OSDN Git Service

mkdosfs (aka mkfs.vfat): new applet by Vladimir
authorDenis Vlasenko <vda.linux@googlemail.com>
Sat, 28 Mar 2009 02:13:01 +0000 (02:13 -0000)
committerDenis Vlasenko <vda.linux@googlemail.com>
Sat, 28 Mar 2009 02:13:01 +0000 (02:13 -0000)
   text    data     bss     dec     hex filename
 821202     476    7616  829294   ca76e busybox_old
 823577     476    7616  831669   cb0b5 busybox_unstripped

include/applets.h
include/libbb.h
include/usage.h
util-linux/Config.in
util-linux/Kbuild
util-linux/mkfs_vfat.c [new file with mode: 0644]

index becb140..1d93258 100644 (file)
@@ -250,11 +250,13 @@ USE_MDEV(APPLET(mdev, _BB_DIR_SBIN, _BB_SUID_NEVER))
 USE_MESG(APPLET(mesg, _BB_DIR_USR_BIN, _BB_SUID_NEVER))
 USE_MICROCOM(APPLET(microcom, _BB_DIR_USR_BIN, _BB_SUID_NEVER))
 USE_MKDIR(APPLET_NOFORK(mkdir, mkdir, _BB_DIR_BIN, _BB_SUID_NEVER, mkdir))
+USE_MKFS_VFAT(APPLET_ODDNAME(mkdosfs, mkfs_vfat, _BB_DIR_SBIN, _BB_SUID_NEVER, mkfs_vfat))
 //USE_MKE2FS(APPLET(mke2fs, _BB_DIR_SBIN, _BB_SUID_NEVER))
 USE_MKFIFO(APPLET(mkfifo, _BB_DIR_USR_BIN, _BB_SUID_NEVER))
 //USE_MKE2FS(APPLET_ODDNAME(mkfs.ext2, mke2fs, _BB_DIR_SBIN, _BB_SUID_NEVER, mkfs_ext2))
 //USE_MKE2FS(APPLET_ODDNAME(mkfs.ext3, mke2fs, _BB_DIR_SBIN, _BB_SUID_NEVER, mkfs_ext3))
 USE_MKFS_MINIX(APPLET_ODDNAME(mkfs.minix, mkfs_minix, _BB_DIR_SBIN, _BB_SUID_NEVER, mkfs_minix))
+USE_MKFS_VFAT(APPLET_ODDNAME(mkfs.vfat, mkfs_vfat, _BB_DIR_SBIN, _BB_SUID_NEVER, mkfs_vfat))
 USE_MKNOD(APPLET(mknod, _BB_DIR_BIN, _BB_SUID_NEVER))
 USE_CRYPTPW(APPLET_ODDNAME(mkpasswd, cryptpw, _BB_DIR_USR_BIN, _BB_SUID_NEVER, mkpasswd))
 USE_MKSWAP(APPLET(mkswap, _BB_DIR_SBIN, _BB_SUID_NEVER))
index 015374b..c2e0344 100644 (file)
@@ -150,6 +150,7 @@ int sysinfo(struct sysinfo* info);
 /* CONFIG_LFS is on */
 # if ULONG_MAX > 0xffffffff
 /* "long" is long enough on this system */
+typedef unsigned long uoff_t;
 #  define XATOOFF(a) xatoul_range(a, 0, LONG_MAX)
 /* usage: sz = BB_STRTOOFF(s, NULL, 10); if (errno || sz < 0) die(); */
 #  define BB_STRTOOFF bb_strtoul
@@ -158,6 +159,7 @@ int sysinfo(struct sysinfo* info);
 #  define OFF_FMT "l"
 # else
 /* "long" is too short, need "long long" */
+typedef unsigned long long uoff_t;
 #  define XATOOFF(a) xatoull_range(a, 0, LLONG_MAX)
 #  define BB_STRTOOFF bb_strtoull
 #  define STRTOOFF strtoull
@@ -168,11 +170,13 @@ int sysinfo(struct sysinfo* info);
 # if UINT_MAX == 0xffffffff
 /* While sizeof(off_t) == sizeof(int), off_t is typedef'ed to long anyway.
