summaryrefslogtreecommitdiffstats
path: root/libblkid/src/superblocks/udf.c
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--libblkid/src/superblocks/udf.c488
1 files changed, 488 insertions, 0 deletions
diff --git a/libblkid/src/superblocks/udf.c b/libblkid/src/superblocks/udf.c
new file mode 100644
index 0000000..97e79da
--- /dev/null
+++ b/libblkid/src/superblocks/udf.c
@@ -0,0 +1,488 @@
+/*
+ * Copyright (C) 1999 by Andries Brouwer
+ * Copyright (C) 1999, 2000, 2003 by Theodore Ts'o
+ * Copyright (C) 2001 by Andreas Dilger
+ * Copyright (C) 2004 Kay Sievers <kay.sievers@vrfy.org>
+ * Copyright (C) 2008 Karel Zak <kzak@redhat.com>
+ * Copyright (C) 2014-2017 Pali Rohár <pali.rohar@gmail.com>
+ *
+ * This file may be redistributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <string.h>
+#include <errno.h>
+#include <ctype.h>
+#include <stdint.h>
+
+#include "superblocks.h"
+
+#define udf_cid_to_enc(cid) ((cid) == 8 ? BLKID_ENC_LATIN1 : (cid) == 16 ? BLKID_ENC_UTF16BE : -1)
+
+struct dstring128 {
+ uint8_t cid;
+ uint8_t c[126];
+ uint8_t clen;
+} __attribute__((packed));
+
+struct dstring32 {
+ uint8_t cid;
+ uint8_t c[30];
+ uint8_t clen;
+} __attribute__((packed));
+
+struct volume_descriptor {
+ struct descriptor_tag {
+ uint16_t id;
+ uint16_t version;
+ uint8_t checksum;
+ uint8_t reserved;
+ uint16_t serial;
+ uint16_t crc;
+ uint16_t crc_len;
+ uint32_t location;
+ } __attribute__((packed)) tag;
+
+ union {
+ struct anchor_descriptor {
+ uint32_t length;
+ uint32_t location;
+ } __attribute__((packed)) anchor;
+
+ struct primary_descriptor {
+ uint32_t seq_num;
+ uint32_t desc_num;
+ struct dstring32 ident;
+ uint16_t vds_num;
+ uint16_t max_vol_seq;
+ uint16_t ichg_lvl;
+ uint16_t max_ichg_lvl;
+ uint32_t charset_list;
+ uint32_t max_charset_list;
+ struct dstring128 volset_id;
+ } __attribute__((packed)) primary;
+
+ struct logical_descriptor {
+ uint32_t seq_num;
+ uint8_t desc_charset[64];
+ struct dstring128 logvol_id;
+ uint32_t logical_blocksize;
+ uint8_t domain_id_flags;
+ char domain_id[23];
+ uint16_t udf_rev;
+ uint8_t domain_suffix_flags;
+ uint8_t reserved[5];
+ uint8_t logical_contents_use[16];
+ uint32_t map_table_length;
+ uint32_t num_partition_maps;
+ uint8_t imp_id[32];
+ uint8_t imp_use[128];
+ uint32_t lvid_length;
+ uint32_t lvid_location;
+ } __attribute__((packed)) logical;
+
+ struct logical_vol_integ_descriptor {
+ uint8_t recording_date[12];
+ uint32_t type;
+ uint32_t next_lvid_length;
+ uint32_t next_lvid_location;
+ uint8_t logical_contents_use[32];
+ uint32_t num_partitions;
+ uint32_t imp_use_length;
+ } __attribute__((packed)) logical_vol_integ;
+ } __attribute__((packed)) type;
+
+} __attribute__((packed));
+
+#define TAG_ID_PVD 1
+#define TAG_ID_AVDP 2
+#define TAG_ID_LVD 6
+#define TAG_ID_TD 8
+#define TAG_ID_LVID 9
+
+struct volume_structure_descriptor {
+ uint8_t type;
+ uint8_t id[5];
+ uint8_t version;
+} __attribute__((packed));
+
+#define UDF_VSD_OFFSET 0x8000LL
+
+struct logical_vol_integ_descriptor_imp_use
+{
+ uint8_t imp_id[32];
+ uint32_t num_files;
+ uint32_t num_dirs;
+ uint16_t min_udf_read_rev;
+ uint16_t min_udf_write_rev;
+ uint16_t max_udf_write_rev;
+} __attribute__ ((packed));
+
+#define UDF_LVIDIU_OFFSET(vd) (sizeof((vd).tag) + sizeof((vd).type.logical_vol_integ) + 2 * 4 * le32_to_cpu((vd).type.logical_vol_integ.num_partitions))
