NetBSD/usr.sbin/sysinst/partman.c

3408 lines
86 KiB
C
Raw Normal View History

/* $NetBSD: partman.c,v 1.48 2020/01/15 19:08:24 martin Exp $ */
/*
* Copyright 2012 Eugene Lozovoy
* 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.
* 3. The name of Eugene Lozovoy may not be used to endorse
* or promote products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY PIERMONT INFORMATION SYSTEMS INC. ``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 PIERMONT INFORMATION SYSTEMS INC. BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* 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)
* 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
/*
* Copyright 2010 The NetBSD Foundation, Inc.
* 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.
*
* THIS SOFTWARE IS PROVIDED BY PIERMONT INFORMATION SYSTEMS INC. ``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 PIERMONT INFORMATION SYSTEMS INC. BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* 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)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGE.
*
*/
/* partman.c - extended partitioning */
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 <assert.h>
#include <fcntl.h>
#include <errno.h>
#include <libgen.h>
#include <paths.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <util.h>
#include "defs.h"
#include "msg_defs.h"
#include "menu_defs.h"
/* XXX: replace all MAX_* defines with vars that depend on kernel settings */
#define MAX_ENTRIES 96
#define MAX_RAID 8
#define MAX_IN_RAID 48
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 raid_comp {
char name[SSTRSIZE]; /* display name for this component */
struct disk_partitions *parts; /* where is this on? */
part_id id; /* which partition in parts */
bool is_spare; /* is this a spare component? */
};
struct raid_desc {
int enabled;
int blocked;
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' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
int node; /* the N in raid${N} */
int numRow, numCol, numSpare;
int sectPerSU, SUsPerParityUnit, SUsPerReconUnit, raid_level;
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 total_size;
struct raid_comp comp[MAX_IN_RAID];
};
struct raid_desc *raids;
#define MAX_VND 4
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 vnd_desc {
int enabled;
int blocked;
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' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
int node; /* the N in vnd${N} */
char filepath[STRSIZE];
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;
int readonly;
int is_exist;
int manual_geom;
int secsize, nsectors, ntracks, ncylinders;
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 pm_devs *pm; /* device this is on */
part_id pm_part; /* which partition (in pm->parts) */
};
struct vnd_desc *vnds;
#define MAX_CGD 4
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 cgd_desc {
int enabled;
int blocked;
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' ... $MAXPART. Get rid of this and replace it with a quite abstract interface that should be able to deal with all variants in partition storage: - partitions are stored in a (partly abstract) struct disk_partitions and most parts of it are only accessed via accessor functions provided by a "partitioning scheme". - implement partitioning schemes for MBR, disklabel and GPT (with likely RDB [amiga] and Apple Partition Map [mac*] to follow soon) - partitioning schemes may be cascaded, e.g. on x86 when using MBR as "outer partitions", we have disklabel as "inner partitions". - all user interface goes via accessor functions in the partitioning scheme, some of which return pointers to special user interface descriptors (e.g. to allow editing partition flags, which are scheme specific) Overall the user interface changes (in this initial step) are minimal but noticable. A new Anita is needed for automatic test setups - many thanks to Andreas Gustafsson for lots of early testing and a new Anita version, and to Manuel Bouyer for cooperation and tests of the Anita release. This work was sponsored by The NetBSD Foundation, Inc.
2019-06-12 09:20:17 +03:00
int node; /* the N in cgd${N} */
char pm_name[SSTRSIZE];
const char *keygen_type;
const char *verify_type;
const char *enc_type;
const char *iv_type;
int key_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
struct pm_devs *pm; /* device this is on */
part_id pm_part; /* which partition (in pm->parts) */
};
struct cgd_desc *cgds;
#define MAX_LVM_VG 16
#define MAX_LVM_PV 255
#define MAX_LVM_LV 255
struct lvm_pv_reg {
struct pm_devs *pm;
daddr_t start;
};
struct lvm_pv_reg lvm_pvs[MAX_LVM_PV]; /* XXX - make dynamic */
typedef struct pv_t {
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 pm_devs *pm;
char pm_name[SSTRSIZE];
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
part_id pm_part;
int metadatasize;
int metadatacopies;
int labelsector;
int setphysicalvolumesize;
} pv_t;
typedef struct lv_t {
int blocked;
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;
char name[SSTRSIZE];
int readonly;
int contiguous;
char extents[SSTRSIZE];
int minor;
int mirrors;
int regionsize;
int persistent;
int readahead;
int stripes;
int stripesize;
int zero;
} lv_t;
typedef struct lvms_t {
int enabled;
int blocked;
char name[SSTRSIZE];
int maxlogicalvolumes;
int maxphysicalvolumes;
int physicalextentsize;
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 total_size;
pv_t pv[MAX_LVM_PV];
lv_t lv[MAX_LVM_LV];
} lvms_t;
lvms_t *lvms;
typedef struct structinfo_t {
int max;
uint entry_size;
uint parent_size;
void *entry_first;
void *entry_enabled;
void *entry_blocked;
void *entry_node;
} structinfo_t;
structinfo_t raids_t_info, vnds_t_info, cgds_t_info, lvms_t_info, lv_t_info;
typedef struct pm_upddevlist_adv_t {
const char *create_msg;
int pe_type;
structinfo_t *s;
int sub_num;
struct pm_upddevlist_adv_t *sub;
} pm_upddevlist_adv_t;
#define MAX_MNTS 48
struct {
char dev[STRSIZE];
const char *mnt_opts, *on;
} *mnts;
static int pm_cursel; /* Number of selected entry in main menu */
static int pm_changed; /* flag indicating that we have unsaved changes */
static int pm_raid_curspare; /* XXX: replace by true way */
static int pm_retvalue;
enum { /* RAIDframe menu enum */
PMR_MENU_DEVS, PMR_MENU_DEVSSPARE, PMR_MENU_RAIDLEVEL, PMR_MENU_NUMROW,
PMR_MENU_NUMCOL, PMR_MENU_NUMSPARE, PMR_MENU_SECTPERSU, PMR_MENU_SUSPERPARITYUNIT,
PMR_MENU_SUSPERRECONUNIT, PMR_MENU_REMOVE, PMR_MENU_END
};
enum { /* VND menu enum */
PMV_MENU_FILEPATH, PMV_MENU_EXIST, PMV_MENU_SIZE, PMV_MENU_RO, PMV_MENU_MGEOM,
PMV_MENU_SECSIZE, PMV_MENU_NSECTORS, PMV_MENU_NTRACKS, PMV_MENU_NCYLINDERS,
PMV_MENU_REMOVE, PMV_MENU_END
};
enum { /* CGD menu enum */
PMC_MENU_DEV, PMC_MENU_ENCTYPE, PMC_MENU_KEYSIZE, PMC_MENU_IVTYPE,
PMC_MENU_KEYGENTYPE, PMC_MENU_VERIFYTYPE, PMC_MENU_REMOVE, PMC_MENU_END
};
enum { /* LVM menu enum */
PML_MENU_PV, PML_MENU_NAME, PML_MENU_MAXLOGICALVOLUMES,
PML_MENU_MAXPHYSICALVOLUMES, PML_MENU_PHYSICALEXTENTSIZE,
PML_MENU_REMOVE, PML_MENU_END
};
enum { /* LVM submenu (logical volumes) enum */
PMLV_MENU_NAME, PMLV_MENU_SIZE, PMLV_MENU_READONLY, PMLV_MENU_CONTIGUOUS,
PMLV_MENU_EXTENTS, PMLV_MENU_MINOR, PMLV_MENU_PERSISTENT,
PMLV_MENU_MIRRORS, PMLV_MENU_REGIONSIZE, PMLV_MENU_READAHEAD,
PMLV_MENU_STRIPES, PMLV_MENU_STRIPESIZE, PMLV_MENU_ZERO,
PMLV_MENU_REMOVE, PMLV_MENU_END
};
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_entry pm_dev_list(int);
static int pm_raid_disk_add(menudesc *, void *);
static int pm_raid_disk_del(menudesc *, 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
static int pm_cgd_disk_set(struct cgd_desc *, struct part_entry *);
static int pm_mount(struct pm_devs *, int);
static int pm_upddevlist(menudesc *, 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
static void pm_select(struct pm_devs *);
static void
pm_edit_size_value(msg prompt_msg, daddr_t cylsec, daddr_t *size)
{
char answer[16], dflt[16];
daddr_t new_size_val, mult;
snprintf(dflt, sizeof dflt, "%" PRIu64 "%s", *size / sizemult,
multname);
msg_prompt_win(prompt_msg, -1, 18, 0, 0, dflt, answer, sizeof answer);
mult = sizemult;
new_size_val = parse_disk_pos(answer, &mult, cylsec, NULL);
if (new_size_val > 0)
*size = new_size_val * mult;
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
}
static const char *
pm_get_mount(struct pm_devs *p, part_id id)
{
if (p->mounted == NULL)
return NULL;
if (id >= p->parts->num_part)
return NULL;
return p->mounted[id];
}
bool pm_set_mount(struct pm_devs *p, part_id id, const char *path);
bool
pm_set_mount(struct pm_devs *p, part_id id, const char *path)
{
if (p->parts == NULL || id >= p->parts->num_part)
return false;
if (p->mounted == NULL) {
p->mounted = calloc(p->parts->num_part, sizeof(char*));
if (p->mounted == NULL)
return false;
}
free(p->mounted[id]);
p->mounted[id] = strdup(path);
return p->mounted[id] != NULL;
}
/* Universal menu for RAID/VND/CGD/LVM entry edit */
static int
pm_edit(int menu_entries_count, void (*menu_fmt)(menudesc *, int, void *),
int (*action)(menudesc *, void *), int (*check_fun)(void *),
void (*entry_init)(void *, void *), void *entry_init_arg,
void *dev_ptr, int dev_ptr_delta, structinfo_t *s)
{
int i, ok = 0;
2019-06-17 20:53:41 +03:00
menu_ent *menu_entries;
if (dev_ptr == NULL) {
/* We should create new device */
for (i = 0; i < s->max && !ok; i++)
if (*(int*)((char*)s->entry_enabled + dev_ptr_delta + s->entry_size * i) == 0) {
dev_ptr = (char*)s->entry_first + dev_ptr_delta + s->entry_size * i;
entry_init(dev_ptr, entry_init_arg);
ok = 1;
}
if (!ok) {
/* We do not have free device slots */
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
hit_enter_to_continue(NULL, MSG_limitcount);
return -1;
}
}
2019-06-17 20:53:41 +03:00
menu_entries = calloc(menu_entries_count, sizeof *menu_entries);
for (i = 0; i < menu_entries_count - 1; i++)
menu_entries[i] = (menu_ent) { .opt_action=action };
menu_entries[i] = (menu_ent) { .opt_name=MSG_fremove,
.opt_flags=OPT_EXIT,
.opt_action=action };
int menu_no = -1;
menu_no = new_menu(NULL, menu_entries, menu_entries_count,
-1, -1, 0, 40, MC_NOCLEAR | MC_SCROLL,
NULL, menu_fmt, NULL, NULL, MSG_DONE);
process_menu(menu_no, dev_ptr);
free_menu(menu_no);
2019-06-17 20:53:41 +03:00
free(menu_entries);
return check_fun(dev_ptr);
}
/* Show filtered partitions menu */
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_entry
pm_dev_list(int type)
{
int dev_num = -1, num_devs = 0;
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 ok;
part_id i;
int menu_no;
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_info info;
menu_ent menu_entries[MAX_DISKS*MAXPARTITIONS];
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_entry disk_entries[MAX_DISKS*MAXPARTITIONS];
struct pm_devs *pm_i;
SLIST_FOREACH(pm_i, &pm_head, l) {
if (pm_i->parts == NULL)
continue;
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
for (i = 0; i < pm_i->parts->num_part; i++) {
ok = false;
if (!pm_i->parts->pscheme->get_part_info(pm_i->parts,
i, &info))
continue;
if (info.flags &
(PTI_WHOLE_DISK|PTI_PSCHEME_INTERNAL|PTI_RAW_PART))
continue;
switch (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
case PM_RAID:
if (info.fs_type == FS_RAID)
ok = 1;
break;
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
case PM_CGD:
if (info.fs_type == FS_CGD)
ok = 1;
break;
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
case PM_LVM:
if (pm_is_lvmpv(pm_i, i, &info))
ok = 1;
break;
}
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
if (!ok)
continue;
if (pm_partusage(pm_i, i, 0) != 0)
continue;
disk_entries[num_devs].dev_ptr = pm_i;
disk_entries[num_devs].id = i;
disk_entries[num_devs].parts = pm_i->parts;
pm_i->parts->pscheme->get_part_device(
pm_i->parts, i, disk_entries[num_devs].fullname,
sizeof disk_entries[num_devs].fullname,
NULL, plain_name, false, true);
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
menu_entries[num_devs] = (struct menu_ent) {
.opt_name = disk_entries[num_devs].fullname,
.opt_action = set_menu_select,
.opt_flags = OPT_EXIT,
};
num_devs++;
}
}
menu_no = new_menu(MSG_avdisks,
menu_entries, num_devs, -1, -1,
(num_devs+1<3)?3:num_devs+1, 13,
MC_SCROLL | MC_NOCLEAR, NULL, NULL, NULL, NULL, MSG_cancel);
if (menu_no == -1)
return (struct part_entry) { };
process_menu(menu_no, &dev_num);
free_menu(menu_no);
if (dev_num < 0 || dev_num >= num_devs)
return (struct part_entry) { };
pm_retvalue = dev_num;
return disk_entries[dev_num];
}
/* Get unused raid*, cgd* or vnd* device */
static int
pm_manage_getfreenode(void *node, const char *d, structinfo_t *s)
{
int i, ii, ok;
char buf[SSTRSIZE];
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 pm_devs *pm_i;
*(int*)node = -1;
for (i = 0; i < s->max; i++) {
ok = 1;
/* Check that node is not already reserved */
for (ii = 0; ii < s->max; ii++) {
if (*(int*)((char*)s->entry_enabled + s->entry_size
* ii) == 0)
continue;
if (*(int*)((char*)s->entry_node + s->entry_size * ii)
== i) {
ok = 0;
break;
}
}
if (! ok)
continue;
/* Check that node is not in the device list */
snprintf(buf, SSTRSIZE, "%s%d", d, i);
SLIST_FOREACH(pm_i, &pm_head, l)
if (! strcmp(pm_i->diskdev, buf)) {
ok = 0;
break;
}
if (ok) {
*(int*)node = i;
return i;
}
}
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
hit_enter_to_continue(NULL, MSG_nofreedev);
return -1;
}
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
/*
* Show a line for a device, usually with full size in the right
* column, alternatively (if != NULL) with no_size_display
* instead in paranthesis (used for error displays or to note
* a action that can be done to this device.
