/* libparted - a library for manipulating disk partitions Copyright (C) 1999-2014, 2019-2023 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see . */ #define PROC_DEVICES_BUFSIZ 16384 #include #include #include #include #include #if defined __s390__ || defined __s390x__ #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include /* for uname() */ #include #include #include #ifdef ENABLE_DEVICE_MAPPER #include #endif #include "../architecture.h" #include "dirname.h" #include "xstrtol.h" #include "xalloc.h" #if ENABLE_NLS # include # define _(String) dgettext (PACKAGE, String) #else # define _(String) (String) #endif /* ENABLE_NLS */ /* The __attribute__ feature is available in gcc versions 2.5 and later. The __-protected variants of the attributes 'format' and 'printf' are accepted by gcc versions 2.6.4 (effectively 2.7) and later. */ #if __GNUC__ > 2 || (__GNUC__ == 2 && __GNUC_MINOR__ >= 7) # define _GL_ATTRIBUTE_FORMAT(spec) __attribute__ ((__format__ spec)) #else # define _GL_ATTRIBUTE_FORMAT(spec) /* empty */ #endif #define STRPREFIX(a, b) (strncmp (a, b, strlen (b)) == 0) #define KERNEL_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c)) #ifndef __NR__llseek #define __NR__llseek 140 #endif #ifndef SCSI_IOCTL_SEND_COMMAND #define SCSI_IOCTL_SEND_COMMAND 1 #endif /* from */ #define HDIO_GETGEO 0x0301 /* get device geometry */ #define HDIO_GET_IDENTITY 0x030d /* get IDE identification info */ #define RD_MODE (O_RDONLY) #define WR_MODE (O_WRONLY) #define RW_MODE (O_RDWR) struct hd_geometry { unsigned char heads; unsigned char sectors; unsigned short cylinders; unsigned long start; }; struct ata7_sectinfo { int valid1:1; int valid2:1; int rsv:26; int multiplier:4; }; /* structure returned by HDIO_GET_IDENTITY, as per ANSI ATA2 rev.2f spec */ struct hd_driveid { unsigned short config; /* lots of obsolete bit flags */ unsigned short cyls; /* "physical" cyls */ unsigned short reserved2; /* reserved (word 2) */ unsigned short heads; /* "physical" heads */ unsigned short track_bytes; /* unformatted bytes per track */ unsigned short sector_bytes; /* unformatted bytes per sector */ unsigned short sectors; /* "physical" sectors per track */ unsigned short vendor0; /* vendor unique */ unsigned short vendor1; /* vendor unique */ unsigned short vendor2; /* vendor unique */ unsigned char serial_no[20]; /* 0 = not_specified */ unsigned short buf_type; unsigned short buf_size; /* 512 byte increments; 0 = not_specified */ unsigned short ecc_bytes; /* for r/w long cmds; 0 = not_specified */ unsigned char fw_rev[8]; /* 0 = not_specified */ char model[40]; /* 0 = not_specified */ unsigned char max_multsect; /* 0=not_implemented */ unsigned char vendor3; /* vendor unique */ unsigned short dword_io; /* 0=not_implemented; 1=implemented */ unsigned char vendor4; /* vendor unique */ unsigned char capability; /* bits 0:DMA 1:LBA 2:IORDYsw 3:IORDYsup*/ unsigned short reserved50; /* reserved (word 50) */ unsigned char vendor5; /* vendor unique */ unsigned char tPIO; /* 0=slow, 1=medium, 2=fast */ unsigned char vendor6; /* vendor unique */ unsigned char tDMA; /* 0=slow, 1=medium, 2=fast */ unsigned short field_valid; /* bits 0:cur_ok 1:eide_ok */ unsigned short cur_cyls; /* logical cylinders */ unsigned short cur_heads; /* logical heads */ unsigned short cur_sectors; /* logical sectors per track */ unsigned short cur_capacity0; /* logical total sectors on drive */ unsigned short cur_capacity1; /* (2 words, misaligned int) */ unsigned char multsect; /* current multiple sector count */ unsigned char multsect_valid; /* when (bit0==1) multsect is ok */ unsigned int lba_capacity; /* total number of sectors */ unsigned short dma_1word; /* single-word dma info */ unsigned short dma_mword; /* multiple-word dma info */ unsigned short eide_pio_modes; /* bits 0:mode3 1:mode4 */ unsigned short eide_dma_min; /* min mword dma cycle time (ns) */ unsigned short eide_dma_time; /* recommended mword dma cycle time (ns) */ unsigned short eide_pio; /* min cycle time (ns), no IORDY */ unsigned short eide_pio_iordy; /* min cycle time (ns), with IORDY */ unsigned short words69_70[2]; /* reserved words 69-70 */ /* HDIO_GET_IDENTITY currently returns only words 0 through 70 */ unsigned short words71_74[4]; /* reserved words 71-74 */ unsigned short queue_depth; /* */ unsigned short words76_79[4]; /* reserved words 76-79 */ unsigned short major_rev_num; /* */ unsigned short minor_rev_num; /* */ unsigned short command_set_1; /* bits 0:Smart 1:Security 2:Removable 3:PM */ unsigned short command_set_2; /* bits 14:Smart Enabled 13:0 zero */ unsigned short cfsse; /* command set-feature supported extensions */ unsigned short cfs_enable_1; /* command set-feature enabled */ unsigned short cfs_enable_2; /* command set-feature enabled */ unsigned short csf_default; /* command set-feature default */ unsigned short dma_ultra; /* */ unsigned short word89; /* reserved (word 89) */ unsigned short word90; /* reserved (word 90) */ unsigned short CurAPMvalues; /* current APM values */ unsigned short word92; /* reserved (word 92) */ unsigned short hw_config; /* hardware config */ unsigned short words94_105[12];/* reserved words 94-105 */ struct ata7_sectinfo ata7_sectinfo; /* ATAPI/ATA7 physical and logical sector size */ unsigned short words107_116[10];/* reserved words 107-116 */ unsigned int logical_sectsize;/* ATAPI/ATA7 logical sector size */ unsigned short words119_125[7];/* reserved words 119-125 */ unsigned short last_lun; /* reserved (word 126) */ unsigned short word127; /* reserved (word 127) */ unsigned short dlf; /* device lock function * 15:9 reserved * 8 security level 1:max 0:high * 7:6 reserved * 5 enhanced erase * 4 expire * 3 frozen * 2 locked * 1 en/disabled * 0 capability */ unsigned short csfo; /* current set features options * 15:4 reserved * 3 auto reassign * 2 reverting * 1 read-look-ahead * 0 write cache */ unsigned short words130_155[26];/* reserved vendor words 130-155 */ unsigned short word156; unsigned short words157_159[3]; /* reserved vendor words 157-159 */ unsigned short words160_255[95];/* reserved words 160-255 */ }; /* from */ #define BLKRRPART _IO(0x12,95) /* re-read partition table */ #define BLKGETSIZE _IO(0x12,96) /* return device size */ #define BLKFLSBUF _IO(0x12,97) /* flush buffer cache */ #define BLKSSZGET _IO(0x12,104) /* get block device sector size */ #define BLKGETLASTSECT _IO(0x12,108) /* get last sector of block device */ #define BLKSETLASTSECT _IO(0x12,109) /* set last sector of block device */ /* return device size in bytes (u64 *arg) */ #define BLKGETSIZE64 _IOR(0x12,114,size_t) struct blkdev_ioctl_param { unsigned int block; size_t content_length; char * block_contents; }; /* from */ #define IDE0_MAJOR 3 #define IDE1_MAJOR 22 #define IDE2_MAJOR 33 #define IDE3_MAJOR 34 #define IDE4_MAJOR 56 #define IDE5_MAJOR 57 #define SCSI_CDROM_MAJOR 11 #define SCSI_DISK0_MAJOR 8 #define SCSI_DISK1_MAJOR 65 #define SCSI_DISK2_MAJOR 66 #define SCSI_DISK3_MAJOR 67 #define SCSI_DISK4_MAJOR 68 #define SCSI_DISK5_MAJOR 69 #define SCSI_DISK6_MAJOR 70 #define SCSI_DISK7_MAJOR 71 #define SCSI_DISK8_MAJOR 128 #define SCSI_DISK9_MAJOR 129 #define SCSI_DISK10_MAJOR 130 #define SCSI_DISK11_MAJOR 131 #define SCSI_DISK12_MAJOR 132 #define SCSI_DISK13_MAJOR 133 #define SCSI_DISK14_MAJOR 134 #define SCSI_DISK15_MAJOR 135 #define COMPAQ_SMART2_MAJOR 72 #define COMPAQ_SMART2_MAJOR1 73 #define COMPAQ_SMART2_MAJOR2 74 #define COMPAQ_SMART2_MAJOR3 75 #define COMPAQ_SMART2_MAJOR4 76 #define COMPAQ_SMART2_MAJOR5 77 #define COMPAQ_SMART2_MAJOR6 78 #define COMPAQ_SMART2_MAJOR7 79 #define COMPAQ_SMART_MAJOR 104 #define COMPAQ_SMART_MAJOR1 105 #define COMPAQ_SMART_MAJOR2 106 #define COMPAQ_SMART_MAJOR3 107 #define COMPAQ_SMART_MAJOR4 108 #define COMPAQ_SMART_MAJOR5 109 #define COMPAQ_SMART_MAJOR6 110 #define COMPAQ_SMART_MAJOR7 111 #define DAC960_MAJOR 48 #define ATARAID_MAJOR 114 #define I2O_MAJOR1 80 #define I2O_MAJOR2 81 #define I2O_MAJOR3 82 #define I2O_MAJOR4 83 #define I2O_MAJOR5 84 #define I2O_MAJOR6 85 #define I2O_MAJOR7 86 #define I2O_MAJOR8 87 #define UBD_MAJOR 98 #define DASD_MAJOR 94 #define VIODASD_MAJOR 112 #define AOE_MAJOR 152 #define SX8_MAJOR1 160 #define SX8_MAJOR2 161 #define XVD_MAJOR 202 #define SDMMC_MAJOR 179 #define LOOP_MAJOR 7 #define MD_MAJOR 9 #define BLKEXT_MAJOR 259 #define RAM_MAJOR 1 #define SCSI_BLK_MAJOR(M) ( \ (M) == SCSI_DISK0_MAJOR \ || (M) == SCSI_CDROM_MAJOR \ || ((M) >= SCSI_DISK1_MAJOR && (M) <= SCSI_DISK7_MAJOR) \ || ((M) >= SCSI_DISK8_MAJOR && (M) <= SCSI_DISK15_MAJOR)) /* Maximum number of partitions supported by linux. */ #define MAX_NUM_PARTS 64 static char* _device_get_part_path (PedDevice const *dev, int num); static int _partition_is_mounted_by_path (const char* path); static unsigned int _device_get_partition_range(PedDevice const* dev); static int _device_open (PedDevice* dev, int flags); static int _device_open_ro (PedDevice* dev); static int _device_close (PedDevice* dev); static int _read_fd (int fd, char **buf) { char* p; size_t size = PROC_DEVICES_BUFSIZ; int s, filesize = 0; *buf = malloc (size * sizeof (char)); if (*buf == 0) { return -1; } do { p = &(*buf) [filesize]; s = read (fd, p, PROC_DEVICES_BUFSIZ); /* exit if there is an error or EOF is reached */ if (s <= 0) break; filesize += s; size += s; char *new_buf = realloc (*buf, size); if (new_buf == NULL) { int saved_errno = errno; free (*buf); errno = saved_errno; return -1; } *buf = new_buf; } while (1); if (filesize == 0 && s < 0) { free (*buf); *buf = NULL; return -1; } else { char *new_buf = realloc (*buf, filesize + 1); if (new_buf == NULL) { int saved_errno = errno; free (*buf); errno = saved_errno; return -1; } *buf = new_buf; (*buf)[filesize] = '\0'; } return filesize; } static int _major_type_in_devices (int major, const char* type) { int fd; char* buf = NULL; char* line; char* end; int bd = 0; char c; fd = open ("/proc/devices", O_RDONLY); if (fd < 0) return 0; if (_read_fd(fd, &buf) < 0) { close(fd); return 0; } line = buf; end = strchr(line, '\n'); while (end) { char *name; int maj; c = *end; *end = '\0'; if (!bd) { if (!strncmp(line, "Block devices:", 14)) bd = 1; goto next; } name = strrchr(line, ' '); if (!name || strcmp(name+1, type)) goto next; maj = strtol(line, &name, 10); if (maj == major) { free(buf); close(fd); return 1; } next: *end = c; line = end+1; end = strchr(line, '\n'); } free(buf); close(fd); return 0; } static int _is_ide_major (int major) { switch (major) { case IDE0_MAJOR: case IDE1_MAJOR: case IDE2_MAJOR: case IDE3_MAJOR: case IDE4_MAJOR: case IDE5_MAJOR: return 1; default: return 0; } } static int _is_cpqarray_major (int major) { return ((COMPAQ_SMART2_MAJOR <= major && major <= COMPAQ_SMART2_MAJOR7) || (COMPAQ_SMART_MAJOR <= major && major <= COMPAQ_SMART_MAJOR7)); } static int _is_i2o_major (int major) { return (I2O_MAJOR1 <= major && major <= I2O_MAJOR8); } static int _is_sx8_major (int major) { return (SX8_MAJOR1 <= major && major <= SX8_MAJOR2); } static int _is_virtblk_major (int major) { return _major_type_in_devices (major, "virtblk"); } static int _is_blkext_major (int major) { return _major_type_in_devices (major, "blkext"); } #ifdef ENABLE_DEVICE_MAPPER static int _dm_task_run_wait (struct dm_task *task, uint32_t cookie) { int rc = 0; rc = dm_task_run (task); dm_udev_wait (cookie); return rc; } static int _is_dm_major (int major) { return _major_type_in_devices (major, "device-mapper"); } static int _dm_maptype (PedDevice *dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); struct dm_task *dmt; uint64_t start, length; char *target_type = NULL; char *params; int r = -1; const char* dev_dir = getenv ("DM_DEV_DIR"); if (dev_dir && *dev_dir && !dm_set_dev_dir(dev_dir)) return r; if (!(dmt = dm_task_create(DM_DEVICE_TABLE))) return r; if (!dm_task_set_major_minor(dmt, arch_specific->major, arch_specific->minor, 0)) goto bad; dm_task_no_open_count(dmt); if (!dm_task_run(dmt)) goto bad; dm_get_next_target(dmt, NULL, &start, &length, &target_type, ¶ms); arch_specific->dmtype = strdup(target_type ? target_type : "NO-TARGET"); if (arch_specific->dmtype == NULL) goto bad; r = 0; bad: dm_task_destroy(dmt); return r; } /* Return nonzero if device-mapper device, DEVPATH, is part of a dmraid array. Use the heuristic of checking for the string "DMRAID-" at the start of its UUID. */ static int _is_dmraid_device (const char *devpath) { int rc = 0; char const *dm_name = strrchr (devpath, '/'); char const *dm_basename = dm_name && *(++dm_name) ? dm_name : devpath; struct dm_task *task = dm_task_create (DM_DEVICE_DEPS); if (!task) return 0; dm_task_set_name (task, dm_basename); if (!dm_task_run (task)) goto err; const char *dmraid_uuid = dm_task_get_uuid (task); if (STRPREFIX (dmraid_uuid, "DMRAID-")) rc = 1; err: dm_task_destroy (task); return rc; } /* We consider a dm device that is a linear mapping with a * * single target that also is a dm device to be a partition */ static int _dm_is_part (const char *path) { int rc = 0; struct dm_task *task = dm_task_create (DM_DEVICE_DEPS); if (!task) return 0; dm_task_set_name(task, path); if (!dm_task_run(task)) goto err; struct dm_info *info = alloca (sizeof *info); memset(info, '\0', sizeof *info); dm_task_get_info (task, info); if (!info->exists) goto err; struct dm_deps *deps = dm_task_get_deps (task); if (!deps) goto err; if (deps->count != 1) goto err; if (!_is_dm_major (major (deps->device[0]))) goto err; dm_task_destroy (task); if (!(task = dm_task_create (DM_DEVICE_TABLE))) return 0; dm_task_set_name (task, path); if (!dm_task_run (task)) goto err; char *target_type = NULL; char *params = NULL; uint64_t start, length; dm_get_next_target (task, NULL, &start, &length, &target_type, ¶ms); if (strcmp (target_type, "linear")) goto err; rc = 1; err: dm_task_destroy(task); return rc; } static int _probe_dm_devices () { DIR* mapper_dir; struct dirent* dent; char buf [512]; /* readdir(3) claims d_name[256] */ struct stat st; mapper_dir = opendir ("/dev/mapper"); if (!mapper_dir) return 0; /* Search the /dev/mapper directory for devices w/ the same major * number that was returned from _probe_lvm_major(). */ while ((dent = readdir (mapper_dir))) { if (strcmp (dent->d_name, ".") == 0 || strcmp (dent->d_name, "..") == 0) continue; snprintf (buf, sizeof (buf), "/dev/mapper/%s", dent->d_name); if (stat (buf, &st) != 0) continue; if (_is_dm_major(major(st.st_rdev)) && _is_dmraid_device (buf) && !_dm_is_part(buf)) _ped_device_probe (buf); } closedir (mapper_dir); return 1; } #endif static int _device_stat (PedDevice* dev, struct stat * dev_stat) { PED_ASSERT (dev != NULL); PED_ASSERT (!dev->external_mode); while (1) { if (!stat (dev->path, dev_stat)) { return 1; } else { if (ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_CANCEL, _("Could not stat device %s - %s."), dev->path, strerror (errno)) != PED_EXCEPTION_RETRY) return 0; } } } static int _device_probe_type (PedDevice* dev) { struct stat dev_stat; int dev_major; int dev_minor; LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); if (!_device_stat (dev, &dev_stat)) return 0; if (!S_ISBLK(dev_stat.st_mode)) { dev->type = PED_DEVICE_FILE; return 1; } arch_specific->major = dev_major = major (dev_stat.st_rdev); arch_specific->minor = dev_minor = minor (dev_stat.st_rdev); if (SCSI_BLK_MAJOR (dev_major) && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_SCSI; } else if (_is_ide_major (dev_major) && (dev_minor % 0x40 == 0)) { dev->type = PED_DEVICE_IDE; } else if (dev_major == DAC960_MAJOR && (dev_minor % 0x8 == 0)) { dev->type = PED_DEVICE_DAC960; } else if (dev_major == ATARAID_MAJOR && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_ATARAID; } else if (dev_major == AOE_MAJOR && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_AOE; } else if (dev_major == DASD_MAJOR && (dev_minor % 0x4 == 0)) { dev->type = PED_DEVICE_DASD; } else if (dev_major == VIODASD_MAJOR && (dev_minor % 0x8 == 0)) { dev->type = PED_DEVICE_VIODASD; } else if (_is_sx8_major(dev_major) && (dev_minor % 0x20 == 0)) { dev->type = PED_DEVICE_SX8; } else if (_is_i2o_major (dev_major) && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_I2O; } else if (_is_cpqarray_major (dev_major) && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_CPQARRAY; } else if (dev_major == UBD_MAJOR && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_UBD; #ifdef ENABLE_DEVICE_MAPPER } else if (_is_dm_major(dev_major)) { dev->type = PED_DEVICE_DM; if (_dm_maptype(dev)) { ped_exception_throw ( PED_EXCEPTION_BUG, PED_EXCEPTION_CANCEL, _("Unable to determine the dm type of %s."), dev->path); } #endif } else if (dev_major == XVD_MAJOR && (dev_minor % 0x10 == 0)) { dev->type = PED_DEVICE_XVD; } else if (dev_major == SDMMC_MAJOR && (dev_minor % 0x08 == 0)) { dev->type = PED_DEVICE_SDMMC; } else if (_is_virtblk_major(dev_major)) { dev->type = PED_DEVICE_VIRTBLK; } else if (dev_major == LOOP_MAJOR) { dev->type = PED_DEVICE_LOOP; } else if (dev_major == MD_MAJOR) { dev->type = PED_DEVICE_MD; } else if (_is_blkext_major(dev_major) && dev->path && strstr(dev->path, "nvme")) { dev->type = PED_DEVICE_NVME; } else if (dev_major == RAM_MAJOR) { dev->type = PED_DEVICE_RAM; } else if (_is_blkext_major(dev_major) && dev->path && strstr(dev->path, "pmem")) { dev->type = PED_DEVICE_PMEM; } else { dev->type = PED_DEVICE_UNKNOWN; } return 1; } static int _get_linux_version () { static int kver = -1; struct utsname uts; unsigned int major = 0; unsigned int minor = 0; unsigned int teeny = 0; if (kver != -1) return kver; if (uname (&uts)) return kver = 0; int n = sscanf (uts.release, "%u.%u.%u", &major, &minor, &teeny); assert (n == 2 || n == 3); return kver = KERNEL_VERSION (major, minor, teeny); } #if USE_BLKID static void get_blkid_topology (LinuxSpecific *arch_specific) { arch_specific->probe = blkid_new_probe (); if (!arch_specific->probe) return; if (blkid_probe_set_device(arch_specific->probe, arch_specific->fd, 0, 0)) return; arch_specific->topology = blkid_probe_get_topology(arch_specific->probe); } #endif static void _device_set_sector_size (PedDevice* dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); int sector_size; dev->sector_size = PED_SECTOR_SIZE_DEFAULT; dev->phys_sector_size = PED_SECTOR_SIZE_DEFAULT; PED_ASSERT (dev->open_count); if (_get_linux_version() < KERNEL_VERSION (2,3,0)) { dev->sector_size = PED_SECTOR_SIZE_DEFAULT; return; } if (ioctl (arch_specific->fd, BLKSSZGET, §or_size)) { ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_OK, _("Could not determine sector size for %s: %s.\n" "Using the default sector size (%lld)."), dev->path, strerror (errno), PED_SECTOR_SIZE_DEFAULT); } else { dev->sector_size = (long long)sector_size; dev->phys_sector_size = dev->sector_size; } #if USE_BLKID get_blkid_topology(arch_specific); if (!arch_specific->topology) { dev->phys_sector_size = 0; } else { dev->phys_sector_size = blkid_topology_get_physical_sector_size( arch_specific->topology); } if (dev->phys_sector_size == 0) { ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_OK, _("Could not determine physical sector size for %s.