kpartx), so + that the device paths stay the same, regardless if crypto is used or not. + +* move discoverable partitions spec into markdown and our tree + +* systemd-repart: by default generate minimized partition tables (i.e. tables + that only covere the space actually used, excluding any free space at the + end), in order to maximize dd'ability. Requires libfdisk work, see + https://github.com/karelzak/util-linux/issues/907 + +* systemd-repart: optionally, allow specifiying a path to initialize new + partitions from, i.e. an fs image file or a source device node. This would + then turn systemd-repart into a simple installer: with a few .repart files + you could replicate the host system on another device. + +* systemd-repart: MBR partition table support. Care needs to be taken regarding + Type=, so that partition definitions can sanely apply to both the GPT and the + MBR case. Idea: accept syntax "Type=gpt:home mbr:0x83" for setting the types + for the two partition types explicitly. And provide an internal mapping so + that "Type=linux-generic" maps to the right types for both partition tables + automatically. + +* systemd-repart: allow sizing partitions as factor of available RAM, so that + we can reasonably size swap partitions for hibernation. + +* when switching root from initrd to host, set the machine_id env var so that + if the host has no machine ID set yet we continue to use the random one the + initrd had set. + * sd-event: add native support for P_ALL waitid() watching, then move PID 1 to it fo reaping assigned but unknown children. This needs to some special care to operate somewhat sensibly in light of priorities: P_ALL will return @@ -175,13 +225,6 @@ Features: * introduce per-unit (i.e. per-slice, per-service) journal log size limits. -* optionally, if a per-partition GPT flag is set for the root/home/… partitions - format the partition on next boot and unset the flag, in order to implement - factory reset. also, add a second flag that simply indicates whether such a - scheme is supported. then, add a tool (or maybe beef up systemd-dissect) to - show state of these flags, and optionally trigger such a factory reset on - next boot by setting the flag. - * sd-boot: automatically load EFI modules from some drop-in dir, so that people can add in file system drivers and such @@ -395,10 +438,6 @@ Features: yogas can be recognized as "convertible" too, even if they predate the DMI "convertible" form factor -* Maybe add a small tool invoked early at boot, that adds in or resizes - partitions automatically, to be used when the media used is actually larger - than the image written onto it is. - * Maybe add PrivatePIDs= as new unit setting, and do minimal PID namespacing after all. Be strict however, only support the equivalent of nspawn's --as-pid2 switch, and sanely proxy sd_notify() messages dropping stuff such diff --git a/fuzzbuzz.yaml b/fuzzbuzz.yaml index 18c70e3555..6dfdd81237 100644 --- a/fuzzbuzz.yaml +++ b/fuzzbuzz.yaml @@ -5,6 +5,7 @@ setup: - sudo apt-get update -y - sudo apt-get build-dep -y systemd - sudo apt-get install -y python3-pip +- sudo apt-get install -y libfdisk-dev libp11-kit-dev libssl-dev # FIXME: temporarily pin the meson version as 0.53 doesn't work with older # python 3.5 # # See: https://github.com/mesonbuild/meson/issues/6427 diff --git a/man/repart.d.xml b/man/repart.d.xml new file mode 100644 index 0000000000..2fe74193d3 --- /dev/null +++ b/man/repart.d.xml @@ -0,0 +1,388 @@ + + + + + + repart.d + systemd + + + + repart.d + 5 + + + + repart.d + Partition Definition Files for Automatic Boot-Time Repartitioning + + + + /etc/repart.d/*.conf +/run/repart.d/*.conf +/usr/lib/repart.d/*.conf + + + + + Description + + repart.d/*.conf files describe basic properties of partitions of block + devices of the local system. They may be used to declare types, names and sizes of partitions that shall + exist. The + systemd-repart8 + service reads these files and attempts to add new partitions currently missing and enlarge existing + partitions according to these definitions. Operation is generally incremental, i.e. when applied, what + exists already is left intact, and partitions are never shrunk, moved or deleted. + + These definition files are useful for implementing operating system images that are prepared and + delivered with minimally sized images (for example lacking any state or swap partitions), and which on + first boot automatically take possession of any remaining disk space following a few basic rules. + + Currently, support for partition definition files is only implemented for GPT partitition + tables. + + Partition files are generally matched against any partitions already existing on disk in a simple + algorithm: the partition files are sorted by their filename (ignoring the directory prefix), and then + compared in order against existing partitions matching the same partition type UUID. Specifically, the + first existing partition with a specific partition type UUID is assigned the first definition file with + the same partition type UUID, and the second existing partition with a specific type UUID the second + partition file with the same type UUID, and so on. Any left-over partition files that have no matching + existing partition are assumed to define new partition that shall be created. Such partitions are + appended to the end of the partition table, in the order defined by their names utilizing the first + partition slot greater than the highest slot number currently in use. Any existing partitions that have + no matching partition file are left as they are. + + Note that these partition definition files do not describe the contents of the partitions, such as + the file system used. Separate mechanisms, such as + systemd-growfs8 and + systemd-makefs maybe be used to initialize or grow the file systems inside of these + partitions. + + + + [Partition] Section Options + + + + Type= + + The GPT partition type UUID to match. This may be a GPT partition type UUID such as + 4f68bce3-e8cd-4db1-96e7-fbcaf984b709, or one of the following special + identifiers: + + + GPT partition type identifiers + + + + + + + + Identifier + Explanation + + + + + + esp + EFI System Partition + + + + xbootldr + Extended Boot Loader Partition + + + + swap + Swap partition + + + + home + Home (/home/) partition + + + + srv + Server data (/srv/) partition + + + + var + Variable data (/var/) partition + + + + tmp + Temporary data (/var/tmp/) partition + + + + linux-generic + Generic Linux file system partition + + + + root + Root file system partition type appropriate for the local architecture (an alias for an architecture root file system partition type listed below, e.g. root-x86-64) + + + + root-verity + Verity data for the root file system partition for the local architecture + + + + root-secondary + Root file system partition of the secondary architecture of the local architecture; usually the matching 32bit architecture for the local 64bit architecture) + + + + root-secondary-verity + Verity data for the root file system partition of the secondary architecture + + + + root-x86 + Root file system partition for the x86 (32bit, aka i386) architecture + + + + root-x86-verity + Verity data for the x86 (32bit) root file system partition + + + + root-x86-64 + Root file system partition for the x86_64 (64bit, aka amd64) architecture + + + + root-x86-64-verity + Verity data for the x86_64 (64bit) root file system partition + + + + root-arm + Root file system partition for the ARM (32bit) architecture + + + + root-arm-verity + Verity data for the ARM (32bit) root file system partition + + + + root-arm64 + Root file system partition for the ARM (64bit, aka aarch64) architecture + + + + root-arm64-verity + Verity data for the ARM (64bit, aka aarch64) root file system partition + + + + root-ia64 + Root file system partition for the ia64 architecture + + + + root-ia64-verity + Verity data for the ia64 root file system partition + + + +
+ + This setting defaults to linux-generic. + + Most of the partition type UUIDs listed above are defined in the Discoverable Partitions + Specification. + + + + Label= + + The textual label to assign to the partition if none is assigned yet. Note that this + setting is not used for matching. It is also not used when a label is already set for an existing + partition. It is thus only used when a partition is newly created or when an existing one had a no + label set (that is: an empty label). If not specified a label derived from the partition type is + automatically used. + + + + Priority= + + A numeric priority to assign to this partition, in the range -2147483648…2147483647, + with smaller values indicating higher priority, and higher values indicating smaller priority. This + priority is used in case the configured size constraints on the defined partitions do not permit + fitting all partitions onto the available disk space. If the partitions do not fit, the highest + numeric partition priority of all defined partitions is determined, and all defined partitions with + this priority are removed from the list of new partitions to create (which may be multiple, if the + same priority is used for multiple partitions). The fitting algorithm is then tried again. If the + partitions still do not fit, the now highest numeric partition priority is determined, and the + matching partitions removed too, and so on. Partitions of a priority of 0 or lower are never + removed. If all partitions with a priority above 0 are removed and the partitions still do not fit on + the device the operation fails. Note that this priority has no effect on ordering partitions, for + that use the alphabetical order of the filenames of the partition definition files. Defaults to + 0. + + + + Weight= + + A numeric weight to assign to this partition in the range 0…1000000. Available disk + space is assigned the defined partitions according to their relative weights (subject to the size + constraints configured with SizeMinBytes=, SizeMaxBytes=), so + that a partition with weight 2000 gets double the space as one with weight 1000, and a partition with + weight 333 a third of that. Defaults to 1000. + + The Weight= setting is used to distribute available disk space in an + "elastic" fashion, based on the disk size and existing partitions. If a partition shall have a fixed + size use both SizeMinBytes= and SizeMaxBytes= with the same + value in order to fixate the size to one value, in which case the weight has no + effect. + + + + PaddingWeight= + + Similar to Weight= but sets a weight for the free space after the + partition (the "padding"). When distributing available space the weights of all partitions and all + defined padding is summed, and then each partition and padding gets the fraction defined by its + weight. Defaults to 0, i.e. by default no padding is applied. + + Padding is useful if empty space shall be left for later additions or a safety margin at the + end of the device or between partitions. + + + + SizeMinBytes= + SizeMaxBytes= + + Specifies minimum and maximum size constraints in bytes. Takes the usual K, M, G, T, + … suffixes (to the base of 1024). If SizeMinBytes= is specified the partition is + created at or grown to at least the specified size. If SizeMaxBytes= is specified + the partition is created at or grown to at most the specified size. The precise size is determined + through the weight value value configured with Weight=, see above. When + SizeMinBytes= is set equal to SizeMaxBytes= the configured + weight has no effect as the partition is explicitly sized to the specified fixed value. Note that + partitions are never created smaller than 4096 bytes, and since partitions are never shrunk the + previous size of the partition (in case the partition already exists) is also enforced as lower bound + for the new size. The values should be specified as multiples of 4096 bytes, and are rounded upwards + (in case of SizeMinBytes=) or downwards (in case of + SizeMaxBytes=) otherwise. If the backing device does not provide enough space to + fulfill the constraints placing the partition will fail. For partitions that shall be created, + depending on the setting of Priority= (see above) the partition might be dropped + and the placing algorithm restarted. By default no size constraints are set. + + + + PaddingMinBytes= + PaddingMaxBytes= + + Specifies minimum and maximum size constrains in bytes for the free space after the + partition (the "padding"). Semantics are similar to SizeMinBytes= and + SizeMaxBytes=, except that unlike partition sizes free space can be shrunk and can + be as small as zero. By default no size constraints on padding are set, so that only + PaddingWeight= determines the size of the padding applied. + + + + FactoryReset= + + Takes a boolean argument. If specified the partition is marked for removal during a + factory reset operation. This functionality is useful to implement schemes where images can be reset + into their original state by removing partitions and creating them anew. Defaults to off. + + + + + + Examples + + + Grow the root partition to the full disk size at first boot + + With the following file the root partition is automatically grown to the full disk if possible during boot. + + # /usr/lib/repart.d/50-root.conf +[Partition] +Type=root + + + + + Create a swap and home partition automatically on boot, if missing + + The home partition gets all available disk space while the swap partition gets 1G at most and 64M + at least. We set a priority > 0 on the swap partition to ensure the swap partition is not used if not + enough space is available. For every three bytes assigned to the home partition the swap partition gets + assigned one. + + # /usr/lib/repart.d/60-home.conf +[Partition] +Type=home + + + # /usr/lib/repart.d/70-swap.conf +[Partition] +Type=swap +SizeMinBytes=64M +SizeMaxBytes=1G +Priority=1 +Weight=333 + + + + + Create B partitions in an A/B Verity setup, if missing + + Let's say the vendor intends to update OS images in an A/B setup, i.e. with two root partitions + (and two matching Verity partitions) that shall be used alternatingly during upgrades. To minimize + image sizes the original image is shipped only with one root and one Verity partition (the "A" set), + and the second root and Verity partitions (the "B" set) shall be created on first boot on the free + space on the medium. + + # /usr/lib/repart.d/50-root.conf +[Partition] +Type=root +SizeMinBytes=512M +SizeMaxBytes=512M + + + # /usr/lib/repart.d/60-root-verity.conf +[Partition] +Type=root-verity +SizeMinBytes=64M +SizeMaxBytes=64M + + + The definitions above cover the "A" set of root partition (of a fixed 512M size) and Verity + partition for the root partition (of a fixed 64M size). Let's use symlinks to create the "B" set of + partitions, since after all they shall have the same properties and sizes as the "A" set. + +# ln -s 50-root.conf /usr/lib/repart.d/70-root-b.conf +# ln -s 60-root-verity.conf /usr/lib/repart.d/80-root-verity-b.conf + + + + + + + See Also + + systemd1, + systemd-repart8, + sfdisk8 + + + + diff --git a/man/rules/meson.build b/man/rules/meson.build index 90376da775..3dc0a045a7 100644 --- a/man/rules/meson.build +++ b/man/rules/meson.build @@ -47,6 +47,7 @@ manpages = [ ['pam_systemd', '8', [], 'HAVE_PAM'], ['portablectl', '1', [], 'ENABLE_PORTABLED'], ['pstore.conf', '5', ['pstore.conf.d'], 'ENABLE_PSTORE'], + ['repart.d', '5', [], ''], ['resolvectl', '1', ['resolvconf'], 'ENABLE_RESOLVE'], ['resolved.conf', '5', ['resolved.conf.d'], 