  * gcc will throw warnings on printf("%d", off_t). Crap... */
+typedef unsigned long uoff_t;
 #  define XATOOFF(a) xatoi_u(a)
 #  define BB_STRTOOFF bb_strtou
 #  define STRTOOFF strtol
 #  define OFF_FMT "l"
 # else
+typedef unsigned long uoff_t;
 #  define XATOOFF(a) xatoul_range(a, 0, LONG_MAX)
 #  define BB_STRTOOFF bb_strtoul
 #  define STRTOOFF strtol
index 2b5d34e..16d5eec 100644 (file)
      "\n       -l FILENAME     Read bad blocks list from FILENAME" \
      "\n       -v              Make version 2 filesystem" \
 
+#define mkfs_vfat_trivial_usage \
+       "[-v] [-n LABEL] FILE_OR_DEVICE [SIZE_IN_KB]"
+/* Accepted but ignored:
+       "[-c] [-C] [-I] [-l bad-block-file] [-b backup-boot-sector] "
+       "[-m boot-msg-file] [-i volume-id] "
+       "[-s sectors-per-cluster] [-S logical-sector-size] [-f number-of-FATs] "
+       "[-h hidden-sectors] [-F fat-size] [-r root-dir-entries] [-R reserved-sectors] "
+*/
+#define mkfs_vfat_full_usage "\n\n" \
+       "Make a FAT32 filesystem\n" \
+     "\nOptions:" \
+/*   "\n       -c      Check device for bad blocks" */ \
+     "\n       -v      Verbose" \
+/*   "\n       -I      Allow to use entire disk device (e.g. /dev/hda)" */ \
+     "\n       -n LBL  Volume label" \
+
 #define mknod_trivial_usage \
        "[OPTIONS] NAME TYPE MAJOR MINOR"
 #define mknod_full_usage "\n\n" \
index 3bba2e2..e5c053f 100644 (file)
@@ -233,6 +233,12 @@ config FEATURE_MINIX2
          this. If you enabled 'mkfs_minix' then you almost certainly want to
          be using the version 2 filesystem support.
 
+config MKFS_VFAT
+       bool "mkfs_vfat"
+       default n
+       help
+         Utility to create FAT32 filesystems.
+
 config GETOPT
        bool "getopt"
        default n
index 842afb7..ac071c6 100644 (file)
@@ -23,6 +23,7 @@ lib-$(CONFIG_IPCS)              += ipcs.o
 lib-$(CONFIG_LOSETUP)           += losetup.o
 lib-$(CONFIG_MDEV)              += mdev.o
 lib-$(CONFIG_MKFS_MINIX)        += mkfs_minix.o
+lib-$(CONFIG_MKFS_VFAT)         += mkfs_vfat.o
 lib-$(CONFIG_MKSWAP)            += mkswap.o
 lib-$(CONFIG_MORE)              += more.o
 lib-$(CONFIG_MOUNT)             += mount.o
diff --git a/util-linux/mkfs_vfat.c b/util-linux/mkfs_vfat.c
new file mode 100644 (file)
index 0000000..930fd45
--- /dev/null
@@ -0,0 +1,593 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * mkfs_vfat: utility to create FAT32 filesystem
+ * inspired by dosfstools
+ *
+ * Busybox'ed (2009) by Vladimir Dronnikov <dronnikov@gmail.com>
+ *
+ * Licensed under GPLv2, see file LICENSE in this tarball for details.
+ */
+#include "libbb.h"
+#include "volume_id/volume_id_internal.h"
+
+#include <linux/hdreg.h> /* HDIO_GETGEO */
+#include <linux/fd.h>    /* FDGETPRM */
+//#include <linux/msdos_fs.h>
+
+#define SECTOR_SIZE             512
+
+#define SECTORS_PER_BLOCK      (BLOCK_SIZE / SECTOR_SIZE)
+
+// M$ says the high 4 bits of a FAT32 FAT entry are reserved
+#define EOF_FAT32       0x0FFFFFF8
+#define BAD_FAT32       0x0FFFFFF7
+#define MAX_CLUST_32    0x0FFFFFF0
+
+#define ATTR_VOLUME     8
+
+#define        NUM_FATS        2
+
+/* FAT32 filesystem looks like this:
+ * sector -nn...-1: "hidden" sectors, all sectors before this partition
+ * (-h hidden-sectors sets it. Useful only for boot loaders,
+ *  they need to know _disk_ offset in order to be able to correctly
+ *  address sectors relative to start of disk)
+ * sector 0: boot sector
+ * sector 1: info sector
+ * sector 2: set aside for boot code which didn't fit into sector 0
+ * ...(zero-filled sectors)...