\n" "Using the logical sector size (%lld)."), dev->path, dev->sector_size); dev->phys_sector_size = dev->sector_size; } #endif #if defined __s390__ || defined __s390x__ /* The real_sector_size is currently needed for DASD layouts, * so we set it unconditionally. In the long run it should * be considered to use the dev->phys_sector_size in label/dasd.c. */ arch_specific->real_sector_size = dev->sector_size; /* Return PED_SECTOR_SIZE_DEFAULT for DASDs. */ if (dev->type == PED_DEVICE_DASD) { dev->sector_size = PED_SECTOR_SIZE_DEFAULT; } #endif } static int _kernel_has_blkgetsize64(void) { int version = _get_linux_version(); if (version >= KERNEL_VERSION (2,5,4)) return 1; if (version < KERNEL_VERSION (2,5,0) && version >= KERNEL_VERSION (2,4,18)) return 1; return 0; } /* TODO: do a binary search if BLKGETSIZE doesn't work?! */ static PedSector _device_get_length (PedDevice* dev) { unsigned long size; LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); uint64_t bytes=0; const char* test_str; PedSector test_size; PED_ASSERT (dev->open_count > 0); PED_ASSERT (dev->sector_size % PED_SECTOR_SIZE_DEFAULT == 0); test_str = getenv ("PARTED_TEST_DEVICE_LENGTH"); if (test_str && xstrtoll (test_str, NULL, 10, &test_size, NULL) == LONGINT_OK) return test_size; if (_kernel_has_blkgetsize64()) { if (ioctl(arch_specific->fd, BLKGETSIZE64, &bytes) == 0) { return bytes / dev->sector_size; } } if (ioctl (arch_specific->fd, BLKGETSIZE, &size)) { ped_exception_throw ( PED_EXCEPTION_BUG, PED_EXCEPTION_CANCEL, _("Unable to determine the size of %s (%s)."), dev->path, strerror (errno)); return 0; } return size; } static int _device_probe_geometry (PedDevice* dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); struct stat dev_stat; struct hd_geometry geometry; int geometry_is_valid = 0; int sector_size = 0; if (!_device_stat (dev, &dev_stat)) return 0; PED_ASSERT (S_ISBLK (dev_stat.st_mode)); _device_set_sector_size (dev); dev->length = _device_get_length (dev); if (!dev->length) return 0; /* initialize the bios_geom values to something */ dev->bios_geom.sectors = 0; dev->bios_geom.heads = 0; dev->bios_geom.cylinders = 0; geometry_is_valid = !ioctl (arch_specific->fd, HDIO_GETGEO, &geometry) && geometry.sectors && geometry.heads; #if defined __s390__ || defined __s390x__ if (geometry_is_valid) { #else if (!ioctl (arch_specific->fd, BLKSSZGET, §or_size)) { /* get the sector count first */ dev->bios_geom.sectors = 1 + (sector_size / PED_SECTOR_SIZE_DEFAULT); dev->bios_geom.heads = 255; } else if (geometry_is_valid) { /* if BLKSSZGET failed, use deprecated HDIO_GETGEO result */ #endif dev->bios_geom.sectors = geometry.sectors; dev->bios_geom.heads = geometry.heads; } else { ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_OK, _("Could not determine sector size for %s: %s.\n" "Using the default sector size (%lld)."), dev->path, strerror (errno), PED_SECTOR_SIZE_DEFAULT); dev->bios_geom.sectors = 2; dev->bios_geom.heads = 255; } dev->bios_geom.cylinders = dev->length / (dev->bios_geom.heads * dev->bios_geom.sectors); dev->hw_geom = dev->bios_geom; return 1; } static char* strip_name(char* str) { int i; int end = 0; for (i = 0; str[i] != 0; i++) { if (!isspace (str[i]) || (isspace (str[i]) && !isspace (str[i+1]) && str[i+1])) { str [end] = str[i]; end++; } } str[end] = 0; return strdup (str); } static int init_ide (PedDevice* dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); struct stat dev_stat; struct hd_driveid hdi; PedExceptionOption ex_status; char hdi_buf[41]; int sector_multiplier = 0; int r; if (!_device_stat (dev, &dev_stat)) goto error; if (!_device_open_ro (dev)) goto error; r = ioctl (arch_specific->fd, HDIO_GET_IDENTITY, &hdi); if (r && errno == EINVAL) { /* silently ignore unsupported ioctl */ dev->model = strdup(_("Generic IDE")); } else if (r) { ex_status = ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_IGNORE_CANCEL, _("Could not get identity of device %s - %s"), dev->path, strerror (errno)); switch (ex_status) { case PED_EXCEPTION_CANCEL: goto error_close_dev; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_IGNORE: dev->model = strdup(_("Generic IDE")); break; default: PED_ASSERT (0); break; } } else { /* hdi.model is not guaranteed to be NULL terminated */ memcpy (hdi_buf, hdi.model, 40); hdi_buf[40] = '\0'; dev->model = strip_name (hdi_buf); if (!hdi.ata7_sectinfo.valid1 && hdi.ata7_sectinfo.valid2) sector_multiplier = hdi.ata7_sectinfo.multiplier; else sector_multiplier = 1; if (sector_multiplier != 1) { ex_status = ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_IGNORE_CANCEL, _("Device %s has multiple (%d) logical sectors " "per physical sector.\n" "GNU Parted supports this EXPERIMENTALLY for " "some special disk label/file system " "combinations, e.g. GPT and ext2/3.\n" "Please consult the web site for up-to-date " "information."), dev->path, sector_multiplier); switch (ex_status) { case PED_EXCEPTION_CANCEL: goto error_close_dev; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_IGNORE: break; default: PED_ASSERT (0); break; } } /* XXX sector_size has not been set yet! */ /* dev->phys_sector_size = dev->sector_size * sector_multiplier;*/ dev->phys_sector_size = PED_SECTOR_SIZE_DEFAULT; } if (!_device_probe_geometry (dev)) goto error_close_dev; _device_close (dev); return 1; error_close_dev: _device_close (dev); error: return 0; } /* This function reads the /sys entry named "file" for device "dev". */ static char * read_device_sysfs_file (PedDevice *dev, const char *file) { FILE *f; char name_buf[128]; char buf[256]; snprintf (name_buf, 127, "/sys/block/%s/device/%s", last_component (dev->path), file); if ((f = fopen (name_buf, "r")) == NULL) return NULL; if (fgets (buf, 255, f) == NULL) { fclose (f); return NULL; } fclose (f); return strip_name (buf); } /* This function sends a query to a SCSI device for vendor and product * information. It uses the deprecated SCSI_IOCTL_SEND_COMMAND to * issue this query. */ static int scsi_query_product_info (PedDevice* dev, char **vendor, char **product) { /* The following are defined by the SCSI-2 specification. */ typedef struct _scsi_inquiry_cmd { uint8_t op; uint8_t lun; /* bits 5-7 denote the LUN */ uint8_t page_code; uint8_t reserved; uint8_t alloc_length; uint8_t control; } __attribute__((packed)) scsi_inquiry_cmd_t; typedef struct _scsi_inquiry_data { uint8_t peripheral_info; uint8_t device_info; uint8_t version_info; uint8_t _field1; uint8_t additional_length; uint8_t _reserved1; uint8_t _reserved2; uint8_t _field2; uint8_t vendor_id[8]; uint8_t product_id[16]; uint8_t product_revision[4]; uint8_t vendor_specific[20]; uint8_t _reserved3[40]; } __attribute__((packed)) scsi_inquiry_data_t; struct scsi_arg { unsigned int inlen; unsigned int outlen; union arg_data { scsi_inquiry_data_t out; scsi_inquiry_cmd_t in; } data; } arg; LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); char buf[32]; *vendor = NULL; *product = NULL; memset (&arg, 0x00, sizeof(struct scsi_arg)); arg.inlen = 0; arg.outlen = sizeof(scsi_inquiry_data_t); arg.data.in.op = INQUIRY; arg.data.in.lun = dev->host << 5; arg.data.in.alloc_length = sizeof(scsi_inquiry_data_t); arg.data.in.page_code = 0; arg.data.in.reserved = 0; arg.data.in.control = 0; if (ioctl (arch_specific->fd, SCSI_IOCTL_SEND_COMMAND, &arg) < 0) return 0; memcpy (buf, arg.data.out.vendor_id, 8); buf[8] = '\0'; *vendor = strip_name (buf); memcpy (buf, arg.data.out.product_id, 16); buf[16] = '\0'; *product = strip_name (buf); return 1; } /* This function provides the vendor and product name for a SCSI device. * It supports both the modern /sys interface and direct queries * via the deprecated ioctl, SCSI_IOCTL_SEND_COMMAND. */ static int scsi_get_product_info (PedDevice* dev, char **vendor, char **product) { *vendor = read_device_sysfs_file (dev, "vendor"); *product = read_device_sysfs_file (dev, "model"); if (*vendor && *product) return 1; return scsi_query_product_info (dev, vendor, product); } static int init_scsi (PedDevice* dev) { struct scsi_idlun { uint32_t dev_id; uint32_t host_unique_id; } idlun; LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); char* vendor; char* product; if (!_device_open_ro (dev)) goto error; if (ioctl (arch_specific->fd, SCSI_IOCTL_GET_IDLUN, &idlun) < 0) { dev->host = 0; dev->did = 0; if (ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_IGNORE_CANCEL, _("Error initialising SCSI device %s - %s"), dev->path, strerror (errno)) != PED_EXCEPTION_IGNORE) goto error_close_dev; if (!_device_probe_geometry (dev)) goto error_close_dev; _device_close (dev); return 1; } dev->host = idlun.host_unique_id; dev->did = idlun.dev_id; dev->model = (char*) ped_malloc (8 + 16 + 2); if (!dev->model) goto error_close_dev; if (scsi_get_product_info (dev, &vendor, &product)) { sprintf (dev->model, "%.8s %.16s", vendor, product); free (vendor); free (product); } else { strcpy (dev->model, "Generic SCSI"); } if (!_device_probe_geometry (dev)) goto error_close_dev; _device_close (dev); return 1; error_close_dev: _device_close (dev); error: return 0; } static int init_file (PedDevice* dev) { struct stat dev_stat; if (!_device_stat (dev, &dev_stat)) goto error; if (!_device_open_ro (dev)) goto error; dev->sector_size = PED_SECTOR_SIZE_DEFAULT; char *p = getenv ("PARTED_SECTOR_SIZE"); if (p) { int s = atoi (p); if (0 < s && s % 512 == 0) dev->sector_size = s; } dev->phys_sector_size = dev->sector_size; if (S_ISBLK(dev_stat.st_mode)) dev->length = _device_get_length (dev); else dev->length = dev_stat.st_size / dev->sector_size; if (dev->length <= 0) { ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_CANCEL, _("The device %s is so small that it cannot possibly " "store a file system or partition table. Perhaps " "you selected the wrong device?"), dev->path); goto error_close_dev; } _device_close (dev); dev->bios_geom.cylinders = dev->length / 4 / 32; dev->bios_geom.heads = 4; dev->bios_geom.sectors = 32; dev->hw_geom = dev->bios_geom; dev->model = strdup (""); return 1; error_close_dev: _device_close (dev); error: return 0; } #if defined __s390__ || defined __s390x__ static int init_dasd (PedDevice* dev, const char* model_name) { struct stat dev_stat; struct hd_geometry geo; dasd_information_t dasd_info; if (!