'ENABLE_RESOLVE'], ['runlevel', '8', [], ''], @@ -769,6 +770,7 @@ manpages = [ 'ENABLE_RANDOMSEED'], ['systemd-rc-local-generator', '8', [], ''], ['systemd-remount-fs.service', '8', ['systemd-remount-fs'], ''], + ['systemd-repart', '8', ['systemd-repart.service'], ''], ['systemd-resolved.service', '8', ['systemd-resolved'], 'ENABLE_RESOLVE'], ['systemd-rfkill.service', '8', diff --git a/man/systemd-makefs@.service.xml b/man/systemd-makefs@.service.xml index 8514af67bc..d07d90315a 100644 --- a/man/systemd-makefs@.service.xml +++ b/man/systemd-makefs@.service.xml @@ -80,6 +80,7 @@ systemd1, systemd.mount8, systemd-fstab-generator8, + systemd-repart8, mkfs.btrfs8, mkfs.cramfs8, mkfs.ext48, diff --git a/man/systemd-repart.xml b/man/systemd-repart.xml new file mode 100644 index 0000000000..cffcb5403a --- /dev/null +++ b/man/systemd-repart.xml @@ -0,0 +1,269 @@ + + + + + + + + systemd-repart + systemd + + + + systemd-repart + 8 + + + + systemd-repart + systemd-repart.service + Automatically grow and add partitions + + + + + systemd-repart + OPTIONS + BLOCKDEVICE + + + systemd-repart.service + + + + Description + + systemd-repart grows and adds partitions to a partition table, based on the + configuration files described in + repart.d5. + + + If invoked with no arguments, it operates on the block device backing the root file system partition + of the OS, thus growing and adding partitions of the booted OS image itself. When called in the initial + RAM disk it operates on the block device backing /sysroot/ instead, i.e. on the + block device the system will soon transition into. The systemd-repart.service + service is generally run at boot in the initial RAM disk, in order to augment the partition table of the + OS before its partitions are mounted. systemd-repart (mostly) operates in a purely + incremental mode: it only grows existing and adds new partitions; it does not shrink, delete or move + existing partitions. The service is intended to be run on every boot, but when it detects that the + partition table already matches the installed repart.d/*.conf configuration + files, it executes no operation. + + systemd-repart is intended to be used when deploying OS images, to automatically + adjust them to the system they are running on, during first boot. This way the deployed image can be + minimal in size and may be augmented automatically at boot when needed, taking possession of disk space + available but not yet used. Specifically the following use cases are among those covered: + + + The root partition may be grown to cover the whole available disk space + A /home/, swap or /srv partition can be added in + A second (or third, …) root partition may be added in, to cover A/B style setups + where a second version of the root file system is alternatingly used for implementing update + schemes. The deployed image would carry only a single partition ("A") but on first boot a second + partition ("B") for this purpose is automatically created. + + + The algorithm executed by systemd-repart is roughly as follows: + + + The repart.d/*.conf configuration files are loaded and parsed, + and ordered by filename (without the directory suffix). + + The partition table already existing on the block device is loaded and + parsed. + + The existing partitions in the partition table are matched up with the + repart.d/*.conf files by GPT partition type UUID. The first existing partition + of a specific type is assigned the first configuration file declaring the same type. The second + existing partition of a specific type is then assigned the second configuration file declaring the same + type, and so on. After this iterative assigning is complete any left-over existing partitions that have + no matching configuration file are considered "foreign" and left as they are. And any configuration + files for which no partition currently exists are understood as a request to create such a + partition. + + Taking the size constraints and weights declared in the configuration files into + account, all partitions that shall be created are now allocated to the disk, taking up all free space, + always respecting the size and padding requests. Similar, existing partitions that are determined to + grow are grown. New partitions are always appended to the end of the existing partition table, taking + the first partition table slot whose index is greater than the indexes of all existing + partitions. Partition table slots are never reordered and thus partition numbers are ensured to remain + stable. Note that this allocation happens in RAM only, the partition table on disk is not updated + yet. + + All existing partitions for which configuration files exist and which currently have no + GPT partition label set will be assigned a label, either explicitly configured in the configuration or + (if that's missing) derived automatically from the partition type. The same is done for all partitions + that are newly created. These assignments are done in RAM only, too, the disk is not updated + yet. + + Similarly, all existing partitions for which configuration files exist and which + currently have an all-zero identifying UUID will be assigned a new UUID. This UUID is cryptographically + hashed from a common seed value together with the partition type UUID (and a counter in case multiple + partitions of the same type are defined), see below. The same is done for all partitions that are + created anew. These assignments are done in RAM only, too, the disk is not updated + yet. + + Similarly, if the disk's volume UUID is all zeroes it is also initialized, also + cryptographically hashed from the same common seed value. Also, in RAM only, too. + + The disk space assigned to new partitions (i.e. what was previously considered free + space but is no longer) is now erased. Specifically, all file system signatures are removed, and if the + device supports it the BLKDISCARD I/O control command is issued to inform the + hardware that the space is empty now. In addition any "padding" between partitions and at the end of + the device is similarly erased. + + The new partition table is finally written to disk. The kernel is asked to reread the + partition table. + + + As exception to the normally strictly incremental operation, when called in a special "factory + reset" mode systemd-repart may also be used to erase select existing partitions to + reset an installation back to vendor defaults. This mode of operation is used when either the + switch is passed on the tool's command line, or the + option specified on the kernel command line, or the + FactoryReset EFI variable (vendor UUID + 8cf2644b-4b0b-428f-9387-6d876050dc67) is set to "yes". It alters the algorithm above + slightly: between the 3rd and the 4th step above the any partition marked explicitly via the + FactoryReset= boolean is deleted, and the algorithm restarted, thus immediately + re-creating these partitions anew empty. + + Note that systemd-repart only changes partition tables, it does not create or + resize any file systems within these partitions. A separate mechanism should be used for that, for + example + systemd-growfs8 and + systemd-makefs. + + The UUIDs identifying the new partitions created (or assigned to existing partitions that have no + UUID yet), as well as the disk as a whole are hashed cryptographically from a common seed value. This + seed value is usually the + machine-id5 of the + system, so that the machine ID reproducibly determines the UUIDs assigned to all partitions. If the + machine ID cannot be read (or the user passes , see below) the seed is + generated randomly instead, so that the partition UUIDs are also effectively random. The seed value may + also be set explicitly, formatted as UUID via the option. By hashing these UUIDs + from a common seed images prepared with this tool become reproducible and the result of the algorithm + above deterministic. + + + + Options + + The following options are understood: + + + + + Takes a boolean. If this switch is not specified is + the implied default. Controls whether systemd-repart executes the requested + re-partition operations or whether it should only show what it would do. Unless + is specified systemd-repart will not actually + touch the device's partition table. + + + + + Takes one of refuse, allow, + require or force. Controls how to operate on block devices that + are entirely empty, i.e. carry no partition table/disk label yet. If this switch is not specified the + implied default is refuse. + + If refuse systemd-repart requires that the block device + it shall operate on already carries a partition table and refuses operation if none is found. If + allow the command will extend an existing partition table or create a new one if + none exists. If require the command will create a new partition table if none + exists so far, and refuse operation if one already exists. If force it will create + a fresh partition table unconditionally, erasing the disk fully in effect. If + force no existing partitions will be taken into account or survive the + operation. Hence: use with care, this is a great way to lose all your data. + + + + + + Takes a boolean. If this switch is not specified is + the implied default. Controls whether to issue the BLKDISCARD I/O control + command on the space taken up by any added partitions or on the space in between them. Usually, it's + a good idea to issue this request since it tells the underlying hardware that the covered blocks + shall be considered empty, improving performance. + + + + + + Takes boolean. If this switch is not specified is + the implied default. Controls whether to operate in "factory reset" mode, see above. If set to true + this will remove all existing partitions marked with FactoryReset= set to yes + early while executing the re-partitioning algorithm. Use with care, this is a great way to lose all + your data. Note that partition files need to explicitly turn FactoryReset= on, as + the option defaults to off. If no partitions are marked for factory reset this switch has no + effect. Note that there are two other methods to request factory reset operation: via the kernel + command line and via an EFI variable, see above. + + + + + + If this switch is specified the disk is not re-partitioned. Instead it is determined + if any existing partitions are marked with FactoryReset=. If there are the tool + will exit with exit status zero, otherwise non-zero. This switch may be used to quickly determine + whether the running system supports a factory reset mechanism built on + systemd-repart. + + + + + + Takes a path to a directory to use as root file system when searching for + repart.d/*.conf files and for the machine ID file to use as seed. By default + when invoked on the regular system this defaults to the host's root file system + /. If invoked from the initial RAM disk this defaults to + /sysroot/, so that the tool operates on the configuration and machine ID stored + in the root file system later transitioned into itself. + + + + + + Takes a UUID as argument or the special value random. If a UUID + is specified the UUIDs to assign to partitions and the partition table itself are derived via + cryptographic hashing from it. If not specified it is attempted to read the machine ID from the host + (or more precisely, the root directory configured via ) and use it as seed + instead, falling back to a randomized seed otherwise. Use to force a + randomized seed. Explicitly specifying the seed may be used to generated strictly reproducible + partition tables. + + + + + + Takes a boolean argument. If this switch is not specified, it defaults to on when + called from an interactive terminal and off otherwise. Controls whether to show a user friendly table + and graphic illustrating the changes applied. + + + + + + Takes a file system path. If specified the *.conf are directly + read from the specified directory instead of searching in + /usr/lib/repart.d/*.conf, /etc/repart.d/*.conf, + /run/repart.d/*.conf. + + + + + + + + + See Also + + systemd1, + repart.d5, + machine-id5 + + + + diff --git a/meson.build b/meson.build index 848140bb03..54820d3f6a 100644 --- a/meson.build +++ b/meson.build @@ -873,6 +873,17 @@ endif libmount = dependency('mount', version : fuzzer_build ? '>= 0' : '>= 2.30') +want_libfdisk = get_option('fdisk') +if want_libfdisk != 'false' and not skip_deps + libfdisk = dependency('fdisk', + required : want_libfdisk == 'true') + have = libfdisk.found() +else + have = false + libfdisk = [] +endif +conf.set10('HAVE_LIBFDISK', have) + want_seccomp = get_option('seccomp') if want_seccomp != 'false' and not skip_deps libseccomp = dependency('libseccomp', @@ -1279,6 +1290,18 @@ conf.set('DEFAULT_DNS_OVER_TLS_MODE', 'DNS_OVER_TLS_' + default_dns_over_tls.underscorify().to_upper()) substs.set('DEFAULT_DNS_OVER_TLS_MODE', default_dns_over_tls) +want_repart = get_option('repart') +if want_repart != 'false' + have = (conf.get('HAVE_OPENSSL') == 1 and + conf.get('HAVE_LIBFDISK') == 1) + if want_repart == 'true' and not have + error('repart support was requested, but dependencies are not available') + endif +else + have = false +endif +conf.set10('ENABLE_REPART', have) + want_importd = get_option('importd') if want_importd != 'false' have = (conf.get('HAVE_LIBCURL') == 1 and @@ -1535,6 +1558,7 @@ subdir('src/coredump') subdir('src/pstore') subdir('src/hostname') subdir('src/import') +subdir('src/partition') subdir('src/kernel-install') subdir('src/locale') subdir('src/machine') @@ -2381,6 +2405,21 @@ if conf.get('ENABLE_BINFMT') == 1 mkdir_p.format(join_paths(sysconfdir, 'binfmt.d'))) endif +if conf.get('ENABLE_REPART') == 1 + executable('systemd-repart', + systemd_repart_sources, + include_directories : includes, + link_with : [libshared], + dependencies : [threads, + libcryptsetup, + libblkid, + libfdisk, + libopenssl], + install_rpath : rootlibexecdir, + install : true, + install_dir : rootbindir) +endif + if conf.get('ENABLE_VCONSOLE') == 1 executable('systemd-vconsole-setup', 'src/vconsole/vconsole-setup.c', @@ -3276,6 +3315,7 @@ foreach tuple : [ ['libiptc'], ['elfutils'], ['binfmt'], + ['repart'], ['vconsole'], ['quotacheck'], ['tmpfiles'], diff --git a/meson_options.txt b/meson_options.txt index 6736240f39..e512d25480 100644 --- a/meson_options.txt +++ b/meson_options.txt @@ -80,6 +80,8 @@ option('environment-d', type : 'boolean', description : 'support for environment.d') option('binfmt', type : 'boolean', description : 'support for custom binary formats') +option('repart', type : 'combo', choices : ['auto', 'true', 'false'], + description : 'install the systemd-repart tool') option('coredump', type : 'boolean', description : 'install the coredump handler') option('pstore', type : 'boolean', @@ -260,6 +262,8 @@ option('audit', type : 'combo', choices : ['auto', 'true', 'false'], description : 'libaudit support') option('blkid', type : 'combo', choices : ['auto', 'true', 'false'], description : 'libblkid support') +option('fdisk', type : 'combo', choices : ['auto', 'true', 'false'], + description : 'libfdisk support') option('kmod', type : 'combo', choices : ['auto', 'true', 'false'], description : 'support for loadable modules') option('pam', type : 'combo', choices : ['auto', 'true', 'false'], diff --git a/semaphoreci/semaphore-runner.sh b/semaphoreci/semaphore-runner.sh index 504ada59ee..a5d8bc26a1 100755 --- a/semaphoreci/semaphore-runner.sh +++ b/semaphoreci/semaphore-runner.sh @@ -35,6 +35,8 @@ while [ -z "\$(ip route list 0/0)" ]; do sleep 1; done apt-get -q --allow-releaseinfo-change update apt-get -y dist-upgrade apt-get install -y eatmydata +# The following four are needed as long as these deps are not covered by Debian's own packaging +apt-get install -y libfdisk-dev libp11-kit-dev libssl-dev apt-get purge --auto-remove -y unattended-upgrades systemctl unmask systemd-networkd systemctl enable systemd-networkd diff --git a/src/basic/locale-util.c b/src/basic/locale-util.c index dc62511935..96151ffbf8 100644 --- a/src/basic/locale-util.c +++ b/src/basic/locale-util.c @@ -345,6 +345,9 @@ const char *special_glyph(SpecialGlyph code) { [SPECIAL_GLYPH_MU] = "u", [SPECIAL_GLYPH_CHECK_MARK] = "+", [SPECIAL_GLYPH_CROSS_MARK] = "-", + [SPECIAL_GLYPH_LIGHT_SHADE] = "-", + [SPECIAL_GLYPH_DARK_SHADE] = "X", + [SPECIAL_GLYPH_SIGMA] = "S", [SPECIAL_GLYPH_ARROW] = "->", [SPECIAL_GLYPH_ELLIPSIS] = "...", [SPECIAL_GLYPH_ECSTATIC_SMILEY] = ":-]", @@ -371,6 +374,9 @@ const char *special_glyph(SpecialGlyph code) { [SPECIAL_GLYPH_MU] = "\316\274", /* μ (actually called: GREEK SMALL LETTER MU) */ [SPECIAL_GLYPH_CHECK_MARK] = "\342\234\223", /* ✓ */ [SPECIAL_GLYPH_CROSS_MARK] = "\342\234\227", /* ✗ (actually called: BALLOT X) */ + [SPECIAL_GLYPH_LIGHT_SHADE] = "\342\226\221", /* ░ */ + [SPECIAL_GLYPH_DARK_SHADE] = "\342\226\223", /* ▒ */ + [SPECIAL_GLYPH_SIGMA] = "\316\243", /* Σ */ /* Single glyph in Unicode, two in ASCII */ [SPECIAL_GLYPH_ARROW] = "\342\206\222", /* → (actually called: RIGHTWARDS ARROW) */ diff --git a/src/basic/locale-util.h b/src/basic/locale-util.h index 1df8ac4cb0..cefc4e7f0e 100644 --- a/src/basic/locale-util.h +++ b/src/basic/locale-util.h @@ -51,6 +51,9 @@ typedef enum { SPECIAL_GLYPH_CROSS_MARK, SPECIAL_GLYPH_ARROW, SPECIAL_GLYPH_ELLIPSIS, + SPECIAL_GLYPH_LIGHT_SHADE, + SPECIAL_GLYPH_DARK_SHADE, + SPECIAL_GLYPH_SIGMA, _SPECIAL_GLYPH_FIRST_SMILEY, SPECIAL_GLYPH_ECSTATIC_SMILEY = _SPECIAL_GLYPH_FIRST_SMILEY, SPECIAL_GLYPH_HAPPY_SMILEY, diff --git a/src/libsystemd/sd-id128/id128-util.c b/src/libsystemd/sd-id128/id128-util.c index 985872b82d..335f22b920 100644 --- a/src/libsystemd/sd-id128/id128-util.c +++ b/src/libsystemd/sd-id128/id128-util.c @@ -190,4 +190,17 @@ int id128_compare_func(const sd_id128_t *a, const sd_id128_t *b) { return memcmp(a, b, 16); } +sd_id128_t id128_make_v4_uuid(sd_id128_t id) { + /* Stolen from generate_random_uuid() of drivers/char/random.c + * in the kernel sources */ + + /* Set UUID version to 4 --- truly random generation */ + id.bytes[6] = (id.bytes[6] & 0x0F) | 0x40; + + /* Set the UUID variant to DCE */ + id.bytes[8] = (id.bytes[8] & 0x3F) | 0x80; + + return id; +} + DEFINE_HASH_OPS(id128_hash_ops, sd_id128_t, id128_hash_func, id128_compare_func); diff --git a/src/libsystemd/sd-id128/id128-util.h b/src/libsystemd/sd-id128/id128-util.h index fe0149a8aa..1901bf119f 100644 --- a/src/libsystemd/sd-id128/id128-util.h +++ b/src/libsystemd/sd-id128/id128-util.h @@ -30,3 +30,5 @@ int id128_write(const char *p, Id128Format f, sd_id128_t id, bool do_sync); void id128_hash_func(const sd_id128_t *p, struct siphash *state); int id128_compare_func(const sd_id128_t *a, const sd_id128_t *b) _pure_; extern const struct hash_ops id128_hash_ops; + +sd_id128_t id128_make_v4_uuid(sd_id128_t id); diff --git a/src/libsystemd/sd-id128/sd-id128.c b/src/libsystemd/sd-id128/sd-id128.c index b331a6b432..9b38ef0c56 100644 --- a/src/libsystemd/sd-id128/sd-id128.c +++ b/src/libsystemd/sd-id128/sd-id128.c @@ -250,19 +250,6 @@ _public_ int sd_id128_get_invocation(sd_id128_t *ret) { return 0; } -static sd_id128_t make_v4_uuid(sd_id128_t id) { - /* Stolen from generate_random_uuid() of drivers/char/random.c - * in the kernel sources */ - - /* Set UUID version to 4 --- truly random generation */ - id.bytes[6] = (id.bytes[6] & 0x0F) | 0x40; - - /* Set the UUID variant to DCE */ - id.bytes[8] = (id.bytes[8] & 0x3F) | 0x80; - - return id; -} - _public_ int sd_id128_randomize(sd_id128_t *ret) { sd_id128_t t; int r; @@ -279,7 +266,7 @@ _public_ int sd_id128_randomize(sd_id128_t *ret) { * only guarantee this for newly generated UUIDs, not for * pre-existing ones. */ - *ret = make_v4_uuid(t); + *ret = id128_make_v4_uuid(t); return 0; } @@ -306,7 +293,7 @@ static int get_app_specific(sd_id128_t base, sd_id128_t app_id, sd_id128_t *ret) /* We chop off the trailing 16 bytes */ memcpy(&result, p, MIN(khash_get_size(h), sizeof(result))); - *ret = make_v4_uuid(result); + *ret = id128_make_v4_uuid(result); return 0; } diff --git a/src/partition/makefs.c b/src/partition/makefs.c index ee4907f73f..951989cbb6 100644 --- a/src/partition/makefs.c +++ b/src/partition/makefs.c @@ -1,6 +1,4 @@ -/*** - SPDX-License-Identifier: LGPL-2.1+ -***/ +/* SPDX-License-Identifier: LGPL-2.1+ */ #include #include diff --git a/src/partition/meson.build b/src/partition/meson.build new file mode 100644 index 0000000000..d0c111a473 --- /dev/null +++ b/src/partition/meson.build @@ -0,0 +1,5 @@ +# SPDX-License-Identifier: LGPL-2.1+ + +systemd_repart_sources = files(''' + repart.c +'''.split()) diff --git a/src/partition/repart.c b/src/partition/repart.c new file mode 100644 index 0000000000..9844de5961 --- /dev/null +++ b/src/partition/repart.c @@ -0,0 +1,3096 @@ +/* SPDX-License-Identifier: LGPL-2.1+ */ + +#if HAVE_VALGRIND_MEMCHECK_H +#include +#endif + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "sd-id128.h" + +#include "alloc-util.h" +#include "blkid-util.h" +#include "blockdev-util.h" +#include "btrfs-util.h" +#include "conf-files.h" +#include "conf-parser.h" +#include "def.h" +#include "efivars.h" +#include "errno-util.h" +#include "fd-util.h" +#include "format-table.h" +#include "format-util.h" +#include "fs-util.h" +#include "gpt.h" +#include "id128-util.h" +#include "list.h" +#include "locale-util.h" +#include "main-func.h" +#include "parse-util.h" +#include "path-util.h" +#include "pretty-print.h" +#include "proc-cmdline.h" +#include "sort-util.h" +#include "stat-util.h" +#include "stdio-util.h" +#include "string-util.h" +#include "strv.h" +#include "terminal-util.h" +#include "utf8.h" + +/* Note: When growing and placing new partitions we always align to 4K sector size. It's how newer hard disks + * are designed, and if everything is aligned to that performance is best. And for older hard disks with 512B + * sector size devices were generally assumed to have an even number of sectors, hence at the worst we'll + * waste 3K per partition, which is probably fine. */ + +static enum { + EMPTY_REFUSE, /* refuse empty disks, never create a partition table */ + EMPTY_ALLOW, /* allow empty disks, create partition table if necessary */ + EMPTY_REQUIRE, /* require an empty disk, create a partition table */ + EMPTY_FORCE, /* make disk empty, erase everything, create a partition table always */ +} arg_empty = EMPTY_REFUSE; + +static bool arg_dry_run = true; +static const char *arg_node = NULL; +static char *arg_root = NULL; +static char *arg_definitions = NULL; +static bool arg_discard = true; +static bool arg_can_factory_reset = false; +static int arg_factory_reset = -1; +static sd_id128_t arg_seed = SD_ID128_NULL; +static bool arg_randomize = false; +static int arg_pretty = -1; + +STATIC_DESTRUCTOR_REGISTER(arg_root, freep); +STATIC_DESTRUCTOR_REGISTER(arg_definitions, freep); + +typedef struct Partition Partition; +typedef struct FreeArea FreeArea; +typedef struct Context Context; + +struct Partition { + char *definition_path; + + sd_id128_t type_uuid; + sd_id128_t current_uuid, new_uuid; + char *current_label, *new_label; + + bool dropped; + bool factory_reset; + int32_t priority; + + uint32_t weight, padding_weight; + + uint64_t current_size, new_size; + uint64_t size_min, size_max; + + uint64_t current_padding, new_padding; + uint64_t padding_min, padding_max; + + uint64_t partno; + uint64_t offset; + + struct fdisk_partition *current_partition; + struct fdisk_partition *new_partition; + FreeArea *padding_area; + FreeArea *allocated_to_area; + + LIST_FIELDS(Partition, partitions); +}; + +#define PARTITION_IS_FOREIGN(p) (!(p)->definition_path) +#define PARTITION_EXISTS(p) (!!(p)->current_partition) + +struct FreeArea { + Partition *after; + uint64_t size; + uint64_t allocated; +}; + +struct Context { + LIST_HEAD(Partition, partitions); + size_t n_partitions; + + FreeArea **free_areas; + size_t n_free_areas, n_allocated_free_areas; + + uint64_t start, end, total; + + struct fdisk_context *fdisk_context; + + sd_id128_t seed; +}; + +static uint64_t round_down_size(uint64_t v, uint64_t p) { + return (v / p) * p; +} + +static uint64_t round_up_size(uint64_t v, uint64_t p) { + + v = DIV_ROUND_UP(v, p); + + if (v > UINT64_MAX / p) + return UINT64_MAX; /* overflow */ + + return v * p; +} + +static Partition *partition_new(void) { + Partition *p; + + p = new(Partition, 1); + if (!p) + return NULL; + + *p = (Partition) { + .weight = 1000, + .padding_weight = 0, + .current_size = UINT64_MAX, + .new_size = UINT64_MAX, + .size_min = UINT64_MAX, + .size_max = UINT64_MAX, + .current_padding = UINT64_MAX, + .new_padding = UINT64_MAX, + .padding_min = UINT64_MAX, + .padding_max = UINT64_MAX, + .partno = UINT64_MAX, + .offset = UINT64_MAX, + }; + + return p; +} + +static Partition* partition_free(Partition *p) { + if (!p) + return NULL; + + free(p->current_label); + free(p->new_label); + free(p->definition_path); + + if (p->current_partition) + fdisk_unref_partition(p->current_partition); + if (p->new_partition) + fdisk_unref_partition(p->new_partition); + + return mfree(p); +} + +static Partition* partition_unlink_and_free(Context *context, Partition *p) { + if (!p) + return NULL; + + LIST_REMOVE(partitions, context->partitions, p); + + assert(context->n_partitions > 0); + context->n_partitions--; + + return partition_free(p); +} + +DEFINE_TRIVIAL_CLEANUP_FUNC(Partition*, partition_free); + +static Context *context_new(sd_id128_t seed) { + Context *context; + + context = new(Context, 1); + if (!context) + return NULL; + + *context = (Context) { + .start = UINT64_MAX, + .end = UINT64_MAX, + .total = UINT64_MAX, + .seed = seed, + }; + + return context; +} + +static void context_free_free_areas(Context *context) { + assert(context); + + for (size_t i = 0; i < context->n_free_areas; i++) + free(context->free_areas[i]); + + context->free_areas = mfree(context->free_areas); + context->n_free_areas = 0; + context->n_allocated_free_areas = 0; +} + +static Context *context_free(Context *context) { + if (!context) + return NULL; + + while (context->partitions) + partition_unlink_and_free(context, context->partitions); + assert(context->n_partitions == 0); + + context_free_free_areas(context); + + if (context->fdisk_context) + fdisk_unref_context(context->fdisk_context); + + return mfree(context); +} + +DEFINE_TRIVIAL_CLEANUP_FUNC(Context*, context_free); + +static int context_add_free_area( + Context *context, + uint64_t size, + Partition *after) { + + FreeArea *a; + + assert(context); + assert(!after || !after->padding_area); + + if (!GREEDY_REALLOC(context->free_areas, context->n_allocated_free_areas, context->n_free_areas + 1)) + return -ENOMEM; + + a = new(FreeArea, 1); + if (!a) + return -ENOMEM; + + *a = (FreeArea) { + .size = size, + .after = after, + }; + + context->free_areas[context->n_free_areas++] = a; + + if (after) + after->padding_area = a; + + return 0; +} + +static bool context_drop_one_priority(Context *context) { + int32_t priority = 0; + Partition *p; + bool exists = false; + + LIST_FOREACH(partitions, p, context->partitions) { + if (p->dropped) + continue; + if (p->priority < priority) + continue; + if (p->priority == priority) { + exists = exists || PARTITION_EXISTS(p); + continue; + } + + priority = p->priority; + exists = PARTITION_EXISTS(p); + } + + /* Refuse to drop partitions with 0 or negative priorities or partitions of priorities that have at + * least one existing priority */ + if (priority <= 0 || exists) + return false; + + LIST_FOREACH(partitions, p, context->partitions) { + if (p->priority < priority) + continue; + + if (p->dropped) + continue; + + p->dropped = true; + log_info("Can't fit partition %s of priority %" PRIi32 ", dropping.", p->definition_path, p->priority); + } + + return true; +} + +static uint64_t partition_min_size(const Partition *p) { + uint64_t sz; + + /* Calculate the disk space we really need at minimum for this partition. If the partition already + * exists the current size is what we really need. If it doesn't exist yet refuse to allocate less + * than 4K. */ + + if (PARTITION_IS_FOREIGN(p)) { + /* Don't allow changing size of partitions not managed by us */ + assert(p->current_size != UINT64_MAX); + return p->current_size; + } + + sz = p->current_size != UINT64_MAX ? p->current_size : 4096; + if (p->size_min != UINT64_MAX) + return MAX(p->size_min, sz); + + return sz; +} + +static uint64_t partition_max_size(const Partition *p) { + /* Calculate how large the partition may become at max. This is generally the configured maximum + * size, except when it already exists and is larger than that. In that case it's the existing size, + * since we never want to shrink partitions. */ + + if (PARTITION_IS_FOREIGN(p)) { + /* Don't allow changing size of partitions not managed by us */ + assert(p->current_size != UINT64_MAX); + return p->current_size; + } + + if (p->current_size != UINT64_MAX) + return MAX(p->current_size, p->size_max); + + return p->size_max; +} + +static uint64_t partition_min_size_with_padding(const Partition *p) { + uint64_t sz; + + /* Calculate the disk space we need for this partition plus any free space coming after it. This + * takes user configured padding into account as well as any additional whitespace needed to align + * the next partition to 4K again. */ + + sz = partition_min_size(p); + + if (p->padding_min != UINT64_MAX) + sz += p->padding_min; + + if (PARTITION_EXISTS(p)) { + /* If the partition wasn't aligned, add extra space so that any we might add will be aligned */ + assert(p->offset != UINT64_MAX); + return round_up_size(p->offset + sz, 4096) - p->offset; + } + + /* If this is a new partition we'll place it aligned, hence we just need to round up the required size here */ + return round_up_size(sz, 4096); +} + +static uint64_t free_area_available(const FreeArea *a) { + assert(a); + + /* Determines how much of this free area is not allocated yet */ + + assert(a->size >= a->allocated); + return a->size - a->allocated; +} + +static uint64_t free_area_available_for_new_partitions(const FreeArea *a) { + uint64_t avail; + + /* Similar to free_area_available(), but takes into account that the required size and padding of the + * preceeding partition is honoured. */ + + avail = free_area_available(a); + if (a->after) { + uint64_t need, space; + + need = partition_min_size_with_padding(a->after); + + assert(a->after->offset != UINT64_MAX); + assert(a->after->current_size != UINT64_MAX); + + space = round_up_size(a->after->offset + a->after->current_size, 4096) - a->after->offset + avail; + if (need >= space) + return 0; + + return space - need; + } + + return avail; +} + +static int free_area_compare(FreeArea *const *a, FreeArea *const*b) { + return CMP(free_area_available_for_new_partitions(*a), + free_area_available_for_new_partitions(*b)); +} + +static uint64_t charge_size(uint64_t total, uint64_t amount) { + uint64_t rounded; + + assert(amount <= total); + + /* Subtract the specified amount from total, rounding up to multiple of 4K if there's room */ + rounded = round_up_size(amount, 4096); + if (rounded >= total) + return 0; + + return total - rounded; +} + +static uint64_t charge_weight(uint64_t total, uint64_t amount) { + assert(amount <= total); + return total - amount; +} + +static bool