+ * sector B: backup copy of sector 0 [B set by -b backup-boot-sector]
+ * sector B+1: backup copy of sector 1
+ * sector B+2: backup copy of sector 2
+ * ...(zero-filled sectors)...
+ * sector R: FAT#1 [R set by -R reserved-sectors]
+ * ...(FAT#1)...
+ * sector R+fat_size: FAT#2
+ * ...(FAT#2)...
+ * sector R+fat_size*2: cluster #2
+ * ...(cluster #2)...
+ * sector R+fat_size*2+clust_size: cluster #3
+ * ...(the rest is filled by clusters till the end)...
+ */
+
+enum {
+// Perhaps this should remain constant
+       info_sector_number = 1,
+// TODO: make these cmdline options
+// dont forget sanity check: backup_boot_sector + 3 <= reserved_sect
+       backup_boot_sector = 3,
+       reserved_sect      = 6,
+};
+
+// how many blocks we try to read while testing
+#define TEST_BUFFER_BLOCKS      16
+
+struct msdos_dir_entry {
+       char     name[11];       /* 000 name and extension */
+       uint8_t  attr;           /* 00b attribute bits */
+       uint8_t  lcase;          /* 00c case for base and extension */
+       uint8_t  ctime_cs;       /* 00d creation time, centiseconds (0-199) */
+       uint16_t ctime;          /* 00e creation time */
+       uint16_t cdate;          /* 010 creation date */
+       uint16_t adate;          /* 012 last access date */
+       uint16_t starthi;        /* 014 high 16 bits of cluster in FAT32 */
+       uint16_t time;           /* 016 time */
+       uint16_t date;           /* 018 date */
+       uint16_t start;          /* 01a first cluster */
+       uint32_t size;           /* 01c file size in bytes */
+} __attribute__ ((packed));
+
+/* Example of boot sector's beginning:
+0000  eb 58 90 4d 53 57 49 4e  34 2e 31 00 02 08 26 00  |...MSWIN4.1...&.|
+0010  02 00 00 00 00 f8 00 00  3f 00 ff 00 3f 00 00 00  |........?...?...|
+0020  54 9b d0 00 0d 34 00 00  00 00 00 00 02 00 00 00  |T....4..........|
+0030  01 00 06 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
+0040  80 00 29 71 df 51 e0 4e  4f 20 4e 41 4d 45 20 20  |..)q.Q.NO NAME  |
+0050  20 20 46 41 54 33 32 20  20 20 33 c9 8e d1 bc f4  |  FAT32   3.....|
+*/
+struct msdos_volume_info { /* (offsets are relative to start of boot sector) */
+       uint8_t  drive_number;    /* 040 BIOS drive number */
+       uint8_t  reserved;        /* 041 unused */
+       uint8_t  ext_boot_sign;   /* 042 0x29 if fields below exist (DOS 3.3+) */
+       uint32_t volume_id32;     /* 043 volume ID number */
+       char     volume_label[11];/* 047 volume label */
+       char     fs_type[8];      /* 052 typically "FATnn" */
+} __attribute__ ((packed));       /* 05a end. Total size 26 (0x1a) bytes */
+
+struct msdos_boot_sector {
+       char     boot_jump[3];       /* 000 short or near jump instruction */
+       char     system_id[8];       /* 003 name - can be used to special case partition manager volumes */
+       uint16_t bytes_per_sect;     /* 00b bytes per logical sector */
+       uint8_t  sect_per_clust;     /* 00d sectors/cluster */
+       uint16_t reserved_sect;      /* 00e reserved sectors (sector offset of 1st FAT relative to volume start) */
+       uint8_t  fats;               /* 010 number of FATs */
+       uint16_t dir_entries;        /* 011 root directory entries */
+       uint16_t volume_size_sect;   /* 013 volume size in sectors */
+       uint8_t  media_byte;         /* 015 media code */
+       uint16_t sect_per_fat;       /* 016 sectors/FAT */
+       uint16_t sect_per_track;     /* 018 sectors per track */
+       uint16_t heads;              /* 01a number of heads */