*/
static void
pm_fmt_disk_line(WINDOW *w, const char *line, const char *on,
daddr_t total, const char *no_size_display)
{
char out[STRSIZE], human[6];
if (on != NULL) {
snprintf(out, sizeof out, "%s %s %s", line,
msg_string(MSG_pm_menu_on), on);
line = out;
}
if (no_size_display != NULL) {
wprintw(w, " %-56s (%s)", line, no_size_display);
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
} else {
humanize_number(human, sizeof(human),
2019-07-15 20:17:59 +03:00
total * 512, "",
HN_AUTOSCALE, HN_B | HN_NOSPACE | HN_DECIMAL);
wprintw(w, " %-56s %s", line, human);
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
}
}
/***
RAIDs
***/
static void
pm_raid_menufmt(menudesc *m, int opt, void *arg)
{
int i, ok = 0;
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
char buf[STRSIZE], rdev[STRSIZE], level[STRSIZE], *line;
struct raid_desc *dev_ptr = ((struct part_entry *)arg)[opt].dev_ptr;
if (dev_ptr->enabled == 0)
return;
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
buf[0] = '\0';
sprintf(rdev, "raid%d", dev_ptr->node);
for (i = 0; i < MAX_IN_RAID; i++) {
if (dev_ptr->comp[i].parts != NULL) {
strlcat(buf, dev_ptr->comp[i].name, sizeof buf);
strlcat(buf, " ", sizeof buf);
ok = 1;
}
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
}
if (ok) {
sprintf(level, "%u", dev_ptr->raid_level);
const char *args[] = { rdev, level };
line = str_arg_subst(msg_string(MSG_raid_menufmt),
__arraycount(args), args);
pm_fmt_disk_line(m->mw, line, buf, dev_ptr->total_size, NULL);
free(line);
} else {
pm_fmt_disk_line(m->mw, buf, NULL, 0,
msg_string(MSG_raid_err_menufmt));
}
}
static void
pm_raid_edit_menufmt(menudesc *m, int opt, void *arg)
{
int i;
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
char buf[STRSIZE];
struct raid_desc *dev_ptr = arg;
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
buf[0] = '\0';
switch (opt) {
case PMR_MENU_DEVS:
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
strlcpy(buf, msg_string(MSG_raid_disks_fmt),
sizeof buf);
strlcat(buf, ": ", sizeof buf);
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
for (i = 0; i < MAX_IN_RAID; i++) {
if (dev_ptr->comp[i].parts == NULL ||
dev_ptr->comp[i].is_spare)
continue;
strlcat(buf, " ", sizeof buf);
strlcat(buf, dev_ptr->comp[i].name, sizeof buf);
}
wprintw(m->mw, "%s", buf);
break;
case PMR_MENU_DEVSSPARE:
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
strlcpy(buf, msg_string(MSG_raid_spares_fmt),
sizeof buf);
strlcat(buf, ": ", sizeof buf);
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
for (i = 0; i < MAX_IN_RAID; i++) {
if (dev_ptr->comp[i].parts == NULL ||
!dev_ptr->comp[i].is_spare)
continue;
strlcat(buf, " ", sizeof buf);
strlcat(buf, dev_ptr->comp[i].name, sizeof buf);
}
wprintw(m->mw, "%s", buf);
break;
case PMR_MENU_RAIDLEVEL:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_level_fmt),
dev_ptr->raid_level);
break;
case PMR_MENU_NUMROW:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_numrow_fmt), dev_ptr->numRow);
break;
case PMR_MENU_NUMCOL:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_numcol_fmt), dev_ptr->numCol);
break;
case PMR_MENU_NUMSPARE:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_numspare_fmt),
dev_ptr->numSpare);
break;
case PMR_MENU_SECTPERSU:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_sectpersu_fmt),
dev_ptr->sectPerSU);
break;
case PMR_MENU_SUSPERPARITYUNIT:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_superpar_fmt),
dev_ptr->SUsPerParityUnit);
break;
case PMR_MENU_SUSPERRECONUNIT:
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
wprintw(m->mw, "%s: %u",
msg_string(MSG_raid_superrec_fmt),
dev_ptr->SUsPerReconUnit);
break;
}
}
static int
pm_raid_set_value(menudesc *m, void *arg)
{
int retvalue = -1;
int *out_var = NULL;
char buf[SSTRSIZE];
const char *msg_to_show = NULL;
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 raid_desc *dev_ptr = arg;
static menu_ent menuent_disk_adddel[] = {
{ .opt_name=MSG_add, .opt_flags=OPT_EXIT,
.opt_action=pm_raid_disk_add },
{ .opt_name=MSG_remove, .opt_flags=OPT_EXIT,
.opt_action=pm_raid_disk_del }
};
static int menu_disk_adddel = -1;
if (menu_disk_adddel == -1) {
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
menu_disk_adddel = new_menu(NULL, menuent_disk_adddel,
__arraycount(menuent_disk_adddel),
-1, -1, 0, 10, MC_NOCLEAR, NULL, NULL, NULL, NULL,
MSG_cancel);
}
switch (m->cursel) {
case PMR_MENU_DEVS:
pm_raid_curspare = 0;
process_menu(menu_disk_adddel, dev_ptr);
return 0;
case PMR_MENU_DEVSSPARE:
pm_raid_curspare = 1;
process_menu(menu_disk_adddel, dev_ptr);
return 0;
case PMR_MENU_RAIDLEVEL:
process_menu(MENU_raidlevel, &retvalue);
if (retvalue >= 0)
dev_ptr->raid_level = retvalue;
return 0;
case PMR_MENU_NUMROW:
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
hit_enter_to_continue(NULL, MSG_raid_nomultidim);
return 0;
#if 0 /* notyet */
msg_to_show = MSG_raid_numrow_ask;
out_var = &(dev_ptr->numRow);
break;
#endif
case PMR_MENU_NUMCOL:
msg_to_show = MSG_raid_numcol_ask;
out_var = &(dev_ptr->numCol);
break;
case PMR_MENU_NUMSPARE:
msg_to_show = MSG_raid_numspare_ask;
out_var = &(dev_ptr->numSpare);
break;
case PMR_MENU_SECTPERSU:
msg_to_show = MSG_raid_sectpersu_ask;
out_var = &(dev_ptr->sectPerSU);
break;
case PMR_MENU_SUSPERPARITYUNIT:
msg_to_show = MSG_raid_superpar_ask;
out_var = &(dev_ptr->SUsPerParityUnit);
break;
case PMR_MENU_SUSPERRECONUNIT:
msg_to_show = MSG_raid_superrec_ask;
out_var = &(dev_ptr->SUsPerReconUnit);
break;
case PMR_MENU_REMOVE:
dev_ptr->enabled = 0;
return 0;
}
if (out_var == NULL || msg_to_show == NULL)
return -1;
snprintf(buf, SSTRSIZE, "%d", *out_var);
msg_prompt_win(msg_to_show, -1, 18, 0, 0, buf, buf, SSTRSIZE);
if (atoi(buf) >= 0)
*out_var = atoi(buf);
return 0;
}
static void
pm_raid_init(void *arg, void *none)
{
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 raid_desc *dev_ptr = arg;
memset(dev_ptr, 0, sizeof(*dev_ptr));
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
*dev_ptr = (struct raid_desc) {
.enabled = 1,
.blocked = 0,
.sectPerSU = 32,
.SUsPerParityUnit = 1,
.SUsPerReconUnit = 1,
};
}
static int
pm_raid_check(void *arg)
{
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
size_t i, dev_num = 0;
daddr_t min_size = 0, cur_size = 0;
struct raid_desc *dev_ptr = arg;
struct disk_part_info info;
struct disk_partitions *parts;
if (dev_ptr->blocked)
return 0;
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
for (i = 0; i < MAX_IN_RAID; i++) {
if (dev_ptr->comp[i].parts != NULL) {
parts = dev_ptr->comp[i].parts;
if (!parts->pscheme->get_part_info(parts,
dev_ptr->comp[i].id, &info))
continue;
cur_size = info.size;
if (cur_size < min_size || dev_num == 0)
min_size = cur_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
if (dev_ptr->comp[i].is_spare)
continue;
dev_num++;
}
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
}
/* Calculate sum of available space */
if (dev_num > 0) {
switch (dev_ptr->raid_level) {
case 0:
dev_ptr->total_size = min_size * dev_num;
break;
case 1:
dev_ptr->total_size = min_size;
break;
case 4:
case 5:
dev_ptr->total_size = min_size * (dev_num - 1);
break;
}
pm_manage_getfreenode(&(dev_ptr->node), "raid", &raids_t_info);
if (dev_ptr->node < 0)
dev_ptr->enabled = 0;
}
else
dev_ptr->enabled = 0;
return dev_ptr->enabled;
}
static int
pm_raid_disk_add(menudesc *m, void *arg)
{
int i;
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 raid_desc *dev_ptr = arg;
struct part_entry disk_entrie = pm_dev_list(PM_RAID);
if (pm_retvalue < 0)
return pm_retvalue;
for (i = 0; i < MAX_IN_RAID; i++)
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
if (dev_ptr->comp[i].parts == NULL) {
dev_ptr->comp[i].parts = disk_entrie.parts;
dev_ptr->comp[i].id = disk_entrie.id;
dev_ptr->comp[i].is_spare = pm_raid_curspare;
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
strlcpy(dev_ptr->comp[i].name, disk_entrie.fullname,
sizeof dev_ptr->comp[i].name);
if (pm_raid_curspare)
dev_ptr->numSpare++;
else
dev_ptr->numCol++;
dev_ptr->numRow = 1;
break;
}
return 0;
}
static int
pm_raid_disk_del(menudesc *m, void *arg)
{
int retvalue = -1, num_devs = 0;
int i, pm_cur;
int menu_no;
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 raid_desc *dev_ptr = arg;
menu_ent menu_entries[MAX_IN_RAID];
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_entry submenu_args[MAX_IN_RAID];
for (i = 0; i < MAX_IN_RAID; i++) {
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
if (dev_ptr->comp[i].parts == NULL ||
dev_ptr->comp[i].is_spare != pm_raid_curspare)
continue;
menu_entries[num_devs] = (struct menu_ent) {
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
.opt_name = dev_ptr->comp[i].name,
.opt_action = set_menu_select,
.opt_flags = OPT_EXIT,
};
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
submenu_args[num_devs].dev_ptr = dev_ptr;
submenu_args[num_devs].index = i;
num_devs++;
}
menu_no = new_menu(MSG_raid_disks,
menu_entries, num_devs, -1, -1,
(num_devs+1<3)?3:num_devs+1, 13,
MC_SCROLL | MC_NOCLEAR, NULL, NULL, NULL, NULL, MSG_cancel);
if (menu_no == -1)
return -1;
process_menu(menu_no, &retvalue);
free_menu(menu_no);
if (retvalue < 0 || retvalue >= num_devs)
return -1;
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
pm_cur = submenu_args[retvalue].index;
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
if (dev_ptr->comp[pm_cur].is_spare)
dev_ptr->numSpare--;
else
dev_ptr->numCol--;
dev_ptr->numRow = (dev_ptr->numCol)?1:0;
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
dev_ptr->comp[pm_cur].parts = NULL;
return 0;
}
static int
pm_raid_commit(void)
{
int i, ii;
FILE *f;
char f_name[STRSIZE], devname[STRSIZE];
for (i = 0; i < MAX_RAID; i++) {
if (! pm_raid_check(&raids[i]))
continue;
/* Generating configure file for our raid */
snprintf(f_name, SSTRSIZE, "/tmp/raid.%d.conf", raids[i].node);
f = fopen(f_name, "w");
if (f == NULL) {
endwin();
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)fprintf(stderr,
"Could not open %s for writing\n", f_name);
if (logfp)
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)fprintf(logfp,
"Could not open %s for writing\n", f_name);
return 1;
}
scripting_fprintf(NULL, "cat <<EOF >%s\n", f_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
scripting_fprintf(f, "START array\n%d %d %d\n",
raids[i].numRow, raids[i].numCol, raids[i].numSpare);
scripting_fprintf(f, "\nSTART disks\n");
for (ii = 0; ii < MAX_IN_RAID; ii++) {
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
if (raids[i].comp[ii].parts != NULL &&
!raids[i].comp[ii].is_spare) {
strcpy(devname, raids[i].comp[ii].name);
if (raids[i].comp[ii].parts != NULL &&
raids[i].comp[ii].id != NO_PART) {
/* wedge may have moved */
raids[i].comp[ii].parts->pscheme->
get_part_device(
raids[i].comp[ii].parts,
raids[i].comp[ii].id,
devname, sizeof devname, NULL,
logical_name, true, true);
raids[i].comp[ii].parts->pscheme->
get_part_device(
raids[i].comp[ii].parts,
raids[i].comp[ii].id,
raids[i].comp[ii].name,
sizeof raids[i].comp[ii].name,
NULL, plain_name, true, true);
}
scripting_fprintf(f, "%s\n", devname);
}
}
scripting_fprintf(f, "\nSTART spare\n");
for (ii = 0; ii < MAX_IN_RAID; ii++) {
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
if (raids[i].comp[ii].parts != NULL &&
raids[i].comp[ii].is_spare) {
strcpy(devname, raids[i].comp[ii].name);
if (raids[i].comp[ii].parts != NULL &&
raids[i].comp[ii].id != NO_PART) {
/* wedge may have moved */
raids[i].comp[ii].parts->pscheme->
get_part_device(
raids[i].comp[ii].parts,
raids[i].comp[ii].id,
devname, sizeof devname, NULL,
logical_name, true, true);
raids[i].comp[ii].parts->pscheme->
get_part_device(
raids[i].comp[ii].parts,
raids[i].comp[ii].id,
raids[i].comp[ii].name,
sizeof raids[i].comp[ii].name,
NULL, plain_name, true, true);
}
scripting_fprintf(f, "%s\n",
devname);
}
}
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
scripting_fprintf(f, "\nSTART layout\n%d %d %d %d\n",
raids[i].sectPerSU, raids[i].SUsPerParityUnit,
raids[i].SUsPerReconUnit, raids[i].raid_level);
scripting_fprintf(f, "\nSTART queue\nfifo 100\n\n");
scripting_fprintf(NULL, "EOF\n");
fclose (f);
fflush(NULL);
/* Raid initialization */
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
if (run_program(RUN_DISPLAY | RUN_PROGRESS,
"raidctl -C %s raid%d", f_name, raids[i].node) == 0 &&
run_program(RUN_DISPLAY | RUN_PROGRESS,
"raidctl -I %d raid%d", rand(), raids[i].node) == 0 &&
run_program(RUN_DISPLAY | RUN_PROGRESS,
"raidctl -vi raid%d", raids[i].node) == 0 &&
run_program(RUN_DISPLAY | RUN_PROGRESS,
"raidctl -v -A yes raid%d", raids[i].node) == 0) {
/*
* RAID creation done, remove it from list to
* prevent its repeated reinitialization
*/
raids[i].blocked = 1;
}
}
return 0;
}
/***
VND
***/
static void
pm_vnd_menufmt(menudesc *m, int opt, void *arg)
{
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 vnd_desc *dev_ptr = ((struct part_entry *)arg)[opt].dev_ptr;
char dev[STRSIZE];
if (dev_ptr->enabled == 0)
return;
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
sprintf(dev, "vnd%d", dev_ptr->node);
if (strlen(dev_ptr->filepath) < 1)
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
pm_fmt_disk_line(m->mw, dev, NULL,
0, msg_string(MSG_vnd_err_menufmt));
else if (dev_ptr->is_exist)
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
pm_fmt_disk_line(m->mw, dev, dev_ptr->filepath,
0, msg_string(MSG_vnd_assign));
else
pm_fmt_disk_line(m->mw, dev, dev_ptr->filepath,
dev_ptr->size, NULL);
}
static int
max_msg_length(const msg *p, size_t cnt)
{
int len, m = 0;
while (cnt > 0) {
len = strlen(msg_string(*p));
if (len > m)
m = len;
cnt--; p++;
}
return m;
}
static void
pm_vnd_edit_menufmt(menudesc *m, int opt, void *arg)
{
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 vnd_desc *dev_ptr = arg;
char buf[SSTRSIZE];
strcpy(buf, "-");
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
static int lcol_width;
if (lcol_width == 0) {
static const msg labels[] = {
MSG_vnd_path_fmt, MSG_vnd_assign_fmt, MSG_vnd_size_fmt,
MSG_vnd_ro_fmt, MSG_vnd_geom_fmt, MSG_vnd_bps_fmt,
MSG_vnd_spt_fmt, MSG_vnd_tpc_fmt, MSG_vnd_cyl_fmt
};
lcol_width = max_msg_length(labels, __arraycount(labels)) + 3;
}
switch (opt) {
case PMV_MENU_FILEPATH:
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_path_fmt), dev_ptr->filepath);
break;
case PMV_MENU_EXIST:
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_assign_fmt),
dev_ptr->is_exist?