context_allocate_partitions(Context *context) { + Partition *p; + + assert(context); + + /* A simple first-fit algorithm, assuming the array of free areas is sorted by size in decreasing + * order. */ + + LIST_FOREACH(partitions, p, context->partitions) { + bool fits = false; + uint64_t required; + FreeArea *a = NULL; + + /* Skip partitions we already dropped or that already exist */ + if (p->dropped || PARTITION_EXISTS(p)) + continue; + + /* Sort by size */ + typesafe_qsort(context->free_areas, context->n_free_areas, free_area_compare); + + /* How much do we need to fit? */ + required = partition_min_size_with_padding(p); + assert(required % 4096 == 0); + + for (size_t i = 0; i < context->n_free_areas; i++) { + a = context->free_areas[i]; + + if (free_area_available_for_new_partitions(a) >= required) { + fits = true; + break; + } + } + + if (!fits) + return false; /* 😢 Oh no! We can't fit this partition into any free area! */ + + /* Assign the partition to this free area */ + p->allocated_to_area = a; + + /* Budget the minimal partition size */ + a->allocated += required; + } + + return true; +} + +static int context_sum_weights(Context *context, FreeArea *a, uint64_t *ret) { + uint64_t weight_sum = 0; + Partition *p; + + assert(context); + assert(a); + assert(ret); + + /* Determine the sum of the weights of all partitions placed in or before the specified free area */ + + LIST_FOREACH(partitions, p, context->partitions) { + if (p->padding_area != a && p->allocated_to_area != a) + continue; + + if (p->weight > UINT64_MAX - weight_sum) + goto overflow_sum; + weight_sum += p->weight; + + if (p->padding_weight > UINT64_MAX - weight_sum) + goto overflow_sum; + weight_sum += p->padding_weight; + } + + *ret = weight_sum; + return 0; + +overflow_sum: + return log_error_errno(SYNTHETIC_ERRNO(EOVERFLOW), "Combined weight of partition exceeds unsigned 64bit range, refusing."); +} + +static int scale_by_weight(uint64_t value, uint64_t weight, uint64_t weight_sum, uint64_t *ret) { + assert(weight_sum >= weight); + assert(ret); + + if (weight == 0) { + *ret = 0; + return 0; + } + + if (value > UINT64_MAX / weight) + return log_error_errno(SYNTHETIC_ERRNO(EOVERFLOW), "Scaling by weight of partition exceeds unsigned 64bit range, refusing."); + + *ret = value * weight / weight_sum; + return 0; +} + +typedef enum GrowPartitionPhase { + /* The first phase: we charge partitions which need more (according to constraints) than their weight-based share. */ + PHASE_OVERCHARGE, + + /* The second phase: we charge partitions which need less (according to constraints) than their weight-based share. */ + PHASE_UNDERCHARGE, + + /* The third phase: we distribute what remains among the remaining partitions, according to the weights */ + PHASE_DISTRIBUTE, +} GrowPartitionPhase; + +static int context_grow_partitions_phase( + Context *context, + FreeArea *a, + GrowPartitionPhase phase, + uint64_t *span, + uint64_t *weight_sum) { + + Partition *p; + int r; + + assert(context); + assert(a); + + /* Now let's look at the intended weights and adjust them taking the minimum space assignments into + * account. i.e. if a partition has a small weight but a high minimum space value set it should not + * get any additional room from the left-overs. Similar, if two partitions have the same weight they + * should get the same space if possible, even if one has a smaller minimum size than the other. */ + LIST_FOREACH(partitions, p, context->partitions) { + + /* Look only at partitions associated with this free area, i.e. immediately + * preceeding it, or allocated into it */ + if (p->allocated_to_area != a && p->padding_area != a) + continue; + + if (p->new_size == UINT64_MAX) { + bool charge = false, try_again = false; + uint64_t share, rsz, xsz; + + /* Calculate how much this space this partition needs if everyone would get + * the weight based share */ + r = scale_by_weight(*span, p->weight, *weight_sum, &share); + if (r < 0) + return r; + + rsz = partition_min_size(p); + xsz = partition_max_size(p); + + if (phase == PHASE_OVERCHARGE && rsz > share) { + /* This partition needs more than its calculated share. Let's assign + * it that, and take this partition out of all calculations and start + * again. */ + + p->new_size = rsz; + charge = try_again = true; + + } else if (phase == PHASE_UNDERCHARGE && xsz != UINT64_MAX && xsz < share) { + /* This partition accepts less than its calculated + * share. Let's assign it that, and take this partition out + * of all calculations and start again. */ + + p->new_size = xsz; + charge = try_again = true; + + } else if (phase == PHASE_DISTRIBUTE) { + /* This partition can accept its calculated share. Let's + * assign it. There's no need to restart things here since + * assigning this shouldn't impact the shares of the other + * partitions. */ + + if (PARTITION_IS_FOREIGN(p)) + /* Never change of foreign partitions (i.e. those we don't manage) */ + p->new_size = p->current_size; + else + p->new_size = MAX(round_down_size(share, 4096), rsz); + + charge = true; + } + + if (charge) { + *span = charge_size(*span, p->new_size); + *weight_sum = charge_weight(*weight_sum, p->weight); + } + + if (try_again) + return 0; /* try again */ + } + + if (p->new_padding == UINT64_MAX) { + bool charge = false, try_again = false; + uint64_t share; + + r = scale_by_weight(*span, p->padding_weight, *weight_sum, &share); + if (r < 0) + return r; + + if (phase == PHASE_OVERCHARGE && p->padding_min != UINT64_MAX && p->padding_min > share) { + p->new_padding = p->padding_min; + charge = try_again = true; + } else if (phase == PHASE_UNDERCHARGE && p->padding_max != UINT64_MAX && p->padding_max < share) { + p->new_padding = p->padding_max; + charge = try_again = true; + } else if (phase == PHASE_DISTRIBUTE) { + + p->new_padding = round_down_size(share, 4096); + if (p->padding_min != UINT64_MAX && p->new_padding < p->padding_min) + p->new_padding = p->padding_min; + + charge = true; + } + + if (charge) { + *span = charge_size(*span, p->new_padding); + *weight_sum = charge_weight(*weight_sum, p->padding_weight); + } + + if (try_again) + return 0; /* try again */ + } + } + + return 1; /* done */ +} + +static int context_grow_partitions_on_free_area(Context *context, FreeArea *a) { + uint64_t weight_sum = 0, span; + int r; + + assert(context); + assert(a); + + r = context_sum_weights(context, a, &weight_sum); + if (r < 0) + return r; + + /* Let's calculate the total area covered by this free area and the partition before it */ + span = a->size; + if (a->after) { + assert(a->after->offset != UINT64_MAX); + assert(a->after->current_size != UINT64_MAX); + + span += round_up_size(a->after->offset + a->after->current_size, 4096) - a->after->offset; + } + + GrowPartitionPhase phase = PHASE_OVERCHARGE; + for (;;) { + r = context_grow_partitions_phase(context, a, phase, &span, &weight_sum); + if (r < 0) + return r; + if (r == 0) /* not done yet, re-run this phase */ + continue; + + if (phase == PHASE_OVERCHARGE) + phase = PHASE_UNDERCHARGE; + else if (phase == PHASE_UNDERCHARGE) + phase = PHASE_DISTRIBUTE; + else if (phase == PHASE_DISTRIBUTE) + break; + } + + /* We still have space left over? Donate to preceeding partition if we have one */ + if (span > 0 && a->after && !PARTITION_IS_FOREIGN(a->after)) { + uint64_t m, xsz; + + assert(a->after->new_size != UINT64_MAX); + m = a->after->new_size + span; + + xsz = partition_max_size(a->after); + if (xsz != UINT64_MAX && m > xsz) + m = xsz; + + span = charge_size(span, m - a->after->new_size); + a->after->new_size = m; + } + + /* What? Even still some space left (maybe because there was no preceeding partition, or it had a + * size limit), then let's donate it to whoever wants it. */ + if (span > 0) { + Partition *p; + + LIST_FOREACH(partitions, p, context->partitions) { + uint64_t m, xsz; + + if (p->allocated_to_area != a) + continue; + + if (PARTITION_IS_FOREIGN(p)) + continue; + + assert(p->new_size != UINT64_MAX); + m = p->new_size + span; + + xsz = partition_max_size(a->after); + if (xsz != UINT64_MAX && m > xsz) + m = xsz; + + span = charge_size(span, m - p->new_size); + p->new_size = m; + + if (span == 0) + break; + } + } + + /* Yuck, still noone? Then make it padding */ + if (span > 0 && a->after) { + assert(a->after->new_padding != UINT64_MAX); + a->after->new_padding += span; + } + + return 0; +} + +static int context_grow_partitions(Context *context) { + Partition *p; + int r; + + assert(context); + + for (size_t i = 0; i < context->n_free_areas; i++) { + r = context_grow_partitions_on_free_area(context, context->free_areas[i]); + if (r < 0) + return r; + } + + /* All existing partitions that have no free space after them can't change size */ + LIST_FOREACH(partitions, p, context->partitions) { + if (p->dropped) + continue; + + if (!PARTITION_EXISTS(p) || p->padding_area) { + /* The algorithm above must have initialized this already */ + assert(p->new_size != UINT64_MAX); + continue; + } + + assert(p->new_size == UINT64_MAX); + p->new_size = p->current_size; + + assert(p->new_padding == UINT64_MAX); + p->new_padding = p->current_padding; + } + + return 0; +} + +static void context_place_partitions(Context *context) { + uint64_t partno = 0; + Partition *p; + + assert(context); + + /* Determine next partition number to assign */ + LIST_FOREACH(partitions, p, context->partitions) { + if (!PARTITION_EXISTS(p)) + continue; + + assert(p->partno != UINT64_MAX); + if (p->partno >= partno) + partno = p->partno + 1; + } + + for (size_t i = 0; i < context->n_free_areas; i++) { + FreeArea *a = context->free_areas[i]; + uint64_t start, left; + + if (a->after) { + assert(a->after->offset != UINT64_MAX); + assert(a->after->new_size != UINT64_MAX); + assert(a->after->new_padding != UINT64_MAX); + + start = a->after->offset + a->after->new_size + a->after->new_padding; + } else + start = context->start; + + start = round_up_size(start, 4096); + left = a->size; + + LIST_FOREACH(partitions, p, context->partitions) { + if (p->allocated_to_area != a) + continue; + + p->offset = start; + p->partno = partno++; + + assert(left >= p->new_size); + start += p->new_size; + left -= p->new_size; + + assert(left >= p->new_padding); + start += p->new_padding; + left -= p->new_padding; + } + } +} + +typedef struct GptPartitionType { + sd_id128_t uuid; + const char *name; +} GptPartitionType; + +static const GptPartitionType gpt_partition_type_table[] = { + { GPT_ROOT_X86, "root-x86" }, + { GPT_ROOT_X86_VERITY, "root-x86-verity" }, + { GPT_ROOT_X86_64, "root-x86-64" }, + { GPT_ROOT_X86_64_VERITY, "root-x86-64-verity" }, + { GPT_ROOT_ARM, "root-arm" }, + { GPT_ROOT_ARM_VERITY, "root-arm-verity" }, + { GPT_ROOT_ARM_64, "root-arm64" }, + { GPT_ROOT_ARM_64_VERITY, "root-arm64-verity" }, + { GPT_ROOT_IA64, "root-ia64" }, + { GPT_ROOT_IA64_VERITY, "root-ia64-verity" }, +#ifdef GPT_ROOT_NATIVE + { GPT_ROOT_NATIVE, "root" }, + { GPT_ROOT_NATIVE_VERITY, "root-verity" }, +#endif +#ifdef GPT_ROOT_SECONDARY + { GPT_ROOT_SECONDARY, "root-secondary" }, + { GPT_ROOT_SECONDARY_VERITY, "root-secondary-verity" }, +#endif + { GPT_ESP, "esp" }, + { GPT_XBOOTLDR, "xbootldr" }, + { GPT_SWAP, "swap" }, + { GPT_HOME, "home" }, + { GPT_SRV, "srv" }, + { GPT_VAR, "var" }, + { GPT_TMP, "tmp" }, + { GPT_LINUX_GENERIC, "linux-generic", }, +}; + +static const char *gpt_partition_type_uuid_to_string(sd_id128_t id) { + for (size_t i = 0; i < ELEMENTSOF(gpt_partition_type_table); i++) + if (sd_id128_equal(id, gpt_partition_type_table[i].uuid)) + return gpt_partition_type_table[i].name; + + return NULL; +} + +static const char *gpt_partition_type_uuid_to_string_harder( + sd_id128_t id, + char buffer[static ID128_UUID_STRING_MAX]) { + + const char *s; + + assert(buffer); + + s = gpt_partition_type_uuid_to_string(id); + if (s) + return s; + + return id128_to_uuid_string(id, buffer); +} + +static int gpt_partition_type_uuid_from_string(const char *s, sd_id128_t *ret) { + assert(s); + assert(ret); + + for (size_t i = 0; i < ELEMENTSOF(gpt_partition_type_table); i++) + if (streq(s, gpt_partition_type_table[i].name)) { + *ret = gpt_partition_type_table[i].uuid; + return 0; + } + + return sd_id128_from_string(s, ret); +} + +static int config_parse_type( + const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + + sd_id128_t *type_uuid = data; + int r; + + assert(rvalue); + assert(type_uuid); + + r = gpt_partition_type_uuid_from_string(rvalue, type_uuid); + if (r < 0) + return log_syntax(unit, LOG_ERR, filename, line, r, "Failed to parse partition type: %s", rvalue); + + return 0; +} + +static int config_parse_label( + const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + + _cleanup_free_ char16_t *recoded = NULL; + char **label = data; + int r; + + assert(rvalue); + assert(label); + + if (!utf8_is_valid(rvalue)) { + log_syntax(unit, LOG_WARNING, filename, line, 0, + "Partition label not valid UTF-8, ignoring: %s", rvalue); + return 0; + } + + recoded = utf8_to_utf16(rvalue, strlen(rvalue)); + if (!recoded) + return log_oom(); + + if (char16_strlen(recoded) > 36) { + log_syntax(unit, LOG_WARNING, filename, line, 0, + "Partition label too long for GPT table, ignoring: %s", rvalue); + return 0; + } + + r = free_and_strdup(label, rvalue); + if (r < 0) + return log_oom(); + + return 0; +} + +static int config_parse_weight( + const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + + uint32_t *priority = data, v; + int r; + + assert(rvalue); + assert(priority); + + r = safe_atou32(rvalue, &v); + if (r < 0) { + log_syntax(unit, LOG_WARNING, filename, line, r, + "Failed to parse weight value, ignoring: %s", rvalue); + return 0; + } + + if (v > 1000U*1000U) { + log_syntax(unit, LOG_WARNING, filename, line, r, + "Weight needs to be in range 0…10000000, ignoring: %" PRIu32, v); + return 0; + } + + *priority = v; + return 0; +} + +static int config_parse_size4096( + const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + + uint64_t *sz = data, parsed; + int r; + + assert(rvalue); + assert(data); + + r = parse_size(rvalue, 1024, &parsed); + if (r < 0) + return log_syntax(unit, LOG_WARNING, filename, line, r, + "Failed to parse size value: %s", rvalue); + + if (ltype > 0) + *sz = round_up_size(parsed, 4096); + else if (ltype < 0) + *sz = round_down_size(parsed, 4096); + else + *sz = parsed; + + if (*sz != parsed) + log_syntax(unit, LOG_NOTICE, filename, line, r, "Rounded %s= size %" PRIu64 " → %" PRIu64 ", a multiple of 4096.", lvalue, parsed, *sz); + + return 0; +} + +static int partition_read_definition(Partition *p, const char *path) { + + ConfigTableItem table[] = { + { "Partition", "Type", config_parse_type, 0, &p->type_uuid }, + { "Partition", "Label", config_parse_label, 0, &p->new_label }, + { "Partition", "Priority", config_parse_int32, 0, &p->priority }, + { "Partition", "Weight", config_parse_weight, 0, &p->weight }, + { "Partition", "PaddingWeight", config_parse_weight, 0, &p->padding_weight }, + { "Partition", "SizeMinBytes", config_parse_size4096, 1, &p->size_min }, + { "Partition", "SizeMaxBytes", config_parse_size4096, -1, &p->size_max }, + { "Partition", "PaddingMinBytes", config_parse_size4096, 1, &p->padding_min }, + { "Partition", "PaddingMaxBytes", config_parse_size4096, -1, &p->padding_max }, + { "Partition", "FactoryReset", config_parse_bool, 0, &p->factory_reset }, + {} + }; + int r; + + r = config_parse(NULL, path, NULL, "Partition\0", config_item_table_lookup, table, CONFIG_PARSE_WARN, p); + if (r < 0) + return r; + + if (p->size_min != UINT64_MAX && p->size_max != UINT64_MAX && p->size_min > p->size_max) + return log_syntax(NULL, LOG_ERR, path, 