NetBSD/usr.sbin/sysinst/partitions.h

648 lines
22 KiB
C
Raw Permalink Normal View History

/* $NetBSD: partitions.h,v 1.29 2023/01/06 18:19:27 martin Exp $ */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
2020-11-06 15:23:10 +03:00
* Copyright (c) 2020 The NetBSD Foundation, Inc.
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
2020-11-06 15:23:10 +03:00
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
2020-11-06 15:23:10 +03:00
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
/*
* Abstract interface to access arbitrary disk partitioning schemes and
* keep Sysinst proper independent of the implementation / on-disk
* details.
*
* NOTE:
* - all sector numbers, alignment and sizes are in units of the
* disks physical sector size (not necessarily 512 bytes)!
* - some interfaces pass the disks sector size (when it is easily
* available at typical callers), but the backends can always
* assume it to be equal to the real physical sector size. If
* no value is passed, the backend can query the disk data
* via get_disk_geom().
* - single exception: disk_partitioning_scheme::size_limit is in 512
* byte sectors (as it is not associated with a concrete disk)
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
#include <sys/types.h>
#include <stdbool.h>
#include "msg_defs.h"
/*
* Import all the file system types, as enum fs_type.
*/
#define FSTYPE_ENUMNAME fs_type
#define FSTYPENAMES
#include <sys/disklabel.h>
#undef FSTYPE_ENUMNAME
/*
* Use random values (outside uint8_t range) to mark special file system
* types that are not in the FSTYPE enumeration.
*/
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
#ifndef FS_TMPFS
#define FS_TMPFS 256 /* tmpfs (prefered for /tmp if available) */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
#endif
#ifndef FS_MFS
#define FS_MFS 257 /* mfs, alternative to tmpfs if that is
not available */
#endif
#ifndef FS_EFI_SP
#define FS_EFI_SP 258 /* EFI system partition, uses FS_MSDOS,
but may have a different partition
type */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
#endif
#define MAX_LABEL_LEN 128 /* max. length of a partition label */
2019-12-26 07:53:11 +03:00
#define MAX_SHORTCUT_LEN 8 /* max. length of a shortcut ("a:") */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* A partition index / handle, identifies a singlepartition within
* a struct disk_partitions. This is just an iterator/index - whenever
* changes to the set of partitions are done, partitions may get a new
* part_id.
* We assume that partitioning schemes keep partitions sorted (with
* key = start address, some schemes will have overlapping partitions,
* like MBR extended partitions).
*/
typedef size_t part_id;
/*
* An invalid value for a partition index / handle
*/
#define NO_PART ((part_id)~0U)
/*
* Intended usage for a partition
*/
enum part_type {
PT_undef, /* invalid value */
PT_unknown, /* anything we can not map to one of these */
PT_root, /* the NetBSD / partition (bootable) */
PT_swap, /* the NetBSD swap partition */
PT_FAT, /* boot partition (e.g. for u-boot) */
PT_EXT2, /* boot partition (for Linux appliances) */
PT_SYSVBFS, /* boot partition (for some SYSV machines) */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
PT_EFI_SYSTEM, /* (U)EFI boot partition */
};
/*
* A generic structure describing partition types for menu/user interface
* purposes. The internal details may be richer and the *pointer* value
* is the unique token - that is: the partitioning scheme will hand out
* pointers to internal data and recognize the exact partition type details
2020-12-02 17:20:19 +03:00
* by pointer comparison.
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
struct part_type_desc {
enum part_type generic_ptype; /* what this maps to in generic terms */
const char *short_desc; /* short type description */
const char *description; /* full description */
};
/* Bits for disk_part_info.flags: */
#define PTI_SEC_CONTAINER 1 /* this covers our secondary
partitions */
#define PTI_WHOLE_DISK 2 /* all of the NetBSD disk */
#define PTI_BOOT 4 /* required for booting */
#define PTI_PSCHEME_INTERNAL 8 /* no user partition, e.g.
MBRs extend partition */
#define PTI_RAW_PART 16 /* total disk */
#define PTI_INSTALL_TARGET 32 /* marks the target partition
* assumed to become / after
* reboot; may not be
* persistent; may only be
* set for a single partition!