_device_stat (dev, &dev_stat)) goto error; if (!_device_open_ro (dev)) goto error; LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); PED_ASSERT (S_ISBLK (dev_stat.st_mode)); _device_set_sector_size (dev); if (!dev->sector_size) goto error_close_dev; dev->length = _device_get_length (dev); if (!dev->length) goto error_close_dev; if (!ioctl (arch_specific->fd, HDIO_GETGEO, &geo)) { dev->hw_geom.sectors = geo.sectors; dev->hw_geom.heads = geo.heads; dev->hw_geom.cylinders = dev->length / (dev->hw_geom.heads * dev->hw_geom.sectors) / (dev->sector_size / PED_SECTOR_SIZE_DEFAULT); dev->bios_geom = dev->hw_geom; } else { dev->bios_geom.sectors = 12; dev->bios_geom.heads = 15; dev->bios_geom.cylinders = dev->length / (dev->hw_geom.heads * dev->hw_geom.sectors) / (dev->sector_size / PED_SECTOR_SIZE_DEFAULT); dev->hw_geom = dev->bios_geom; } if (!ioctl(arch_specific->fd, BIODASDINFO, &dasd_info)) { arch_specific->devno = dasd_info.devno; } else { arch_specific->devno = arch_specific->major * 256 + arch_specific->minor; } dev->model = strdup (model_name); _device_close (dev); return 1; error_close_dev: _device_close (dev); error: return 0; } #endif static int init_generic (PedDevice* dev, const char* model_name) { struct stat dev_stat; PedExceptionOption ex_status; if (!_device_stat (dev, &dev_stat)) goto error; if (!_device_open_ro (dev)) goto error; ped_exception_fetch_all (); if (_device_probe_geometry (dev)) { ped_exception_leave_all (); } else { if (!_device_get_length (dev)) { ped_exception_catch (); ped_exception_leave_all (); goto error_close_dev; } /* hack to allow use of files, for testing */ ped_exception_catch (); ped_exception_leave_all (); ex_status = ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_IGNORE_CANCEL, _("Unable to determine geometry of " "file/device %s. You should not use Parted " "unless you REALLY know what you're doing!"), dev->path); switch (ex_status) { case PED_EXCEPTION_CANCEL: goto error_close_dev; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_IGNORE: break; default: PED_ASSERT (0); break; } /* what should we stick in here? */ dev->length = dev_stat.st_size / PED_SECTOR_SIZE_DEFAULT; dev->bios_geom.cylinders = dev->length / 4 / 32; dev->bios_geom.heads = 4; dev->bios_geom.sectors = 32; dev->sector_size = PED_SECTOR_SIZE_DEFAULT; dev->phys_sector_size = PED_SECTOR_SIZE_DEFAULT; } dev->model = strdup (model_name); _device_close (dev); return 1; error_close_dev: _device_close (dev); error: return 0; } static int sdmmc_get_product_info (PedDevice* dev, char **type, char **name) { *type = read_device_sysfs_file (dev, "type"); *name = read_device_sysfs_file (dev, "name"); if (*type && *name) return 1; return 0; } static int init_sdmmc (PedDevice* dev) { char id[128]; char *type = NULL; char *name = NULL; if (sdmmc_get_product_info (dev, &type, &name)) { snprintf (id, sizeof(id) - 1, "%s %s", type, name); } else { snprintf (id, sizeof(id) - 1, "%s", _("Generic SD/MMC Storage Card")); } free (type); free (name); return init_generic(dev, id); } static int init_nvme (PedDevice* dev) { int ret; char *model = read_device_sysfs_file (dev, "model"); if (!model) ret = init_generic (dev, _("NVMe Device")); else { ret = init_generic (dev, model); free (model); } return ret; } static PedDevice* linux_new (const char* path) { PedDevice* dev; LinuxSpecific* arch_specific; PED_ASSERT (path != NULL); dev = (PedDevice*) ped_malloc (sizeof (PedDevice)); if (!dev) goto error; dev->path = strdup (path); if (!dev->path) goto error_free_dev; dev->arch_specific = (LinuxSpecific*) ped_malloc (sizeof (LinuxSpecific)); if (!dev->arch_specific) goto error_free_path; arch_specific = LINUX_SPECIFIC (dev); arch_specific->dmtype = NULL; #if USE_BLKID arch_specific->probe = NULL; arch_specific->topology = NULL; #endif dev->open_count = 0; dev->read_only = 0; dev->external_mode = 0; dev->dirty = 0; dev->boot_dirty = 0; #ifdef ENABLE_DEVICE_MAPPER dm_udev_set_sync_support(1); #endif if (!_device_probe_type (dev)) goto error_free_arch_specific; switch (dev->type) { case PED_DEVICE_IDE: if (!init_ide (dev)) goto error_free_arch_specific; break; case PED_DEVICE_SCSI: if (!init_scsi (dev)) goto error_free_arch_specific; break; case PED_DEVICE_DAC960: if (!init_generic (dev, _("DAC960 RAID controller"))) goto error_free_arch_specific; break; case PED_DEVICE_SX8: if (!init_generic (dev, _("Promise SX8 SATA Device"))) goto error_free_arch_specific; break; case PED_DEVICE_AOE: if (!init_generic (dev, _("ATA over Ethernet Device"))) goto error_free_arch_specific; break; #if defined __s390__ || defined __s390x__ case PED_DEVICE_DASD: if (!init_dasd (dev, _("IBM S390 DASD drive"))) goto error_free_arch_specific; break; #endif case PED_DEVICE_VIODASD: if (!init_generic (dev, _("IBM iSeries Virtual DASD"))) goto error_free_arch_specific; break; case PED_DEVICE_CPQARRAY: if (!init_generic (dev, _("Compaq Smart Array"))) goto error_free_arch_specific; break; case PED_DEVICE_NVME: if (!init_nvme (dev)) goto error_free_arch_specific; break; case PED_DEVICE_PMEM: if (!init_generic (dev, _("NVDIMM Device"))) goto error_free_arch_specific; break; case PED_DEVICE_ATARAID: if (!init_generic (dev, _("ATARAID Controller"))) goto error_free_arch_specific; break; case PED_DEVICE_I2O: if (!init_generic (dev, _("I2O Controller"))) goto error_free_arch_specific; break; case PED_DEVICE_UBD: if (!init_generic (dev, _("User-Mode Linux UBD"))) goto error_free_arch_specific; break; case PED_DEVICE_FILE: if (!init_file (dev)) goto error_free_arch_specific; break; case PED_DEVICE_LOOP: if (!init_generic (dev, _("Loopback device"))) goto error_free_arch_specific; break; case PED_DEVICE_DM: { char* type; if (arch_specific->dmtype == NULL || asprintf(&type, _("Linux device-mapper (%s)"), arch_specific->dmtype) == -1) goto error_free_arch_specific; bool ok = init_generic (dev, type); free (type); if (!ok) goto error_free_arch_specific; break; } case PED_DEVICE_XVD: if (!init_generic (dev, _("Xen Virtual Block Device"))) goto error_free_arch_specific; break; case PED_DEVICE_UNKNOWN: if (!init_generic (dev, _("Unknown"))) goto error_free_arch_specific; break; case PED_DEVICE_SDMMC: if (!init_sdmmc (dev)) goto error_free_arch_specific; break; case PED_DEVICE_VIRTBLK: if (!init_generic(dev, _("Virtio Block Device"))) goto error_free_arch_specific; break; case PED_DEVICE_MD: if (!init_generic(dev, _("Linux Software RAID Array"))) goto error_free_arch_specific; break; case PED_DEVICE_RAM: if (!init_generic (dev, _("RAM Drive"))) goto error_free_arch_specific; break; default: ped_exception_throw (PED_EXCEPTION_NO_FEATURE, PED_EXCEPTION_CANCEL, _("ped_device_new() Unsupported device type")); goto error_free_arch_specific; } return dev; error_free_arch_specific: free (dev->arch_specific); error_free_path: free (dev->path); error_free_dev: free (dev); error: return NULL; } static void linux_destroy (PedDevice* dev) { LinuxSpecific *arch_specific = LINUX_SPECIFIC(dev); void *p = arch_specific->dmtype; #if USE_BLKID if (arch_specific->probe) blkid_free_probe(arch_specific->probe); #endif free (p); free (dev->arch_specific); free (dev->path); free (dev->model); free (dev); } static int linux_is_busy (PedDevice* dev) { int i; char* part_name; if (_partition_is_mounted_by_path (dev->path)) return 1; for (i = 0; i < 32; i++) { int status; part_name = _device_get_part_path (dev, i); if (!part_name) return 1; status = _partition_is_mounted_by_path (part_name); free (part_name); if (status) return 1; } return 0; } /* we need to flush the master device, and all the partition devices, * because there is no coherency between the caches. * We should only flush unmounted partition devices, because: * - there is never a need to flush them (we're not doing IO there) * - flushing a device that is mounted causes unnecessary IO, and can * even screw journaling & friends up. Even cause oopsen! */ static void _flush_cache (PedDevice* dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); int i; int lpn = _device_get_partition_range(dev); if (dev->read_only || dev->type == PED_DEVICE_RAM) return; dev->dirty = 0; ioctl (arch_specific->fd, BLKFLSBUF); for (i = 1; i < lpn; i++) { char* name; int fd; name = _device_get_part_path (dev, i); if (!name) break; if (!