msg_string(MSG_No) : msg_string(MSG_Yes));
break;
case PMV_MENU_SIZE:
if (!dev_ptr->is_exist)
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
snprintf(buf, SSTRSIZE, "%" PRIu64,
dev_ptr->size / sizemult);
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_size_fmt), buf);
break;
case PMV_MENU_RO:
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_ro_fmt),
dev_ptr->readonly?
msg_string(MSG_Yes) : msg_string(MSG_No));
break;
case PMV_MENU_MGEOM:
if (!dev_ptr->is_exist)
snprintf(buf, SSTRSIZE, "%s",
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
dev_ptr->manual_geom?
msg_string(MSG_Yes) : msg_string(MSG_No));
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_geom_fmt), buf);
break;
case PMV_MENU_SECSIZE:
if (dev_ptr->manual_geom && !dev_ptr->is_exist)
snprintf(buf, SSTRSIZE, "%d", dev_ptr->secsize);
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_bps_fmt), buf);
break;
case PMV_MENU_NSECTORS:
if (dev_ptr->manual_geom && !dev_ptr->is_exist)
snprintf(buf, SSTRSIZE, "%d",
dev_ptr->nsectors);
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_spt_fmt), buf);
break;
case PMV_MENU_NTRACKS:
if (dev_ptr->manual_geom && !dev_ptr->is_exist)
snprintf(buf, SSTRSIZE, "%d", dev_ptr->ntracks);
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_tpc_fmt), buf);
break;
case PMV_MENU_NCYLINDERS:
if (dev_ptr->manual_geom && !dev_ptr->is_exist)
snprintf(buf, SSTRSIZE, "%d",
dev_ptr->ncylinders);
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
wprintw(m->mw, "%*s %s", -lcol_width,
msg_string(MSG_vnd_cyl_fmt), buf);
break;
}
}
static int
pm_vnd_set_value(menudesc *m, void *arg)
{
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 vnd_desc *dev_ptr = arg;
char buf[STRSIZE];
const char *msg_to_show = NULL;
int *out_var = NULL;
switch (m->cursel) {
case PMV_MENU_FILEPATH:
msg_prompt_win(MSG_vnd_path_ask, -1, 18, 0, 0,
dev_ptr->filepath, dev_ptr->filepath, STRSIZE);
if (dev_ptr->filepath[0] != '/') {
strlcpy(buf, dev_ptr->filepath, MOUNTLEN);
snprintf(dev_ptr->filepath, MOUNTLEN, "/%s",
buf);
}
if (dev_ptr->filepath[strlen(dev_ptr->filepath) - 1]
== '/')
dev_ptr->filepath[strlen(dev_ptr->filepath)
- 1] = '\0';
return 0;
case PMV_MENU_EXIST:
dev_ptr->is_exist = !dev_ptr->is_exist;
return 0;
case PMV_MENU_SIZE:
if (dev_ptr->is_exist)
return 0;
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
pm_edit_size_value(MSG_vnd_size_ask, pm->dlcylsize,
&dev_ptr->size);
break;
case PMV_MENU_RO:
dev_ptr->readonly = !dev_ptr->readonly;
return 0;
case PMV_MENU_MGEOM:
if (dev_ptr->is_exist)
return 0;
dev_ptr->manual_geom = !dev_ptr->manual_geom;
return 0;
case PMV_MENU_SECSIZE:
if (!dev_ptr->manual_geom || dev_ptr->is_exist)
return 0;
msg_to_show = MSG_vnd_bps_ask;
out_var = &(dev_ptr->secsize);
break;
case PMV_MENU_NSECTORS:
if (!dev_ptr->manual_geom || dev_ptr->is_exist)
return 0;
msg_to_show = MSG_vnd_spt_ask;
out_var = &(dev_ptr->nsectors);
break;
case PMV_MENU_NTRACKS:
if (!dev_ptr->manual_geom || dev_ptr->is_exist)
return 0;
msg_to_show = MSG_vnd_tpc_ask;
out_var = &(dev_ptr->ntracks);
break;
case PMV_MENU_NCYLINDERS:
if (!dev_ptr->manual_geom || dev_ptr->is_exist)
return 0;
msg_to_show = MSG_vnd_cyl_ask;
out_var = &(dev_ptr->ncylinders);
break;
case PMV_MENU_REMOVE:
dev_ptr->enabled = 0;
return 0;
}
if (out_var == NULL || msg_to_show == NULL)
return -1;
snprintf(buf, SSTRSIZE, "%d", *out_var);
msg_prompt_win(msg_to_show, -1, 18, 0, 0, buf, buf, SSTRSIZE);
if (atoi(buf) >= 0)
*out_var = atoi(buf);
return 0;
}
static void
pm_vnd_init(void *arg, void *none)
{
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 vnd_desc *dev_ptr = arg;
memset(dev_ptr, 0, sizeof(*dev_ptr));
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
*dev_ptr = (struct vnd_desc) {
.enabled = 1,
.blocked = 0,
.filepath[0] = '\0',
.is_exist = 0,
.size = 1024,
.readonly = 0,
.manual_geom = 0,
.secsize = 512,
.nsectors = 18,
.ntracks = 2,
.ncylinders = 80
};
}
static int
pm_vnd_check(void *arg)
{
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 vnd_desc *dev_ptr = arg;
if (dev_ptr->blocked)
return 0;
if (strlen(dev_ptr->filepath) < 1 ||
dev_ptr->size < 1)
dev_ptr->enabled = 0;
else {
pm_manage_getfreenode(&(dev_ptr->node), "vnd", &vnds_t_info);
if (dev_ptr->node < 0)
dev_ptr->enabled = 0;
}
return dev_ptr->enabled;
}
/* XXX: vndconfig always return 0? */
static int
pm_vnd_commit(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
int i, error;
char r_o[3], buf[MOUNTLEN+3], resultpath[STRSIZE];
const char *mpath, *mp_suit = NULL, *rp;
struct pm_devs *pm_i, *pm_suit = NULL;
part_id id, part_suit = NO_PART;
struct disk_part_info info;
for (i = 0; i < MAX_VND; i++) {
error = 0;
if (! pm_vnd_check(&vnds[i]))
continue;
/* Trying to assign target device */
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
SLIST_FOREACH(pm_i, &pm_head, l) {
for (id = 0; id < pm_i->parts->num_part; id++) {
if (!pm_i->parts->pscheme->get_part_info(
pm_i->parts, id, &info))
continue;
if (info.flags & (PTI_SEC_CONTAINER|
PTI_WHOLE_DISK|PTI_PSCHEME_INTERNAL|
PTI_RAW_PART))
continue;
if (info.last_mounted == NULL ||
info.last_mounted[0] == 0)
continue;
mpath = info.last_mounted;
strcpy(buf, mpath);
if (buf[strlen(buf)-1] != '/')
strcat(buf, "/");
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
if (strstr(vnds[i].filepath, buf) !=
vnds[i].filepath)
continue;
if (part_suit == NO_PART || pm_suit == NULL ||
strlen(buf) > strlen(mp_suit)) {
part_suit = id;
pm_suit = pm_i;
mp_suit = mpath;
}
}
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
}
if (part_suit == NO_PART || pm_suit == NULL ||
mp_suit == NULL)
continue;
/* Mounting assigned partition and try to get real file path*/
if (pm_mount(pm_suit, part_suit) != 0)
continue;
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
rp = pm_get_mount(pm_suit, part_suit);
snprintf(resultpath, STRSIZE, "%s/%s",
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
rp,
&(vnds[i].filepath[strlen(rp)]));
strcpy(r_o, vnds[i].readonly?"-r":"");
/* If this is a new image */
if (!vnds[i].is_exist) {
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
run_program(RUN_DISPLAY | RUN_PROGRESS, "mkdir -p %s ",
dirname(resultpath));
if (error == 0)
error = run_program(RUN_DISPLAY | RUN_PROGRESS,
"dd if=/dev/zero of=%s bs=1m "
"count=% " PRIi64 " progress=100 "
"msgfmt=human",
resultpath, vnds[i].size*(MEG/512));
}
if (error)
continue;
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
/* If this is a new image with manual geometry */
if (!vnds[i].is_exist && vnds[i].manual_geom)
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
error = run_program(RUN_DISPLAY | RUN_PROGRESS,
"vndconfig %s vnd%d %s %d %d %d %d",
r_o, vnds[i].node,
resultpath, vnds[i].secsize,
vnds[i].nsectors,
vnds[i].ntracks, vnds[i].ncylinders);
else
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
/* If this is a existing image or image without manual
* geometry */
error = run_program(RUN_DISPLAY | RUN_PROGRESS,
"vndconfig %s vnd%d %s",
r_o, vnds[i].node, resultpath);
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
if (error == 0) {
vnds[i].blocked = 1;
vnds[i].pm_part = part_suit;
vnds[i].pm = pm_suit;
vnds[i].pm->blocked++;
}
}
return 0;
}
/***
CGD
***/
static void
pm_cgd_menufmt(menudesc *m, int opt, void *arg)
{
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 cgd_desc *dev_ptr = ((struct part_entry *)arg)[opt].dev_ptr;
char desc[STRSIZE];
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_info info;
if (dev_ptr->enabled == 0)
return;
if (dev_ptr->pm == NULL)
wprintw(m->mw, "%s", msg_string(MSG_cgd_err_menufmt));
else {
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
snprintf(desc, sizeof desc, "cgd%d (%s-%d)",
dev_ptr->node, dev_ptr->enc_type, dev_ptr->key_size);
dev_ptr->pm->parts->pscheme->get_part_info(dev_ptr->pm->parts,
dev_ptr->pm_part, &info);
pm_fmt_disk_line(m->mw, desc, dev_ptr->pm_name,
info.size, NULL);
}
}
static void
pm_cgd_edit_menufmt(menudesc *m, int opt, void *arg)
{
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 cgd_desc *dev_ptr = arg;
switch (opt) {
case PMC_MENU_DEV:
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
wprintw(m->mw, "%-15s: %s",
msg_string(MSG_cgd_dev_fmt), dev_ptr->pm_name);
break;
case PMC_MENU_ENCTYPE:
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
wprintw(m->mw, "%-15s: %s",
msg_string(MSG_cgd_enc_fmt), dev_ptr->enc_type);
break;
case PMC_MENU_KEYSIZE:
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
wprintw(m->mw, "%-15s: %d",
msg_string(MSG_cgd_key_fmt), dev_ptr->key_size);
break;
case PMC_MENU_IVTYPE:
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
wprintw(m->mw, "%-15s: %s",
msg_string(MSG_cgd_iv_fmt), dev_ptr->iv_type);
break;
case PMC_MENU_KEYGENTYPE:
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
wprintw(m->mw, "%-15s: %s",
msg_string(MSG_cgd_keygen_fmt), dev_ptr->keygen_type);
break;
case PMC_MENU_VERIFYTYPE:
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
wprintw(m->mw, "%-15s: %s",
msg_string(MSG_cgd_verif_fmt), dev_ptr->verify_type);
break;
}
}
static int
pm_cgd_set_value(menudesc *m, void *arg)
{
char *retstring;
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 cgd_desc *dev_ptr = arg;
switch (m->cursel) {
case PMC_MENU_DEV:
pm_cgd_disk_set(dev_ptr, NULL);
return 0;
case PMC_MENU_ENCTYPE:
process_menu(MENU_cgd_enctype, &retstring);
dev_ptr->enc_type = retstring;
2016-12-11 03:56:34 +03:00
if (! strcmp(retstring, "aes-xts"))
dev_ptr->key_size = 256;
if (! strcmp(retstring, "aes-cbc"))
dev_ptr->key_size = 192;
if (! strcmp(retstring, "blowfish-cbc"))
dev_ptr->key_size = 128;
if (! strcmp(retstring, "3des-cbc"))
dev_ptr->key_size = 192;
return 0;
case PMC_MENU_KEYSIZE:
2016-12-11 03:56:34 +03:00
if (! strcmp(dev_ptr->enc_type, "aes-xts"))
dev_ptr->key_size +=
(dev_ptr->key_size < 512)? 256 : -256;
if (! strcmp(dev_ptr->enc_type, "aes-cbc"))
dev_ptr->key_size +=
(dev_ptr->key_size < 256)? 64 : -128;
if (! strcmp(dev_ptr->enc_type, "blowfish-cbc"))
dev_ptr->key_size = 128;
if (! strcmp(dev_ptr->enc_type, "3des-cbc"))
dev_ptr->key_size = 192;
return 0;
case PMC_MENU_IVTYPE:
process_menu(MENU_cgd_ivtype, &retstring);
dev_ptr->iv_type = retstring;
return 0;
case PMC_MENU_KEYGENTYPE:
process_menu(MENU_cgd_keygentype, &retstring);
dev_ptr->keygen_type = retstring;
return 0;
case PMC_MENU_VERIFYTYPE:
process_menu(MENU_cgd_verifytype, &retstring);
dev_ptr->verify_type = retstring;
return 0;
case PMC_MENU_REMOVE:
dev_ptr->enabled = 0;
return 0;
}
return -1;
}
static void
pm_cgd_init(void *arg1, void *arg2)
{
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 cgd_desc *dev_ptr = arg1;
struct part_entry *disk_entrie = arg2;
memset(dev_ptr, 0, sizeof(*dev_ptr));