*/
#define PTI_SPECIAL_PARTS \
(PTI_PSCHEME_INTERNAL|PTI_WHOLE_DISK|PTI_SEC_CONTAINER|PTI_RAW_PART)
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/* A single partition */
struct disk_part_info {
daddr_t start, size; /* start and size on disk */
uint32_t flags; /* active PTI_ flags */
const struct part_type_desc *nat_type; /* native partition type */
/*
* The following will only be available
* a) for a small subset of file system types
* b) if the partition (in this state) has already been
* used before
* It is OK to leave all these zeroed / NULL when setting
* partition data - or leave them at the last values a get operation
* returned. Backends can not rely on them to be valid.
*/
const char *last_mounted; /* last mount point or NULL */
unsigned int fs_type, fs_sub_type, /* FS_* type of filesystem
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* and for some FS a sub
* type (e.g. FFSv1 vs. FFSv2)
*/
fs_opt1, fs_opt2, fs_opt3; /* FS specific option, used
* for FFS block/fragsize
* and inodes
*/
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
};
/* An unused area that may be used for new partitions */
struct disk_part_free_space {
daddr_t start, size;
};
/*
* Some partition schemes define additional data that needs to be edited.
* These attributes are described in this structure and referenced by
* their index into the fixed list of available attributes.
*/
enum custom_attr_type { pet_bool, pet_cardinal, pet_str };
struct disk_part_custom_attribute {
msg label; /* Name, like "active partition" */
enum custom_attr_type type; /* bool, long, char* */
size_t strlen; /* maximum length if pet_str */
};
/*
* When displaying a partition editor, we have standard columns, but
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* partitioning schemes add custom columns to the table as well.
* There is a fixed number of columns and they are described by this
* structure:
*/
struct disk_part_edit_column_desc {
msg title;
unsigned int width;
};
struct disk_partitions; /* in-memory representation of a set of partitions */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* When querying partition "device" names, we may ask for:
*/
enum dev_name_usage {
parent_device_only, /* wd0 instead of wd0i, no path */
logical_name, /* NAME=my-root instead of dk7 */
plain_name, /* e.g. /dev/wd0i or /dev/dk7 */
raw_dev_name, /* e.g. /dev/rwd0i or /dev/rdk7 */
};
/*
* A scheme how to store partitions on-disk, and methods to read/write
* them to/from our abstract internal presentation.
*/
struct disk_partitioning_scheme {
/* name of the on-disk scheme, retrieved via msg_string */
msg name, short_name;
/* prompt shown when creating custom partition types */
msg new_type_prompt;
/* description of scheme specific partition flags */
msg part_flag_desc;
/*
* size restrictions for this partitioning scheme (number
* of 512 byte sectors max)
*/
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
daddr_t size_limit; /* 0 if not limited */
/*
* If this scheme allows sub-partitions (i.e. MBR -> disklabel),
* this is a pointer to the (potential/optional) secondary
* scheme. Depending on partitioning details it may not be
* used in the end.
* This link is only here for better help messages.
* See *secondary_partitions further below for actually accessing
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* secondary partitions.
*/
const struct disk_partitioning_scheme *secondary_scheme;
/*
* Partition editor colum descriptions for whatever the scheme
* needs to display (see format_partition_table_str below).
*/
size_t edit_columns_count;
const struct disk_part_edit_column_desc *edit_columns;
/*
* Custom attributes editable by the partitioning scheme (but of
* no particular meaning for sysinst)
*/
size_t custom_attribute_count;
const struct disk_part_custom_attribute *custom_attributes;
/*
* Partition types supported by this scheme,
* first function gets the number, second queries single elements
*/
size_t (*get_part_types_count)(void);
const struct part_type_desc * (*get_part_type)(size_t ndx);
/*
* Get the preferred native representation for a generic partition type
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
const struct part_type_desc * (*get_generic_part_type)(enum part_type);
/*
* Get the preferred native partition type for a specific file system
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* type (FS_*) and subtype (fs specific value)
*/
const struct part_type_desc * (*get_fs_part_type)(
enum part_type, unsigned, unsigned);
/*
* Optional: inverse to above: given a part_type_desc, set default
* fstype and subtype.