_partition_is_mounted_by_path (name)) { fd = open (name, WR_MODE, 0); if (fd > -1) { ioctl (fd, BLKFLSBUF); retry: if (fsync (fd) < 0 || close (fd) < 0) if (ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_RETRY + PED_EXCEPTION_IGNORE, _("Error fsyncing/closing %s: %s"), name, strerror (errno)) == PED_EXCEPTION_RETRY) goto retry; } } free (name); } } static int _device_open_ro (PedDevice* dev) { int rc = _device_open (dev, RD_MODE); if (rc) dev->open_count++; return rc; } static int linux_open (PedDevice* dev) { return _device_open (dev, RW_MODE); } static int _device_open (PedDevice* dev, int flags) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); retry: arch_specific->fd = open (dev->path, flags); if (arch_specific->fd == -1) { char* rw_error_msg = strerror (errno); arch_specific->fd = open (dev->path, RD_MODE); if (arch_specific->fd == -1) { if (ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_CANCEL, _("Error opening %s: %s"), dev->path, strerror (errno)) != PED_EXCEPTION_RETRY) { return 0; } else { goto retry; } } else { ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_OK, _("Unable to open %s read-write (%s). %s has " "been opened read-only."), dev->path, rw_error_msg, dev->path); dev->read_only = 1; } } else { dev->read_only = 0; } _flush_cache (dev); return 1; } static int linux_refresh_open (PedDevice* dev) { return 1; } static int linux_close (PedDevice* dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); if (dev->dirty) _flush_cache (dev); retry: if (fsync (arch_specific->fd) < 0 || close (arch_specific->fd) < 0) if (ped_exception_throw ( PED_EXCEPTION_WARNING, PED_EXCEPTION_RETRY + PED_EXCEPTION_IGNORE, _("Error fsyncing/closing %s: %s"), dev->path, strerror (errno)) == PED_EXCEPTION_RETRY) goto retry; return 1; } static int linux_refresh_close (PedDevice* dev) { if (dev->dirty) _flush_cache (dev); return 1; } static int _device_close (PedDevice* dev) { int rc = linux_close (dev); if (dev->open_count > 0) dev->open_count--; return rc; } #if SIZEOF_OFF_T < 8 static _syscall5(int,_llseek, unsigned int, fd, unsigned long, offset_high, unsigned long, offset_low, loff_t*, result, unsigned int, origin) loff_t llseek (unsigned int fd, loff_t offset, unsigned int whence) { loff_t result; int retval; retval = _llseek(fd, ((unsigned long long)offset) >> 32, ((unsigned long long)offset) & 0xffffffff, &result, whence); return (retval==-1 ? (loff_t) retval : result); } #endif /* SIZEOF_OFF_T < 8 */ static int _device_seek (const PedDevice* dev, PedSector sector) { LinuxSpecific* arch_specific; PED_ASSERT (dev->sector_size % PED_SECTOR_SIZE_DEFAULT == 0); PED_ASSERT (dev != NULL); PED_ASSERT (!dev->external_mode); arch_specific = LINUX_SPECIFIC (dev); #if SIZEOF_OFF_T < 8 if (sizeof (off_t) < 8) { loff_t pos = (loff_t)(sector * dev->sector_size); return llseek (arch_specific->fd, pos, SEEK_SET) == pos; } else #endif { off_t pos = sector * dev->sector_size; return lseek (arch_specific->fd, pos, SEEK_SET) == pos; } } static int _read_lastoddsector (const PedDevice* dev, void* buffer) { LinuxSpecific* arch_specific; struct blkdev_ioctl_param ioctl_param; PED_ASSERT(dev != NULL); PED_ASSERT(buffer != NULL); arch_specific = LINUX_SPECIFIC (dev); retry: ioctl_param.block = 0; /* read the last sector */ ioctl_param.content_length = dev->sector_size; ioctl_param.block_contents = buffer; if (ioctl(arch_specific->fd, BLKGETLASTSECT, &ioctl_param) == -1) { PedExceptionOption opt; opt = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, _("%s during read on %s"), strerror (errno), dev->path); if (opt == PED_EXCEPTION_CANCEL) return 0; if (opt == PED_EXCEPTION_RETRY) goto retry; } return 1; } static int linux_read (const PedDevice* dev, void* buffer, PedSector start, PedSector count) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); PedExceptionOption ex_status; void* diobuf = NULL; PED_ASSERT (dev != NULL); PED_ASSERT (dev->sector_size % PED_SECTOR_SIZE_DEFAULT == 0); if (_get_linux_version() < KERNEL_VERSION (2,6,0)) { /* Kludge. This is necessary to read/write the last block of an odd-sized disk, until Linux 2.5.x kernel fixes. */ if (dev->type != PED_DEVICE_FILE && (dev->length & 1) && start + count - 1 == dev->length - 1) return ped_device_read (dev, buffer, start, count - 1) && _read_lastoddsector ( dev, (char *) buffer + (count-1) * 512); } while (1) { if (_device_seek (dev, start)) break; ex_status = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, _("%s during seek for read on %s"), strerror (errno), dev->path); switch (ex_status) { case PED_EXCEPTION_IGNORE: return 1; case PED_EXCEPTION_RETRY: break; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_CANCEL: return 0; default: PED_ASSERT (0); break; } } size_t read_length = count * dev->sector_size; if (posix_memalign (&diobuf, dev->sector_size, read_length) != 0) return 0; while (1) { ssize_t status = read (arch_specific->fd, diobuf, read_length); if (status > 0) memcpy(buffer, diobuf, status); if (status == (ssize_t) read_length) break; if (status > 0) { read_length -= status; buffer = (char *) buffer + status; continue; } ex_status = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, (status == 0 ? _("%0.0send of file while reading %s") : _("%s during read on %s")), strerror (errno), dev->path); switch (ex_status) { case PED_EXCEPTION_IGNORE: free(diobuf); return 1; case PED_EXCEPTION_RETRY: break; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_CANCEL: free(diobuf); return 0; default: PED_ASSERT (0); break; } } free (diobuf); return 1; } static int _write_lastoddsector (PedDevice* dev, const void* buffer) { LinuxSpecific* arch_specific; struct blkdev_ioctl_param ioctl_param; PED_ASSERT(dev != NULL); PED_ASSERT(buffer != NULL); arch_specific = LINUX_SPECIFIC (dev); retry: ioctl_param.block = 0; /* write the last sector */ ioctl_param.content_length = dev->sector_size; ioctl_param.block_contents = (void*) buffer; if (ioctl(arch_specific->fd, BLKSETLASTSECT, &ioctl_param) == -1) { PedExceptionOption opt; opt = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, _("%s during write on %s"), strerror (errno), dev->path); if (opt == PED_EXCEPTION_CANCEL) return 0; if (opt == PED_EXCEPTION_RETRY) goto retry; } return 1; } static int linux_write (PedDevice* dev, const void* buffer, PedSector start, PedSector count) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); PedExceptionOption ex_status; void* diobuf; void* diobuf_start; PED_ASSERT(dev->sector_size % PED_SECTOR_SIZE_DEFAULT == 0); if (dev->read_only) { if (ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_IGNORE_CANCEL, _("Can't write to %s, because it is opened read-only."), dev->path) != PED_EXCEPTION_IGNORE) return 0; else return 1; } if (_get_linux_version() < KERNEL_VERSION (2,6,0)) { /* Kludge. This is necessary to read/write the last block of an odd-sized disk, until Linux 2.5.x kernel fixes. */ if (dev->type != PED_DEVICE_FILE && (dev->length & 1) && start + count - 1 == dev->length - 1) return ped_device_write (dev, buffer, start, count - 1) && _write_lastoddsector ( dev, ((char*) buffer + (count-1) * dev->sector_size)); } while (1) { if (_device_seek (dev, start)) break; ex_status = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, _("%s during seek for write on %s"), strerror (errno), dev->path); switch (ex_status) { case PED_EXCEPTION_IGNORE: return 1; case PED_EXCEPTION_RETRY: break; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_CANCEL: return 0; default: PED_ASSERT (0); break; } } #ifdef READ_ONLY printf ("ped_device_write (\"%s\", %p, %d, %d)\n", dev->path, buffer, (int) start, (int) count); #else size_t write_length = count * dev->sector_size; dev->dirty = 1; if (posix_memalign(&diobuf, dev->sector_size, write_length) != 0) return 0; memcpy(diobuf, buffer, write_length); diobuf_start = diobuf; while (1) { ssize_t status = write (arch_specific->fd, diobuf, write_length); if (status == write_length) break; if (status > 0) { write_length -= status; diobuf = (char *) diobuf + status; continue; } ex_status = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, _("%s during write on %s"), strerror (errno), dev->path); switch (ex_status) { case PED_EXCEPTION_IGNORE: free(diobuf_start); return 1; case PED_EXCEPTION_RETRY: break; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_CANCEL: free(diobuf_start); return 0; default: PED_ASSERT (0); break; } } free(diobuf_start); #endif /* !READ_ONLY */ return 1; } /* returns the number of sectors that are ok. */ static PedSector linux_check (PedDevice* dev, void* buffer, PedSector start, PedSector count) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); PedSector done = 0; int status; void* diobuf; PED_ASSERT(dev != NULL); if (!_device_seek (dev, start)) return 0; if (posix_memalign(&diobuf, PED_SECTOR_SIZE_DEFAULT, count * PED_SECTOR_SIZE_DEFAULT) != 0) return 0; for (done = 0; done < count; done += status / dev->sector_size) { status = read (arch_specific->fd, diobuf, (size_t) ((count-done) * dev->sector_size)); if (status > 0) memcpy(buffer, diobuf, status); if (status < 0) break; } free(diobuf); return done; } static int _do_fsync (PedDevice* dev) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); int status; PedExceptionOption ex_status; while (1) { status = fsync (arch_specific->fd); if (status >= 0) break; ex_status = ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_RETRY_IGNORE_CANCEL, _("%s during write on %s"), strerror (errno), dev->path); switch (ex_status) { case PED_EXCEPTION_IGNORE: return 1; case PED_EXCEPTION_RETRY: break; case PED_EXCEPTION_UNHANDLED: ped_exception_catch (); /* FALLTHROUGH */ case PED_EXCEPTION_CANCEL: return 0; default: PED_ASSERT (0); break; } } return 1; } static int linux_sync (PedDevice* dev) { PED_ASSERT (dev != NULL); PED_ASSERT (!dev->external_mode); if (dev->read_only) return 1; if (!_do_fsync (dev)) return 0; _flush_cache (dev); return 1; } static int linux_sync_fast (PedDevice* dev) { PED_ASSERT (dev != NULL); PED_ASSERT (!dev->external_mode); if (dev->read_only) return 1; if (!_do_fsync (dev)) return 0; /* no cache flush... */ return 1; } static inline int _compare_digit_state (char ch, int need_digit) { return !!isdigit (ch) == need_digit; } /* matches the regexp "[^0-9]+[0-9]+[^0-9]+[0-9]+$". * Motivation: accept devices looking like /dev/rd/c0d0, but * not looking like /dev/hda1 and /dev/rd/c0d0p1 */ static int _GL_ATTRIBUTE_PURE _match_rd_device (const char* name) { const char* pos; int state; /* exclude directory names from test */ pos = strrchr(name, '/') ?: name; /* states: * 0 non-digits * 1 digits * 2 non-digits * 3 digits */ for (state = 0; state < 4; state++) { int want_digits = (state % 2 == 1); do { if (!