lvm2/lib/device/dev-type.c

/*
 * Copyright (C) 2013 Red Hat, Inc. All rights reserved.
 *
 * This file is part of LVM2.
 *
 * This copyrighted material is made available to anyone wishing to use,
 * modify, copy, or redistribute it subject to the terms and conditions
 * of the GNU Lesser General Public License v.2.1.
 *
 * You should have received a copy of the GNU Lesser General Public License
 * along with this program; if not, write to the Free Software Foundation,
 * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */

#include "lib.h"
#include "dev-type.h"
#include "xlate.h"
#include "config.h"
#include "metadata.h"

#include <libgen.h>
#include <ctype.h>

#ifdef BLKID_WIPING_SUPPORT
#include <blkid/blkid.h>
#endif

#include "device-types.h"

struct dev_types *create_dev_types(const char *proc_dir,
				   const struct dm_config_node *cn)
{
	struct dev_types *dt;
	char line[80];
	char proc_devices[PATH_MAX];
	FILE *pd = NULL;
	int i, j = 0;
	int line_maj = 0;
	int blocksection = 0;
	size_t dev_len = 0;
	const struct dm_config_value *cv;
	const char *name;
	char *nl;

	if (!(dt = dm_zalloc(sizeof(struct dev_types)))) {
		log_error("Failed to allocate device type register.");
		return NULL;
	}

	if (!*proc_dir) {
		log_verbose("No proc filesystem found: using all block device types");
		for (i = 0; i < NUMBER_OF_MAJORS; i++)
			dt->dev_type_array[i].max_partitions = 1;
		return dt;
	}

	if (dm_snprintf(proc_devices, sizeof(proc_devices),
			 "%s/devices", proc_dir) < 0) {
		log_error("Failed to create /proc/devices string");
		goto bad;
	}

	if (!(pd = fopen(proc_devices, "r"))) {
		log_sys_error("fopen", proc_devices);
		goto bad;
	}

	while (fgets(line, sizeof(line), pd) != NULL) {
		i = 0;
		while (line[i] == ' ')
			i++;

		/* If it's not a number it may be name of section */
		line_maj = atoi(((char *) (line + i)));

		if (line_maj < 0 || line_maj >= NUMBER_OF_MAJORS) {
			/*
			 * Device numbers shown in /proc/devices are actually direct
			 * numbers passed to registering function, however the kernel
			 * uses only 12 bits, so use just 12 bits for major.
			 */
			if ((nl = strchr(line, '\n'))) *nl = '\0';
			log_warn("WARNING: /proc/devices line: %s, replacing major with %d.",
				 line, line_maj & (NUMBER_OF_MAJORS - 1));
			line_maj &= (NUMBER_OF_MAJORS - 1);
		}

		if (!line_maj) {
			blocksection = (line[i] == 'B') ? 1 : 0;
			continue;
		}

		/* We only want block devices ... */
		if (!blocksection)
			continue;

		/* Find the start of the device major name */
		while (line[i] != ' ' && line[i] != '\0')
			i++;
		while (line[i] == ' ')
			i++;

		/* Look for md device */
		if (!strncmp("md", line + i, 2) && isspace(*(line + i + 2)))
			dt->md_major = line_maj;

		/* Look for blkext device */
		if (!strncmp("blkext", line + i, 6) && isspace(*(line + i + 6)))
			dt->blkext_major = line_maj;

		/* Look for drbd device */
		if (!strncmp("drbd", line + i, 4) && isspace(*(line + i + 4)))
			dt->drbd_major = line_maj;

		/* Look for EMC powerpath */
		if (!strncmp("emcpower", line + i, 8) && isspace(*(line + i + 8)))
			dt->emcpower_major = line_maj;

		if (!strncmp("power2", line + i, 6) && isspace(*(line + i + 6)))
			dt->power2_major = line_maj;

		/* Look for device-mapper device */
		/* FIXME Cope with multiple majors */
		if (!strncmp("device-mapper", line + i, 13) && isspace(*(line + i + 13)))
			dt->device_mapper_major = line_maj;