*/
bool (*get_default_fstype)(const struct part_type_desc *,
unsigned *fstype, unsigned *fs_sub_type);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Create a custom partition type. If the type already exists
* (or there is a collision), the old existing type will be
* returned and no new type created. This is not considered
* an error (to keep the user interface simple).
* On failure NULL is returned and (if passed != NULL)
* *err_msg is set to a message describing the error.
*/
const struct part_type_desc * (*create_custom_part_type)
(const char *custom, const char **err_msg);
/*
* Return a usable internal partition type representation
* for types that are not otherwise mappable.
* This could be FS_OTHER for disklabel, or a randomly
* created type guid for GPT. This type may or may not be
* in the regular type list. If not, it needs to behave like a
* custom type.
*/
const struct part_type_desc * (*create_unknown_part_type)(void);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Global attributes
*/
/*
* Get partition alignment suggestion. The schemen may enforce
* additional/different alignment for some partitions.
*/
daddr_t (*get_part_alignment)(const struct disk_partitions*);
/*
* Methods to manipulate the in-memory abstract representation
*/
/* Retrieve data about a single partition, identified by the part_id.
* Fill the disk_part_info structure
*/
bool (*get_part_info)(const struct disk_partitions*, part_id,
struct disk_part_info*);
/* Optional: fill an attribute string describing the given partition */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
bool (*get_part_attr_str)(const struct disk_partitions*, part_id,
char *str, size_t avail_space);
/* Format a partition editor element for the "col" column in
* edit_columns. Used e.g. with MBR to set "active" flags.
*/
bool (*format_partition_table_str)(const struct disk_partitions*,
part_id, size_t col, char *outstr, size_t outspace);
2022-05-24 09:18:34 +03:00
/* is the type of this partition changeable? */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
bool (*part_type_can_change)(const struct disk_partitions*,
part_id);
/* can we add further partitions? */
bool (*can_add_partition)(const struct disk_partitions*);
2022-05-24 09:18:34 +03:00
/* is the custom attribute changeable? */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
bool (*custom_attribute_writable)(const struct disk_partitions*,
part_id, size_t attr_no);
/*
* Output formatting for custom attributes.
* If "info" is != NULL, use (where it makes sense)
* values from that structure, as if a call to set_part_info
* would have been done before this call.
*/
bool (*format_custom_attribute)(const struct disk_partitions*,
part_id, size_t attr_no, const struct disk_part_info *info,
char *out, size_t out_space);
/* value setter functions for custom attributes */
/* pet_bool: */
bool (*custom_attribute_toggle)(struct disk_partitions*,
part_id, size_t attr_no);
/* pet_cardinal: */
bool (*custom_attribute_set_card)(struct disk_partitions*,
part_id, size_t attr_no, long new_val);
/* pet_str or pet_cardinal: */
bool (*custom_attribute_set_str)(struct disk_partitions*,
part_id, size_t attr_no, const char *new_val);
/*
* Optional: additional user information when showing the size
* editor (especially for existing unknown partitions)
*/
const char * (*other_partition_identifier)(const struct
disk_partitions*, part_id);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/* Retrieve device and partition names, e.g. for checking
* against kern.root_device or invoking newfs.
* For disklabel partitions, "part" will be set to the partition
* index (a = 0, b = 1, ...), for others it will get set to -1.
* If dev_name_usage is parent_device_only, the device name will
* not include a partition letter - obviously this only makes a
* difference with disklabel partitions.
* If dev_name_usage is logical_name instead of a device name
* a given name may be returned in NAME= syntax.
* If with_path is true (and the returned value is a device
* node), include the /dev/ prefix in the result string
* (this is ignored when returning NAME= syntax for /etc/fstab).
* If life is true, the device must be made available under
* that name (only makes a difference for NAME=syntax if
* no wedge has been created yet,) - implied for all variants
* where dev_name_usage != logical_name.