+       uint32_t hidden;             /* 01c hidden sectors (sector offset of volume within physical disk) */
+       uint32_t fat32_volume_size_sect; /* 020 volume size in sectors (if volume_size_sect == 0) */
+       uint32_t fat32_sect_per_fat; /* 024 sectors/FAT */
+       uint16_t fat32_flags;        /* 028 bit 8: fat mirroring, low 4: active fat */
+       uint8_t  fat32_version[2];   /* 02a major, minor filesystem version (I see 0,0) */
+       uint32_t fat32_root_cluster; /* 02c first cluster in root directory */
+       uint16_t fat32_info_sector;  /* 030 filesystem info sector (usually 1) */
+       uint16_t fat32_backup_boot;  /* 032 backup boot sector (usually 6) */
+       uint32_t reserved2[3];       /* 034 unused */
+       struct msdos_volume_info vi; /* 040 */
+       char     boot_code[0x200 - 0x5a - 2]; /* 05a */
+#define BOOT_SIGN 0xAA55
+       uint16_t boot_sign;          /* 1fe */
+} __attribute__ ((packed));
+
+#define FAT_FSINFO_SIG1 0x41615252
+#define FAT_FSINFO_SIG2 0x61417272
+struct fat32_fsinfo {
+       uint32_t signature1;         /* 0x52,0x52,0x41,0x61, "RRaA" */
+       uint32_t reserved1[128 - 8];
+       uint32_t signature2;         /* 0x72,0x72,0x61,0x41, "rrAa" */
+       uint32_t free_clusters;      /* free cluster count.  -1 if unknown */
+       uint32_t next_cluster;       /* most recently allocated cluster */
+       uint32_t reserved2[3];
+       uint16_t reserved3;          /* 1fc */
+       uint16_t boot_sign;          /* 1fe */
+} __attribute__ ((packed));
+
+struct bug_check {
+       char BUG1[sizeof(struct msdos_dir_entry  ) == 0x20 ? 1 : -1];
+       char BUG2[sizeof(struct msdos_volume_info) == 0x1a ? 1 : -1];
+       char BUG3[sizeof(struct msdos_boot_sector) == 0x200 ? 1 : -1];
+       char BUG4[sizeof(struct fat32_fsinfo     ) == 0x200 ? 1 : -1];
+};
+
+static const char boot_code[] ALIGN1 =
+       "\x0e"          /* 05a:         push  cs */
+       "\x1f"          /* 05b:         pop   ds */
+       "\xbe\x77\x7c"  /*  write_msg:  mov   si, offset message_txt */
+       "\xac"          /* 05f:         lodsb */
+       "\x22\xc0"      /* 060:         and   al, al */
+       "\x74\x0b"      /* 062:         jz    key_press */
+       "\x56"          /* 064:         push  si */
+       "\xb4\x0e"      /* 065:         mov   ah, 0eh */
+       "\xbb\x07\x00"  /* 067:         mov   bx, 0007h */
+       "\xcd\x10"      /* 06a:         int   10h */
+       "\x5e"          /* 06c:         pop   si */
+       "\xeb\xf0"      /* 06d:         jmp   write_msg */
+       "\x32\xe4"      /*  key_press:  xor   ah, ah */
+       "\xcd\x16"      /* 071:         int   16h */
+       "\xcd\x19"      /* 073:         int   19h */
+       "\xeb\xfe"      /*  foo:        jmp   foo */
+       /* 077: message_txt: */
+       "This is not a bootable disk\r\n";
+
+
+// mark specified cluster as having a particular value
+#define        mark_cluster(cluster, value) \
+       ((uint32_t *)fat)[cluster] = cpu_to_le32(value)
+
+void BUG_unsupported_field_size(void);
+#define STORE_LE(field, value) \
+do { \
+       if (sizeof(field) == 4) \
+               field = cpu_to_le32(value); \
+       else if (sizeof(field) == 2) \
+               field = cpu_to_le16(value); \
+       else if (sizeof(field) == 1) \
+               field = (value); \
+       else \
+               BUG_unsupported_field_size(); \
+} while (0)
+
+/* compat:
+ * mkdosfs 2.11 (12 Mar 2005)
+ * Usage: mkdosfs [-A] [-c] [-C] [-v] [-I] [-l bad-block-file]