		/* Major is SCSI device */
		if (!strncmp("sd", line + i, 2) && isspace(*(line + i + 2)))
			dt->dev_type_array[line_maj].flags |= PARTITION_SCSI_DEVICE;

		/* Go through the valid device names and if there is a
		   match store max number of partitions */
		for (j = 0; _dev_known_types[j].name[0]; j++) {
			dev_len = strlen(_dev_known_types[j].name);
			if (dev_len <= strlen(line + i) &&
			    !strncmp(_dev_known_types[j].name, line + i, dev_len) &&
			    (line_maj < NUMBER_OF_MAJORS)) {
				dt->dev_type_array[line_maj].max_partitions =
					_dev_known_types[j].max_partitions;
				break;
			}
		}

		if (!cn)
			continue;

		/* Check devices/types for local variations */
		for (cv = cn->v; cv; cv = cv->next) {
			if (cv->type != DM_CFG_STRING) {
				log_error("Expecting string in devices/types "
					  "in config file");
				if (fclose(pd))
					log_sys_error("fclose", proc_devices);
				goto bad;
			}
			dev_len = strlen(cv->v.str);
			name = cv->v.str;
			cv = cv->next;
			if (!cv || cv->type != DM_CFG_INT) {
				log_error("Max partition count missing for %s "
					  "in devices/types in config file",
					  name);
				if (fclose(pd))
					log_sys_error("fclose", proc_devices);
				goto bad;
			}
			if (!cv->v.i) {
				log_error("Zero partition count invalid for "
					  "%s in devices/types in config file",
					  name);
				if (fclose(pd))
					log_sys_error("fclose", proc_devices);
				goto bad;
			}
			if (dev_len <= strlen(line + i) &&
			    !strncmp(name, line + i, dev_len) &&
			    (line_maj < NUMBER_OF_MAJORS)) {
				dt->dev_type_array[line_maj].max_partitions = cv->v.i;
				break;
			}
		}
	}

	if (fclose(pd))
		log_sys_error("fclose", proc_devices);

	return dt;
bad:
	dm_free(dt);
	return NULL;
}

int dev_subsystem_part_major(struct dev_types *dt, struct device *dev)
{
	dev_t primary_dev;

	if (MAJOR(dev->dev) == dt->device_mapper_major)
		return 1;

	if (MAJOR(dev->dev) == dt->drbd_major)
		return 1;

	if (MAJOR(dev->dev) == dt->emcpower_major)
		return 1;

	if (MAJOR(dev->dev) == dt->power2_major)
		return 1;

	if ((MAJOR(dev->dev) == dt->blkext_major) &&
	    dev_get_primary_dev(dt, dev, &primary_dev) &&
	    (MAJOR(primary_dev) == dt->md_major))
		return 1;

	return 0;
}

const char *dev_subsystem_name(struct dev_types *dt, struct device *dev)
{
	if (MAJOR(dev->dev) == dt->md_major)
		return "MD";

	if (MAJOR(dev->dev) == dt->drbd_major)
		return "DRBD";

	if (MAJOR(dev->dev) == dt->emcpower_major)
		return "EMCPOWER";

	if (MAJOR(dev->dev) == dt->power2_major)
		return "POWER2";

	if (MAJOR(dev->dev) == dt->blkext_major)
		return "BLKEXT";

	return "";
}

int major_max_partitions(struct dev_types *dt, int major)
{
	if (major >= NUMBER_OF_MAJORS)
		return 0;

	return dt->dev_type_array[major].max_partitions;
}

int major_is_scsi_device(struct dev_types *dt, int major)
{
	if (major >= NUMBER_OF_MAJORS)
		return 0;

	return (dt->dev_type_array[major].flags & PARTITION_SCSI_DEVICE) ? 1 : 0;
}

/* See linux/genhd.h and fs/partitions/msdos */
#define PART_MAGIC 0xAA55
#define PART_MAGIC_OFFSET UINT64_C(0x1FE)
#define PART_OFFSET UINT64_C(0x1BE)

struct partition {
	uint8_t boot_ind;
	uint8_t head;
	uint8_t sector;
	uint8_t cyl;
	uint8_t sys_ind;	/* partition type */
	uint8_t end_head;
	uint8_t end_sector;
	uint8_t end_cyl;
	uint32_t start_sect;
	uint32_t nr_sects;
} __attribute__((packed));

static int _is_partitionable(struct dev_types *dt, struct device *dev)
{
	int parts = major_max_partitions(dt, MAJOR(dev->dev));