+#define UDF_LVIDIU_LENGTH(vd) (le32_to_cpu((vd).type.logical_vol_integ.imp_use_length))
+
+static inline int gen_uuid_from_volset_id(unsigned char uuid[17], struct dstring128 *volset_id)
+{
+ int enc;
+ size_t i;
+ size_t len;
+ size_t clen;
+ size_t nonhexpos;
+ unsigned char buf[17];
+
+ memset(buf, 0, sizeof(buf));
+
+ clen = volset_id->clen;
+ if (clen > 0)
+ --clen;
+ if (clen > sizeof(volset_id->c))
+ clen = sizeof(volset_id->c);
+
+ enc = udf_cid_to_enc(volset_id->cid);
+ if (enc == -1)
+ return -1;
+
+ len = blkid_encode_to_utf8(enc, buf, sizeof(buf), volset_id->c, clen);
+ if (len < 8)
+ return -1;
+
+ nonhexpos = 16;
+ for (i = 0; i < 16; ++i) {
+ if (!isxdigit(buf[i])) {
+ nonhexpos = i;
+ break;
+ }
+ }
+
+ if (nonhexpos < 8) {
+ snprintf((char *) uuid, 17, "%02x%02x%02x%02x%02x%02x%02x%02x",
+ buf[0], buf[1], buf[2], buf[3],
+ buf[4], buf[5], buf[6], buf[7]);
+ } else if (nonhexpos < 16) {
+ for (i = 0; i < 8; ++i)
+ uuid[i] = tolower(buf[i]);
+ snprintf((char *) uuid + 8, 9, "%02x%02x%02x%02x",
+ buf[8], buf[9], buf[10], buf[11]);
+ } else {
+ for (i = 0; i < 16; ++i)
+ uuid[i] = tolower(buf[i]);
+ uuid[16] = 0;
+ }
+
+ return 0;
+}
+
+static int probe_udf(blkid_probe pr,
+ const struct blkid_idmag *mag __attribute__((__unused__)))
+{
+ struct volume_descriptor *vd;
+ struct volume_structure_descriptor *vsd;
+ struct logical_vol_integ_descriptor_imp_use *lvidiu;
+ uint32_t lvid_len = 0;
+ uint32_t lvid_loc = 0;
+ uint32_t bs;
+ uint32_t b;
+ uint16_t type;
+ uint32_t count;
+ uint32_t loc;
+ size_t i;
+ uint32_t vsd_len;
+ uint16_t udf_rev = 0;
+ int vsd_2048_valid = -1;
+ int have_label = 0;
+ int have_uuid = 0;
+ int have_logvolid = 0;
+ int have_volid = 0;
+ int have_volsetid = 0;
+
+ /* The block size of a UDF filesystem is that of the underlying
+ * storage; we check later on for the special case of image files,
+ * which may have any block size valid for UDF filesystem */
+ uint32_t pbs[] = { 0, 512, 1024, 2048, 4096 };
+ pbs[0] = blkid_probe_get_sectorsize(pr);
+
+ for (i = 0; i < ARRAY_SIZE(pbs); i++) {
+ /* Do not try with block size same as sector size two times */
+ if (i != 0 && pbs[0] == pbs[i])
+ continue;
+
+ /* ECMA-167 2/8.4, 2/9.1: Each VSD is either 2048 bytes long or
+ * its size is same as blocksize (for blocksize > 2048 bytes)
+ * plus padded with zeros */
+ vsd_len = pbs[i] > 2048 ? pbs[i] : 2048;
+
+ /* Process 2048 bytes long VSD only once */
+ if (vsd_len == 2048) {
+ if (vsd_2048_valid == 0)
+ continue;
+ else if (vsd_2048_valid == 1)
+ goto anchor;
+ }
+
+ /* Check for a Volume Structure Descriptor (VSD) */
+ for (b = 0; b < 64; b++) {
+ vsd = (struct volume_structure_descriptor *)
+ blkid_probe_get_buffer(pr,
+ UDF_VSD_OFFSET + b * vsd_len,
+ sizeof(*vsd));
+ if (!vsd)
+ return errno ? -errno : 1;
+ if (vsd->id[0] == '\0')
+ break;
+ if (memcmp(vsd->id, "NSR02", 5) == 0 ||
+ memcmp(vsd->id, "NSR03", 5) == 0)
+ goto anchor;
+ else if (memcmp(vsd->id, "BEA01", 5) != 0 &&
+ memcmp(vsd->id, "BOOT2", 5) != 0 &&
+ memcmp(vsd->id, "CD001", 5) != 0 &&
+ memcmp(vsd->id, "CDW02", 5) != 0 &&
+ memcmp(vsd->id, "TEA01", 5) != 0)
+ /* ECMA-167 2/8.3.1: The volume recognition sequence is
+ * terminated by the first sector which is not a valid
+ * descriptor.