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
*dev_ptr = (struct cgd_desc) {
.enabled = 1,
.blocked = 0,
.pm = NULL,
.pm_name[0] = '\0',
.pm_part = 0,
.keygen_type = "pkcs5_pbkdf2/sha1",
.verify_type = "disklabel",
2016-12-11 03:56:34 +03:00
.enc_type = "aes-xts",
.iv_type = "encblkno1",
2016-12-11 03:56:34 +03:00
.key_size = 256,
};
if (disk_entrie != NULL) {
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
disk_entrie->parts->pscheme->get_part_device(
disk_entrie->parts, disk_entrie->id,
disk_entrie->fullname, sizeof(disk_entrie->fullname),
NULL, logical_name, false, true);
pm_cgd_disk_set(dev_ptr, disk_entrie);
}
}
static int
pm_cgd_check(void *arg)
{
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 cgd_desc *dev_ptr = arg;
if (dev_ptr->blocked)
return 0;
if (dev_ptr->pm == NULL)
dev_ptr->enabled = 0;
else {
pm_manage_getfreenode(&(dev_ptr->node), "cgd", &cgds_t_info);
if (dev_ptr->node < 0)
dev_ptr->enabled = 0;
}
return dev_ptr->enabled;
}
static int
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
pm_cgd_disk_set(struct cgd_desc *dev_ptr, struct part_entry *disk_entrie)
{
int alloc_disk_entrie = 0;
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
if (disk_entrie == NULL) {
alloc_disk_entrie = 1;
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
disk_entrie = malloc (sizeof(struct part_entry));
if (disk_entrie == NULL)
return -2;
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
*disk_entrie = pm_dev_list(PM_CGD);
if (pm_retvalue < 0) {
free(disk_entrie);
return -1;
}
}
dev_ptr->pm = disk_entrie->dev_ptr;
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
dev_ptr->pm_part = disk_entrie->id;
strlcpy(dev_ptr->pm_name, disk_entrie->fullname, SSTRSIZE);
if (alloc_disk_entrie)
free(disk_entrie);
return 0;
}
int
pm_cgd_edit_new(struct pm_devs *mypm, part_id id)
{
struct part_entry pe = { .id = id, .parts = mypm->parts,
.dev_ptr = mypm, .type = PM_CGD };
return pm_edit(PMC_MENU_END, pm_cgd_edit_menufmt,
pm_cgd_set_value, pm_cgd_check, pm_cgd_init,
&pe, NULL, 0, &cgds_t_info);
}
int
pm_cgd_edit_old(struct part_entry *pe)
{
return pm_edit(PMC_MENU_END, pm_cgd_edit_menufmt,
pm_cgd_set_value, pm_cgd_check, pm_cgd_init,
pe->dev_ptr != NULL ? pe : NULL,
pe->dev_ptr, 0, &cgds_t_info);
}
static int
pm_cgd_commit(void)
{
char devname[STRSIZE];
int i, error = 0;
for (i = 0; i < MAX_CGD; i++) {
if (! pm_cgd_check(&cgds[i]))
continue;
if (run_program(RUN_DISPLAY | RUN_PROGRESS,
"cgdconfig -g -V %s -i %s -k %s -o /tmp/cgd.%d.conf"
" %s %d", cgds[i].verify_type,
cgds[i].iv_type, cgds[i].keygen_type, cgds[i].node,
cgds[i].enc_type, cgds[i].key_size) != 0) {
error++;
continue;
}
if (cgds[i].pm != NULL && cgds[i].pm->parts != NULL) {
/* wedge device names may have changed */
cgds[i].pm->parts->pscheme->get_part_device(
cgds[i].pm->parts, cgds[i].pm_part,
devname, sizeof devname, NULL,
logical_name, true, true);
cgds[i].pm->parts->pscheme->get_part_device(
cgds[i].pm->parts, cgds[i].pm_part,
cgds[i].pm_name, sizeof cgds[i].pm_name, NULL,
plain_name, false, true);
} else {
continue;
}
if (run_program(RUN_DISPLAY | RUN_PROGRESS,
"cgdconfig -V re-enter cgd%d '%s' /tmp/cgd.%d.conf",
cgds[i].node, devname, cgds[i].node) != 0) {
error++;
continue;
}
cgds[i].pm->blocked++;
cgds[i].blocked = 1;
}
return error;
}
/***
LVM
***/
/* Add lvm logical volumes to pm list */
/* XXX: rewrite */
static void
pm_lvm_find(void)
{
int i, ii, already_found;
char dev[STRSIZE];
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 pm_devs *pm_i;
for (i = 0; i < MAX_LVM_VG; i++) {
if (! lvms[i].blocked)
continue;
for (ii = 0; ii < MAX_LVM_LV; ii++) {
if (! lvms[i].lv[ii].blocked || lvms[i].lv[ii].size < 1)
continue;
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
snprintf(dev, STRSIZE, "%s/%s", lvms[i].name,
lvms[i].lv[ii].name);
already_found = 0;
SLIST_FOREACH(pm_i, &pm_head, l)
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
if (!already_found && strcmp(pm_i->diskdev,
dev) == 0) {
pm_i->found = 1;
already_found = 1;
}
if (already_found)
/* We already added this device, skipping */
continue;
pm_new->found = 1;
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
pm_new->ptstart = 0;
pm_new->ptsize = 0;
pm_new->no_part = true;
pm_new->refdev = &lvms[i].lv[ii];
pm_new->sectorsize = 1;
pm_new->dlcylsize = MEG;
strlcpy(pm_new->diskdev, dev, SSTRSIZE);
strlcpy(pm_new->diskdev_descr, dev, STRSIZE);
if (SLIST_EMPTY(&pm_head))
SLIST_INSERT_HEAD(&pm_head, pm_new, l);
else
SLIST_INSERT_AFTER(pm_i, pm_new, l);
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
pm_new = malloc(sizeof (struct pm_devs));
memset(pm_new, 0, sizeof *pm_new);
}
}
}
static int
pm_lvm_disk_add(menudesc *m, void *arg)
{
int i;
lvms_t *dev_ptr = arg;
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_entry disk_entrie = pm_dev_list(PM_LVM);
if (pm_retvalue < 0)
return pm_retvalue;
for (i = 0; i < MAX_LVM_PV; i++) {
if (dev_ptr->pv[i].pm == NULL) {
dev_ptr->pv[i].pm = disk_entrie.dev_ptr;
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
dev_ptr->pv[i].pm_part = disk_entrie.id;
strlcpy(dev_ptr->pv[i].pm_name, disk_entrie.fullname,
sizeof(dev_ptr->pv[i].pm_name));
break;
}
}
pm_retvalue = 1;
return 0;
}
static int
pm_lvm_disk_del(menudesc *m, void *arg)
{
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
int retvalue = -1, num_devs = 0;
size_t i;
int menu_no;
lvms_t *dev_ptr = arg;
menu_ent menu_entries[MAX_LVM_PV];
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_entry submenu_args[MAX_LVM_PV];
for (i = 0; i < MAX_LVM_PV; i++) {
if (dev_ptr->pv[i].pm == NULL)
continue;
menu_entries[num_devs] = (struct menu_ent) {
.opt_name = dev_ptr->pv[i].pm_name,
.opt_action = set_menu_select,
.opt_flags = OPT_EXIT,
};
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
submenu_args[num_devs].index = i;
num_devs++;
}
menu_no = new_menu(MSG_lvm_disks,
menu_entries, num_devs, -1, -1,
(num_devs+1<3)?3:num_devs+1, 13,
MC_SCROLL | MC_NOCLEAR, NULL, NULL, NULL, NULL, MSG_cancel);
if (menu_no == -1)
return -1;
process_menu(menu_no, &retvalue);
free_menu(menu_no);
if (retvalue < 0 || retvalue >= num_devs)
return -1;
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
dev_ptr->pv[submenu_args[retvalue].index].pm = NULL;
return 0;
}
static void
pm_lvm_menufmt(menudesc *m, int opt, void *arg)
{
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
int i, ok = 0;
daddr_t used_size = 0;
char buf1[STRSIZE]; buf1[0] = 0;
char buf2[STRSIZE]; buf2[0] = 0;
char devs[STRSIZE]; devs[0] = 0;
lvms_t *dev_ptr = ((struct part_entry *)arg)[opt].dev_ptr;
if (dev_ptr->enabled == 0)
return;
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
snprintf(buf1, STRSIZE, "VG '%s'", dev_ptr->name);
for (i = 0; i < MAX_LVM_PV; i++)
if (dev_ptr->pv[i].pm != NULL) {
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
strlcat(devs, dev_ptr->pv[i].pm_name, STRSIZE);
strlcat(devs, " ", STRSIZE);
ok = 1;
}
for (i = 0; i < MAX_LVM_LV; i++)
used_size += dev_ptr->lv[i].size;
if (ok) {
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
snprintf(buf2, STRSIZE, "%" PRIi64 "/%" PRIi64,
dev_ptr->total_size - used_size, dev_ptr->total_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
pm_fmt_disk_line(m->mw, buf1, devs, 0, buf2);
} else {
pm_fmt_disk_line(m->mw, buf1, NULL, 0,
msg_string(MSG_lvm_err_menufmt));
}
}
static void
pm_lvm_edit_menufmt(menudesc *m, int opt, void *arg)
{
int i;
char buf[STRSIZE];
lvms_t *dev_ptr = arg;
strlcpy(buf, msg_string(MSG_auto), sizeof buf);
switch (opt) {
case PML_MENU_PV:
buf[0] = '\0';
for (i = 0; i < MAX_LVM_PV; i++) {
if (dev_ptr->pv[i].pm != NULL) {
strlcat(buf, " ", sizeof buf);
strlcat(buf, dev_ptr->pv[i].pm_name,
sizeof buf);
}
}
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvm_disks_fmt), buf);
break;
case PML_MENU_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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvm_name_fmt), dev_ptr->name);
break;
case PML_MENU_MAXLOGICALVOLUMES:
if (dev_ptr->maxlogicalvolumes > 0)
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
snprintf(buf, STRSIZE, "%d",
dev_ptr->maxlogicalvolumes);
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvm_maxlv_fmt), buf);
break;
case PML_MENU_MAXPHYSICALVOLUMES:
if (dev_ptr->maxphysicalvolumes > 0)
snprintf(buf, STRSIZE, "%d",
dev_ptr->maxphysicalvolumes);
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvm_maxpv_fmt), buf);
break;
case PML_MENU_PHYSICALEXTENTSIZE:
if (dev_ptr->physicalextentsize > 0)
snprintf(buf, STRSIZE, "%dM",
dev_ptr->physicalextentsize);
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvm_extsiz_fmt), buf);
break;
}
}
static int
pm_lvm_set_value(menudesc *m, void *arg)
{
char buf[STRSIZE];
const char *msg_to_show = NULL;
int *out_var = NULL;
lvms_t *dev_ptr = arg;
static menu_ent menuent_disk_adddel[] = {
{ .opt_name=MSG_add, .opt_flags=OPT_EXIT,
.opt_action=pm_lvm_disk_add },
{ .opt_name=MSG_remove, .opt_flags=OPT_EXIT,
.opt_action=pm_lvm_disk_del }
};
static int menu_disk_adddel = -1;
if (menu_disk_adddel == -1) {
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
menu_disk_adddel = new_menu(NULL, menuent_disk_adddel,
__arraycount(menuent_disk_adddel),
-1, -1, 0, 10, MC_NOCLEAR, NULL, NULL, NULL, NULL,
MSG_cancel);
}
switch (m->cursel) {
case PML_MENU_PV:
process_menu(menu_disk_adddel, arg);
return 0;
case PML_MENU_NAME:
msg_prompt_win(MSG_lvm_name_ask, -1, 18, 0, 0,
dev_ptr->name, dev_ptr->name, SSTRSIZE);
return 0;
case PML_MENU_MAXLOGICALVOLUMES:
msg_to_show = MSG_lvm_maxlv_ask;
out_var = &(dev_ptr->maxlogicalvolumes);
break;
case PML_MENU_MAXPHYSICALVOLUMES:
msg_to_show = MSG_lvm_maxpv_ask;
out_var = &(dev_ptr->maxphysicalvolumes);
break;
case PML_MENU_PHYSICALEXTENTSIZE:
msg_to_show = MSG_lvm_extsiz_ask;
out_var = &(dev_ptr->physicalextentsize);
break;
case PML_MENU_REMOVE:
dev_ptr->enabled = 0;
return 0;
}
if (out_var == NULL || msg_to_show == NULL)
return -1;
snprintf(buf, SSTRSIZE, "%d", *out_var);
msg_prompt_win(msg_to_show, -1, 18, 0, 0, buf, buf, SSTRSIZE);
if (atoi(buf) >= 0)
*out_var = atoi(buf);
return 0;
}
static void
pm_lvm_init(void *arg, void* none)
{
lvms_t *dev_ptr = arg;
memset(dev_ptr, 0, sizeof *dev_ptr);
*dev_ptr = (struct lvms_t) {
.enabled = 1,
.blocked = 0,
.maxlogicalvolumes = MAX_LVM_PV,
.maxphysicalvolumes = MAX_LVM_LV,
.physicalextentsize = -1,
};
sprintf(dev_ptr->name, "vg%.2d", rand()%100);
}
static int
pm_lvm_check(void *arg)
{
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
size_t i;
bool ok = false;
lvms_t *dev_ptr = arg;
dev_ptr->total_size = 0;
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_info info;
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
for (i = 0; i < MAX_LVM_PV; i++) {