+ *        [-b backup-boot-sector]
+ *        [-m boot-msg-file] [-n volume-name] [-i volume-id]
+ *        [-s sectors-per-cluster] [-S logical-sector-size]
+ *        [-f number-of-FATs]
+ *        [-h hidden-sectors] [-F fat-size] [-r root-dir-entries]
+ *        [-R reserved-sectors]
+ *        /dev/name [blocks]
+ */
+int mkfs_vfat_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
+int mkfs_vfat_main(int argc UNUSED_PARAM, char **argv)
+{
+       struct stat st;
+       const char *volume_label = "";
+       char *buf;
+       char *device_name;
+       uoff_t volume_size_bytes;
+       uoff_t volume_size_blocks;
+       uoff_t volume_size_sect;
+       uint32_t total_clust;
+       uint32_t volume_id;
+       int dev;
+       unsigned bytes_per_sect;
+       unsigned sect_per_fat;
+       unsigned opts;
+       uint16_t sect_per_track;
+       uint8_t media_byte;
+       uint8_t sect_per_clust;
+       uint8_t heads;
+       enum {
+               OPT_A = 1 << 0,  // [IGNORED] atari format
+               OPT_b = 1 << 1,  // [IGNORED] location of backup boot sector
+               OPT_c = 1 << 2,  // [IGNORED] check filesystem
+               OPT_C = 1 << 3,  // [IGNORED] create a new file
+               OPT_f = 1 << 4,  // [IGNORED] number of FATs
+               OPT_F = 1 << 5,  // [IGNORED, implied 32] choose FAT size
+               OPT_h = 1 << 6,  // [IGNORED] number of hidden sectors
+               OPT_I = 1 << 7,  // [IGNORED] don't bark at entire disk devices
+               OPT_i = 1 << 8,  // [IGNORED] volume ID
+               OPT_l = 1 << 9,  // [IGNORED] bad block filename
+               OPT_m = 1 << 10, // [IGNORED] message file
+               OPT_n = 1 << 11, // volume label
+               OPT_r = 1 << 12, // [IGNORED] root directory entries
+               OPT_R = 1 << 13, // [IGNORED] number of reserved sectors
+               OPT_s = 1 << 14, // [IGNORED] sectors per cluster
+               OPT_S = 1 << 15, // [IGNORED] sector size
+               OPT_v = 1 << 16, // verbose
+       };
+
+       opt_complementary = "-1";//:b+:f+:F+:h+:r+:R+:s+:S+:vv:c--l:l--c";
+       opts = getopt32(argv, "Ab:cCf:F:h:Ii:l:m:n:r:R:s:S:v",
+               NULL, NULL, NULL, NULL, NULL,
+               NULL, NULL, &volume_label, NULL, NULL, NULL, NULL);
+       argv += optind;
+
+       // cache device name
+       device_name = argv[0];
+       // default volume ID = creation time
+       volume_id = time(NULL);
+
+       dev = xopen(device_name, O_EXCL | O_RDWR);
+       if (fstat(dev, &st) < 0)
+               bb_simple_perror_msg_and_die(device_name);
+
+       //
+       // Get image size and sector size
+       //
+       bytes_per_sect = SECTOR_SIZE;
+       volume_size_bytes = st.st_size;
+       if (!S_ISBLK(st.st_mode)) {
+               if (!S_ISREG(st.st_mode)) {
+                       if (!argv[1])
+                               bb_error_msg_and_die("image size must be specified");
+               }
+               // not a block device, skip bad sectors check
+               opts &= ~OPT_c;
+       } else {
+               int min_bytes_per_sect;
+
+               // more portable than BLKGETSIZE[64]
+               volume_size_bytes = xlseek(dev, 0, SEEK_END);
+               xlseek(dev, 0, SEEK_SET);
+#if 0
+               unsigned device_num;
+               // for true block devices we do check sanity
+               device_num = st.st_rdev & 0xff3f;
+               // do we allow to format the whole disk device?