	/* All MD devices are partitionable via blkext (as of 2.6.28) */
	if (MAJOR(dev->dev) == dt->md_major)
		return 1;

	if ((parts <= 1) || (MINOR(dev->dev) % parts))
		return 0;

	return 1;
}

static int _has_partition_table(struct device *dev)
{
	int ret = 0;
	unsigned p;
	struct {
		uint8_t skip[PART_OFFSET];
		struct partition part[4];
		uint16_t magic;
	} __attribute__((packed)) buf; /* sizeof() == SECTOR_SIZE */

	if (!dev_read(dev, UINT64_C(0), sizeof(buf), &buf))
		return_0;

	/* FIXME Check for other types of partition table too */

	/* Check for msdos partition table */
	if (buf.magic == xlate16(PART_MAGIC)) {
		for (p = 0; p < 4; ++p) {
			/* Table is invalid if boot indicator not 0 or 0x80 */
			if (buf.part[p].boot_ind & 0x7f) {
				ret = 0;
				break;
			}
			/* Must have at least one non-empty partition */
			if (buf.part[p].nr_sects)
				ret = 1;
		}
	}

	return ret;
}

int dev_is_partitioned(struct dev_types *dt, struct device *dev)
{
	if (!_is_partitionable(dt, dev))
		return 0;

	return _has_partition_table(dev);
}

/*
 * Get primary dev for the dev supplied.
 *
 * We can get a primary device for a partition either by:
 *   A: knowing the number of partitions allowed for the dev and also
 *      which major:minor number represents the primary and partition device
 *      (by using the dev_types->dev_type_array)
 *   B: by the existence of the 'partition' sysfs attribute
 *      (/dev/block/<major>:<minor>/partition)
 *
 * Method A is tried first, then method B as a fallback if A fails.
 *
 * N.B. Method B can only do the decision based on the pure existence of
 *      the 'partition' sysfs item. There's no direct scan for partition
 *      tables whatsoever!
 *
 * Returns:
 *   0 on error
 *   1 if the dev is already a primary dev, primary dev in 'result'
 *   2 if the dev is a partition, primary dev in 'result'
 */
int dev_get_primary_dev(struct dev_types *dt, struct device *dev, dev_t *result)
{
	const char *sysfs_dir = dm_sysfs_dir();
	int major = (int) MAJOR(dev->dev);
	int minor = (int) MINOR(dev->dev);
	char path[PATH_MAX+1];
	char temp_path[PATH_MAX+1];
	char buffer[64];
	struct stat info;
	FILE *fp = NULL;
	int parts, residue, size, ret = 0;

	/*
	 * Try to get the primary dev out of the
	 * list of known device types first.
	 */
	if ((parts = dt->dev_type_array[major].max_partitions) > 1) {
		if ((residue = minor % parts)) {
			*result = MKDEV((dev_t)major, (minor - residue));
			ret = 2;
		} else {
			*result = dev->dev;
			ret = 1; /* dev is not a partition! */
		}
		goto out;
	}

	/*
	 * If we can't get the primary dev out of the list of known device
	 * types, try to look at sysfs directly then. This is more complex
	 * way and it also requires certain sysfs layout to be present
	 * which might not be there in old kernels!
	 */

	/* check if dev is a partition */
	if (dm_snprintf(path, PATH_MAX, "%s/dev/block/%d:%d/partition",
			sysfs_dir, major, minor) < 0) {
		log_error("dm_snprintf partition failed");
		goto out;
	}

	if (stat(path, &info) == -1) {
		if (errno != ENOENT)
			log_sys_error("stat", path);
		*result = dev->dev;
		ret = 1;
		goto out; /* dev is not a partition! */