1, SYNTHETIC_ERRNO(EINVAL), + "SizeMinBytes= larger than SizeMaxBytes=, refusing."); + + if (p->padding_min != UINT64_MAX && p->padding_max != UINT64_MAX && p->padding_min > p->padding_max) + return log_syntax(NULL, LOG_ERR, path, 1, SYNTHETIC_ERRNO(EINVAL), + "PaddingMinBytes= larger than PaddingMaxBytes=, refusing."); + + if (sd_id128_is_null(p->type_uuid)) + return log_syntax(NULL, LOG_ERR, path, 1, SYNTHETIC_ERRNO(EINVAL), + "Type= not defined, refusing."); + + return 0; +} + +static int context_read_definitions( + Context *context, + const char *directory, + const char *root) { + + _cleanup_strv_free_ char **files = NULL; + Partition *last = NULL; + char **f; + int r; + + assert(context); + + if (directory) + r = conf_files_list_strv(&files, ".conf", NULL, CONF_FILES_REGULAR|CONF_FILES_FILTER_MASKED, (const char**) STRV_MAKE(directory)); + else + r = conf_files_list_strv(&files, ".conf", root, CONF_FILES_REGULAR|CONF_FILES_FILTER_MASKED, (const char**) CONF_PATHS_STRV("repart.d")); + if (r < 0) + return log_error_errno(r, "Failed to enumerate *.conf files: %m"); + + STRV_FOREACH(f, files) { + _cleanup_(partition_freep) Partition *p = NULL; + + p = partition_new(); + if (!p) + return log_oom(); + + p->definition_path = strdup(*f); + if (!p->definition_path) + return log_oom(); + + r = partition_read_definition(p, *f); + if (r < 0) + return r; + + LIST_INSERT_AFTER(partitions, context->partitions, last, p); + last = TAKE_PTR(p); + context->n_partitions++; + } + + return 0; +} + +DEFINE_TRIVIAL_CLEANUP_FUNC(struct fdisk_context*, fdisk_unref_context); +DEFINE_TRIVIAL_CLEANUP_FUNC(struct fdisk_partition*, fdisk_unref_partition); +DEFINE_TRIVIAL_CLEANUP_FUNC(struct fdisk_parttype*, fdisk_unref_parttype); +DEFINE_TRIVIAL_CLEANUP_FUNC(struct fdisk_table*, fdisk_unref_table); + +static int determine_current_padding( + struct fdisk_context *c, + struct fdisk_table *t, + struct fdisk_partition *p, + uint64_t *ret) { + + size_t n_partitions; + uint64_t offset, next = UINT64_MAX; + + assert(c); + assert(t); + assert(p); + + if (!fdisk_partition_has_end(p)) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Partition has no end!"); + + offset = fdisk_partition_get_end(p); + assert(offset < UINT64_MAX / 512); + offset *= 512; + + n_partitions = fdisk_table_get_nents(t); + for (size_t i = 0; i < n_partitions; i++) { + struct fdisk_partition *q; + uint64_t start; + + q = fdisk_table_get_partition(t, i); + if (!q) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Failed to read partition metadata: %m"); + + if (fdisk_partition_is_used(q) <= 0) + continue; + + if (!fdisk_partition_has_start(q)) + continue; + + start = fdisk_partition_get_start(q); + assert(start < UINT64_MAX / 512); + start *= 512; + + if (start >= offset && (next == UINT64_MAX || next > start)) + next = start; + } + + if (next == UINT64_MAX) { + /* No later partition? In that case check the end of the usable area */ + next = fdisk_get_last_lba(c); + assert(next < UINT64_MAX); + next++; /* The last LBA is one sector before the end */ + + assert(next < UINT64_MAX / 512); + next *= 512; + + if (offset > next) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Partition end beyond disk end."); + } + + assert(next >= offset); + offset = round_up_size(offset, 4096); + next = round_down_size(next, 4096); + + if (next >= offset) /* Check again, rounding might have fucked things up */ + *ret = next - offset; + else + *ret = 0; + + return 0; +} + +static int fdisk_ask_cb(struct fdisk_context *c, struct fdisk_ask *ask, void *data) { + _cleanup_free_ char *ids = NULL; + int r; + + if (fdisk_ask_get_type(ask) != FDISK_ASKTYPE_STRING) + return -EINVAL; + + ids = new(char, ID128_UUID_STRING_MAX); + if (!ids) + return -ENOMEM; + + r = fdisk_ask_string_set_result(ask, id128_to_uuid_string(*(sd_id128_t*) data, ids)); + if (r < 0) + return r; + + TAKE_PTR(ids); + return 0; +} + +static int fdisk_set_disklabel_id_by_uuid(struct fdisk_context *c, sd_id128_t id) { + int r; + + r = fdisk_set_ask(c, fdisk_ask_cb, &id); + if (r < 0) + return r; + + r = fdisk_set_disklabel_id(c); + if (r < 0) + return r; + + return fdisk_set_ask(c, NULL, NULL); +} + +#define DISK_UUID_TOKEN "disk-uuid" + +static int disk_acquire_uuid(Context *context, sd_id128_t *ret) { + union { + unsigned char md[SHA256_DIGEST_LENGTH]; + sd_id128_t id; + } result; + + assert(context); + assert(ret); + + /* Calculate the HMAC-SHA256 of the string "disk-uuid", keyed off the machine ID. We use the machine + * ID as key (and not as cleartext!) since it's the machine ID we don't want to leak. */ + + if (!HMAC(EVP_sha256(), + &context->seed, sizeof(context->seed), + (const unsigned char*) DISK_UUID_TOKEN, strlen(DISK_UUID_TOKEN), + result.md, NULL)) + return log_error_errno(SYNTHETIC_ERRNO(ENOTRECOVERABLE), "HMAC-SHA256 calculation failed."); + + /* Take the first half, mark it as v4 UUID */ + assert_cc(sizeof(result.md) == sizeof(result.id) * 2); + *ret = id128_make_v4_uuid(result.id); + return 0; +} + +static int context_load_partition_table(Context *context, const char *node) { + _cleanup_(fdisk_unref_contextp) struct fdisk_context *c = NULL; + _cleanup_(fdisk_unref_tablep) struct fdisk_table *t = NULL; + uint64_t left_boundary = UINT64_MAX, first_lba, last_lba, nsectors; + _cleanup_free_ char *disk_uuid_string = NULL; + bool from_scratch = false; + sd_id128_t disk_uuid; + size_t n_partitions; + int r; + + assert(context); + assert(node); + + c = fdisk_new_context(); + if (!c) + return log_oom(); + + r = fdisk_assign_device(c, node, arg_dry_run); + if (r < 0) + return log_error_errno(r, "Failed to open device: %m"); + + /* Tell udev not to interfere while we are processing the device */ + if (flock(fdisk_get_devfd(c), arg_dry_run ? LOCK_SH : LOCK_EX) < 0) + return log_error_errno(errno, "Failed to lock block device: %m"); + + switch (arg_empty) { + + case EMPTY_REFUSE: + /* Refuse empty disks, insist on an existing GPT partition table */ + if (!fdisk_is_labeltype(c, FDISK_DISKLABEL_GPT)) + return log_notice_errno(SYNTHETIC_ERRNO(EHWPOISON), "Disk %s has no GPT disk label, not repartitioning.", node); + + break; + + case EMPTY_REQUIRE: + /* Require an empty disk, refuse any existing partition table */ + r = fdisk_has_label(c); + if (r < 0) + return log_error_errno(r, "Failed to determine whether disk %s has a disk label: %m", node); + if (r > 0) + return log_notice_errno(SYNTHETIC_ERRNO(EHWPOISON), "Disk %s already has a disk label, refusing.", node); + + from_scratch = true; + break; + + case EMPTY_ALLOW: + /* Allow both an empty disk and an existing partition table, but only GPT */ + r = fdisk_has_label(c); + if (r < 0) + return log_error_errno(r, "Failed to determine whether disk %s has a disk label: %m", node); + if (r > 0) { + if (!fdisk_is_labeltype(c, FDISK_DISKLABEL_GPT)) + return log_notice_errno(SYNTHETIC_ERRNO(EHWPOISON), "Disk %s has non-GPT disk label, not repartitioning.", node); + } else + from_scratch = true; + + break; + + case EMPTY_FORCE: + /* Always reinitiaize the disk, don't consider what there was on the disk before */ + from_scratch = true; + break; + } + + if (from_scratch) { + r = fdisk_enable_wipe(c, true); + if (r < 0) + return log_error_errno(r, "Failed to enable wiping of disk signature: %m"); + + r = fdisk_create_disklabel(c, "gpt"); + if (r < 0) + return log_error_errno(r, "Failed to create GPT disk label: %m"); + + r = disk_acquire_uuid(context, &disk_uuid); + if (r < 0) + return log_error_errno(r, "Failed to acquire disk GPT uuid: %m"); + + r = fdisk_set_disklabel_id_by_uuid(c, disk_uuid); + if (r < 0) + return log_error_errno(r, "Failed to set GPT disk label: %m"); + + goto add_initial_free_area; + } + + r = fdisk_get_disklabel_id(c, &disk_uuid_string); + if (r < 0) + return log_error_errno(r, "Failed to get current GPT disk label UUID: %m"); + + r = sd_id128_from_string(disk_uuid_string, &disk_uuid); + if (r < 0) + return log_error_errno(r, "Failed to parse current GPT disk label UUID: %m"); + + if (sd_id128_is_null(disk_uuid)) { + r = disk_acquire_uuid(context, &disk_uuid); + if (r < 0) + return log_error_errno(r, "Failed to acquire disk GPT uuid: %m"); + + r = fdisk_set_disklabel_id(c); + if (r < 0) + return log_error_errno(r, "Failed to set GPT disk label: %m"); + } + + r = fdisk_get_partitions(c, &t); + if (r < 0) + return log_error_errno(r, "Failed to acquire partition table: %m"); + + n_partitions = fdisk_table_get_nents(t); + for (size_t i = 0; i < n_partitions; i++) { + _cleanup_free_ char *label_copy = NULL; + Partition *pp, *last = NULL; + struct fdisk_partition *p; + struct fdisk_parttype *pt; + const char *pts, *ids, *label; + uint64_t sz, start; + bool found = false; + sd_id128_t ptid, id; + size_t partno; + + p = fdisk_table_get_partition(t, i); + if (!p) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Failed to read partition metadata: %m"); + + if (fdisk_partition_is_used(p) <= 0) + continue; + + if (fdisk_partition_has_start(p) <= 0 || + fdisk_partition_has_size(p) <= 0 || + fdisk_partition_has_partno(p) <= 0) + return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Found a partition without a position, size or number."); + + pt = fdisk_partition_get_type(p); + if (!pt) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Failed to acquire type of partition: %m"); + + pts = fdisk_parttype_get_string(pt); + if (!pts) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Failed to acquire type of partition as string: %m"); + + r = sd_id128_from_string(pts, &ptid); + if (r < 0) + return log_error_errno(r, "Failed to parse partition type UUID %s: %m", pts); + + ids = fdisk_partition_get_uuid(p); + if (!ids) + return log_error_errno(SYNTHETIC_ERRNO(EINVAL), "Found a partition without a UUID."); + + r = sd_id128_from_string(ids, &id); + if (r < 0) + return log_error_errno(r, "Failed to parse partition UUID %s: %m", ids); + + label = fdisk_partition_get_name(p); + if (!isempty(label)) { + label_copy = strdup(label); + if (!label_copy) + return log_oom(); + } + + sz = fdisk_partition_get_size(p); + assert_se(sz <= UINT64_MAX/512); + sz *= 512; + + start = fdisk_partition_get_start(p); + assert_se(start <= UINT64_MAX/512); + start *= 512; + + partno = fdisk_partition_get_partno(p); + + if (left_boundary == UINT64_MAX || left_boundary > start) + left_boundary = start; + + /* Assign this existing partition to the first partition of the right type that doesn't have + * an existing one assigned yet. */ + LIST_FOREACH(partitions, pp, context->partitions) { + last = pp; + + if (!sd_id128_equal(pp->type_uuid, ptid)) + continue; + + if (!pp->current_partition) { + pp->current_uuid = id; + pp->current_size = sz; + pp->offset = start; + pp->partno = partno; + pp->current_label = TAKE_PTR(label_copy); + + pp->current_partition = p; + fdisk_ref_partition(p); + + r = determine_current_padding(c, t, p, &pp->current_padding); + if (r < 0) + return r; + + if (pp->current_padding > 0) { + r = context_add_free_area(context, pp->current_padding, pp); + if (r < 0) + return r; + } + + found = true; + break; + } + } + + /* If we have no matching definition, create a new one. */ + if (!found) { + _cleanup_(partition_freep) Partition *np = NULL; + + np = partition_new(); + if (!np) + return log_oom(); + + np->current_uuid = id; + np->type_uuid = ptid; + np->current_size = sz; + np->offset = start; + np->partno = partno; + np->current_label = TAKE_PTR(label_copy); + + np->current_partition = p; + fdisk_ref_partition(p); + + r = determine_current_padding(c, t, p, &np->current_padding); + if (r < 0) + return r; + + if (np->current_padding > 0) { + r = context_add_free_area(context, np->current_padding, np); + if (r < 0) + return r; + } + + LIST_INSERT_AFTER(partitions, context->partitions, last, TAKE_PTR(np)); + context->n_partitions++; + } + } + +add_initial_free_area: + nsectors = fdisk_get_nsectors(c); + assert(nsectors <= UINT64_MAX/512); + nsectors *= 512; + + first_lba = fdisk_get_first_lba(c); + assert(first_lba <= UINT64_MAX/512); + first_lba *= 512; + + last_lba = fdisk_get_last_lba(c); + assert(last_lba < UINT64_MAX); + last_lba++; + assert(last_lba <= UINT64_MAX/512); + last_lba *= 512; + + assert(last_lba >= first_lba); + + if (left_boundary == UINT64_MAX) { + /* No partitions at all? Then the whole disk is up for grabs. */ + + first_lba = round_up_size(first_lba, 4096); + last_lba = round_down_size(last_lba, 4096); + + if (last_lba > first_lba) { + r = context_add_free_area(context, last_lba - first_lba, NULL); + if (r < 0) + return r; + } + } else { + /* Add space left of first partition */ + assert(left_boundary >= first_lba); + + first_lba = round_up_size(first_lba, 4096); + left_boundary = round_down_size(left_boundary, 4096); + last_lba = round_down_size(last_lba, 4096); + + if (left_boundary > first_lba) { + r = context_add_free_area(context, left_boundary - first_lba, NULL); + if (r < 0) + return r; + } + } + + context->start = first_lba; + context->end = last_lba; + context->total = nsectors; + context->fdisk_context = TAKE_PTR(c); + + return from_scratch; +} + +static void context_unload_partition_table(Context *context) { + Partition *p, *next; + + assert(context); + + LIST_FOREACH_SAFE(partitions, p, next, context->partitions) { + + /* Entirely remove partitions that have no configuration */ + if (PARTITION_IS_FOREIGN(p)) { + partition_unlink_and_free(context, p); + continue; + } + + /* Otherwise drop all data we read off the block device and everything we might have + * calculated based on it */ + + p->dropped = false; + p->current_size = UINT64_MAX; + p->new_size = UINT64_MAX; + p->current_padding = UINT64_MAX; + p->new_padding = UINT64_MAX; + p->partno = UINT64_MAX; + p->offset = UINT64_MAX; + + if (p->current_partition) { + fdisk_unref_partition(p->current_partition); + p->current_partition = NULL; + } + + if (p->new_partition) { + fdisk_unref_partition(p->new_partition); + p->new_partition = NULL; + } + + p->padding_area = NULL; + p->allocated_to_area = NULL; + + p->current_uuid = p->new_uuid = SD_ID128_NULL; + } + + context->start = UINT64_MAX; + context->end = UINT64_MAX; + context->total = UINT64_MAX; + + if (context->fdisk_context) { + fdisk_unref_context(context->fdisk_context); + context->fdisk_context = NULL; + } + + context_free_free_areas(context); +} + +static int format_size_change(uint64_t from, uint64_t to, char **ret) { + char format_buffer1[FORMAT_BYTES_MAX], format_buffer2[FORMAT_BYTES_MAX], *buf; + + if (from != UINT64_MAX) + format_bytes(format_buffer1, sizeof(format_buffer1), from); + if (to != UINT64_MAX) + format_bytes(format_buffer2, sizeof(format_buffer2), to); + + if (from != UINT64_MAX) { + if (from == to || to == UINT64_MAX) + buf = strdup(format_buffer1); + else + buf = strjoin(format_buffer1, " ", special_glyph(SPECIAL_GLYPH_ARROW), " ", format_buffer2); + } else if (to != UINT64_MAX) + buf = strjoin(special_glyph(SPECIAL_GLYPH_ARROW), " ", format_buffer2); + else { + *ret = NULL; + return 0; + } + + if (!buf) + return log_oom(); + + *ret = TAKE_PTR(buf); + return 1; +} + +static const char *partition_label(const Partition *p) { + assert(p); + + if (p->new_label) + return p->new_label; + + if (p->current_label) + return p->current_label; + + return gpt_partition_type_uuid_to_string(p->type_uuid); +} + +static int context_dump_partitions(Context *context, const char *node) { + _cleanup_(table_unrefp) Table *t = NULL; + uint64_t sum_padding = 0, sum_size = 0; + Partition *p; + int r; + + t = table_new("type", "label", "uuid", "file", "node", "offset", "raw size", "size", "raw padding", "padding"); + if (!t) + return log_oom(); + + if (!DEBUG_LOGGING) + (void) table_set_display(t, 