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
bool (*get_part_device)(const struct disk_partitions*,
part_id, char *devname, size_t max_devname_len, int *part,
enum dev_name_usage, bool with_path, bool life);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* How big could we resize the given position (start of existing
* partition or free space)
*/
daddr_t (*max_free_space_at)(const struct disk_partitions*, daddr_t);
/*
* Provide a list of free spaces usable for further partitioning,
* assuming the given partition alignment.
* If start is > 0 no space with lower sector numbers will
* be found.
* If ignore is > 0, any partition starting at that sector will
* be considered "free", this is used e.g. when moving an existing
* partition around.
*/
size_t (*get_free_spaces)(const struct disk_partitions*,
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
struct disk_part_free_space *result, size_t max_num_result,
daddr_t min_space_size, daddr_t align, daddr_t start,
daddr_t ignore /* -1 */);
/*
* Translate a partition description from a foreign partitioning
* scheme as close as possible to what we can handle in add_partition.
* This mostly adjusts flags and partition type pointers (using
* more lose matching than add_partition would do).
*/
bool (*adapt_foreign_part_info)(
const struct disk_partitions *myself, struct disk_part_info *dest,
const struct disk_partitioning_scheme *src_scheme,
const struct disk_part_info *src);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Update data for an existing partition
*/
bool (*set_part_info)(struct disk_partitions*, part_id,
const struct disk_part_info*, const char **err_msg);
/* Add a new partition and return its part_id. */
part_id (*add_partition)(struct disk_partitions*,
const struct disk_part_info*, const char **err_msg);
/*
* Optional: add a partition from an outer scheme, accept all
* details w/o verification as best as possible.
*/
part_id (*add_outer_partition)(struct disk_partitions*,
const struct disk_part_info*, const char **err_msg);
/* Delete all partitions */
bool (*delete_all_partitions)(struct disk_partitions*);
/* Optional: delete any partitions inside the given range */
bool (*delete_partitions_in_range)(struct disk_partitions*,
daddr_t start, daddr_t size);
/* Delete the specified partition */
bool (*delete_partition)(struct disk_partitions*, part_id,
const char **err_msg);
/*
* Methods for the whole set of partitions
*/
/*
* If this scheme only creates a singly NetBSD partition, which
* then is sub-partitioned (usually by disklabel), this returns a
* pointer to the secondary partition set.
* Otherwise NULL is returned, e.g. when there is no
* NetBSD partition defined (so this might change over time).
* Schemes that NEVER use a secondary scheme set this
* function pointer to NULL.
*
* If force_empty = true, ignore all on-disk contents and just
* create a new disk_partitions structure for the secondary scheme
* (this is used after deleting all partitions and setting up
* things for "use whole disk").
*
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
* The returned pointer is always owned by the primary partitions,
* caller MUST never free it, but otherwise can manipulate it
* arbitrarily.
*/
struct disk_partitions *
(*secondary_partitions)(struct disk_partitions *, daddr_t start,
bool force_empty);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Write the whole set (in new_state) back to disk.
*/
bool (*write_to_disk)(struct disk_partitions *new_state);
/*
* Try to read partitions from a disk, return NULL if this is not
* the partitioning scheme in use on that device.
* Usually start and len are 0 (and ignored).
* If this is about a part of a disk (like only the NetBSD
* MBR partition, start and len are the valid part of the
* disk.
*/
struct disk_partitions * (*read_from_disk)(const char *,
daddr_t start, daddr_t len, size_t bytes_per_sec,
const struct disk_partitioning_scheme *);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Set up all internal data for a new disk.
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
struct disk_partitions * (*create_new_for_disk)(const char *,
daddr_t start, daddr_t len, bool is_boot_drive,
struct disk_partitions *parent);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional: this scheme may be used to boot from the given disk
*/
bool (*have_boot_support)(const char *disk);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional: try to guess disk geometry from the partition information
*/
int (*guess_disk_geom)(struct disk_partitions *,
int *cyl, int *head, int *sec);
/*
* Return a "cylinder size" (in number of blocks) - whatever that
* means to a particular partitioning scheme.
*/
size_t (*get_cylinder_size)(const struct disk_partitions *);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional: change used geometry info and update internal state
*/
bool (*change_disk_geom)(struct disk_partitions *,
int cyl, int head, int sec);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional:
* Get or set a name for the whole disk (most partitioning
* schemes do not provide this). Used for disklabel "pack names",
* which then may be used for aut-discovery of wedges, so it
* makes sense for the user to edit them.