if (dev_ptr->pv[i].pm != NULL) {
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
if (!dev_ptr->pv[i].pm->parts->pscheme->get_part_info(
dev_ptr->pv[i].pm->parts, dev_ptr->pv[i].pm_part,
&info))
continue;
ok = 1;
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
dev_ptr->total_size += info.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
}
if (!ok)
dev_ptr->enabled = 0;
return dev_ptr->enabled;
}
static void
pm_lvmlv_menufmt(menudesc *m, int opt, void *arg)
{
char buf[STRSIZE];
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
lv_t *dev_ptr = ((struct part_entry *)arg)[opt].dev_ptr;
if (dev_ptr->size > 0) {
snprintf(buf, STRSIZE, "'%s'", dev_ptr->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
pm_fmt_disk_line(m->mw, buf, NULL, dev_ptr->size, NULL);
}
}
static void
pm_lvmlv_edit_menufmt(menudesc *m, int opt, void *arg)
{
lv_t *dev_ptr = arg;
char buf[STRSIZE];
strlcpy(buf, msg_string(MSG_auto), STRSIZE);
switch (opt) {
case PMLV_MENU_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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_name_fmt), dev_ptr->name);
break;
case PMLV_MENU_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
wprintw(m->mw, "%-20s: %" PRIi64,
msg_string(MSG_lvmlv_size_fmt), dev_ptr->size);
break;
case PMLV_MENU_READONLY:
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_ro_fmt),
dev_ptr->readonly? msg_string(MSG_Yes) : msg_string(MSG_No));
break;
case PMLV_MENU_CONTIGUOUS:
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_cont_fmt),
dev_ptr->contiguous? msg_string(MSG_Yes) : msg_string(MSG_No));
break;
case PMLV_MENU_EXTENTS:
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_extnum_fmt),
(strlen(dev_ptr->extents) > 0) ?
dev_ptr->extents : msg_string(MSG_auto));
break;
case PMLV_MENU_MINOR:
if (dev_ptr->minor > 0)
snprintf(buf, STRSIZE, "%dK", dev_ptr->minor);
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_minor_fmt), buf);
break;
case PMLV_MENU_MIRRORS:
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
wprintw(m->mw, "%-20s: %d",
msg_string(MSG_lvmlv_mirrors_fmt),
dev_ptr->mirrors);
break;
case PMLV_MENU_REGIONSIZE:
if (dev_ptr->regionsize > 0)
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
snprintf(buf, STRSIZE, "%dM",
dev_ptr->regionsize);
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_regsiz_fmt), buf);
break;
case PMLV_MENU_PERSISTENT:
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_pers_fmt),
dev_ptr->persistent ?
msg_string(MSG_Yes) : msg_string(MSG_No));
break;
case PMLV_MENU_READAHEAD:
if (dev_ptr->readahead > 0)
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
snprintf(buf, STRSIZE, "%d",
dev_ptr->readahead);
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_readahsect_fmt), buf);
break;
case PMLV_MENU_STRIPES:
if (dev_ptr->stripes > 0)
snprintf(buf, STRSIZE, "%d", dev_ptr->stripes);
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_stripes_fmt), buf);
break;
case PMLV_MENU_STRIPESIZE:
if (dev_ptr->stripesize > 0)
snprintf(buf, STRSIZE, "%dK", dev_ptr->stripesize);
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_stripesiz_fmt), buf);
break;
case PMLV_MENU_ZERO:
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
wprintw(m->mw, "%-20s: %s",
msg_string(MSG_lvmlv_zero_fmt),
dev_ptr->zero ?
msg_string(MSG_Yes) : msg_string(MSG_No));
break;
}
}
static int
pm_lvmlv_set_value(menudesc *m, void *arg)
{
char buf[STRSIZE];
const char *msg_to_show = NULL;
int *out_var = NULL;
lv_t *dev_ptr = arg;
switch (m->cursel) {
case PMLV_MENU_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
msg_prompt_win(MSG_lvmlv_name_ask, -1, 18, 0, 0,
dev_ptr->name, dev_ptr->name, SSTRSIZE);
return 0;
case PMLV_MENU_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
pm_edit_size_value(MSG_lvmlv_size_ask,
pm->dlcylsize, &dev_ptr->size); /* XXX cylsize? */
break;
case PMLV_MENU_READONLY:
dev_ptr->readonly = !dev_ptr->readonly;
return 0;
case PMLV_MENU_CONTIGUOUS:
dev_ptr->contiguous = !dev_ptr->contiguous;
return 0;
case PMLV_MENU_EXTENTS:
msg_prompt_win(MSG_lvmlv_extnum_ask, -1, 18, 0, 0,
dev_ptr->extents, dev_ptr->extents, SSTRSIZE);
return 0;
case PMLV_MENU_MINOR:
msg_to_show = MSG_lvmlv_minor_ask;
out_var = &(dev_ptr->minor);
break;
case PMLV_MENU_MIRRORS:
msg_to_show = MSG_lvmlv_mirrors_ask;
out_var = &(dev_ptr->mirrors);
break;
case PMLV_MENU_REGIONSIZE:
msg_to_show = MSG_lvmlv_regsiz_ask;
out_var = &(dev_ptr->regionsize);
break;
case PMLV_MENU_PERSISTENT:
dev_ptr->persistent = !dev_ptr->persistent;
return 0;
case PMLV_MENU_READAHEAD:
msg_to_show = MSG_lvmlv_readahsect_ask;
out_var = &(dev_ptr->readahead);
break;
case PMLV_MENU_STRIPES:
msg_to_show = MSG_lvmlv_stripes_ask;
out_var = &(dev_ptr->stripes);
break;
case PMLV_MENU_STRIPESIZE:
if (dev_ptr->stripesize << 1 > 512)
dev_ptr->stripesize = 4;
else
dev_ptr->stripesize <<= 1;
return 0;
case PMLV_MENU_ZERO:
dev_ptr->zero = !dev_ptr->zero;
return 0;
case PMLV_MENU_REMOVE:
dev_ptr->size = 0;
return 0;
}
if (out_var == NULL || msg_to_show == NULL)
return -1;
snprintf(buf, SSTRSIZE, "%d", *out_var);
msg_prompt_win(msg_to_show, -1, 18, 0, 0, buf, buf, SSTRSIZE);
if (atoi(buf) >= 0)
*out_var = atoi(buf);
return 0;
}
static void
pm_lvmlv_init(void *arg, void *none)
{
lv_t *dev_ptr = arg;
memset(dev_ptr, 0, sizeof *(dev_ptr));
*dev_ptr = (struct lv_t) {
.blocked = 0,
.size = 1024,
.stripesize = 64,
};
sprintf (dev_ptr->name, "lvol%.2d", rand()%100);
}
static int
pm_lvmlv_check(void *arg)
{
lv_t *dev_ptr = arg;
if (dev_ptr->size > 0 && strlen(dev_ptr->name) > 0)
return 1;
else {
dev_ptr->size = 0;
return 0;
}
}
static int
pm_lvm_commit(void)
{
int i, ii, error;
uint used_size = 0;
char params[STRSIZE*3], devs[STRSIZE*3], arg[STRSIZE];
for (i = 0; i < MAX_LVM_VG; i++) {
/* Stage 0: checks */
if (! pm_lvm_check(&lvms[i]))
continue;
for (ii = 0; ii < MAX_LVM_LV; ii++)
used_size += lvms[i].lv[ii].size;
if (used_size > lvms[i].total_size)
continue;
params[0] = '\0';
devs[0] = '\0';
error = 0;
/* Stage 1: creating Physical Volumes (PV's) */
for (ii = 0; ii < MAX_LVM_PV && ! error; ii++)
if (lvms[i].pv[ii].pm != NULL) {
run_program(RUN_SILENT | RUN_ERROR_OK,
"lvm pvremove -ffy /dev/r%s",
(char*)lvms[i].pv[ii].pm_name);
error += run_program(RUN_DISPLAY | RUN_PROGRESS,
"lvm pvcreate -ffy /dev/r%s",
(char*)lvms[i].pv[ii].pm_name);
if (error)
break;
strlcat(devs, " /dev/r", sizeof devs);
strlcat(devs, lvms[i].pv[ii].pm_name,
sizeof devs);
}
if (error)
continue;
/* Stage 2: creating Volume Groups (VG's) */
if (lvms[i].maxlogicalvolumes > 0) {
snprintf(arg, sizeof arg, " -l %d",
lvms[i].maxlogicalvolumes);
strlcat(params, arg, sizeof params);
}
if (lvms[i].maxphysicalvolumes > 0) {
snprintf(arg, sizeof arg, " -p %d",
lvms[i].maxphysicalvolumes);
strlcat(params, arg, sizeof params);
}
if (lvms[i].physicalextentsize > 0) {
snprintf(arg, sizeof arg, " -s %d",
lvms[i].physicalextentsize);
strlcat(params, arg, sizeof params);
}
error += run_program(RUN_DISPLAY | RUN_PROGRESS,
"lvm vgcreate %s %s %s", params, lvms[i].name, devs);
if (error)
continue;
/* Stage 3: creating Logical Volumes (LV's) */
for (ii = 0; ii < MAX_LVM_LV; ii++) {
if (lvms[i].lv[ii].size <= 0)
continue;
params[0] = '\0';
snprintf(arg, sizeof arg, " -C %c",
lvms[i].lv[ii].contiguous?'y':'n');
strlcat(params, arg, sizeof params);
snprintf(arg, sizeof arg, " -M %c",
lvms[i].lv[ii].persistent?'y':'n');
strlcat(params, arg, sizeof params);
snprintf(arg, sizeof arg, " -p %s",
lvms[i].lv[ii].readonly?"r":"rw");
strlcat(params, arg, sizeof params);
snprintf(arg, sizeof arg, " -Z %c",
lvms[i].lv[ii].zero?'y':'n');
strlcat(params, arg, sizeof params);
if (strlen(lvms[i].lv[ii].name) > 0) {
snprintf(arg, sizeof arg, " -n %s",
lvms[i].lv[ii].name);
strlcat(params, arg, sizeof params);
}
if (strlen(lvms[i].lv[ii].extents) > 0) {
snprintf(arg, sizeof arg, " -l %s",
lvms[i].lv[ii].extents);
strlcat(params, arg, sizeof params);
}
if (lvms[i].lv[ii].minor > 0) {
snprintf(arg, sizeof arg, " --minor %d",
lvms[i].lv[ii].minor);
strlcat(params, arg, sizeof params);
}
if (lvms[i].lv[ii].mirrors > 0) {
snprintf(arg, sizeof arg, " -m %d",
lvms[i].lv[ii].mirrors);
strlcat(params, arg, sizeof params);
if (lvms[i].lv[ii].regionsize > 0) {
snprintf(arg, sizeof arg, " -R %d",
lvms[i].lv[ii].regionsize);
strlcat(params, arg, sizeof params);
}
}
if (lvms[i].lv[ii].readahead > 0) {
snprintf(arg, sizeof arg, " -r %d",
lvms[i].lv[ii].readahead);
strlcat(params, arg, sizeof params);
}
if (lvms[i].lv[ii].stripes > 0) {
snprintf(arg, sizeof arg, " -i %d",
lvms[i].lv[ii].stripes);
strlcat(params, arg, sizeof params);
if (lvms[i].lv[ii].stripesize > 0) {
snprintf(arg, sizeof arg, " -I %d",
lvms[i].lv[ii].stripesize);
strlcat(params, arg, sizeof params);
}
}
snprintf(arg, sizeof arg, " -L %" PRIi64 "M",
lvms[i].lv[ii].size);
strlcat(params, arg, sizeof params);
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
error += run_program(RUN_DISPLAY | RUN_PROGRESS,
"lvm lvcreate %s %s", params, lvms[i].name);
}
if (! error) {
lvms[i].blocked = 1;
for (ii = 0; ii < MAX_LVM_PV; ii++)
if (lvms[i].pv[ii].pm != NULL)
lvms[i].pv[ii].pm->blocked++;
for (ii = 0; ii < MAX_LVM_LV; ii++)
if (lvms[i].lv[ii].size > 0)
lvms[i].lv[ii].blocked = 1;
}
}
return 0;
}
/***
Partman generic functions
***/
int
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
pm_getrefdev(struct pm_devs *pm_cur)
{
int i, ii, dev_num, num_devs, num_devs_s;
char descr[MENUSTRSIZE], dev[MENUSTRSIZE] = "";
pm_cur->refdev = NULL;
if (! strncmp(pm_cur->diskdev, "cgd", 3)) {
dev_num = pm_cur->diskdev[3] - '0';
for (i = 0; i < MAX_CGD; i++)
if (cgds[i].blocked && cgds[i].node == dev_num) {
pm_cur->refdev = &cgds[i];
snprintf(descr, sizeof descr,
" (%s, %s-%d)", cgds[i].pm_name,
cgds[i].enc_type, cgds[i].key_size);
strlcat(pm_cur->diskdev_descr, descr,
sizeof(pm_cur->diskdev_descr));
break;
}
} else if (! strncmp(pm_cur->diskdev, "vnd", 3)) {
dev_num = pm_cur->diskdev[3] - '0';
for (i = 0; i < MAX_VND; i++)
if (vnds[i].blocked && vnds[i].node == dev_num) {
pm_cur->refdev = &vnds[i];
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
vnds[i].pm->parts->pscheme->get_part_device(
vnds[i].pm->parts, vnds[i].pm_part,
dev, sizeof dev, NULL, plain_name, false,
true);
snprintf(descr, sizeof descr, " (%s, %s)",
dev, vnds[i].filepath);
strlcat(pm_cur->diskdev_descr, descr,
sizeof(pm_cur->diskdev_descr));
break;
}
} else if (! strncmp(pm_cur->diskdev, "raid", 4)) {
dev_num = pm_cur->diskdev[4] - '0';
for (i = 0; i < MAX_RAID; i++)
if (raids[i].blocked && raids[i].node == dev_num) {
pm_cur->refdev = &raids[i];
num_devs = 0; num_devs_s = 0;
for (ii = 0; ii < MAX_IN_RAID; ii++)
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