0, 1, 2, 3, 4, 7, 9, (size_t) -1); + + (void) table_set_align_percent(t, table_get_cell(t, 0, 4), 100); + (void) table_set_align_percent(t, table_get_cell(t, 0, 5), 100); + + LIST_FOREACH(partitions, p, context->partitions) { + _cleanup_free_ char *size_change = NULL, *padding_change = NULL, *partname = NULL; + char uuid_buffer[ID128_UUID_STRING_MAX]; + const char *label; + + if (p->dropped) + continue; + + label = partition_label(p); + partname = p->partno != UINT64_MAX ? fdisk_partname(node, p->partno+1) : NULL; + + r = format_size_change(p->current_size, p->new_size, &size_change); + if (r < 0) + return r; + + r = format_size_change(p->current_padding, p->new_padding, &padding_change); + if (r < 0) + return r; + + if (p->new_size != UINT64_MAX) + sum_size += p->new_size; + if (p->new_padding != UINT64_MAX) + sum_padding += p->new_padding; + + r = table_add_many( + t, + TABLE_STRING, gpt_partition_type_uuid_to_string_harder(p->type_uuid, uuid_buffer), + TABLE_STRING, label ?: "-", TABLE_SET_COLOR, label ? NULL : ansi_grey(), + TABLE_UUID, sd_id128_is_null(p->new_uuid) ? p->current_uuid : p->new_uuid, + TABLE_STRING, p->definition_path ? basename(p->definition_path) : "-", TABLE_SET_COLOR, p->definition_path ? NULL : ansi_grey(), + TABLE_STRING, partname ?: "no", TABLE_SET_COLOR, partname ? NULL : ansi_highlight(), + TABLE_UINT64, p->offset, + TABLE_UINT64, p->new_size, + TABLE_STRING, size_change, TABLE_SET_COLOR, !p->partitions_next && sum_size > 0 ? ansi_underline() : NULL, + TABLE_UINT64, p->new_padding, + TABLE_STRING, padding_change, TABLE_SET_COLOR, !p->partitions_next && sum_padding > 0 ? ansi_underline() : NULL); + if (r < 0) + return log_error_errno(r, "Failed to add row to table: %m"); + } + + if (sum_padding > 0 || sum_size > 0) { + char s[FORMAT_BYTES_MAX]; + const char *a, *b; + + a = strjoina(special_glyph(SPECIAL_GLYPH_SIGMA), " = ", format_bytes(s, sizeof(s), sum_size)); + b = strjoina(special_glyph(SPECIAL_GLYPH_SIGMA), " = ", format_bytes(s, sizeof(s), sum_padding)); + + r = table_add_many( + t, + TABLE_EMPTY, + TABLE_EMPTY, + TABLE_EMPTY, + TABLE_EMPTY, + TABLE_EMPTY, + TABLE_EMPTY, + TABLE_EMPTY, + TABLE_STRING, a, + TABLE_EMPTY, + TABLE_STRING, b); + if (r < 0) + return log_error_errno(r, "Failed to add row to table: %m"); + } + + r = table_print(t, stdout); + if (r < 0) + return log_error_errno(r, "Failed to dump table: %m"); + + return 0; +} + +static void context_bar_char_process_partition( + Context *context, + Partition *bar[], + size_t n, + Partition *p, + size_t *ret_start) { + + uint64_t from, to, total; + size_t x, y; + + assert(context); + assert(bar); + assert(n > 0); + assert(p); + + if (p->dropped) + return; + + assert(p->offset != UINT64_MAX); + assert(p->new_size != UINT64_MAX); + + from = p->offset; + to = from + p->new_size; + + assert(context->end >= context->start); + total = context->end - context->start; + + assert(from >= context->start); + assert(from <= context->end); + x = (from - context->start) * n / total; + + assert(to >= context->start); + assert(to <= context->end); + y = (to - context->start) * n / total; + + assert(x <= y); + assert(y <= n); + + for (size_t i = x; i < y; i++) + bar[i] = p; + + *ret_start = x; +} + +static int partition_hint(const Partition *p, const char *node, char **ret) { + _cleanup_free_ char *buf = NULL; + char ids[ID128_UUID_STRING_MAX]; + const char *label; + sd_id128_t id; + + /* Tries really hard to find a suitable description for this partition */ + + if (p->definition_path) { + buf = strdup(basename(p->definition_path)); + goto done; + } + + label = partition_label(p); + if (!isempty(label)) { + buf = strdup(label); + goto done; + } + + if (p->partno != UINT64_MAX) { + buf = fdisk_partname(node, p->partno+1); + goto done; + } + + if (!sd_id128_is_null(p->new_uuid)) + id = p->new_uuid; + else if (!sd_id128_is_null(p->current_uuid)) + id = p->current_uuid; + else + id = p->type_uuid; + + buf = strdup(id128_to_uuid_string(id, ids)); + +done: + if (!buf) + return -ENOMEM; + + *ret = TAKE_PTR(buf); + return 0; +} + +static int context_dump_partition_bar(Context *context, const char *node) { + _cleanup_free_ Partition **bar = NULL; + _cleanup_free_ size_t *start_array = NULL; + Partition *p, *last = NULL; + bool z = false; + size_t c, j = 0; + + assert((c = columns()) >= 2); + c -= 2; /* We do not use the leftmost and rightmost character cell */ + + bar = new0(Partition*, c); + if (!bar) + return log_oom(); + + start_array = new(size_t, context->n_partitions); + if (!start_array) + return log_oom(); + + LIST_FOREACH(partitions, p, context->partitions) + context_bar_char_process_partition(context, bar, c, p, start_array + j++); + + putc(' ', stdout); + + for (size_t i = 0; i < c; i++) { + if (bar[i]) { + if (last != bar[i]) + z = !z; + + fputs(z ? ansi_green() : ansi_yellow(), stdout); + fputs(special_glyph(SPECIAL_GLYPH_DARK_SHADE), stdout); + } else { + fputs(ansi_normal(), stdout); + fputs(special_glyph(SPECIAL_GLYPH_LIGHT_SHADE), stdout); + } + + last = bar[i]; + } + + fputs(ansi_normal(), stdout); + putc('\n', stdout); + + for (size_t i = 0; i < context->n_partitions; i++) { + _cleanup_free_ char **line = NULL; + + line = new0(char*, c); + if (!line) + return log_oom(); + + j = 0; + LIST_FOREACH(partitions, p, context->partitions) { + _cleanup_free_ char *d = NULL; + j++; + + if (i < context->n_partitions - j) { + + if (line[start_array[j-1]]) { + const char *e; + + /* Upgrade final corner to the right with a branch to the right */ + e = startswith(line[start_array[j-1]], special_glyph(SPECIAL_GLYPH_TREE_RIGHT)); + if (e) { + d = strjoin(special_glyph(SPECIAL_GLYPH_TREE_BRANCH), e); + if (!d) + return log_oom(); + } + } + + if (!d) { + d = strdup(special_glyph(SPECIAL_GLYPH_TREE_VERTICAL)); + if (!d) + return log_oom(); + } + + } else if (i == context->n_partitions - j) { + _cleanup_free_ char *hint = NULL; + + (void) partition_hint(p, node, &hint); + + if (streq_ptr(line[start_array[j-1]], special_glyph(SPECIAL_GLYPH_TREE_VERTICAL))) + d = strjoin(special_glyph(SPECIAL_GLYPH_TREE_BRANCH), " ", strna(hint)); + else + d = strjoin(special_glyph(SPECIAL_GLYPH_TREE_RIGHT), " ", strna(hint)); + + if (!d) + return log_oom(); + } + + if (d) + free_and_replace(line[start_array[j-1]], d); + } + + putc(' ', stdout); + + j = 0; + while (j < c) { + if (line[j]) { + fputs(line[j], stdout); + j += utf8_console_width(line[j]); + } else { + putc(' ', stdout); + j++; + } + } + + putc('\n', stdout); + + for (j = 0; j < c; j++) + free(line[j]); + } + + return 0; +} + +static bool context_changed(const Context *context) { + Partition *p; + + LIST_FOREACH(partitions, p, context->partitions) { + if (p->dropped) + continue; + + if (p->allocated_to_area) + return true; + + if (p->new_size != p->current_size) + return true; + } + + return false; +} + +static int context_wipe_partition(Context *context, Partition *p) { + _cleanup_(blkid_free_probep) blkid_probe probe = NULL; + int r; + + assert(context); + assert(p); + assert(!PARTITION_EXISTS(p)); /* Safety check: never wipe existing partitions */ + + probe = blkid_new_probe(); + if (!probe) + return log_oom(); + + assert(p->offset != UINT64_MAX); + assert(p->new_size != UINT64_MAX); + + errno = 0; + r = blkid_probe_set_device(probe, fdisk_get_devfd(context->fdisk_context), p->offset, p->new_size); + if (r < 0) + return log_error_errno(errno ?: SYNTHETIC_ERRNO(EIO), "Failed to allocate device probe for partition %" PRIu64 ".", p->partno); + + errno = 0; + if (blkid_probe_enable_superblocks(probe, true) < 0 || + blkid_probe_set_superblocks_flags(probe, BLKID_SUBLKS_MAGIC|BLKID_SUBLKS_BADCSUM) < 0 || + blkid_probe_enable_partitions(probe, true) < 0 || + blkid_probe_set_partitions_flags(probe, BLKID_PARTS_MAGIC) < 0) + return log_error_errno(errno ?: SYNTHETIC_ERRNO(EIO), "Failed to enable superblock and partition probing for partition %" PRIu64 ".", p->partno); + + for (;;) { + errno = 0; + r = blkid_do_probe(probe); + if (r < 0) + return log_error_errno(errno ?: SYNTHETIC_ERRNO(EIO), "Failed to probe for file systems."); + if (r > 0) + break; + + errno = 0; + if (blkid_do_wipe(probe, false) < 0) + return log_error_errno(errno ?: SYNTHETIC_ERRNO(EIO), "Failed to wipe file system signature."); + } + + log_info("Successfully wiped file system signatures from partition %" PRIu64 ".", p->partno); + return 0; +} + +static int context_discard_range(Context *context, uint64_t offset, uint64_t size) { + struct stat st; + int fd; + + assert(context); + assert(offset != UINT64_MAX); + assert(size != UINT64_MAX); + + if (size <= 0) + return 0; + + fd = fdisk_get_devfd(context->fdisk_context); + assert(fd >= 0); + + if (fstat(fd, &st) < 0) + return -errno; + + if (S_ISREG(st.st_mode)) { + if (fallocate(fd, FALLOC_FL_PUNCH_HOLE|FALLOC_FL_KEEP_SIZE, offset, size) < 0) { + if (ERRNO_IS_NOT_SUPPORTED(errno)) + return -EOPNOTSUPP; + + return -errno; + } + + return 1; + } + + if (S_ISBLK(st.st_mode)) { + uint64_t range[2], end; + + range[0] = round_up_size(offset, 512); + + end = offset + size; + if (end <= range[0]) + return 0; + + range[1] = round_down_size(end - range[0], 512); + if (range[1] <= 0) + return 0; + + if (ioctl(fd, BLKDISCARD, range) < 0) { + if (ERRNO_IS_NOT_SUPPORTED(errno)) + return -EOPNOTSUPP; + + return -errno; + } + + return 1; + } + + return -EOPNOTSUPP; +} + +static int context_discard_partition(Context *context, Partition *p) { + int r; + + assert(context); + assert(p); + + assert(p->offset != UINT64_MAX); + assert(p->new_size != UINT64_MAX); + assert(!PARTITION_EXISTS(p)); /* Safety check: never discard existing partitions */ + + if (!arg_discard) + return 0; + + r = context_discard_range(context, p->offset, p->new_size); + if (r == -EOPNOTSUPP) { + log_info("Storage does not support discarding, not discarding data in new partition %" PRIu64 ".", p->partno); + return 0; + } + if (r == 0) { + log_info("Partition %" PRIu64 " too short for discard, skipping.", p->partno); + return 0; + } + if (r < 0) + return log_error_errno(r, "Failed to discard data for new partition %" PRIu64 ".", p->partno); + + log_info("Successfully discarded data from partition %" PRIu64 ".", p->partno); + return 1; +} + +static int context_discard_gap_after(Context *context, Partition *p) { + uint64_t gap, next = UINT64_MAX; + Partition *q; + int r; + + assert(context); + assert(!p || (p->offset != UINT64_MAX && p->new_size != UINT64_MAX)); + + if (p) + gap = p->offset + p->new_size; + else + gap = context->start; + + LIST_FOREACH(partitions, q, context->partitions) { + if (q->dropped) + continue; + + assert(q->offset != UINT64_MAX); + assert(q->new_size != UINT64_MAX); + + if (q->offset < gap) + continue; + + if (next == UINT64_MAX || q->offset < next) + next = q->offset; + } + + if (next == UINT64_MAX) { + next = context->end; + if (gap > next) + return log_error_errno(SYNTHETIC_ERRNO(EIO), "Partition end beyond disk end."); + } + + assert(next >= gap); + r = context_discard_range(context, gap, next - gap); + if (r == -EOPNOTSUPP) { + if (p) + log_info("Storage does not support discarding, not discarding gap after partition %" PRIu64 ".", p->partno); + else + log_info("Storage does not support discarding, not discarding gap at beginning of disk."); + return 0; + } + if (r == 0) /* Too short */ + return 0; + if (r < 0) { + if (p) + return log_error_errno(r, "Failed to discard gap after partition %" PRIu64 ".", p->partno); + else + return log_error_errno(r, "Failed to discard gap at beginning of disk."); + } + + if (p) + log_info("Successfully discarded gap after partition %" PRIu64 ".", p->partno); + else + log_info("Successfully discarded gap at beginning of disk."); + + return 0; +} + +static int context_wipe_and_discard(Context *context, bool from_scratch) { + Partition *p; + int r; + + assert(context); + + /* Wipe and discard the contents of all partitions we are about to create. We skip the discarding if + * we were supposed to start from scratch anyway, as in that case we just discard the whole block + * device in one go early on. */ + + LIST_FOREACH(partitions, p, context->partitions) { + + if (!p->allocated_to_area) + continue; + + if (!from_scratch) { + r = context_discard_partition(context, p); + if (r < 0) + return r; + } + + r = context_wipe_partition(context, p); + if (r < 0) + return r; + + if (!from_scratch) { + r = context_discard_gap_after(context, p); + if (r < 0) + return r; + } + } + + if (!from_scratch) { + r = context_discard_gap_after(context, NULL); + if (r < 0) + return r; + } + + return 0; +} + +static int partition_acquire_uuid(Context *context, Partition *p, sd_id128_t *ret) { + struct { + sd_id128_t type_uuid; + uint64_t counter; + } _packed_ plaintext = {}; + union { + unsigned char md[SHA256_DIGEST_LENGTH]; + sd_id128_t id; + } result; + + uint64_t k = 0; + Partition *q; + int r; + + assert(context); + assert(p); + assert(ret); + + /* Calculate a good UUID for the indicated partition. We want a certain degree of reproducibility, + * hence we won't generate the UUIDs randomly. Instead we use a cryptographic hash (precisely: + * HMAC-SHA256) to derive them from a single seed. The seed is generally the machine ID of the + * installation we are processing, but if random behaviour is desired can be random, too. We use the + * seed value as key for the HMAC (since the machine ID is something we generally don't want to leak) + * and the partition type as plaintext. The partition type is suffixed with a counter (only for the + * second and later partition of the same type) if we have more than one partition of the same + * time. Or in other words: + * + * With: + * SEED := /etc/machine-id + * + * If first partition instance of type TYPE_UUID: + * PARTITION_UUID := HMAC-SHA256(SEED, TYPE_UUID) + * + * For all later partition instances of type TYPE_UUID with INSTANCE being the LE64 encoded instance number: + * PARTITION_UUID := HMAC-SHA256(SEED, TYPE_UUID || INSTANCE) + */ + + LIST_FOREACH(partitions, q, context->partitions) { + if (p == q) + break; + + if (!sd_id128_equal(p->type_uuid, q->type_uuid)) + continue; + + k++; + } + + plaintext.type_uuid = p->type_uuid; + plaintext.counter = htole64(k); + + if (!HMAC(EVP_sha256(), + &context->seed, sizeof(context->seed), + (const unsigned char*) &plaintext, k == 0 ? sizeof(sd_id128_t) : sizeof(plaintext), + result.md, NULL)) + return log_error_errno(SYNTHETIC_ERRNO(ENOTRECOVERABLE), "SHA256 calculation failed."); + + /* Take the first half, mark it as v4 UUID */ + assert_cc(sizeof(result.md) == sizeof(result.id) * 2); + result.id = id128_make_v4_uuid(result.id); + + /* Ensure this partition UUID is actually unique, and there's no remaining partition from an earlier run? */ + LIST_FOREACH(partitions, q, context->partitions) { + if (p == q) + continue; + + if (sd_id128_equal(q->current_uuid, result.id) || + sd_id128_equal(q->new_uuid, result.id)) { + log_warning("Partition UUID calculated from seed for partition %" PRIu64 " exists already, reverting to randomized UUID.", p->partno); + + r = sd_id128_randomize(&result.id); + if (r < 0) + return log_error_errno(r, "Failed to generate randomized UUID: %m"); + + break; + } + } + + *ret = result.id; + return 0; +} + +static int partition_acquire_label(Context *context, Partition *p, char **ret) { + _cleanup_free_ char *label = NULL; + const char *prefix; + unsigned k = 1; + + assert(context); + assert(p); + assert(ret); + + prefix = gpt_partition_type_uuid_to_string(p->type_uuid); + if (!prefix) + prefix = "linux"; + + for (;;) { + const char *ll = label ?: prefix; + bool retry = false; + Partition *q; + + LIST_FOREACH(partitions, q, context->partitions) { + if (p == q) + break; + + if (streq_ptr(ll, q->current_label) || + streq_ptr(ll, q->new_label)) { + retry = true; + break; + } + } + + if (!retry) + break; + + label = mfree(label); + + + if (asprintf(&label, "%s-%u", prefix, ++k) < 0) + return log_oom(); + } + + if (!label) { + label = strdup(prefix); + if (!label) + return log_oom(); + } + + *ret = TAKE_PTR(label); + return 0; +} + +static int context_acquire_partition_uuids_and_labels(Context *context) { + Partition *p; + int r; + + assert(context); + + LIST_FOREACH(partitions, p, context->partitions) { + assert(sd_id128_is_null(p->new_uuid)); + assert(!p->new_label); + + /* Never touch foreign partitions */ + if (PARTITION_IS_FOREIGN(p)) { + p->new_uuid = p->current_uuid; + + if (p->current_label) { + p->new_label = strdup(p->current_label); + if (!p->new_label) + return log_oom(); + } + + continue; + } + + if (!sd_id128_is_null(p->current_uuid)) + p->new_uuid = p->current_uuid; /* Never change initialized UUIDs */ + else { + r = partition_acquire_uuid(context, p, &p->new_uuid); + if (r < 0) + return r; + } + + if (!isempty(p->current_label)) { + p->new_label = strdup(p->current_label); /* never change initialized labels */ + if (!p->new_label) + return log_oom(); + } else { + r = partition_acquire_label(context, p, &p->new_label); + if (r < 0) + return r; + } + } + + return 0; +} + +static int device_kernel_partitions_supported(int fd) { + struct loop_info64 info; + struct stat st; + + assert(fd >= 0); + + if (fstat(fd, &st) < 0) + return log_error_errno(fd, "Failed to fstat() image file: %m"); + if (!S_ISBLK(st.st_mode)) + return false; + + if (ioctl(fd, LOOP_GET_STATUS64, &info) < 0) { + + if (ERRNO_IS_NOT_SUPPORTED(errno) || errno == EINVAL) + return true; /* not a loopback device, let's assume partition are supported */ + + return log_error_errno(fd, "Failed to issue LOOP_GET_STATUS64 on block device: %m"); + } + +#if HAVE_VALGRIND_MEMCHECK_H + /* Valgrind currently doesn't know LOOP_GET_STATUS64. Remove this once it does */ + VALGRIND_MAKE_MEM_DEFINED(&info, sizeof(info)); +#endif + + return FLAGS_SET(info.lo_flags, LO_FLAGS_PARTSCAN); +} + +static int context_write_partition_table( + Context *context, + const char *node, + bool from_scratch) { + + _cleanup_(fdisk_unref_tablep) struct fdisk_table *original_table = NULL; + int capable, r; + Partition *p; + + assert(context); + + if (arg_pretty > 0 || + (arg_pretty < 0 && isatty(STDOUT_FILENO) > 0)) { + + if (context->n_partitions == 0) + puts("Empty partition table."); + else + (void) context_dump_partitions(context, node); + + putc('\n', stdout); + + (void) context_dump_partition_bar(context, node); + putc('\n', stdout); + fflush(stdout); + } + + if (!from_scratch && !context_changed(context)) { + log_info("No changes."); + return 0; + } + + if (arg_dry_run) { + log_notice("Refusing to repartition, please re-run with --dry-run=no."); + return 0; + } + + log_info("Applying changes."); + + if (from_scratch) { + r = context_discard_range(context, 0, context->total); + if (r == -EOPNOTSUPP) + log_info("Storage does not support discarding, not discarding entire block device data."); + else if (r < 0) + return log_error_errno(r, "Failed to discard entire block device: %m"); + else if (r > 0) + log_info("Discarded entire block device."); + } + + r = fdisk_get_partitions(context->fdisk_context, &original_table); + if (r < 0) + return log_error_errno(r, "Failed to acquire partition table: %m"); + + /* Wipe fs signatures and discard sectors where the new partitions are going to be placed and in the + * gaps between partitions, just to be sure. */ + r = context_wipe_and_discard(context, from_scratch); + if (r < 0) + return r; + + LIST_FOREACH(partitions, p, context->partitions) { + if (p->dropped) + continue; + + assert(p->new_size != UINT64_MAX); + assert(p->offset != UINT64_MAX); + assert(p->partno != UINT64_MAX); + + if (PARTITION_EXISTS(p)) { + bool changed = false; + + assert(p->current_partition); + + if (p->new_size != p->current_size) { + assert(p->new_size >= p->current_size); + assert(p->new_size % 512 == 0); + + r = fdisk_partition_size_explicit(p->current_partition, true); + if (r < 0) + return log_error_errno(r, "Failed to enable explicit sizing: %m"); + + r = fdisk_partition_set_size(p->current_partition, p->new_size / 512); + if (r < 0) + return log_error_errno(r, "Failed to grow partition: %m"); + + log_info("Growing existing partition %" PRIu64 ".", p->partno); + changed = true; + } + + if (!sd_id128_equal(p->new_uuid, p->current_uuid)) { + char buf[ID128_UUID_STRING_MAX]; + + assert(!sd_id128_is_null(p->new_uuid)); + + r = fdisk_partition_set_uuid(p->current_partition, id128_to_uuid_string(p->new_uuid, buf)); + if (r < 0) + return log_error_errno(r, "Failed to set partition UUID: %m"); + + log_info("Initializing UUID of existing partition %" PRIu64 ".", p->partno); + changed = true; + } + + if (!streq_ptr(p->new_label, p->current_label)) { + assert(!isempty(p->new_label)); + + r = fdisk_partition_set_name(p->current_partition, p->new_label); + if (r < 0) + return log_error_errno(r, "Failed to set partition label: %m"); + + log_info("Setting partition label of existing partition %" PRIu64 ".", p->partno); + changed = true; + } + + if (changed) { + assert(!PARTITION_IS_FOREIGN(p)); /* never touch foreign partitions */ + + r = fdisk_set_partition(context->fdisk_context, p->partno, p->current_partition); + if (r < 0) + return log_error_errno(r, "Failed to update partition: %m"); + } + } else { + _cleanup_(fdisk_unref_partitionp) struct fdisk_partition *q = NULL; + _cleanup_(fdisk_unref_parttypep) struct fdisk_parttype *t = NULL; + char ids[ID128_UUID_STRING_MAX]; + + assert(!p->new_partition); + assert(p->offset % 512 == 0); + assert(p->new_size % 512 == 0); + assert(!sd_id128_is_null(p->new_uuid)); + assert(!isempty(p->new_label)); + + t = fdisk_new_parttype(); + if (!t) + return log_oom(); + + r = fdisk_parttype_set_typestr(t, id128_to_uuid_string(p->type_uuid, ids)); + if (r < 0) + return log_error_errno(r, "Failed to initialize partition type: %m"); + + q = fdisk_new_partition(); + if (!q) + return log_oom(); + + r = fdisk_partition_set_type(q, t); + if (r < 0) + return log_error_errno(r, "Failed to set partition type: %m"); + + r = fdisk_partition_size_explicit(q, true); + if (r < 0) + return log_error_errno(r, "Failed to enable explicit sizing: %m"); + + r = fdisk_partition_set_start(q, p->offset / 512); + if (r < 0) + return log_error_errno(r, "Failed to position partition: %m"); + + r = fdisk_partition_set_size(q, p->new_size / 512); + if (r < 0) + return log_error_errno(r, "Failed to grow partition: %m"); + + r = fdisk_partition_set_partno(q, p->partno); + if (r < 0) + return log_error_errno(r, "Failed to set partition number: %m"); + + r = fdisk_partition_set_uuid(q, id128_to_uuid_string(p->new_uuid, ids)); + if (r < 0) + return log_error_errno(r, "Failed to set partition UUID: %m"); + + r = fdisk_partition_set_name(q, p->new_label); + if (r < 0) + return log_error_errno(r, "Failed to set partition label: %m"); + + log_info("Creating new partition %" PRIu64 ".", p->partno); + + r = fdisk_add_partition(context->fdisk_context, q, NULL); + if (r < 0) + return log_error_errno(r, "Failed to add partition: %m"); + + assert(!p->new_partition); + p->new_partition = TAKE_PTR(q); + } + } + + log_info("Writing new partition table."); + + r = fdisk_write_disklabel(context->fdisk_context); + if (r < 0) + return log_error_errno(r, "Failed to write partition table: %m"); + + capable = device_kernel_partitions_supported(fdisk_get_devfd(context->fdisk_context)); + if (capable < 0) + return capable; + if (capable > 0) { + log_info("Telling kernel to reread partition table."); + + if (from_scratch) + r = fdisk_reread_partition_table(context->fdisk_context); + else + r = fdisk_reread_changes(context->fdisk_context, original_table); + if (r < 0) + return log_error_errno(r, "Failed to reread partition table: %m"); + } else + log_notice("Not telling kernel to reread partition table, because selected image does not support kernel partition block devices."); + + log_info("All done."); + + return 0; +} + +static int context_read_seed(Context *context, const char *root) { + int r; + + assert(context); + + if (!sd_id128_is_null(context->seed)) + return 0; + + if (!arg_randomize) { + _cleanup_close_ int fd = -1; + + fd = chase_symlinks_and_open("/etc/machine-id", root, CHASE_PREFIX_ROOT, O_RDONLY|O_CLOEXEC, NULL); + if (fd == -ENOENT) + log_info("No machine ID set, using randomized partition UUIDs."); + else if (fd < 0) + return log_error_errno(fd, "Failed to determine machine ID of image: %m"); + else { + r = id128_read_fd(fd, ID128_PLAIN, &context->seed); + if (r == -ENOMEDIUM) + log_info("No machine ID set, using randomized partition UUIDs."); + else if (r < 0) + return log_error_errno(r, "Failed to parse machine ID of image: %m"); + + return 0; + } + } + + r = sd_id128_randomize(&context->seed); + if (r < 0) + return log_error_errno(r, "Failed to generate randomized seed: %m"); + + return 0; +} + +static int context_factory_reset(Context *context, bool from_scratch) { + Partition *p; + size_t n = 0; + int r; + + assert(context); + + if (arg_factory_reset <= 0) + return 0; + + if (from_scratch) /* Nothing to reset if we start from scratch */ + return 0; + + if (arg_dry_run) { + log_notice("Refusing to factory reset, please re-run with --dry-run=no."); + return 0; + } + + log_info("Applying factory reset."); + + LIST_FOREACH(partitions, p, context->partitions) { + + if (!p->factory_reset || !PARTITION_EXISTS(p)) + continue; + + assert(p->partno != UINT64_MAX); + + log_info("Removing partition %" PRIu64 " for factory reset.", p->partno); + + r = fdisk_delete_partition(context->fdisk_context, p->partno); + if (r < 0) + return log_error_errno(r, "Failed to remove partition %" PRIu64 ": %m", p->partno); + + n++; + } + + if (n == 0) { + log_info("Factory reset requested, but no partitions to delete found."); + return 0; + } + + r = fdisk_write_disklabel(context->fdisk_context); + if (r < 0) + return log_error_errno(r, "Failed to write disk label: %m"); + + log_info("Successfully deleted %zu partitions.", n); + return 1; +} + +static int context_can_factory_reset(Context *context) { + Partition *p; + + assert(context); + + LIST_FOREACH(partitions, p, context->partitions) + if (p->factory_reset && PARTITION_EXISTS(p)) + return true; + + return false; +} + +static int help(void) { + _cleanup_free_ char *link = NULL; + int r; + + r = terminal_urlify_man("systemd-repart", "1", &link); + if (r < 0) + return log_oom(); + + printf("%s [OPTIONS...] [DEVICE]\n" + "\n%sGrow and add partitions to partition table.%s\n\n" + " -h --help Show this help\n" + " --version Show package version\n" + " --dry-run=BOOL Whether to run dry-run operation\n" + " --empty=MODE One of refuse, allow, require, force; controls how to\n" + " handle empty disks lacking partition table\n" + " --discard=BOOL Whether to discard backing blocks for new partitions\n" + " --pretty=BOOL Whether to show pretty summary before executing operation\n" + " --factory-reset=BOOL Whether to remove data partitions before recreating\n" + " them\n" + " --can-factory-reset Test whether factory reset is defined\n" + " --root=PATH Operate relative to root path\n" + " --definitions=DIR Find partitions in specified directory\n" + " --seed=UUID 128bit seed UUID to derive all UUIDs from\n" + "\nSee the %s for details.\n" + , program_invocation_short_name + , ansi_highlight(), ansi_normal() + , link + ); + + return 0; +} + +static int parse_argv(int argc, char *argv[]) { + + enum { + ARG_VERSION = 0x100, + ARG_DRY_RUN, + ARG_EMPTY, + ARG_DISCARD, + ARG_FACTORY_RESET, + ARG_CAN_FACTORY_RESET, + ARG_ROOT, + ARG_SEED, + ARG_PRETTY, + ARG_DEFINITIONS, + }; + + static const struct option options[] = { + { "help", no_argument, NULL, 'h' }, + { "version", no_argument, NULL, ARG_VERSION }, + { "dry-run", required_argument, NULL, ARG_DRY_RUN }, + { "empty", required_argument, NULL, ARG_EMPTY }, + { "discard", required_argument, NULL, ARG_DISCARD }, + { "factory-reset", required_argument, NULL, ARG_FACTORY_RESET }, + { "can-factory-reset", no_argument, NULL, ARG_CAN_FACTORY_RESET }, + { "root", required_argument, NULL, ARG_ROOT }, + { "seed", required_argument, NULL, ARG_SEED }, + { "pretty", required_argument, NULL, ARG_PRETTY }, + { "definitions", required_argument, NULL, ARG_DEFINITIONS }, + {} + }; + + int c, r; + + assert(argc >= 0); + assert(argv); + + while ((c = getopt_long(argc, argv, "h", options, NULL)) >= 0) + + switch (c) { + + case 'h': + return help(); + + case ARG_VERSION: + return version(); + + case ARG_DRY_RUN: + r = parse_boolean(optarg); + if (r < 0) + return log_error_errno(r, "Failed to parse --dry-run= parameter: %s", optarg); + + arg_dry_run = r; + break; + + case ARG_EMPTY: + if (isempty(optarg) || streq(optarg, "refuse")) + arg_empty = EMPTY_REFUSE; + else if (streq(optarg, "allow")) + arg_empty = EMPTY_ALLOW; + else if (streq(optarg, "require")) + arg_empty = EMPTY_REQUIRE; + else if (streq(optarg, "force")) + arg_empty = EMPTY_FORCE; + else + return log_error_errno(SYNTHETIC_ERRNO(EINVAL), + "Failed to parse --empty= parameter: %s", optarg); + break; + + case ARG_DISCARD: + r = parse_boolean(optarg); + if (r < 0) + return log_error_errno(r, "Failed to parse --discard= parameter: %s", optarg); + + arg_discard = r; + break; + + case ARG_FACTORY_RESET: + r = parse_boolean(optarg); + if (r < 0) + return log_error_errno(r, "Failed to parse --factory-reset= parameter: %s", optarg); + + arg_factory_reset = r; + break; + + case ARG_CAN_FACTORY_RESET: + arg_can_factory_reset = true; + break; + + case ARG_ROOT: + r = parse_path_argument_and_warn(optarg, false, &arg_root); + if (r < 0) + return r; + break; + + case ARG_SEED: + if (isempty(optarg)) { + arg_seed = SD_ID128_NULL; + arg_randomize = false; + } else if (streq(optarg, "random")) + arg_randomize = true; + else { + r = sd_id128_from_string(optarg, &arg_seed); + if (r < 0) + return log_error_errno(r, "Failed to parse seed: %s", optarg); + + arg_randomize = false; + } + + break; + + case ARG_PRETTY: + r = parse_boolean(optarg); + if (r < 0) + return log_error_errno(r, "Failed to parse --pretty= parameter: %s", optarg); + + arg_pretty = r; + break; + + case ARG_DEFINITIONS: + r = parse_path_argument_and_warn(optarg, false, &arg_definitions); + if (r < 0) + return r; + break; + + case '?': + return -EINVAL; + + default: + assert_not_reached("Unhandled option"); + } + + if (argc - optind > 1) + return log_error_errno(SYNTHETIC_ERRNO(EINVAL), + "Expected at most one argument, the path to the block device."); + + if (arg_factory_reset > 0 && IN_SET(arg_empty, EMPTY_FORCE, EMPTY_REQUIRE)) + return log_error_errno(SYNTHETIC_ERRNO(EINVAL), + "Combination of --factory-reset=yes and --empty=force/--empty=require is invalid."); + + if (arg_can_factory_reset) + arg_dry_run = true; + + arg_node = argc > optind ? argv[optind] : NULL; + return 1; +} + +static int parse_proc_cmdline_factory_reset(void) { + bool b; + int r; + + if (arg_factory_reset >= 0) /* Never override what is specified on the process command line */ + return 0; + + if (!in_initrd()) /* Never honour kernel command line factory reset request outside of the initrd */ + return 0; + + r = proc_cmdline_get_bool("systemd.factory_reset", &b); + if (r < 0) + return log_error_errno(r, "Failed to parse systemd.factory_reset kernel command line argument: %m"); + if (r > 0) { + arg_factory_reset = b; + + if (b) + log_notice("Honouring