*pos) return 0; if (!_compare_digit_state (*pos, want_digits)) return 0; pos++; } while (_compare_digit_state (*pos, want_digits)); } return *pos == 0; } static int _probe_proc_partitions () { FILE* proc_part_file; int major, minor, size; char buf [512]; char part_name [256]; char dev_name [256]; int ok = 0; proc_part_file = fopen ("/proc/partitions", "r"); if (!proc_part_file) return 0; if (fgets (buf, 256, proc_part_file) == NULL) goto done; if (fgets (buf, 256, proc_part_file) == NULL) goto done; while (fgets (buf, 512, proc_part_file) && sscanf (buf, "%d %d %d %255s", &major, &minor, &size, part_name) == 4) { /* Heuristic for telling partitions and devices apart * Probably needs to be improved */ if (!_match_rd_device (part_name) && isdigit (part_name [strlen (part_name) - 1])) continue; strcpy (dev_name, "/dev/"); strcat (dev_name, part_name); _ped_device_probe (dev_name); } ok = 1; done: fclose (proc_part_file); return ok; } struct _entry { const char *name; size_t len; }; static int _GL_ATTRIBUTE_PURE _skip_entry (const char *name) { struct _entry *i; static struct _entry entries[] = { { ".", sizeof (".") - 1 }, { "..", sizeof ("..") - 1 }, { "dm-", sizeof ("dm-") - 1 }, { "loop", sizeof ("loop") - 1 }, { "ram", sizeof ("ram") - 1 }, { "fd", sizeof ("fd") - 1 }, { 0, 0 }, }; for (i = entries; i->name != 0; i++) { if (strncmp (name, i->name, i->len) == 0) return 1; } return 0; } static int _probe_sys_block () { DIR *blockdir; struct dirent *dirent; char dev_name [256]; char *ptr; if (!(blockdir = opendir ("/sys/block"))) return 0; while ((dirent = readdir (blockdir))) { if (_skip_entry (dirent->d_name)) continue; if (strlen (dirent->d_name) > sizeof (dev_name) - 6) continue; /* device name too long! */ strcpy (dev_name, "/dev/"); strcat (dev_name, dirent->d_name); /* in /sys/block, '/'s are replaced with '!' */ for (ptr = dev_name; *ptr != '\0'; ptr++) { if (*ptr == '!') *ptr = '/'; } _ped_device_probe (dev_name); } closedir (blockdir); return 1; } static int _probe_standard_devices () { _ped_device_probe ("/dev/hda"); _ped_device_probe ("/dev/hdb"); _ped_device_probe ("/dev/hdc"); _ped_device_probe ("/dev/hdd"); _ped_device_probe ("/dev/hde"); _ped_device_probe ("/dev/hdf"); _ped_device_probe ("/dev/hdg"); _ped_device_probe ("/dev/hdh"); _ped_device_probe ("/dev/sda"); _ped_device_probe ("/dev/sdb"); _ped_device_probe ("/dev/sdc"); _ped_device_probe ("/dev/sdd"); _ped_device_probe ("/dev/sde"); _ped_device_probe ("/dev/sdf"); return 1; } static void linux_probe_all () { /* we should probe the standard devs too, even with /proc/partitions, * because /proc/partitions might return devfs stuff, and we might not * have devfs available */ _probe_standard_devices (); #ifdef ENABLE_DEVICE_MAPPER /* device-mapper devices aren't listed in /proc/partitions; or, if * they are, they're listed as dm-X. So, instead of relying on that, * we do our own checks. */ _probe_dm_devices (); #endif /* /sys/block is more reliable and consistent; fall back to using * /proc/partitions if the former is unavailable, however. */ if (!_probe_sys_block ()) _probe_proc_partitions (); } static char * _GL_ATTRIBUTE_FORMAT ((__printf__, 1, 2)) zasprintf (const char *format, ...) { va_list args; char *resultp; va_start (args, format); int r = vasprintf (&resultp, format, args); va_end (args); return r < 0 ? NULL : resultp; } #ifdef ENABLE_DEVICE_MAPPER static char * dm_canonical_path (PedDevice const *dev) { LinuxSpecific const *arch_specific = LINUX_SPECIFIC (dev); /* Get map name from devicemapper */ struct dm_task *task = dm_task_create (DM_DEVICE_INFO); if (!task) goto err; if (!dm_task_set_major_minor (task, arch_specific->major, arch_specific->minor, 0)) goto err; if (!dm_task_run(task)) goto err; char *dev_name = zasprintf ("/dev/mapper/%s", dm_task_get_name (task)); if (dev_name == NULL) goto err; dm_task_destroy (task); return dev_name; err: return NULL; } #endif static char* _device_get_part_path (PedDevice const *dev, int num) { char *devpath; size_t path_len; char *result; #ifdef ENABLE_DEVICE_MAPPER devpath = (dev->type == PED_DEVICE_DM ? dm_canonical_path (dev) : dev->path); #else devpath = dev->path; #endif if (!devpath) return NULL; path_len = strlen (devpath); /* Check for devfs-style /disc => /partN transformation unconditionally; the system might be using udev with devfs rules, and if not the test is harmless. */ if (5 < path_len && !strcmp (devpath + path_len - 5, "/disc")) { /* replace /disc with /part%d */ result = zasprintf ("%.*s/part%d", (int) (path_len - 5), devpath, num); } else { char const *p = (dev->type == PED_DEVICE_DAC960 || dev->type == PED_DEVICE_CPQARRAY || dev->type == PED_DEVICE_ATARAID || isdigit (devpath[path_len - 1]) ? "p" : ""); result = zasprintf ("%s%s%d", devpath, p, num); } #ifdef ENABLE_DEVICE_MAPPER if (dev->type == PED_DEVICE_DM) free (devpath); #endif return result; } static char* linux_partition_get_path (const PedPartition* part) { /* loop label means use the whole disk */ if (strcmp (part->disk->type->name, "loop") == 0) return xstrdup (part->disk->dev->path); return _device_get_part_path (part->disk->dev, part->num); } static int _mount_table_search (const char* file_name, dev_t dev) { struct stat part_stat; char line[512]; char part_name[512]; FILE* file; file = fopen (file_name, "r"); if (!file) return 0; while (fgets (line, 512, file)) { if (sscanf (line, "%s", part_name) == 1 && stat (part_name, &part_stat) == 0) { if (part_stat.st_rdev == dev) { fclose (file); return 1; } } } fclose (file); return 0; } static int _partition_is_mounted_by_dev (dev_t dev) { return _mount_table_search( "/proc/mounts", dev) || _mount_table_search( "/proc/swaps", dev) || _mount_table_search( "/etc/mtab", dev); } static int _partition_is_mounted_by_path (const char *path) { struct stat part_stat; if (stat (path, &part_stat) != 0) return 0; if (!S_ISBLK(part_stat.st_mode)) return 0; return _partition_is_mounted_by_dev (part_stat.st_rdev); } /* If partition PART is mounted, or if we encounter an out-of-memory error while trying to determine its status, return 1. Otherwise, return 0. */ static int _partition_is_mounted (const PedPartition *part) { if (!ped_partition_is_active (part)) return 0; char *part_name = _device_get_part_path (part->disk->dev, part->num); if (!part_name) return 1; int status = _partition_is_mounted_by_path (part_name); free (part_name); return !!status; } static int linux_partition_is_busy (const PedPartition* part) { PedPartition* walk; PED_ASSERT (part != NULL); if (strcmp (part->disk->type->name, "loop") == 0) return linux_is_busy (part->disk->dev); if (_partition_is_mounted (part)) return 1; if (part->type == PED_PARTITION_EXTENDED) { for (walk = part->part_list; walk; walk = walk->next) { if (linux_partition_is_busy (walk)) return 1; } } return 0; } static int _blkpg_part_command (PedDevice* dev, struct blkpg_partition* part, int op) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (dev); struct blkpg_ioctl_arg ioctl_arg; ioctl_arg.op = op; ioctl_arg.flags = 0; ioctl_arg.datalen = sizeof (struct blkpg_partition); ioctl_arg.data = (void*) part; return ioctl (arch_specific->fd, BLKPG, &ioctl_arg) == 0; } static int _blkpg_add_partition (PedDisk* disk, const PedPartition *part) { struct blkpg_partition linux_part; const char* vol_name; char* dev_name; PED_ASSERT(disk != NULL); PED_ASSERT(disk->dev->sector_size % PED_SECTOR_SIZE_DEFAULT == 0); if (ped_disk_type_check_feature (disk->type, PED_DISK_TYPE_PARTITION_NAME)) vol_name = ped_partition_get_name (part); else vol_name = NULL; dev_name = _device_get_part_path (disk->dev, part->num); if (!dev_name) return 0; memset (&linux_part, 0, sizeof (linux_part)); linux_part.start = part->geom.start * disk->dev->sector_size; /* see fs/partitions/msdos.c:msdos_partition(): "leave room for LILO" */ if (part->type & PED_PARTITION_EXTENDED) { linux_part.length = 1; if (disk->dev->sector_size == 512) { if (linux_part.length == 1) linux_part.length = 2; PedPartition *walk; /* if the second sector is claimed by a logical partition, then there's just no room for lilo, so don't try to use it */ for (walk = part->part_list; walk; walk = walk->next) { if (walk->geom.start == part->geom.start+1) linux_part.length = 1; } } linux_part.length *= disk->dev->sector_size; } else { linux_part.length = part->geom.length * disk->dev->sector_size; } linux_part.pno = part->num; strncpy (linux_part.devname, dev_name, BLKPG_DEVNAMELTH-1); linux_part.devname[BLKPG_DEVNAMELTH-1] = '\0'; if (vol_name) { strncpy (linux_part.volname, vol_name, BLKPG_VOLNAMELTH-1); linux_part.volname[BLKPG_VOLNAMELTH-1] = '\0'; } free (dev_name); if (!_blkpg_part_command (disk->dev, &linux_part, BLKPG_ADD_PARTITION)) { return 0; } return 1; } static int _blkpg_remove_partition (PedDisk* disk, int n) { struct blkpg_partition linux_part; memset (&linux_part, 0, sizeof (linux_part)); linux_part.pno = n; return _blkpg_part_command (disk->dev, &linux_part, BLKPG_DEL_PARTITION); } #ifdef BLKPG_RESIZE_PARTITION static int _blkpg_resize_partition (PedDisk* disk, const PedPartition *part) { struct blkpg_partition linux_part; char* dev_name; PED_ASSERT(disk != NULL); PED_ASSERT(disk->dev->sector_size % PED_SECTOR_SIZE_DEFAULT == 0); dev_name = _device_get_part_path (disk->dev, part->num); if (!dev_name) return 0; memset (&linux_part, 0, sizeof (linux_part)); linux_part.start = part->geom.start * disk->dev->sector_size; /* see fs/partitions/msdos.c:msdos_partition(): "leave room for LILO" */ if (part->type & PED_PARTITION_EXTENDED) { if (disk->dev->sector_size == 512) { linux_part.length = 2; PedPartition *walk; /* if the second sector is claimed by a logical partition, then there's just no room for lilo, so don't try to use it */ for (walk = part->part_list; walk; walk = walk->next) { if (walk->geom.start == part->geom.start+1) linux_part.length = 1; } } else { linux_part.length = 1; } linux_part.length *= disk->dev->sector_size; } else linux_part.length = part->geom.length * disk->dev->sector_size; linux_part.pno = part->num; strncpy (linux_part.devname, dev_name, BLKPG_DEVNAMELTH-1); linux_part.devname[BLKPG_DEVNAMELTH-1] = '\0'; free (dev_name); if (!