if (raids[i].comp[ii].parts != NULL) {
if(raids[i].comp[ii].is_spare)
num_devs_s++;
else
num_devs++;
}
snprintf(descr, sizeof descr,
" (lvl %d, %d disks, %d spare)",
raids[i].raid_level, num_devs,
num_devs_s);
strlcat(pm_cur->diskdev_descr, descr,
sizeof(pm_cur->diskdev_descr));
break;
}
} else
return -1;
return 0;
}
/*
* Enable/disable items in the extended partition disk/partition action
* menu
*/
void
pmdiskentry_enable(menudesc *menu, struct part_entry *pe)
{
int i;
menu_ent *m;
bool enable;
for (i = 0; i < menu->numopts; i++) {
m = &menu->opts[i];
enable = false;
if (m->opt_name == MSG_unconfig) {
if (pe->type == PM_DISK)
enable = ((struct pm_devs *)pe->dev_ptr)
->refdev != NULL;
} else if (m->opt_name == MSG_undo) {
if (pe->type != PM_DISK)
continue;
enable = ((struct pm_devs *)pe->dev_ptr)->unsaved;
} else if (m->opt_name == MSG_switch_parts) {
enable = pm_from_pe(pe)->parts != NULL;
} else {
continue;
}
if (enable)
m->opt_flags &= ~OPT_IGNORE;
else
m->opt_flags |= OPT_IGNORE;
}
}
/* Detect that partition is in use */
int
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
pm_partusage(struct pm_devs *pm_cur, int part_num, int do_del)
{
int i, ii, retvalue = 0;
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_info info;
part_id id;
if (pm_cur->parts == NULL)
return -1; /* nothing can be in use */
if (part_num < 0) {
/* Check all partitions on device */
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
for (id = 0; id < pm_cur->parts->num_part; id++) {
if (!pm_cur->parts->pscheme->get_part_info(
pm_cur->parts, id, &info))
continue;
if (info.flags & (PTI_SEC_CONTAINER|
PTI_WHOLE_DISK|PTI_PSCHEME_INTERNAL|
PTI_RAW_PART))
continue;
retvalue += pm_partusage(pm_cur, id, do_del);
}
return retvalue;
}
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
id = (part_id)part_num;
if (id >= pm_cur->parts->num_part)
return 0;
for (i = 0; i < MAX_CGD; i++)
if (cgds[i].enabled &&
cgds[i].pm == pm_cur &&
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
cgds[i].pm_part == id) {
if (do_del) {
cgds[i].pm = NULL;
cgds[i].pm_name[0] = '\0';
}
return 1;
}
for (i = 0; i < MAX_RAID; i++)
for (ii = 0; ii < MAX_IN_RAID; ii++)
if (raids[i].enabled &&
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
raids[i].comp[ii].parts == pm_cur->parts &&
raids[i].comp[ii].id == id) {
if (do_del)
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
raids[i].comp[ii].parts = NULL;
return 1;
}
for (i = 0; i < MAX_LVM_VG; i++)
for (ii = 0; ii < MAX_LVM_PV; ii++)
if (lvms[i].enabled &&
lvms[i].pv[ii].pm == pm_cur &&
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
lvms[i].pv[ii].pm_part == id) {
if (do_del)
lvms[i].pv[ii].pm = NULL;
return 1;
}
return 0;
}
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
static void
pm_destroy_one(struct pm_devs *pm_i)
{
part_id i;
if (pm_i->parts != NULL) {
if (pm_i->mounted != NULL) {
for (i = 0; i < pm_i->parts->num_part; i++)
free(pm_i->mounted[i]);
}
pm_i->parts->pscheme->free(pm_i->parts);
}
free(pm_i);
}
/* Clean up removed devices */
static int
pm_clean(void)
{
int count = 0;
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 pm_devs *pm_i, *tmp;
SLIST_FOREACH_SAFE(pm_i, &pm_head, l, tmp)
if (! pm_i->found) {
count++;
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
SLIST_REMOVE(&pm_head, pm_i, pm_devs, l);
pm_destroy_one(pm_i);
}
return count;
}
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
/* Free all pm storage */
void
pm_destroy_all(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
struct pm_devs *pm_i, *tmp;
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
if (pm_new != pm)
pm_destroy_one(pm_new);
SLIST_FOREACH_SAFE(pm_i, &pm_head, l, tmp) {
SLIST_REMOVE(&pm_head, pm_i, pm_devs, l);
pm_destroy_one(pm_i);
}
}
void
pm_set_lvmpv(struct pm_devs *my_pm, part_id pno, bool add)
{
size_t i;
struct disk_part_info info;
if (!my_pm->parts->pscheme->get_part_info(my_pm->parts, pno, &info))
return;
if (add) {
for (i = 0; i < __arraycount(lvm_pvs); i++)
if (lvm_pvs[i].pm == NULL && lvm_pvs[i].start == 0)
break;
if (i >= __arraycount(lvm_pvs))
return;
lvm_pvs[i].pm = my_pm;
lvm_pvs[i].start = info.start;
return;
} else {
for (i = 0; i < __arraycount(lvm_pvs); i++)
if (lvm_pvs[i].pm == my_pm &&
lvm_pvs[i].start == info.start)
break;
if (i >= __arraycount(lvm_pvs))
return;
lvm_pvs[i].pm = NULL;
lvm_pvs[i].start = 0;
}
}
bool
pm_is_lvmpv(struct pm_devs *my_pm, part_id id,
const struct disk_part_info *info)
{
size_t i;
for (i = 0; i < __arraycount(lvm_pvs); i++) {
if (lvm_pvs[i].pm != my_pm)
continue;
if (lvm_pvs[i].start == info->start)
return true;
}
return false;
}
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
pm_setfstype(struct pm_devs *pm_cur, part_id id, int fstype, int fs_subtype)
{
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_info info;
if (!pm_cur->parts->pscheme->get_part_info(pm_cur->parts, id, &info))
return;
info.nat_type = pm_cur->parts->pscheme->get_fs_part_type(PT_root,
fstype, fs_subtype);
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
if (info.nat_type == NULL)
return;
info.fs_type = fstype;
info.fs_sub_type = fs_subtype;
pm_cur->parts->pscheme->set_part_info(pm_cur->parts, id, &info, NULL);
}
static 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
pm_select(struct pm_devs *pm_devs_in)
{
pm = pm_devs_in;
if (logfp)
(void)fprintf(logfp, "Partman device: %s\n", pm->diskdev);
}
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
pm_rename(struct pm_devs *pm_cur)
{
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
#if 0 // XXX - convert to custom attribute or handle similar
pm_select(pm_cur);
msg_prompt_win(MSG_packname, -1, 18, 0, 0, pm_cur->bsddiskname,
pm_cur->bsddiskname, sizeof pm_cur->bsddiskname);
#ifndef NO_DISKLABEL
(void) savenewlabel(pm_cur->bsdlabel, MAXPARTITIONS);
#endif
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
}
int
pm_editpart(int part_num)
{
struct partition_usage_set pset = {};
usage_set_from_parts(&pset, pm->parts);
edit_ptn(&(struct menudesc){.cursel = part_num}, &pset);
free_usage_set(&pset);
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
if (checkoverlap(pm->parts)) {
hit_enter_to_continue(MSG_cantsave, NULL);
return -1;
}
pm->unsaved = 1;
return 0;
}
/* Safe erase of disk */
void
pm_shred(struct part_entry *pe, int shredtype)
{
const char *srcdev;
char dev[SSTRSIZE];
struct pm_devs *my_pm;
my_pm = pe->dev_ptr;
if (pe->type == PM_DISK) {
snprintf(dev, sizeof dev,
_PATH_DEV "r%s%c", my_pm->diskdev, 'a' + RAW_PART);
if (pe->parts != NULL) {
pe->parts->pscheme->free(pe->parts);
pe->parts = NULL;
my_pm->parts = NULL;
}
} else if (pe->type == PM_PART) {
pe->parts->pscheme->get_part_device(pe->parts, pe->id,
dev, sizeof dev, NULL, raw_dev_name, true, true);
}
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
switch (shredtype) {
case SHRED_ZEROS:
srcdev = _PATH_DEVZERO;
break;
case SHRED_RANDOM:
srcdev = _PATH_URANDOM;
break;
default:
return;
}
run_program(RUN_DISPLAY | RUN_PROGRESS,
"progress -f %s -b 1m dd bs=1m of=%s", srcdev, dev);
pm_partusage(my_pm, -1, 1);
}
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
#if 0 // XXX
static int
pm_mountall_sort(const void *a, const void *b)
{
return strcmp(mnts[*(const int *)a].on, mnts[*(const int *)b].on);
}
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
#if 0 // XXX
/* Mount all available partitions */
static int
pm_mountall(void)
{
int num_devs = 0;
int mnts_order[MAX_MNTS];
int i, ii, error, ok;
char dev[SSTRSIZE]; dev[0] = '\0';
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 pm_devs *pm_i;
localfs_dev[0] = '\0';
if (mnts == NULL)
mnts = calloc(MAX_MNTS, sizeof(*mnts));
SLIST_FOREACH(pm_i, &pm_head, l) {
ok = 0;
for (i = 0; i < MAXPARTITIONS; i++) {
if (!(pm_i->bsdlabel[i].pi_flags & PIF_MOUNT &&
pm_i->bsdlabel[i].mnt_opts != NULL))
continue;
mnts[num_devs].mnt_opts = pm_i->bsdlabel[i].mnt_opts;
if (strlen(pm_i->bsdlabel[i].mounted) > 0) {
/* Device is already mounted. So, doing mount_null */
strlcpy(mnts[num_devs].dev, pm_i->bsdlabel[i].mounted, MOUNTLEN);
mnts[num_devs].mnt_opts = "-t null";
} else {
pm_getdevstring(dev, SSTRSIZE, pm_i, i);
snprintf(mnts[num_devs].dev, STRSIZE, "/dev/%s", dev);
}
mnts[num_devs].on = pm_i->bsdlabel[i].pi_mount;
if (strcmp(pm_i->bsdlabel[i].pi_mount, "/") == 0) {
/* Use disk with / as a default if the user has
the sets on a local disk */
strlcpy(localfs_dev, pm_i->diskdev, SSTRSIZE);
}
num_devs++;
ok = 1;
}
if (ok)
md_pre_mount(NULL, 0);
}
if (strlen(localfs_dev) < 1) {
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
hit_enter_to_continue(MSG_noroot, NULL);
return -1;
}
for (i = 0; i < num_devs; i++)
mnts_order[i] = i;
qsort(mnts_order, num_devs, sizeof mnts_order[0], pm_mountall_sort);
for (i = 0; i < num_devs; i++) {
ii = mnts_order[i];
make_target_dir(mnts[ii].on);
error = target_mount_do(mnts[ii].mnt_opts, mnts[ii].dev, mnts[ii].on);
if (error)
return error;
}
return 0;
}
#endif
/* Mount partition bypassing ordinary order */
static int
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
pm_mount(struct pm_devs *pm_cur, int part_num)
{
int error = 0;
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
#if 0 // XXX
char buf[MOUNTLEN];
if (strlen(pm_cur->bsdlabel[part_num].mounted) > 0)
return 0;
snprintf(buf, sizeof(buf), "/tmp/%s%c", pm_cur->diskdev,
part_num + 'a');
if (! dir_exists_p(buf))
run_program(RUN_DISPLAY | RUN_PROGRESS, "/bin/mkdir -p %s", buf);
if (pm_cur->bsdlabel[part_num].pi_flags & PIF_MOUNT &&
pm_cur->bsdlabel[part_num].mnt_opts != NULL &&
strlen(pm_cur->bsdlabel[part_num].mounted) < 1)
error += run_program(RUN_DISPLAY | RUN_PROGRESS, "/sbin/mount %s /dev/%s%c %s",
pm_cur->bsdlabel[part_num].mnt_opts,
pm_cur->diskdev, part_num + 'a', buf);
if (error)
pm_cur->bsdlabel[part_num].mounted[0] = '\0';
else {
strlcpy(pm_cur->bsdlabel[part_num].mounted, buf, MOUNTLEN);
pm_cur->blocked++;
}
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
return error;
}
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
pm_umount(struct pm_devs *pm_cur, int part_num)
{
char buf[SSTRSIZE]; buf[0] = '\0';
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
part_id id;
if (part_num < 0)
return;
id = (part_id)part_num;
pm_cur->parts->pscheme->get_part_device(pm_cur->parts, id, buf,
sizeof buf, NULL, plain_name, false, true);
if (run_program(RUN_DISPLAY | RUN_PROGRESS,
"umount -f /dev/%s", buf) == 0) {
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
free(pm_cur->mounted[id]);
pm_cur->mounted[id] = NULL;
if (pm_cur->blocked > 0)
pm_cur->blocked--;
}
}
int
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
pm_unconfigure(struct pm_devs *pm_cur)
{
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