	}

	/*
	 * extract parent's path from the partition's symlink, e.g.:
	 * - readlink /sys/dev/block/259:0 = ../../block/md0/md0p1
	 * - dirname ../../block/md0/md0p1 = ../../block/md0
	 * - basename ../../block/md0/md0  = md0
	 * Parent's 'dev' sysfs attribute  = /sys/block/md0/dev
	 */
	if ((size = readlink(dirname(path), temp_path, PATH_MAX)) < 0) {
		log_sys_error("readlink", path);
		goto out;
	}

	temp_path[size] = '\0';

	if (dm_snprintf(path, PATH_MAX, "%s/block/%s/dev",
			sysfs_dir, basename(dirname(temp_path))) < 0) {
		log_error("dm_snprintf dev failed");
		goto out;
	}

	/* finally, parse 'dev' attribute and create corresponding dev_t */
	if (stat(path, &info) == -1) {
		if (errno == ENOENT)
			log_error("sysfs file %s does not exist", path);
		else
			log_sys_error("stat", path);
		goto out;
	}

	fp = fopen(path, "r");
	if (!fp) {
		log_sys_error("fopen", path);
		goto out;
	}

	if (!fgets(buffer, sizeof(buffer), fp)) {
		log_sys_error("fgets", path);
		goto out;
	}

	if (sscanf(buffer, "%d:%d", &major, &minor) != 2) {
		log_error("sysfs file %s not in expected MAJ:MIN format: %s",
			  path, buffer);
		goto out;
	}
	*result = MKDEV((dev_t)major, minor);
	ret = 2;
out:
	if (fp && fclose(fp))
		log_sys_error("fclose", path);

	return ret;
}

#ifdef BLKID_WIPING_SUPPORT

static int _blkid_wipe(blkid_probe probe, struct device *dev, const char *name,
		       int exclude_lvm_member, int yes, force_t force)
{
	static const char* msg_failed_offset = "Failed to get offset of the %s signature on %s.";
	static const char* msg_failed_length = "Failed to get length of the %s signature on %s.";
	const char *offset = NULL, *type = NULL, *magic = NULL,
		   *usage = NULL, *label = NULL, *uuid = NULL;
	loff_t offset_value;
	size_t len;

	if (!blkid_probe_lookup_value(probe, "TYPE", &type, NULL)) {
		if (exclude_lvm_member &&
		    (!strcmp(type, "LVM1_member") || !strcmp(type, "LVM2_member")))
			return 1;
		if (blkid_probe_lookup_value(probe, "SBMAGIC_OFFSET", &offset, NULL)) {
			log_error(msg_failed_offset, type, name);
			return 0;
		}
		if (blkid_probe_lookup_value(probe, "SBMAGIC", &magic, &len)) {
			log_error(msg_failed_length, type, name);
			return 0;
		}
	} else if (!blkid_probe_lookup_value(probe, "PTTYPE", &type, NULL)) {
		if (blkid_probe_lookup_value(probe, "PTMAGIC_OFFSET", &offset, NULL)) {
			log_error(msg_failed_offset, type, name);
			return 0;
		}
		if (blkid_probe_lookup_value(probe, "PTMAGIC", &magic, &len)) {
			log_error(msg_failed_length, type, name);
			return 0;
		}
		usage = "partition table";
	} else
		return_0;

	offset_value = strtoll(offset, NULL, 10);

	if (!usage)
		blkid_probe_lookup_value(probe, "USAGE", &usage, NULL);
	blkid_probe_lookup_value(probe, "LABEL", &label, NULL);
	blkid_probe_lookup_value(probe, "UUID", &uuid, NULL);

	log_verbose("Found existing signature on %s at offset %s: LABEL=\"%s\" "
		    "UUID=\"%s\" TYPE=\"%s\" USAGE=\"%s\"",
		     name, offset, label, uuid, type, usage);