+ * UDF-2.60 2.1.7: UDF 2.00 and lower revisions do not
+ * have requirement that NSR descritor is in Extended Area
+ * (between BEA01 and TEA01) and that there is only one
+ * Extended Area. So do not stop scanning after TEA01. */
+ break;
+ }
+
+ if (vsd_len == 2048)
+ vsd_2048_valid = 0;
+
+ /* NSR was not found, try with next block size */
+ continue;
+
+anchor:
+ if (vsd_len == 2048)
+ vsd_2048_valid = 1;
+
+ /* Read Anchor Volume Descriptor (AVDP), detect block size */
+ vd = (struct volume_descriptor *)
+ blkid_probe_get_buffer(pr, 256 * pbs[i], sizeof(*vd));
+ if (!vd)
+ return errno ? -errno : 1;
+
+ /* Check that we read correct sector and detected correct block size */
+ if (le32_to_cpu(vd->tag.location) != 256)
+ continue;
+
+ type = le16_to_cpu(vd->tag.id);
+ if (type == TAG_ID_AVDP)
+ goto real_blksz;
+
+ }
+ return 1;
+
+real_blksz:
+ /* Use the actual block size from here on out */
+ bs = pbs[i];
+
+ /* get descriptor list address and block count */
+ count = le32_to_cpu(vd->type.anchor.length) / bs;
+ loc = le32_to_cpu(vd->type.anchor.location);
+
+ /* pick the primary descriptor from the list and read UDF identifiers */
+ for (b = 0; b < count; b++) {
+ vd = (struct volume_descriptor *)
+ blkid_probe_get_buffer(pr,
+ (uint64_t) (loc + b) * bs,
+ sizeof(*vd));
+ if (!vd)
+ return errno ? -errno : 1;
+ type = le16_to_cpu(vd->tag.id);
+ if (type == 0)
+ break;
+ if (le32_to_cpu(vd->tag.location) != loc + b)
+ break;
+ if (type == TAG_ID_TD)
+ break;
+ if (type == TAG_ID_PVD) {
+ if (!have_volid) {
+ int enc = udf_cid_to_enc(vd->type.primary.ident.cid);
+ uint8_t clen = vd->type.primary.ident.clen;
+ if (clen > 0)
+ --clen;
+ if (clen > sizeof(vd->type.primary.ident.c))
+ clen = sizeof(vd->type.primary.ident.c);
+ if (enc != -1)
+ have_volid = !blkid_probe_set_utf8_id_label(pr, "VOLUME_ID",
+ vd->type.primary.ident.c, clen, enc);
+ }
+ if (!have_uuid) {
+ /* VolumeSetIdentifier in UDF 2.01 specification:
+ * =================================================================================
+ * 2.2.2.5 dstring VolumeSetIdentifier
+ *
+ * Interpreted as specifying the identifier for the volume set.
+ *
+ * The first 16 characters of this field should be set to a unique value. The
+ * remainder of the field may be set to any allowed value. Specifically, software
+ * generating volumes conforming to this specification shall not set this field to a
+ * fixed or trivial value. Duplicate disks which are intended to be identical may
+ * contain the same value in this field.
+ *
+ * NOTE: The intended purpose of this is to guarantee Volume Sets with unique
+ * identifiers. The first 8 characters of the unique part should come from a CS0
+ * hexadecimal representation of a 32-bit time value. The remaining 8 characters
+ * are free for implementation use.