int error = 0;
if (! strncmp(pm_cur->diskdev, "cgd", 3)) {
error = run_program(RUN_DISPLAY | RUN_PROGRESS, "cgdconfig -u %s", pm_cur->diskdev);
if (! error && pm_cur->refdev != NULL) {
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 cgd_desc*)pm_cur->refdev)->pm->blocked--;
((struct cgd_desc*)pm_cur->refdev)->blocked = 0;
}
} else if (! strncmp(pm_cur->diskdev, "vnd", 3)) {
error = run_program(RUN_DISPLAY | RUN_PROGRESS, "vndconfig -u %s", pm_cur->diskdev);
if (! error && pm_cur->refdev != NULL) {
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 vnd_desc*)pm_cur->refdev)->pm->blocked--;
((struct vnd_desc*)pm_cur->refdev)->blocked = 0;
}
} else if (! strncmp(pm_cur->diskdev, "raid", 4)) {
error = run_program(RUN_DISPLAY | RUN_PROGRESS, "raidctl -u %s", pm_cur->diskdev);
if (! error && pm_cur->refdev != NULL) {
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 raid_desc*)pm_cur->refdev)->blocked = 0;
#if 0 // XXX
for (i = 0; i < MAX_IN_RAID; i++)
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
if (((struct raid_desc*)pm_cur->refdev)->comp[i].parts != NULL)
((raids_t*)pm_cur->refdev)->pm[i]->blocked--;
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
}
} else if (! strncmp(pm_cur->diskdev, "dk", 2)) {
if (pm_cur->refdev == NULL)
return -2;
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
/* error = */
run_program(RUN_DISPLAY | RUN_PROGRESS, "dkctl %s delwedge %s",
((struct pm_devs*)pm_cur->refdev)->diskdev, pm_cur->diskdev);
#if 0 // XXX
if (! error) {
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
if (pm_cur->refdev != NULL && ((struct pm_devs*)pm_cur->refdev)->blocked > 0)
((struct pm_devs*)pm_cur->refdev)->blocked--;
sscanf(pm_cur->diskdev, "dk%d", &num);
if (num >= 0 && num < MAX_WEDGES)
wedges[num].allocated = 0;
}
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
} /* XXX: lvm */
else
error = run_program(RUN_DISPLAY | RUN_PROGRESS, "eject -t disk /dev/%sd",
pm_cur->diskdev);
if (!error)
pm_cur->found = 0;
return error;
}
/* Last checks before leaving partition manager */
#if 0
static int
pm_lastcheck(void)
{
FILE *file_tmp = fopen(concat_paths(targetroot_mnt, "/etc/fstab"), "r");
if (file_tmp == NULL)
return 1;
fclose(file_tmp);
return 0;
}
#endif
/* Are there unsaved changes? */
static int
pm_needsave(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
struct pm_devs *pm_i;
SLIST_FOREACH(pm_i, &pm_head, l)
if (pm_i->unsaved) {
/* Oops, we have unsaved changes */
pm_changed = 1;
msg_display(MSG_saveprompt);
return ask_yesno(NULL);
}
return 0;
}
/* Write all changes to disk */
static int
pm_commit(menudesc *m, void *arg)
{
int retcode;
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 pm_devs *pm_i;
struct disk_partitions *secondary;
pm_retvalue = -1;
SLIST_FOREACH(pm_i, &pm_head, l) {
if (! pm_i->unsaved)
continue;
if (pm_i->parts == NULL) {
pm_i->unsaved = false;
continue;
}
if (!pm_i->parts->pscheme->write_to_disk(pm_i->parts)) {
if (logfp)
fprintf(logfp, "partitining error %s\n",
pm_i->diskdev);
return -1;
}
if (pm_i->parts->pscheme->secondary_scheme != NULL) {
secondary = pm_i->parts->pscheme->
secondary_partitions(pm_i->parts, -1, false);
if (secondary != NULL) {
if (!secondary->pscheme->write_to_disk(
secondary)) {
if (logfp)
fprintf(logfp,
"partitining error %s\n",
pm_i->diskdev);
return -1;
}
}
}
}
/* Call all functions that may create new devices */
if ((retcode = pm_raid_commit()) != 0) {
if (logfp)
fprintf(logfp, "RAIDframe configuring error #%d\n", retcode);
return -1;
}
if ((retcode = pm_cgd_commit()) != 0) {
if (logfp)
fprintf(logfp, "CGD configuring error #%d\n", retcode);
return -1;
}
if ((retcode = pm_lvm_commit()) != 0) {
if (logfp)
fprintf(logfp, "LVM configuring error #%d\n", retcode);
return -1;
}
if ((retcode = pm_vnd_commit()) != 0) {
if (logfp)
fprintf(logfp, "VND configuring error #%d\n", retcode);
return -1;
}
if (m != NULL && arg != NULL)
pm_upddevlist(m, arg);
if (logfp)
fflush (logfp);
pm_retvalue = 0;
return 0;
}
#if 0 // XXX
static int
pm_savebootsector(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
struct pm_devs *pm_i;
SLIST_FOREACH(pm_i, &pm_head, l)
if (pm_i->bootable) {
if (! strncmp("raid", pm_i->diskdev, 4))
if (run_program(RUN_DISPLAY | RUN_PROGRESS,
"raidctl -v -A root %s", pm_i->diskdev) != 0) {
if (logfp)
fprintf(logfp, "Error writting RAID bootsector to %s\n",
pm_i->diskdev);
continue;
}
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
if (pm_i->no_part) {
if (pm_i->rootpart < 0 ||
run_program(RUN_DISPLAY | RUN_PROGRESS,
"gpt biosboot -i %d %s",
pm_i->rootpart + 1, pm_i->diskdev) != 0) {
if (logfp)
fprintf(logfp, "Error writting GPT bootsector to %s\n",
pm_i->diskdev);
continue;
}
} else {
pm_select(pm_i);
if (
#ifndef NO_DISKLABEL
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
!check_partitions(pm_i, pm_i->parts) ||
#endif
md_post_newfs() != 0) {
if (logfp)
fprintf(logfp, "Error writting bootsector to %s\n",
pm_i->diskdev);
continue;
}
}
}
return 0;
}
#endif
/* Function for 'Enter'-menu */
static int
pm_submenu(menudesc *m, void *arg)
{
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 pm_devs *pm_cur = NULL;
pm_retvalue = m->cursel + 1;
struct part_entry *cur_pe = (struct part_entry *)arg + m->cursel;
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
switch (cur_pe->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
case PM_DISK:
case PM_PART:
case PM_SPEC:
if (cur_pe->dev_ptr != NULL) {
pm_cur = cur_pe->dev_ptr;
if (pm_cur == NULL)
return -1;
if (pm_cur->blocked) {
clear();
refresh();
msg_display(MSG_wannaunblock);
if (!ask_noyes(NULL))
return -2;
pm_cur->blocked = 0;
}
pm_select(pm_cur);
}
default:
break;
}
switch (cur_pe->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
case PM_DISK:
process_menu(MENU_pmdiskentry, cur_pe);
break;
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
case PM_PART:
process_menu(MENU_pmpartentry, cur_pe);
break;
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
case PM_SPEC:
process_menu(MENU_pmpartentry, cur_pe);
break;
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
case PM_RAID:
pm_edit(PMR_MENU_END, pm_raid_edit_menufmt,
pm_raid_set_value, pm_raid_check, pm_raid_init,
NULL, cur_pe->dev_ptr, 0, &raids_t_info);
break;
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
case PM_VND:
return pm_edit(PMV_MENU_END, pm_vnd_edit_menufmt,
pm_vnd_set_value, pm_vnd_check, pm_vnd_init,
NULL, cur_pe->dev_ptr, 0, &vnds_t_info);
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
case PM_CGD:
pm_cgd_edit_old(cur_pe);
break;
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
case PM_LVM:
return pm_edit(PML_MENU_END, pm_lvm_edit_menufmt,
pm_lvm_set_value, pm_lvm_check, pm_lvm_init,
NULL, cur_pe->dev_ptr, 0, &lvms_t_info);
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
case PM_LVMLV:
return pm_edit(PMLV_MENU_END, pm_lvmlv_edit_menufmt,
pm_lvmlv_set_value, pm_lvmlv_check, pm_lvmlv_init,
NULL, cur_pe->dev_ptr,
cur_pe->dev_ptr_delta, &lv_t_info);
}
return 0;
}
/* Functions that generate menu entries text */
static void
pm_menufmt(menudesc *m, int opt, void *arg)
{
const char *dev_status = "";
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
char buf[STRSIZE], dev[STRSIZE];
part_id part_num = ((struct part_entry *)arg)[opt].id;
struct pm_devs *pm_cur = ((struct part_entry *)arg)[opt].dev_ptr;
struct disk_partitions *parts = ((struct part_entry *)arg)[opt].parts;
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_info info;
const char *mount_point, *fstype;
switch (((struct part_entry *)arg)[opt].type) {
case PM_DISK:
if (pm_cur->blocked)
dev_status = msg_string(MSG_pmblocked);
else if (! pm_cur->unsaved)
dev_status = msg_string(MSG_pmunchanged);
else
dev_status = msg_string(MSG_pmused);
wprintw(m->mw, "%-43.42s %25.24s",
pm_cur->diskdev_descr,
dev_status);
break;
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
case PM_PART:
if (parts->pscheme->get_part_device != NULL)
parts->pscheme->get_part_device(
parts, part_num,
dev, sizeof dev, NULL, plain_name, false,
true);
else
strcpy(dev, "-");
parts->pscheme->get_part_info(parts,
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
part_num, &info);
if (pm_cur->mounted != NULL &&
pm_cur->mounted[part_num] != NULL &&
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
pm_cur->mounted[part_num][0] != 0)
mount_point = msg_string(MSG_pmmounted);
else
mount_point = msg_string(MSG_pmunused);
fstype = getfslabelname(info.fs_type,
info.fs_sub_type);
if (info.last_mounted != NULL) {
snprintf(buf, STRSIZE, "%s (%s) %s",
info.last_mounted, fstype,
mount_point);
pm_fmt_disk_line(m->mw, dev, buf,
info.size, NULL);
} else {
if (fstype != NULL) {
strlcat(dev, " (", sizeof(dev));
strlcat(dev, fstype, sizeof(dev));
strlcat(dev, ")", sizeof(dev));
}
pm_fmt_disk_line(m->mw, dev, NULL,
info.size, NULL);
}
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
break;
case PM_SPEC:
/* XXX ? */
pm_fmt_disk_line(m->mw, pm_cur->diskdev_descr, NULL,
pm_cur->dlsize, NULL);
break;
case PM_RAID:
pm_raid_menufmt(m, opt, arg);
break;
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
case PM_VND:
pm_vnd_menufmt(m, opt, arg);
break;
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
case PM_CGD:
pm_cgd_menufmt(m, opt, arg);
break;
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
case PM_LVM:
pm_lvm_menufmt(m, opt, arg);
break;
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
case PM_LVMLV:
pm_lvmlv_menufmt(m, opt, arg);
break;
}
}
/* Submenu for RAID/LVM/CGD/VND */
static void
pm_upddevlist_adv(menudesc *m, void *arg, int *i,
pm_upddevlist_adv_t *d)
{
int ii;
if (d->create_msg != NULL) {
/* We want to have menu entry that creates a new device */
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_entry *)arg)[*i].type = d->pe_type;
((struct part_entry *)arg)[*i].dev_ptr = NULL;
((struct part_entry *)arg)[*i].dev_ptr_delta = d->s->parent_size * d->sub_num;
m->opts[(*i)++] = (struct menu_ent) {
.opt_name = d->create_msg,
.opt_action = pm_submenu,
};
}
for (ii = 0; ii < d->s->max; ii++) {
if (d->s->entry_enabled == NULL ||
d->s->entry_blocked == NULL ||
*(int*)((char*)d->s->entry_enabled + d->s->entry_size * ii +
d->s->parent_size * d->sub_num) == 0 ||
*(int*)((char*)d->s->entry_blocked + d->s->entry_size * ii +
d->s->parent_size * d->sub_num) != 0)
continue;
/* We have a entry for displaying */
pm_changed = 1;
m->opts[*i] = (struct menu_ent) {
.opt_name = NULL,
.opt_action = pm_submenu,
};
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_entry *)arg)[*i].type = d->pe_type;
((struct part_entry *)arg)[*i].dev_ptr = (char*)d->s->entry_first +
d->s->entry_size * ii + d->s->parent_size * d->sub_num;
(*i)++;