factory reset requested via kernel command line."); + } + + return 0; +} + +static int parse_efi_variable_factory_reset(void) { + _cleanup_free_ char *value = NULL; + int r; + + if (arg_factory_reset >= 0) /* Never override what is specified on the process command line */ + return 0; + + if (!in_initrd()) /* Never honour EFI variable factory reset request outside of the initrd */ + return 0; + + r = efi_get_variable_string(EFI_VENDOR_SYSTEMD, "FactoryReset", &value); + if (r == -ENOENT || ERRNO_IS_NOT_SUPPORTED(r)) + return 0; + if (r < 0) + return log_error_errno(r, "Failed to read EFI variable FactoryReset: %m"); + + r = parse_boolean(value); + if (r < 0) + return log_error_errno(r, "Failed to parse EFI variable FactoryReset: %m"); + + arg_factory_reset = r; + if (r) + log_notice("Honouring factory reset requested via EFI variable FactoryReset: %m"); + + return 0; +} + +static int remove_efi_variable_factory_reset(void) { + int r; + + r = efi_set_variable(EFI_VENDOR_SYSTEMD, "FactoryReset", NULL, 0); + if (r == -ENOENT || ERRNO_IS_NOT_SUPPORTED(r)) + return 0; + if (r < 0) + return log_error_errno(r, "Failed to remove EFI variable FactoryReset: %m"); + + log_info("Successfully unset EFI variable FactoryReset."); + return 0; +} + +static int acquire_root_devno(const char *p, int mode, char **ret) { + _cleanup_close_ int fd = -1; + struct stat st; + dev_t devno; + int r; + + fd = open(p, mode); + if (fd < 0) + return -errno; + + if (fstat(fd, &st) < 0) + return -errno; + + if (S_ISREG(st.st_mode)) { + char *s; + + s = strdup(p); + if (!s) + return log_oom(); + + *ret = s; + return 0; + } + + if (S_ISBLK(st.st_mode)) + devno = st.st_rdev; + else if (S_ISDIR(st.st_mode)) { + + devno = st.st_dev; + + if (major(st.st_dev) == 0) { + r = btrfs_get_block_device_fd(fd, &devno); + if (r == -ENOTTY) /* not btrfs */ + return -ENODEV; + if (r < 0) + return r; + } + + } else + return -ENOTBLK; + + /* From dm-crypt to backing partition */ + r = block_get_originating(devno, &devno); + if (r < 0) + log_debug_errno(r, "Failed to find underlying block device for '%s', ignoring: %m", p); + + /* From partition to whole disk containing it */ + r = block_get_whole_disk(devno, &devno); + if (r < 0) + log_debug_errno(r, "Failed to find whole disk block device for '%s', ingoring: %m", p); + + return device_path_make_canonical(S_IFBLK, devno, ret); +} + +static int find_root(char **ret) { + const char *t; + int r; + + if (arg_node) { + r = acquire_root_devno(arg_node, O_RDONLY|O_CLOEXEC, ret); + if (r < 0) + return log_error_errno(r, "Failed to determine backing device of %s: %m", arg_node); + + return 0; + } + + /* Let's search for the root device. We look for two cases here: first in /, and then in /usr. The + * latter we check for cases where / is a tmpfs and only /usr is an actual persistent block device + * (think: volatile setups) */ + + FOREACH_STRING(t, "/", "/usr") { + _cleanup_free_ char *j = NULL; + const char *p; + + if (in_initrd()) { + j = path_join("/sysroot", t); + if (!j) + return log_oom(); + + p = j; + } else + p = t; + + r = acquire_root_devno(p, O_RDONLY|O_DIRECTORY|O_CLOEXEC, ret); + if (r < 0) { + if (r != -ENODEV) + return log_error_errno(r, "Failed to determine backing device of %s: %m", p); + } else + return 0; + } + + return log_error_errno(SYNTHETIC_ERRNO(ENODEV), "Failed to discover root block device."); +} + +static int run(int argc, char *argv[]) { + _cleanup_(context_freep) Context* context = NULL; + _cleanup_free_ char *node = NULL; + bool from_scratch; + int r; + + log_show_color(true); + log_parse_environment(); + log_open(); + + if (in_initrd()) { + /* Default to operation on /sysroot when invoked in the initrd! */ + arg_root = strdup("/sysroot"); + if (!arg_root) + return log_oom(); + } + + r = parse_argv(argc, argv); + if (r <= 0) + return r; + + r = parse_proc_cmdline_factory_reset(); + if (r < 0) + return r; + + r = parse_efi_variable_factory_reset(); + if (r < 0) + return r; + + r = find_root(&node); + if (r < 0) + return r; + + context = context_new(arg_seed); + if (!context) + return log_oom(); + + r = context_read_definitions(context, arg_definitions, arg_root); + if (r < 0) + return r; + + r = context_load_partition_table(context, node); + if (r == -EHWPOISON) + return 77; /* Special return value which means "Not GPT, so not doing anything". This isn't + * really an error when called at boot. */ + if (r < 0) + return r; + from_scratch = r > 0; /* Starting from scratch */ + + if (arg_can_factory_reset) { + r = context_can_factory_reset(context); + if (r < 0) + return r; + if (r == 0) + return EXIT_FAILURE; + + return 0; + } + + r = context_factory_reset(context, from_scratch); + if (r < 0) + return r; + if (r > 0) { + /* We actually did a factory reset! */ + r = remove_efi_variable_factory_reset(); + if (r < 0) + return r; + + /* Reload the reduced partition table */ + context_unload_partition_table(context); + r = context_load_partition_table(context, node); + if (r < 0) + return r; + } + +#if 0 + (void) context_dump_partitions(context, node); + putchar('\n'); +#endif + + r = context_read_seed(context, arg_root); + if (r < 0) + return r; + + /* First try to fit new partitions in, dropping by priority until it fits */ + for (;;) { + if (context_allocate_partitions(context)) + break; /* Success! */ + + if (!context_drop_one_priority(context)) + return log_error_errno(SYNTHETIC_ERRNO(ENOSPC), + "Can't fit requested partitions into free space, refusing."); + } + + /* Now assign free space according to the weight logic */ + r = context_grow_partitions(context); + if (r < 0) + return r; + + /* Now calculate where each partition gets placed */ + context_place_partitions(context); + + /* Make sure each partition has a unique UUID and unique label */ + r = context_acquire_partition_uuids_and_labels(context); + if (r < 0) + return r; + + r = context_write_partition_table(context, node, from_scratch); + if (r < 0) + return r; + + return 0; +} + +DEFINE_MAIN_FUNCTION_WITH_POSITIVE_FAILURE(run); diff --git a/src/shared/conf-parser.c b/src/shared/conf-parser.c index 463db1cb4d..318588cc1f 100644 --- a/src/shared/conf-parser.c +++ b/src/shared/conf-parser.c @@ -515,6 +515,7 @@ DEFINE_PARSER(long, long, safe_atoli); DEFINE_PARSER(uint8, uint8_t, safe_atou8); DEFINE_PARSER(uint16, uint16_t, safe_atou16); DEFINE_PARSER(uint32, uint32_t, safe_atou32); +DEFINE_PARSER(int32, int32_t, safe_atoi32); DEFINE_PARSER(uint64, uint64_t, safe_atou64); DEFINE_PARSER(unsigned, unsigned, safe_atou); DEFINE_PARSER(double, double, safe_atod); diff --git a/src/shared/conf-parser.h b/src/shared/conf-parser.h index 136ee32426..2f3cb4217d 100644 --- a/src/shared/conf-parser.h +++ b/src/shared/conf-parser.h @@ -115,6 +115,7 @@ CONFIG_PARSER_PROTOTYPE(config_parse_long); CONFIG_PARSER_PROTOTYPE(config_parse_uint8); CONFIG_PARSER_PROTOTYPE(config_parse_uint16); CONFIG_PARSER_PROTOTYPE(config_parse_uint32); +CONFIG_PARSER_PROTOTYPE(config_parse_int32); CONFIG_PARSER_PROTOTYPE(config_parse_uint64); CONFIG_PARSER_PROTOTYPE(config_parse_double); CONFIG_PARSER_PROTOTYPE(config_parse_iec_size); diff --git a/src/shared/format-table.c b/src/shared/format-table.c index d7cb976757..4250130464 100644 --- a/src/shared/format-table.c +++ b/src/shared/format-table.c @@ -4,12 +4,15 @@ #include #include +#include "sd-id128.h" + #include "alloc-util.h" #include "fd-util.h" #include "fileio.h" #include "format-table.h" #include "format-util.h" #include "gunicode.h" +#include "id128-util.h" #include "in-addr-util.h" #include "locale-util.h" #include "memory-util.h" @@ -94,6 +97,7 @@ typedef struct TableData { int percent; /* we use 'int' as datatype for percent values in order to match the result of parse_percent() */ int ifindex; union in_addr_union address; + sd_id128_t id128; /* … add more here as we start supporting more cell data types … */ }; } TableData; @@ -289,6 +293,10 @@ static size_t table_data_size(TableDataType type, const void *data) { case TABLE_IN6_ADDR: return sizeof(struct in6_addr); + case TABLE_UUID: + case TABLE_ID128: + return sizeof(sd_id128_t); + default: assert_not_reached("Uh? Unexpected cell type"); } @@ -335,7 +343,6 @@ static bool table_data_matches( k = table_data_size(type, data); l = table_data_size(d->type, d->data); - if (k != l) return false; @@ -778,6 +785,7 @@ int table_add_many_internal(Table *t, TableDataType first_type, ...) { int ifindex; bool b; union in_addr_union address; + sd_id128_t id128; } buffer; switch (type) { @@ -901,6 +909,12 @@ int table_add_many_internal(Table *t, TableDataType first_type, ...) { data = &buffer.address.in6; break; + case TABLE_UUID: + case TABLE_ID128: + buffer.id128 = va_arg(ap, sd_id128_t); + data = &buffer.id128; + break; + case TABLE_SET_MINIMUM_WIDTH: { size_t w = va_arg(ap, size_t); @@ -1137,6 +1151,10 @@ static int cell_data_compare(TableData *a, size_t index_a, TableData *b, size_t case TABLE_IN6_ADDR: return memcmp(&a->address.in6, &b->address.in6, FAMILY_ADDRESS_SIZE(AF_INET6)); + case TABLE_UUID: + case TABLE_ID128: + return memcmp(&a->id128, &b->id128, sizeof(sd_id128_t)); + default: ; } @@ -1451,6 +1469,28 @@ static const char *table_data_format(Table *t, TableData *d) { break; } + case TABLE_ID128: { + char *p; + + p = new(char, SD_ID128_STRING_MAX); + if (!p) + return NULL; + + d->formatted = sd_id128_to_string(d->id128, p); + break; + } + + case TABLE_UUID: { + char *p; + + p = new(char, ID128_UUID_STRING_MAX); + if (!p) + return NULL; + + d->formatted = id128_to_uuid_string(d->id128, p); + break; + } + default: assert_not_reached("Unexpected type?"); } @@ -2155,6 +2195,16 @@ static int table_data_to_json(TableData *d, JsonVariant **ret) { case TABLE_IN6_ADDR: return json_variant_new_array_bytes(ret, &d->address, FAMILY_ADDRESS_SIZE(AF_INET6)); + case TABLE_ID128: { + char buf[SD_ID128_STRING_MAX]; + return json_variant_new_string(ret, sd_id128_to_string(d->id128, buf)); + } + + case TABLE_UUID: { + char buf[ID128_UUID_STRING_MAX]; + return json_variant_new_string(ret, id128_to_uuid_string(d->id128, buf)); + } + default: return -EINVAL; } diff --git a/src/shared/format-table.h b/src/shared/format-table.h index fa7a2bd6d6..870a29d385 100644 --- a/src/shared/format-table.h +++ b/src/shared/format-table.h @@ -35,6 +35,8 @@ typedef enum TableDataType { TABLE_IFINDEX, TABLE_IN_ADDR, /* Takes a union in_addr_union (or a struct in_addr) */ TABLE_IN6_ADDR, /* Takes a union in_addr_union (or a struct in6_addr) */ + TABLE_ID128, + TABLE_UUID, _TABLE_DATA_TYPE_MAX, /* The following are not really data types, but commands for table_add_cell_many() to make changes to diff --git a/test/TEST-45-REPART/Makefile b/test/TEST-45-REPART/Makefile new file mode 120000 index 0000000000..e9f93b1104 --- /dev/null +++ b/test/TEST-45-REPART/Makefile @@ -0,0 +1 @@ +../TEST-01-BASIC/Makefile \ No newline at end of file diff --git a/test/TEST-45-REPART/test.sh b/test/TEST-45-REPART/test.sh new file mode 100755 index 0000000000..cdec7c4bdc --- /dev/null +++ b/test/TEST-45-REPART/test.sh @@ -0,0 +1,37 @@ +#!/bin/bash +set -e +TEST_DESCRIPTION="test systemd-repart" + +. $TEST_BASE_DIR/test-functions + +test_setup() { + create_empty_image_rootdir + + ( + LOG_LEVEL=5 + eval $(udevadm info --export --query=env --name=${LOOPDEV}p2) + + setup_basic_environment + + mask_supporting_services + dracut_install truncate sfdisk + + # setup the testsuite service + cat >$initdir/etc/systemd/system/testsuite.service < /testok + exit 0 +fi + +systemd-analyze log-level debug + +truncate -s 1G /tmp/zzz + +SEED=e2a40bf9-73f1-4278-9160-49c031e7aef8 + +systemd-repart /tmp/zzz --empty=force --dry-run=no --seed=$SEED + +sfdisk -d /tmp/zzz > /tmp/empty + +cmp /tmp/empty - < /tmp/definitions/root.conf < /tmp/definitions/home.conf < /tmp/definitions/swap.conf < /tmp/populated + +cmp /tmp/populated - < /tmp/definitions/swap.conf < /tmp/definitions/extra.conf < /tmp/populated2 + +cmp /tmp/populated2 - < /tmp/populated3 + +cmp /tmp/populated3 - < /testok + +exit 0 diff --git a/travis-ci/managers/debian.sh b/travis-ci/managers/debian.sh index 6a6923fbc8..dedddab3b5 100755 --- a/travis-ci/managers/debian.sh +++ b/travis-ci/managers/debian.sh @@ -18,7 +18,10 @@ REPO_ROOT="${REPO_ROOT:-$PWD}" ADDITIONAL_DEPS=(python3-libevdev python3-pyparsing clang - perl) + perl + libfdisk-dev + libp11-kit-dev + libssl-dev) function info() { echo -e "\033[33;1m$1\033[0m" diff --git a/travis-ci/managers/fedora.sh b/travis-ci/managers/fedora.sh index e07b4938df..dbd484c25c 100755 --- a/travis-ci/managers/fedora.sh +++ b/travis-ci/managers/fedora.sh @@ -23,7 +23,10 @@ ADDITIONAL_DEPS=(dnf-plugins-core libubsan clang llvm - perl) + perl + libfdisk-devel + openssl-devel + p11-kit-devel) function info() { echo -e "\033[33;1m$1\033[0m" diff --git a/travis-ci/managers/fuzzbuzz.sh b/travis-ci/managers/fuzzbuzz.sh index 1541b46652..21b736cb60 100755 --- a/travis-ci/managers/fuzzbuzz.sh +++ b/travis-ci/managers/fuzzbuzz.sh @@ -10,6 +10,8 @@ sudo bash -c "echo 'deb-src http://archive.ubuntu.com/ubuntu/ xenial main restri sudo apt-get update -y sudo apt-get build-dep systemd -y sudo apt-get install -y ninja-build python3-pip python3-setuptools quota +# The following should be dropped when debian packaging has been updated to include them +sudo apt-get install -y libfdisk-dev libp11-kit-dev libssl-dev # FIXME: temporarily pin the meson version as 0.53 doesn't work with older python 3.5 # See: https://github.com/mesonbuild/meson/issues/6427 pip3 install meson==0.52.1 diff --git a/travis-ci/managers/fuzzit.sh b/travis-ci/managers/fuzzit.sh index 376761e20c..044d126e15 100755 --- a/travis-ci/managers/fuzzit.sh +++ b/travis-ci/managers/fuzzit.sh @@ -14,6 +14,8 @@ sudo bash -c "echo 'deb-src http://archive.ubuntu.com/ubuntu/ xenial main restri sudo apt-get update -y sudo apt-get build-dep systemd -y sudo apt-get install -y ninja-build python3-pip python3-setuptools +# The following should be dropped when debian packaging has been updated to include them +sudo apt-get install -y libfdisk-dev libp11-kit-dev libssl-dev # FIXME: temporarily pin the meson version as 0.53 doesn't work with older python 3.5 # See: https://github.com/mesonbuild/meson/issues/6427 pip3 install meson==0.52.1 diff --git a/units/meson.build b/units/meson.build index f7653c920c..581f44f99e 100644 --- a/units/meson.build +++ b/units/meson.build @@ -220,6 +220,8 @@ in_units = [ 'multi-user.target.wants/'], ['systemd-vconsole-setup.service', 'ENABLE_VCONSOLE'], ['systemd-volatile-root.service', ''], + ['systemd-repart.service', 'ENABLE_REPART', + 'sysinit.target.wants/ initrd-root-fs.target.wants/'], ['user-runtime-dir@.service', ''], ['user@.service', ''], ] diff --git a/units/systemd-repart.service.in b/units/systemd-repart.service.in new file mode 100644 index 0000000000..7ce6aefd29 --- /dev/null +++ b/units/systemd-repart.service.in @@ -0,0 +1,25 @@ +# SPDX-License-Identifier: LGPL-2.1+ +# +# This file is part of systemd. +# +# systemd is free software; you can redistribute it and/or modify it +# under the terms of the GNU Lesser General Public License as published by +# the Free Software Foundation; either version 2.1 of the License, or +# (at your option) any later version. + +[Unit] +Description=Repartition Root Disk +Documentation=man:systemd-repart.service(8) +DefaultDependencies=no +Conflicts=shutdown.target +After=sysroot.mount +Before=initrd-root-fs.target shutdown.target +ConditionVirtualization=!container + +[Service] +Type=oneshot +RemainAfterExit=yes +ExecStart=@rootbindir@/systemd-repart --dry-run=no + +# The tool returns 77 if there's no GPT partition table pre-existing +SuccessExitStatus=77