*/
bool (*get_disk_pack_name)(const struct disk_partitions *,
char *, size_t);
bool (*set_disk_pack_name)(struct disk_partitions *, const char *);
/*
* Optional:
* Find a partition by name (as used in /etc/fstab NAME= entries)
*/
part_id (*find_by_name)(struct disk_partitions *, const char *name);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional:
* Try to guess install target partition from internal data,
* returns true if a safe match was found and sets start/size
* to the target partition.
*/
bool (*guess_install_target)(const struct disk_partitions *,
daddr_t *start, daddr_t *size);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional: verify that the whole set of partitions would be bootable,
* fix up any issues (with user interaction) where needed.
* If "quiet" is true, fix up everything silently if possible
* and never return 1.
* Returns:
* 0: abort install
* 1: re-edit partitions
* 2: use anyway (continue)
*/
int (*post_edit_verify)(struct disk_partitions *, bool quiet);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Optional: called during updates, before mounting the target disk(s),
* before md_pre_update() is called. Can be used to fixup
* partition info for historic errors (e.g. i386 changing MBR
* partition type from 165 to 169), similar to post_edit_verify.
* Returns:
* true if the partition info has changed (write back required)
* false if nothing further needs to be done.
*/
bool (*pre_update_verify)(struct disk_partitions *);
/* Free all the data */
void (*free)(struct disk_partitions*);
/* Wipe all on-disk state, leave blank disk - and free data */
void (*destroy_part_scheme)(struct disk_partitions*);
/* Scheme global cleanup */
void (*cleanup)(void);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
};
/*
* The in-memory representation of all partitions on a concrete disk,
* tied to the partitioning scheme in use.
*
* Concrete schemes will derive from the abstract disk_partitions
* structure (by aggregation), but consumers of the API will only
* ever see this public part.
*/
struct disk_partitions {
/* which partitioning scheme is in use */
const struct disk_partitioning_scheme *pscheme;
/* the disk device this came from (or should go to) */
const char *disk;
/* global/public disk data */
/*
* The basic unit of size used for this disk (all "start",
* "size" and "align" values are in this unit).
*/
size_t bytes_per_sector; /* must be 2^n and >= 512 */
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Valid partitions may have IDs in the range 0 .. num_part (excl.)
*/
part_id num_part;
/*
* If this is a sub-partitioning, the start of the "disk" is
* some arbitrary partition in the parent. Sometimes we need
* to be able to calculate absoluted offsets.
*/
daddr_t disk_start;
/*
* Total size of the disk (usable for partitioning)
*/
daddr_t disk_size;
/*
* Space not yet allocated
*/
daddr_t free_space;
/*
* If this is the secondary partitioning scheme, pointer to
* the outer one. Otherwise NULL.
*/
struct disk_partitions *parent;
};
/*
* A list of partitioning schemes, so we can iterate over everything
* supported (e.g. when partitioning a new disk). NULL terminated.
*/
extern const struct disk_partitioning_scheme **available_part_schemes;
extern size_t num_available_part_schemes;
/*
* Generic reader - query a disk device and read all partitions from it
*/
struct disk_partitions *
partitions_read_disk(const char *, daddr_t disk_size,
size_t bytes_per_sector, bool no_mbr);
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
/*
* Generic part info adaption, may be overridden by individual partitioning
* schemes
*/
bool generic_adapt_foreign_part_info(
const struct disk_partitions *myself, struct disk_part_info *dest,
const struct disk_partitioning_scheme *src_scheme,
const struct disk_part_info *src);
/*
* One time initialization and cleanup
Rework internal data structures and "interfaces to user interface" functions to get rid of all disklabel assumptions. Previously (even for GPT partitioning) struct disklabel was used, which obviously breaks large disk setups. Also many MD parts and parts of the user interface assumed (a) a struct disklabel is used internally to store partitioning information and (b) partitions are named 'a' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
*/
void partitions_init(void);
void partitions_cleanup(void);