/* We should show submenu for current entry */
if (d->sub != NULL) {
d->sub->sub_num = ii;
pm_upddevlist_adv(m, arg, i, d->sub);
}
}
}
/* Update partman main menu with devices list */
static int
pm_upddevlist(menudesc *m, void *arg)
{
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
int i = 0;
size_t ii;
struct pm_devs *pm_i;
struct disk_partitions *secondary;
const struct disk_partitioning_scheme *ps;
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_info info;
if (arg != NULL)
pm_retvalue = m->cursel + 1;
pm_changed = 0;
/* Mark all devices as not found */
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
SLIST_FOREACH(pm_i, &pm_head, l) {
if (pm_i->parts != NULL && !pm_i->unsaved) {
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
pm_i->parts->pscheme->free(pm_i->parts);
pm_i->parts = NULL;
}
if (pm_i->found > 0)
pm_i->found = 0;
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
}
/* Detect all present devices */
(void)find_disks("partman", false);
if (have_lvm)
pm_lvm_find();
pm_clean();
if (m == NULL || arg == NULL)
return -1;
SLIST_FOREACH(pm_i, &pm_head, l) {
memset(&m->opts[i], 0, sizeof m->opts[i]);
m->opts[i].opt_action = pm_submenu;
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_entry *)arg)[i].dev_ptr = pm_i;
((struct part_entry *)arg)[i].id = NO_PART;
if (pm_i->no_part)
((struct part_entry *)arg)[i].type = PM_SPEC;
else {
ps = pm_i->parts != NULL ? pm_i->parts->pscheme : NULL;
secondary = NULL;
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_entry *)arg)[i].type = PM_DISK;
for (ii = 0; pm_i->parts != NULL &&
ii < pm_i->parts->num_part; ii++) {
if (!ps->get_part_info(
pm_i->parts, ii, &info))
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
continue;
if (info.flags & PTI_SEC_CONTAINER) {
if (secondary == NULL &&
ps->secondary_scheme != NULL)
secondary = ps->
secondary_partitions(
pm_i->parts,
info.start, false);
continue;
}
if (info.flags & (PTI_WHOLE_DISK|
PTI_PSCHEME_INTERNAL|PTI_RAW_PART))
continue;
if (info.fs_type == FS_UNUSED)
continue;
if (i >= MAX_ENTRIES)
break;
i++;
memset(&m->opts[i], 0, sizeof m->opts[i]);
m->opts[i].opt_action = pm_submenu;
((struct part_entry *)arg)[i].parts =
pm_i->parts;
((struct part_entry *)arg)[i].dev_ptr = pm_i;
((struct part_entry *)arg)[i].id = ii;
((struct part_entry *)arg)[i].type = PM_PART;
}
for (ii = 0; secondary != NULL &&
ii < secondary->num_part; ii++) {
if (!secondary->pscheme->get_part_info(
secondary, ii, &info))
continue;
if (info.flags & (PTI_WHOLE_DISK|
PTI_PSCHEME_INTERNAL|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
continue;
if (info.fs_type == FS_UNUSED)
continue;
if (i >= MAX_ENTRIES)
break;
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
i++;
memset(&m->opts[i], 0, sizeof m->opts[i]);
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
m->opts[i].opt_action = pm_submenu;
((struct part_entry *)arg)[i].parts = secondary;
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_entry *)arg)[i].dev_ptr = pm_i;
((struct part_entry *)arg)[i].id = ii;
((struct part_entry *)arg)[i].type = PM_PART;
}
}
i++;
}
for (ii = 0; ii <= (size_t)i; ii++) {
m->opts[ii].opt_flags = OPT_EXIT;
}
if (have_cgd) {
pm_upddevlist_adv(m, arg, &i,
&(pm_upddevlist_adv_t) {MSG_create_cgd, PM_CGD,
&cgds_t_info, 0, NULL});
}
pm_upddevlist_adv(m, arg, &i,
&(pm_upddevlist_adv_t) {MSG_create_vnd, PM_VND,
&vnds_t_info, 0, NULL});
if (have_lvm) {
pm_upddevlist_adv(m, arg, &i,
&(pm_upddevlist_adv_t) {MSG_create_vg, PM_LVM,
&lvms_t_info, 0,
&(pm_upddevlist_adv_t) {MSG_create_lv, PM_LVMLV,
&lv_t_info, 0,
NULL}});
}
if (have_raid) {
pm_upddevlist_adv(m, arg, &i,
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
&(pm_upddevlist_adv_t) {MSG_create_raid, PM_RAID, &raids_t_info, 0, NULL});
}
m->opts[i++] = (struct menu_ent) {
.opt_name = MSG_updpmlist,
.opt_action = pm_upddevlist,
};
m->opts[i ] = (struct menu_ent) {
.opt_name = MSG_savepm,
.opt_action = pm_commit,
};
if (pm_retvalue >= 0)
m->cursel = pm_retvalue - 1;
return i;
}
static void
pm_menuin(menudesc *m, void *arg)
{
if (pm_cursel > m->numopts)
m->cursel = m->numopts;
else if (pm_cursel < 0)
m->cursel = 0;
else
m->cursel = pm_cursel;
}
static void
pm_menuout(menudesc *m, void *arg)
{
pm_cursel = m->cursel;
}
/* Main partman function */
int
partman(void)
{
int menu_no, menu_num_entries;
static int firstrun = 1;
menu_ent menu_entries[MAX_ENTRIES+6];
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_entry args[MAX_ENTRIES];
if (firstrun) {
check_available_binaries();
if (!have_raid)
remove_raid_options();
else if (!(raids = calloc(MAX_RAID, sizeof(*raids))))
have_raid = 0;
#define remove_vnd_options() (void)0
if (!have_vnd)
remove_vnd_options();
else if (!(vnds = calloc(MAX_VND, sizeof(*vnds))))
have_vnd = 0;
if (!have_cgd)
remove_cgd_options();
else if (!(cgds = calloc(MAX_CGD, sizeof(*cgds))))
have_cgd = 0;
if (!have_lvm)
remove_lvm_options();
else if (!(lvms = calloc(MAX_LVM_VG, sizeof(*lvms))))
have_lvm = 0;
raids_t_info = (structinfo_t) {
.max = MAX_RAID,
.entry_size = sizeof raids[0],
.entry_first = &raids[0],
.entry_enabled = &(raids[0].enabled),
.entry_blocked = &(raids[0].blocked),
.entry_node = &(raids[0].node),
};
vnds_t_info = (structinfo_t) {
.max = MAX_VND,
.entry_size = sizeof vnds[0],
.entry_first = &vnds[0],
.entry_enabled = &(vnds[0].enabled),
.entry_blocked = &(vnds[0].blocked),
.entry_node = &(vnds[0].node),
};
cgds_t_info = (structinfo_t) {
.max = MAX_CGD,
.entry_size = sizeof cgds[0],
.entry_first = &cgds[0],
.entry_enabled = &(cgds[0].enabled),
.entry_blocked = &(cgds[0].blocked),
.entry_node = &(cgds[0].node),
};
lvms_t_info = (structinfo_t) {
.max = MAX_LVM_VG,
.entry_size = sizeof lvms[0],
.entry_first = &lvms[0],
.entry_enabled = &(lvms[0].enabled),
.entry_blocked = &(lvms[0].blocked),
.entry_node = NULL,
};
lv_t_info = (structinfo_t) {
.max = MAX_LVM_LV,
.entry_size = sizeof lvms[0].lv[0],
.entry_first = &lvms[0].lv[0],
.entry_enabled = &(lvms[0].lv[0].size),
.entry_blocked = &(lvms[0].lv[0].blocked),
.parent_size = sizeof lvms[0],
};
pm_cursel = 0;
pm_changed = 0;
firstrun = 0;
}
do {
menu_num_entries = pm_upddevlist(&(menudesc){.opts = menu_entries}, args);
menu_no = new_menu(MSG_partman_header,
menu_entries, menu_num_entries+1, 1, 1, 0, 75, /* Fixed width */
MC_ALWAYS_SCROLL | MC_NOBOX | MC_NOCLEAR,
pm_menuin, pm_menufmt, pm_menuout, NULL, MSG_finishpm);
if (menu_no == -1)
pm_retvalue = -1;
else {
pm_retvalue = 0;
clear();
refresh();
process_menu(menu_no, &args);
free_menu(menu_no);
}
if (pm_retvalue == 0 && pm->parts != NULL)
if (pm_needsave())
pm_commit(NULL, NULL);
} while (pm_retvalue > 0);
/* retvalue <0 - error, retvalue ==0 - user quits, retvalue >0 - all ok */
return (pm_retvalue >= 0)?0:-1;
}
void
update_wedges(const char *disk)
{
check_available_binaries();
if (!have_dk)
return;
run_program(RUN_SILENT | RUN_ERROR_OK,
"dkctl %s makewedges", 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
bool
pm_force_parts(struct pm_devs *my_pm)
{
if (my_pm == NULL)
return false;
if (my_pm->parts != NULL)
return true;
const struct disk_partitioning_scheme *ps =
select_part_scheme(my_pm, NULL, false, NULL);
if (ps == NULL)
return false;
struct disk_partitions *parts =
(*ps->create_new_for_disk)(my_pm->diskdev, 0, my_pm->dlsize,
my_pm->dlsize, false, NULL);
if (parts == NULL)
return false;
my_pm->parts = parts;
if (pm->dlsize > ps->size_limit)
pm->dlsize = ps->size_limit;
return true;
}
void
pm_edit_partitions(struct part_entry *pe)
{
struct pm_devs *my_pm = pm_from_pe(pe);
struct partition_usage_set pset = { 0 };
struct disk_partitions *parts;
if (!my_pm)
return;
if (!pm_force_parts(my_pm))
return;
parts = my_pm->parts;
clear();
refresh();
if (my_pm->parts->pscheme->secondary_scheme != NULL) {
if (!edit_outer_parts(my_pm->parts))
goto done;
parts = get_inner_parts(parts);
}
usage_set_from_parts(&pset, parts);
edit_and_check_label(my_pm, &pset, false);
free_usage_set(&pset);
done:
pm_partusage(my_pm, -1, -1);
my_pm->unsaved = true;
pm_retvalue = 1;
}
part_id
pm_whole_disk(struct part_entry *pe, int t)
{
struct pm_devs *my_pm = pm_from_pe(pe);
struct disk_partitions *parts, *outer;
struct disk_part_info info, oinfo;
struct disk_part_free_space space;
daddr_t align;
int fst;
struct partition_usage_set pset = { 0 };
part_id new_part, id;
size_t i, cnt;
if (!my_pm)
return NO_PART;
if (!pm_force_parts(my_pm))
return NO_PART;
parts = my_pm->parts;
parts->pscheme->delete_all_partitions(parts);
if (parts->pscheme->secondary_scheme != NULL) {
outer = parts;
parts = parts->pscheme->secondary_partitions(outer,
0, true);
if (parts == NULL) {
parts = outer;
} else {
if (outer->pscheme->write_to_disk(outer))
my_pm->parts = parts;
}
}
align = parts->pscheme->get_part_alignment(parts);
memset(&info, 0, sizeof info);
switch (t) {
case SY_NEWRAID:
fst = FS_RAID;
break;
case SY_NEWLVM:
fst = FS_BSDFFS;
break;
case SY_NEWCGD:
fst = FS_CGD;
break;
default:
assert(false);
return NO_PART;
}
info.nat_type = parts->pscheme->get_fs_part_type(PT_root, fst, 0);
if (info.nat_type != NULL && parts->pscheme->get_default_fstype != NULL)
parts->pscheme->get_default_fstype(info.nat_type,
&info.fs_type, &info.fs_sub_type);
if (parts->pscheme->get_free_spaces(parts, &space, 1,
5*align, align, -1, -1) != 1)
return NO_PART;
info.start = space.start;
info.size = space.size;
new_part = parts->pscheme->add_partition(parts, &info, NULL);
if (new_part == NO_PART)
return NO_PART;
parts->pscheme->get_part_info(parts, new_part, &oinfo);
clear();
refresh();
usage_set_from_parts(&pset, parts);
edit_and_check_label(my_pm, &pset, false);
free_usage_set(&pset);
/*
* Try to match our new partition after user edit
*/
new_part = NO_PART;
for (cnt = i = 0; i < parts->num_part; i++) {
if (!parts->pscheme->get_part_info(parts,i, &info))
continue;
if (info.flags & (PTI_SEC_CONTAINER|PTI_WHOLE_DISK|
PTI_PSCHEME_INTERNAL|PTI_RAW_PART))
continue;
if (info.nat_type != oinfo.nat_type)
continue;
if (new_part == NO_PART)
new_part = i;
cnt++;
}
if (cnt > 1) {
/* multiple matches, retry matching with start */
id = NO_PART;
for (cnt = i = 0; i < parts->num_part; i++) {
if (!parts->pscheme->get_part_info(parts, i, &info))
continue;
if (info.flags & (PTI_SEC_CONTAINER|PTI_WHOLE_DISK|
PTI_PSCHEME_INTERNAL|PTI_RAW_PART))
continue;
if (info.nat_type != oinfo.nat_type)
continue;
if (info.start != oinfo.start)
continue;
if (id == NO_PART)
id = i;
cnt++;
}
if (id != NO_PART)
new_part = id;
}
clear();
refresh();
pm_partusage(my_pm, -1, -1);
my_pm->unsaved = true;
pm_retvalue = 1;
return new_part;
}
struct pm_devs *
pm_from_pe(struct part_entry *pe)
{
switch (pe->type) {
case PM_DISK:
return pe->dev_ptr;
default:
assert(false);
}
return NULL;
}