	if (!yes && (force == PROMPT) &&
	    yes_no_prompt("WARNING: %s signature detected on %s at offset %s. "
			  "Wipe it? [y/n] ", type, name, offset) != 'y')
		return_0;

	log_print_unless_silent("Wiping %s signature on %s.", type, name);
	if (!dev_set(dev, offset_value, len, 0)) {
		log_error("Failed to wipe %s signature on %s.", type, name);
		return 0;
	}

	return 1;
}

static int _wipe_known_signatures_with_blkid(struct device *dev, const char *name,
					     int exclude_lvm_member,
					     int yes, force_t force)
{
	blkid_probe probe = NULL;
	int found = 0, wiped = 0, left = 0;
	int r = 0;

	/* TODO: Should we check for valid dev - _dev_is_valid(dev)? */

	if (!(probe = blkid_new_probe_from_filename(dev_name(dev)))) {
		log_error("Failed to create a new blkid probe for device %s.", dev_name(dev));
		goto out;
	}

	blkid_probe_enable_partitions(probe, 1);
	blkid_probe_set_partitions_flags(probe, BLKID_PARTS_MAGIC);

	blkid_probe_enable_superblocks(probe, 1);
	blkid_probe_set_superblocks_flags(probe, BLKID_SUBLKS_LABEL |
						 BLKID_SUBLKS_UUID |
						 BLKID_SUBLKS_TYPE |
						 BLKID_SUBLKS_USAGE |
						 BLKID_SUBLKS_VERSION |
						 BLKID_SUBLKS_MAGIC |
						 BLKID_SUBLKS_BADCSUM);

	while (!blkid_do_probe(probe)) {
		found++;
		if (_blkid_wipe(probe, dev, name, exclude_lvm_member, yes, force))
			wiped++;
	}

	if (!found)
		r = 1;

	left = found - wiped;
	if (!left)
		r = 1;
	else
		log_warn("%d existing signature%s left on the device.",
			  left, left > 1 ? "s" : "");
out:
	if (probe)
		blkid_free_probe(probe);
	return r;
}

#endif /* BLKID_WIPING_SUPPORT */

static int _wipe_signature(struct device *dev, const char *type, const char *name,
			   int wipe_len, int yes, force_t force,
			   int (*signature_detection_fn)(struct device *dev, uint64_t *offset_found))
{
	int wipe;
	uint64_t offset_found;

	wipe = signature_detection_fn(dev, &offset_found);
	if (wipe == -1) {
		log_error("Fatal error while trying to detect %s on %s.",
			  type, name);
		return 0;
	}

	if (wipe == 0)
		return 1;

	/* Specifying --yes => do not ask. */
	if (!yes && (force == PROMPT) &&
	    yes_no_prompt("WARNING: %s detected on %s. Wipe it? [y/n] ",
			  type, name) != 'y')
		return_0;

	log_print_unless_silent("Wiping %s on %s.", type, name);
	if (!dev_set(dev, offset_found, wipe_len, 0)) {
		log_error("Failed to wipe %s on %s.", type, name);
		return 0;
	}

	return 1;
}

static int _wipe_known_signatures_with_lvm(struct device *dev, const char *name,
					   int exclude_lvm_member,
					   int yes, force_t force)
{
	if (!_wipe_signature(dev, "software RAID md superblock", name, 4, yes, force, dev_is_md) ||
	    !_wipe_signature(dev, "swap signature", name, 10, yes, force, dev_is_swap) ||
	    !_wipe_signature(dev, "LUKS signature", name, 8, yes, force, dev_is_luks))
		return 0;

	return 1;
}

int wipe_known_signatures(struct cmd_context *cmd, struct device *dev,
			  const char *name, int exclude_lvm_member,
			  int yes, force_t force)
{
#ifdef BLKID_WIPING_SUPPORT
	if (find_config_tree_bool(cmd, allocation_use_blkid_wiping_CFG, NULL))
		return _wipe_known_signatures_with_blkid(dev, name,
				exclude_lvm_member, yes, force);
#endif
	return _wipe_known_signatures_with_lvm(dev, name,
			exclude_lvm_member, yes, force);
}