+ * =================================================================================
+ *
+ * Implementation in libblkid:
+ * The first 16 (Unicode) characters of VolumeSetIdentifier are encoded to UTF-8
+ * and then first 16 UTF-8 bytes are used to generate UUID. If all 16 bytes are
+ * hexadecimal digits then their lowercase variants are used as UUID. If one of
+ * the first 8 bytes (time value) is not hexadecimal digit then first 8 bytes are
+ * encoded to their hexadecimal representations, resulting in 16 characters and
+ * set as UUID. If all first 8 bytes (time value) are hexadecimal digits but some
+ * remaining not then lowercase variant of the first 8 bytes are used as first
+ * part of UUID and next 4 bytes encoded in hexadecimal representations (resulting
+ * in 8 characters) are used as second part of UUID string.
+ */
+ unsigned char uuid[17];
+ if (gen_uuid_from_volset_id(uuid, &vd->type.primary.volset_id) == 0)
+ have_uuid = !blkid_probe_strncpy_uuid(pr, uuid, sizeof(uuid));
+ }
+ if (!have_volsetid) {
+ int enc = udf_cid_to_enc(vd->type.primary.volset_id.cid);
+ uint8_t clen = vd->type.primary.volset_id.clen;
+ if (clen > 0)
+ --clen;
+ if (clen > sizeof(vd->type.primary.volset_id.c))
+ clen = sizeof(vd->type.primary.volset_id.c);
+ if (enc != -1)
+ have_volsetid = !blkid_probe_set_utf8_id_label(pr, "VOLUME_SET_ID",
+ vd->type.primary.volset_id.c, clen, enc);
+ }
+ } else if (type == TAG_ID_LVD) {
+ if (!lvid_len || !lvid_loc) {
+ uint32_t num_partition_maps = le32_to_cpu(vd->type.logical.num_partition_maps);
+ /* ECMA-167 3/10.6.12: If num_partition_maps is 0, then no LVID is specified */
+ if (num_partition_maps) {
+ lvid_len = le32_to_cpu(vd->type.logical.lvid_length);
+ lvid_loc = le32_to_cpu(vd->type.logical.lvid_location);
+ }
+ }
+ if (!udf_rev) {
+ /* UDF-2.60: 2.1.5.3: UDF revision field shall indicate revision of UDF document
+ * We use this field as fallback value for ID_FS_VERSION when LVIDIU is missing */
+ if (strncmp(vd->type.logical.domain_id, "*OSTA UDF Compliant", sizeof(vd->type.logical.domain_id)) == 0)
+ udf_rev = le16_to_cpu(vd->type.logical.udf_rev);
+ }
+ if (!have_logvolid || !have_label) {
+ /* LogicalVolumeIdentifier in UDF 2.01 specification:
+ * ===============================================================
+ * 2. Basic Restrictions & Requirements
+ *
+ * Logical Volume Descriptor
+ *
+ * There shall be exactly one prevailing Logical Volume
+ * Descriptor recorded per Volume Set.
+ *
+ * The LogicalVolumeIdentifier field shall not be null and
+ * should contain an identifier that aids in the identification of
+ * the logical volume. Specifically, software generating
+ * volumes conforming to this specification shall not set this
+ * field to a fixed or trivial value. Duplicate disks, which are
+ * intended to be identical, may contain the same value in this
+ * field. This field is extremely important in logical volume
+ * identification when multiple media are present within a
+ * jukebox. This name is typically what is displayed to the user.
+ * ===============================================================
+ *
+ * Implementation in libblkid:
+ * The LogicalVolumeIdentifier field is used for LABEL. MS Windows
+ * read Volume Label also from LogicalVolumeIdentifier. Grub2 read
+ * LABEL also from this field. Program newfs_udf (from UDFclient)
+ * when formatting disk set this field from user option Disc Name.