_blkpg_part_command (disk->dev, &linux_part, BLKPG_RESIZE_PARTITION)) { return ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_IGNORE_CANCEL, _("Error informing the kernel about modifications to " "partition %s -- %s. This means Linux won't know " "about any changes you made to %s until you reboot " "-- so you shouldn't mount it or use it in any way " "before rebooting."), linux_part.devname, strerror (errno), linux_part.devname) == PED_EXCEPTION_IGNORE; } return 1; } #endif /* Read the integer from /sys/block/DEV_BASE/ENTRY and set *VAL to that value, where DEV_BASE is the last component of DEV->path. Upon success, return true. Otherwise, return false. */ static bool _sysfs_int_entry_from_dev(PedDevice const* dev, const char *entry, int *val) { char path[128]; int r = snprintf(path, sizeof(path), "/sys/block/%s/%s", last_component(dev->path), entry); if (r < 0 || r >= sizeof(path)) return false; FILE *fp = fopen(path, "r"); if (!fp) return false; bool ok = fscanf(fp, "%d", val) == 1; fclose(fp); return ok; } /* Read the unsigned long long from /sys/block/DEV_BASE/PART_BASE/ENTRY and set *VAL to that value, where DEV_BASE is the last component of path to block device corresponding to PART and PART_BASE is the sysfs name of PART. Upon success, return true. Otherwise, return false. */ static bool _sysfs_ull_entry_from_part(PedPartition const* part, const char *entry, unsigned long long *val) { char path[128]; char *part_name = _device_get_part_path (part->disk->dev, part->num); if (!part_name) return false; int r = snprintf(path, sizeof(path), "/sys/block/%s/%s/%s", last_component(part->disk->dev->path), last_component(part_name), entry); free(part_name); if (r < 0 || r >= sizeof(path)) return false; FILE *fp = fopen(path, "r"); if (!fp) return false; bool ok = fscanf(fp, "%llu", val) == 1; fclose(fp); return ok; } /* Get the starting sector and length of a partition PART within a block device Use blkpg if available, then check sysfs and then use HDIO_GETGEO and BLKGETSIZE64 ioctls as fallback. Upon success, return true. Otherwise, return false. */ static bool _kernel_get_partition_start_and_length(PedPartition const *part, unsigned long long *start, unsigned long long *length) { PED_ASSERT(part); PED_ASSERT(start); PED_ASSERT(length); char *dev_name = _device_get_part_path (part->disk->dev, part->num); if (!dev_name) return false; int ok = _sysfs_ull_entry_from_part (part, "start", start); if (!ok) { struct hd_geometry geom; int dev_fd = open (dev_name, O_RDONLY); if (dev_fd != -1 && ioctl (dev_fd, HDIO_GETGEO, &geom)) { *start = geom.start; close (dev_fd); ok = true; } else { if (dev_fd != -1) close(dev_fd); free (dev_name); return false; } } *start = (*start * 512) / part->disk->dev->sector_size; ok = _sysfs_ull_entry_from_part (part, "size", length); int fd; if (!ok) { fd = open (dev_name, O_RDONLY); if (fd != -1 && ioctl (fd, BLKGETSIZE64, length)) ok = true; } else { fd = -1; *length *= 512; } *length /= part->disk->dev->sector_size; if (fd != -1) close (fd); if (!ok) ped_exception_throw ( PED_EXCEPTION_BUG, PED_EXCEPTION_CANCEL, _("Unable to determine the start and length of %s."), dev_name); free (dev_name); return ok; } /* * The number of partitions that a device can have depends on the kernel. * If we don't find this value in /sys/block/DEV/ext_range, we will use our own * value. */ static unsigned int _device_get_partition_range(PedDevice const* dev) { int range; if (dev->type == PED_DEVICE_DM) return MAX_NUM_PARTS; bool ok = _sysfs_int_entry_from_dev(dev, "ext_range", &range); if (!ok) return MAX_NUM_PARTS; /* both 0 and 1 mean no partitions */ return range > 1 ? range : 0; } #ifdef ENABLE_DEVICE_MAPPER static int _dm_remove_partition(PedDisk* disk, int partno) { int rc = 0; uint32_t cookie = 0; char *part_name = _device_get_part_path (disk->dev, partno); int fd = open (part_name, O_RDONLY | O_EXCL); if (fd == -1) { if (errno == ENOENT) errno = ENXIO; /* nothing to remove, device already doesn't exist */ goto err; } close (fd); struct dm_task *task = dm_task_create(DM_DEVICE_REMOVE); if (!task) goto err; dm_task_set_name (task, part_name); dm_task_retry_remove(task); if (!dm_task_set_cookie (task, &cookie, 0)) goto err; rc = _dm_task_run_wait (task, cookie); dm_task_update_nodes(); dm_task_destroy(task); err: free (part_name); return rc; } static bool _dm_get_partition_start_and_length(PedPartition const *part, unsigned long long *start, unsigned long long *length) { struct dm_task* task = NULL; int rc = 0; if (!(task = dm_task_create(DM_DEVICE_TABLE))) return 0; char *path = _device_get_part_path (part->disk->dev, part->num); PED_ASSERT(path); /* libdevmapper likes to complain on stderr instead of quietly returning ENOENT or ENXIO, so try to stat first */ struct stat st; if (stat(path, &st)) goto err; dm_task_set_name(task, path); if (!dm_task_run(task)) goto err; int major, minor; char *params; char *target_type; dm_get_next_target(task, NULL, (uint64_t *)start, (uint64_t *)length, &target_type, ¶ms); if (sscanf (params, "%d:%d %Lu", &major, &minor, start) != 3) goto err; rc = 1; /* device-mapper uses 512b units, make sure we return length and start in terms of the device's * sector size. */ *start /= (part->disk->dev->sector_size / PED_SECTOR_SIZE_DEFAULT); *length /= (part->disk->dev->sector_size / PED_SECTOR_SIZE_DEFAULT); err: free (path); dm_task_destroy(task); return rc; } static int _dm_add_partition (PedDisk* disk, const PedPartition* part) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (disk->dev); char* params = NULL; char* vol_name = NULL; const char* dev_name = NULL; char* vol_uuid = NULL; const char* dev_uuid = NULL; uint32_t cookie = 0; /* Get map name from devicemapper */ struct dm_task *task = dm_task_create (DM_DEVICE_INFO); if (!task) goto err; if (!dm_task_set_major_minor (task, arch_specific->major, arch_specific->minor, 0)) goto err; if (!dm_task_run(task)) goto err; dev_name = dm_task_get_name (task); size_t name_len = strlen (dev_name); vol_name = zasprintf ("%s%s%d", dev_name, isdigit (dev_name[name_len - 1]) ? "p" : "", part->num); if (vol_name == NULL) goto err; dev_uuid = dm_task_get_uuid (task); if (dev_uuid && (strlen(dev_uuid) > 0) && !(vol_uuid = zasprintf ("part%d-%s", part->num, dev_uuid))) goto err; /* Caution: dm_task_destroy frees dev_name. */ dm_task_destroy (task); task = NULL; /* device-mapper uses 512b units, not the device's sector size */ if ( ! (params = zasprintf ("%d:%d %lld", arch_specific->major, arch_specific->minor, part->geom.start * (disk->dev->sector_size / PED_SECTOR_SIZE_DEFAULT)))) goto err; task = dm_task_create (DM_DEVICE_CREATE); if (!task) goto err; dm_task_set_name (task, vol_name); if (vol_uuid) dm_task_set_uuid (task, vol_uuid); /* device-mapper uses 512b units, not the device's sector size */ dm_task_add_target (task, 0, part->geom.length * (disk->dev->sector_size / PED_SECTOR_SIZE_DEFAULT), "linear", params); if (!dm_task_set_cookie (task, &cookie, 0)) goto err; if (_dm_task_run_wait (task, cookie)) { dm_task_update_nodes (); dm_task_destroy (task); free (params); free (vol_uuid); free (vol_name); return 1; } else { _dm_remove_partition (disk, part->num); } err: dm_task_update_nodes(); if (task) dm_task_destroy (task); free (params); free (vol_uuid); free (vol_name); return 0; } static int _dm_resize_partition (PedDisk* disk, const PedPartition* part) { LinuxSpecific* arch_specific = LINUX_SPECIFIC (disk->dev); char* params = NULL; char* vol_name = NULL; const char* dev_name = NULL; uint32_t cookie = 0; int rc = 0; /* Get map name from devicemapper */ struct dm_task *task = dm_task_create (DM_DEVICE_INFO); if (!task) goto err; if (!dm_task_set_major_minor (task, arch_specific->major, arch_specific->minor, 0)) goto err; if (!dm_task_run(task)) goto err; dev_name = dm_task_get_name (task); size_t name_len = strlen (dev_name); vol_name = zasprintf ("%s%s%d", dev_name, isdigit (dev_name[name_len - 1]) ? "p" : "", part->num); if (vol_name == NULL) goto err; /* Caution: dm_task_destroy frees dev_name. */ dm_task_destroy (task); task = NULL; /* device-mapper uses 512b units, not the device's sector size */ if ( ! (params = zasprintf ("%d:%d %lld", arch_specific->major, arch_specific->minor, part->geom.start * (disk->dev->sector_size / PED_SECTOR_SIZE_DEFAULT)))) goto err; task = dm_task_create (DM_DEVICE_RELOAD); if (!task) goto err; dm_task_set_name (task, vol_name); /* device-mapper uses 512b units, not the device's sector size */ dm_task_add_target (task, 0, part->geom.length * (disk->dev->sector_size / PED_SECTOR_SIZE_DEFAULT), "linear", params); /* NOTE: DM_DEVICE_RELOAD doesn't generate udev events, so no cookie is needed (it will freeze). * DM_DEVICE_RESUME does, so get a cookie and synchronize with udev. */ if (dm_task_run (task)) { dm_task_destroy (task); task = dm_task_create (DM_DEVICE_RESUME); if (!task) goto err; dm_task_set_name (task, vol_name); if (!dm_task_set_cookie (task, &cookie, 0)) goto err; if (_dm_task_run_wait (task, cookie)) { rc = 1; } } err: dm_task_update_nodes(); if (task) dm_task_destroy (task); free (params); free (vol_name); return rc; } #endif /* * Sync the partition table in two step process: * 1. Remove all of the partitions from the kernel's tables, but do not attempt * removal of any partition for which the corresponding ioctl call fails. * 2. Add all the partitions that we hold in disk, throwing a warning * if we cannot because step 1 failed to remove it and it is not being * added back with the same start and length. * * To achieve this two step process we must calculate the minimum number of * maximum possible partitions between what linux supports and what the label * type supports. EX: * * number=MIN(max_parts_supported_in_linux,max_parts_supported_in_msdos_tables) */ static int _disk_sync_part_table (PedDisk* disk) { PED_ASSERT(disk != NULL); PED_ASSERT(disk->dev != NULL); int lpn, lpn2; unsigned int part_range = _device_get_partition_range(disk->dev); int (*add_partition)(PedDisk* disk, const PedPartition *part); int (*resize_partition)(PedDisk* disk, const PedPartition *part); int (*remove_partition)(PedDisk* disk, int partno); bool (*get_partition_start_and_length)(PedPartition const *part, unsigned long long *start, unsigned long long *length); #ifdef ENABLE_DEVICE_MAPPER if (disk->dev->type == PED_DEVICE_DM) { add_partition = _dm_add_partition; remove_partition = _dm_remove_partition; resize_partition = _dm_resize_partition; get_partition_start_and_length = _dm_get_partition_start_and_length; } else #endif { add_partition = _blkpg_add_partition; remove_partition = _blkpg_remove_partition; #ifdef BLKPG_RESIZE_PARTITION resize_partition = _blkpg_resize_partition; #else resize_partition = NULL; #endif get_partition_start_and_length = _kernel_get_partition_start_and_length; } /* lpn = largest partition number. * for remove pass, use greater of device or label limit */ if (ped_disk_get_max_supported_partition_count(disk, &lpn)) lpn = PED_MAX(lpn, part_range); else lpn = part_range; /* for add pass, use lesser of device or label limit */ if (ped_disk_get_max_supported_partition_count(disk, &lpn2)) lpn2 = PED_MIN(lpn2, part_range); else lpn2 = part_range; /* Its not possible to support largest_partnum < 0. * largest_partnum == 0 would mean does not support partitions. * */ if (lpn < 1) return 0; int ret = 0; int *ok = calloc (lpn, sizeof *ok); if (!ok) return 0; int *errnums = ped_malloc(sizeof(int) * lpn); if (!errnums) goto cleanup; int i; /* remove old partitions first */ for (i = 1; i <= lpn; i++) { PedPartition *part = ped_disk_get_partition (disk, i); if (part) { unsigned long long length; unsigned long long start; /* get start and length of existing partition */ if (get_partition_start_and_length(part, &start, &length) && start == part->geom.start && (length == part->geom.length || (resize_partition && part->num < lpn2))) { /* partition is unchanged, or will be resized so nothing to do */ ok[i - 1] = 1; continue; } } /* Attempt to remove the partition, retrying for up to max_sleep_seconds upon any failure due to EBUSY. */ unsigned int sleep_microseconds = 10000; unsigned int max_sleep_seconds = 1; unsigned int n_sleep = (max_sleep_seconds * 1000000 / sleep_microseconds); do { ok[i - 1] = remove_partition (disk, i); errnums[i - 1] = errno; if (ok[i - 1] || errnums[i - 1] != EBUSY) break; usleep (sleep_microseconds); } while (n_sleep--); if (!ok[i - 1] && errnums[i - 1] == ENXIO) ok[i - 1] = 1; /* it already doesn't exist */ } lpn = lpn2; /* don't actually add partitions for loop */ if (strcmp (disk->type->name, "loop") == 0) lpn = 0; for (i = 1; i <= lpn; i++) { PedPartition *part = ped_disk_get_partition (disk, i); if (!part) continue; unsigned long long length; unsigned long long start; /* get start and length of existing partition */ if (get_partition_start_and_length(part, &start, &length) && start == part->geom.start) { if (length == part->geom.length) { ok[i - 1] = 1; /* partition is unchanged, so nothing to do */ continue; } if (resize_partition && start == part->geom.start) { /* try to resize */ if (resize_partition (disk, part)) { ok[i - 1] = 1; continue; } } } /* add the (possibly modified or new) partition */ if (!add_partition (disk, part)) { ok[i - 1] = 0; errnums[i - 1] = errno; } } char *bad_part_list = NULL; /* now warn about any errors */ for (i = 1; i <= lpn; i++) { if (ok[i - 1] || errnums[i - 1] == ENXIO) continue; if (bad_part_list == NULL) { bad_part_list = malloc (lpn * 5); if (!bad_part_list) goto cleanup; bad_part_list[0] = 0; } sprintf (bad_part_list + strlen (bad_part_list), "%d, ", i); } if (bad_part_list == NULL) ret = 1; else { bad_part_list[strlen (bad_part_list) - 2] = 0; if (ped_exception_throw ( PED_EXCEPTION_ERROR, PED_EXCEPTION_IGNORE_CANCEL, _("Partition(s) %s on %s have been written, but we have " "been unable to inform the kernel of the change, " "probably because it/they are in use. As a result, " "the old partition(s) will remain in use. You " "should reboot now before making further changes."), bad_part_list, disk->dev->path) == PED_EXCEPTION_IGNORE) ret = 1; free (bad_part_list); } cleanup: free (errnums); free (ok); return ret; } static int _have_blkpg () { static int have_blkpg = -1; int kver; if (have_blkpg != -1) return have_blkpg; kver = _get_linux_version(); return have_blkpg = kver >= KERNEL_VERSION (2,4,0) ? 1 : 0; } /* Return nonzero upon success, 0 if something fails. */ static int linux_disk_commit (PedDisk* disk) { if (disk->dev->type != PED_DEVICE_FILE) { /* We now require BLKPG support. If this assertion fails, please write to the mailing list describing your system. Assuming it's never triggered, ... FIXME: remove this assertion in 2012. */ assert (_have_blkpg ()); if (!_disk_sync_part_table (disk)) return 0; } return 1; } #if USE_BLKID static PedAlignment* linux_get_minimum_alignment(const PedDevice *dev) { blkid_topology tp = LINUX_SPECIFIC(dev)->topology; if (!tp) return NULL; if (blkid_topology_get_minimum_io_size(tp) == 0) return ped_alignment_new( blkid_topology_get_alignment_offset(tp) / dev->sector_size, dev->phys_sector_size / dev->sector_size); return ped_alignment_new( blkid_topology_get_alignment_offset(tp) / dev->sector_size, blkid_topology_get_minimum_io_size(tp) / dev->sector_size); } static PedAlignment* linux_get_optimum_alignment(const PedDevice *dev) { blkid_topology tp = LINUX_SPECIFIC(dev)->topology; if (!tp) return NULL; /* When PED_DEFAULT_ALIGNMENT is divisible by the *_io_size or there are no *_io_size values, use the PED_DEFAULT_ALIGNMENT If one or the other will not divide evenly, fall through to previous logic. */ unsigned long optimal_io = blkid_topology_get_optimal_io_size(tp); unsigned long minimum_io = blkid_topology_get_minimum_io_size(tp); if ( (!optimal_io && !minimum_io) || (optimal_io && PED_DEFAULT_ALIGNMENT % optimal_io == 0 && minimum_io && PED_DEFAULT_ALIGNMENT % minimum_io == 0) || (!minimum_io && optimal_io && PED_DEFAULT_ALIGNMENT % optimal_io == 0) || (!optimal_io && minimum_io && PED_DEFAULT_ALIGNMENT % minimum_io == 0) ) return ped_alignment_new( blkid_topology_get_alignment_offset(tp) / dev->sector_size, PED_DEFAULT_ALIGNMENT / dev->sector_size); /* If optimal_io_size is 0 and we don't meet the other criteria for using the device.c default, return the minimum alignment. */ if (blkid_topology_get_optimal_io_size(tp) == 0) return linux_get_minimum_alignment(dev); return ped_alignment_new( blkid_topology_get_alignment_offset(tp) / dev->sector_size, blkid_topology_get_optimal_io_size(tp) / dev->sector_size); } #endif #if defined __s390__ || defined __s390x__ /** * Check whether this device could be a DASD * * The device probing yields PED_DEVICE_DASD for native DASD transport * If the block device uses a different transport (e.g. virtio) * a simplified heuristic (assuming a model 3390 with 4K sectors) * is applied (only) on s390x systems for this check. * * \return 1 if the geometry indicates this could be a DASD * and 0 otherwise */ static int _ped_device_like_dasd(const PedDevice *dev) { return (dev->type == PED_DEVICE_DASD) || (dev->hw_geom.heads == 15 && dev->hw_geom.sectors == 12 && (dev->hw_geom.cylinders * dev->hw_geom.heads * dev->hw_geom.sectors * dev->phys_sector_size == dev->length * dev->sector_size)); } static PedAlignment* s390_get_minimum_alignment(const PedDevice *dev) { #if USE_BLKID return linux_get_minimum_alignment(dev); #else return ped_alignment_new(0, dev->phys_sector_size / dev->sector_size); #endif } static PedAlignment* s390_get_optimum_alignment(const PedDevice *dev) { /* DASD needs to use minimum alignment */ if (_ped_device_like_dasd(dev)) return s390_get_minimum_alignment(dev); #if USE_BLKID return linux_get_optimum_alignment(dev); #else return NULL; #endif } #endif static PedDeviceArchOps linux_dev_ops = { _new: linux_new, destroy: linux_destroy, is_busy: linux_is_busy, open: linux_open, refresh_open: linux_refresh_open, close: linux_close, refresh_close: linux_refresh_close, read: linux_read, write: linux_write, check: linux_check, sync: linux_sync, sync_fast: linux_sync_fast, probe_all: linux_probe_all, #if defined __s390__ || defined __s390x__ get_minimum_alignment: s390_get_minimum_alignment, get_optimum_alignment: s390_get_optimum_alignment, #elif USE_BLKID get_minimum_alignment: linux_get_minimum_alignment, get_optimum_alignment: linux_get_optimum_alignment, #endif }; PedDiskArchOps linux_disk_ops = { partition_get_path: linux_partition_get_path, partition_is_busy: linux_partition_is_busy, disk_commit: linux_disk_commit }; PedArchitecture ped_linux_arch = { dev_ops: &linux_dev_ops, disk_ops: &linux_disk_ops };