#ifdef __linux__

static int _snprintf_attr(char *buf, size_t buf_size, const char *sysfs_dir,
			 const char *attribute, dev_t dev)
{
	if (dm_snprintf(buf, buf_size, "%s/dev/block/%d:%d/%s", sysfs_dir,
			(int)MAJOR(dev), (int)MINOR(dev),
			attribute) < 0) {
		log_warn("dm_snprintf %s failed.", attribute);
		return 0;
	}

	return 1;
}

static unsigned long _dev_topology_attribute(struct dev_types *dt,
					     const char *attribute,
					     struct device *dev)
{
	const char *sysfs_dir = dm_sysfs_dir();
	char path[PATH_MAX], buffer[64];
	FILE *fp;
	struct stat info;
	dev_t uninitialized_var(primary);
	unsigned long result = 0UL;

	if (!attribute || !*attribute)
		return_0;

	if (!sysfs_dir || !*sysfs_dir)
		return_0;

	if (!_snprintf_attr(path, sizeof(path), sysfs_dir, attribute, dev->dev))
                return_0;

	/*
	 * check if the desired sysfs attribute exists
	 * - if not: either the kernel doesn't have topology support
	 *   or the device could be a partition
	 */
	if (stat(path, &info) == -1) {
		if (errno != ENOENT) {
			log_sys_error("stat", path);
			return 0;
		}
		if (!dev_get_primary_dev(dt, dev, &primary))
			return 0;

		/* get attribute from partition's primary device */
		if (!_snprintf_attr(path, sizeof(path), sysfs_dir, attribute, primary))
			return_0;

		if (stat(path, &info) == -1) {
			if (errno != ENOENT)
				log_sys_error("stat", path);
			return 0;
		}
	}

	if (!(fp = fopen(path, "r"))) {
		log_sys_error("fopen", path);
		return 0;
	}

	if (!fgets(buffer, sizeof(buffer), fp)) {
		log_sys_error("fgets", path);
		goto out;
	}

	if (sscanf(buffer, "%lu", &result) != 1) {
		log_error("sysfs file %s not in expected format: %s", path,
			  buffer);
		goto out;
	}

	log_very_verbose("Device %s %s is %lu bytes.",
			 dev_name(dev), attribute, result);

out:
	if (fclose(fp))
		log_sys_error("fclose", path);

	return result >> SECTOR_SHIFT;
}

unsigned long dev_alignment_offset(struct dev_types *dt, struct device *dev)
{
	return _dev_topology_attribute(dt, "alignment_offset", dev);
}

unsigned long dev_minimum_io_size(struct dev_types *dt, struct device *dev)
{
	return _dev_topology_attribute(dt, "queue/minimum_io_size", dev);
}

unsigned long dev_optimal_io_size(struct dev_types *dt, struct device *dev)
{
	return _dev_topology_attribute(dt, "queue/optimal_io_size", dev);
}

unsigned long dev_discard_max_bytes(struct dev_types *dt, struct device *dev)
{
	return _dev_topology_attribute(dt, "queue/discard_max_bytes", dev);
}

unsigned long dev_discard_granularity(struct dev_types *dt, struct device *dev)
{
	return _dev_topology_attribute(dt, "queue/discard_granularity", dev);
}

#else

int dev_get_primary_dev(struct dev_types *dt, struct device *dev, dev_t *result)
{
	return 0;
}

unsigned long dev_alignment_offset(struct dev_types *dt, struct device *dev)
{
	return 0UL;
}

unsigned long dev_minimum_io_size(struct dev_types *dt, struct device *dev)
{
	return 0UL;
}

unsigned long dev_optimal_io_size(struct dev_types *dt, struct device *dev)
{
	return 0UL;
}

unsigned long dev_discard_max_bytes(struct dev_types *dt, struct device *dev)
{
	return 0UL;
}

unsigned long dev_discard_granularity(struct dev_types *dt, struct device *dev)
{
	return 0UL;
}

#endif