+ */
+ int enc = udf_cid_to_enc(vd->type.logical.logvol_id.cid);
+ uint8_t clen = vd->type.logical.logvol_id.clen;
+ if (clen > 0)
+ --clen;
+ if (clen > sizeof(vd->type.logical.logvol_id.c))
+ clen = sizeof(vd->type.logical.logvol_id.c);
+ if (enc != -1) {
+ if (!have_label)
+ have_label = !blkid_probe_set_utf8label(pr,
+ vd->type.logical.logvol_id.c, clen, enc);
+ if (!have_logvolid)
+ have_logvolid = !blkid_probe_set_utf8_id_label(pr, "LOGICAL_VOLUME_ID",
+ vd->type.logical.logvol_id.c, clen, enc);
+ }
+ }
+ }
+ if (have_volid && have_uuid && have_volsetid && have_logvolid && have_label && lvid_len && lvid_loc)
+ break;
+ }
+
+ /* Pick the first logical volume integrity descriptor and read UDF revision */
+ if (lvid_loc && lvid_len >= sizeof(*vd)) {
+ vd = (struct volume_descriptor *)
+ blkid_probe_get_buffer(pr,
+ (uint64_t) lvid_loc * bs,
+ sizeof(*vd));
+ if (!vd)
+ return errno ? -errno : 1;
+ type = le16_to_cpu(vd->tag.id);
+ if (type == TAG_ID_LVID &&
+ le32_to_cpu(vd->tag.location) == lvid_loc &&
+ UDF_LVIDIU_LENGTH(*vd) >= sizeof(*lvidiu)) {
+ /* ECMA-167 3/8.8.2: There is stored sequence of LVIDs and valid is just last
+ * one. So correctly we should jump to next_lvid_location and read next LVID
+ * until we find last one. This could be time consuming process and could
+ * lead to scanning lot of disk blocks. Because we use LVID only for UDF
+ * version, in the worst case we would report only wrong ID_FS_VERSION. */
+ uint16_t lvidiu_udf_rev;
+ lvidiu = (struct logical_vol_integ_descriptor_imp_use *)
+ blkid_probe_get_buffer(pr,
+ (uint64_t) lvid_loc * bs + UDF_LVIDIU_OFFSET(*vd),
+ sizeof(*lvidiu));
+ if (!lvidiu)
+ return errno ? -errno : 1;
+ /* Use Minimum UDF Read Revision as ID_FS_VERSION */
+ lvidiu_udf_rev = le16_to_cpu(lvidiu->min_udf_read_rev);
+ if (lvidiu_udf_rev)
+ udf_rev = lvidiu_udf_rev;
+ /* UDF-2.60: 2. Basic Restrictions & Requirements:
+ * The Minimum UDF Read Revision value shall be at most #0250
+ * for all media with a UDF 2.60 file system.
+ * So in this case use Minimum UDF Write Revision as ID_FS_VERSION
+ * to distinguish between UDF 2.50 and UDF 2.60 discs. */
+ if (lvidiu_udf_rev == 0x250) {
+ lvidiu_udf_rev = le16_to_cpu(lvidiu->min_udf_write_rev);
+ if (lvidiu_udf_rev > 0x250)
+ udf_rev = lvidiu_udf_rev;
+ }
+ }
+ }
+
+ if (udf_rev)
+ /* UDF revision is stored as decimal number in hexadecimal format.
+ * E.g. number 0x0150 is revision 1.50, number 0x0201 is revision 2.01. */
+ blkid_probe_sprintf_version(pr, "%x.%02x", (unsigned int)(udf_rev >> 8), (unsigned int)(udf_rev & 0xFF));
+
+ return 0;
+}
+
+
+const struct blkid_idinfo udf_idinfo =
+{
+ .name = "udf",
+ .usage = BLKID_USAGE_FILESYSTEM,
+ .probefunc = probe_udf,
+ .flags = BLKID_IDINFO_TOLERANT,
+ .magics =
+ {
+ { .magic = "BEA01", .len = 5, .kboff = 32, .sboff = 1 },
+ { .magic = "BOOT2", .len = 5, .kboff = 32, .sboff = 1 },
+ { .magic = "CD001", .len = 5, .kboff = 32, .sboff = 1 },
+ { .magic = "CDW02", .len = 5, .kboff = 32, .sboff = 1 },
+ { .magic = "NSR02", .len = 5, .kboff = 32, .sboff = 1 },
+ { .magic = "NSR03", .len = 5, .kboff = 32, .sboff = 1 },
+ { .magic = "TEA01", .len = 5, .kboff = 32, .sboff = 1 },
+ { NULL }
+ }
+};