+               if (!(opts & OPT_I) && (
+                       device_num == 0x0300 || // hda, hdb
+                       (device_num & 0xff0f) == 0x0800 || // sd
+                       device_num == 0x0d00 || // xd
+                       device_num == 0x1600 )  // hdc, hdd
+               )
+                       bb_error_msg_and_die("Will not try to make filesystem on full-disk device (use -I if wanted)");
+               // can't work on mounted filesystems
+               if (find_mount_point(device_name, NULL))
+                       bb_error_msg_and_die("Can't format mounted filesystem");
+#endif
+               // get true sector size
+               // (parameter must be int*, not long* or size_t*)
+               xioctl(dev, BLKSSZGET, &min_bytes_per_sect);
+               if (min_bytes_per_sect > SECTOR_SIZE) {
+                       bytes_per_sect = min_bytes_per_sect;
+                       bb_error_msg("for this device sector size is %u", min_bytes_per_sect);
+               }
+       }
+       if (argv[1]) {
+               volume_size_bytes = XATOOFF(argv[1]);
+               if (volume_size_bytes >= MAXINT(off_t) / 1024)
+                       bb_error_msg_and_die("image size is too big");
+               volume_size_bytes *= 1024;
+       }
+       volume_size_blocks = (volume_size_bytes >> BLOCK_SIZE_BITS);
+       volume_size_sect = volume_size_bytes / bytes_per_sect;
+
+       //
+       // Find out or guess media parameters
+       //
+       media_byte = 0xf8;
+       heads = 255;
+       sect_per_track = 63;
+       sect_per_clust = 1;
+       {
+               struct hd_geometry geometry;
+               // size (in sectors), sect (per track), head
+               struct floppy_struct param;
+
+               // N.B. whether to use HDIO_GETGEO or HDIO_REQ?
+               if (ioctl(dev, HDIO_GETGEO, &geometry) == 0
+                && geometry.sectors
+                && geometry.heads
+               ) {
+                       // hard drive
+                       sect_per_track = geometry.sectors;
+                       heads = geometry.heads;
+
+ set_cluster_size:
+                       /* For FAT32, try to do the same as M$'s format command
+                        * (see http://www.win.tue.nl/~aeb/linux/fs/fat/fatgen103.pdf p. 20):
+                        * fs size <= 260M: 0.5k clusters
+                        * fs size <=   8G: 4k clusters
+                        * fs size <=  16G: 8k clusters
+                        * fs size >   16G: 16k clusters
+                        */
+                       sect_per_clust = volume_size_bytes > ((off_t)16)*1024*1024*1024 ? 32 :
+                                       volume_size_bytes > ((off_t)8)*1024*1024*1024 ? 16 :
+                                       volume_size_bytes >        260*1024*1024 ? 8 : 1;
+               } else {
+                       int not_floppy = ioctl(dev, FDGETPRM, &param);
+                       if (not_floppy == 0) {
+                               // floppy disk
+                               sect_per_track = param.sect;
+                               heads = param.head;
+                               volume_size_sect = param.size;
+                               volume_size_bytes = param.size * SECTOR_SIZE;
+                       }
+                       // setup the media descriptor byte
+                       switch (volume_size_sect) {
+                       case 2*360:     // 5.25", 2, 9, 40 - 360K
+                               media_byte = 0xfd;
+                               break;
+                       case 2*720:     // 3.5", 2, 9, 80 - 720K
+                       case 2*1200:    // 5.25", 2, 15, 80 - 1200K
+                               media_byte = 0xf9;
+                               break;
+                       default:        // anything else
+                               if (not_floppy)
+                                       goto set_cluster_size;
+                       case 2*1440:    // 3.5", 2, 18, 80 - 1440K
+                       case 2*2880:    // 3.5", 2, 36, 80 - 2880K
+                               media_byte = 0xf0;
+                               break;
+                       }
+                       // not floppy, but size matches floppy exactly.
+                       // perhaps it is a floppy image.
+                       // we already set media_byte as if it is a floppy,
+                       // now set sect_per_track and heads.
+                       heads = 2;
+                       sect_per_track = (unsigned)volume_size_sect / 160;
+                       if (sect_per_track < 9)
+                               sect_per_track = 9;
+               }
+       }
+
+       //
+       // Calculate number of clusters, sectors/cluster, sectors/FAT
+       // (an initial guess for sect_per_clust should already be set)
+       //
+       if ((off_t)(volume_size_sect - reserved_sect) < 16) // arbitrary limit
+               bb_error_msg_and_die("the image is too small for FAT32");
+       sect_per_fat = 1;
+       while (1) {
+               while (1) {
+                       unsigned spf;
+                       off_t tcl = (volume_size_sect - reserved_sect - NUM_FATS * sect_per_fat) / sect_per_clust;
+                       // tcl may be > MAX_CLUST_32 here, but it may be
+                       // because sect_per_fat is underestimated,
+                       // and with increased sect_per_fat it still may become
+                       // <= MAX_CLUST_32. Therefore, we do not check
+                       // against MAX_CLUST_32, but against a bigger const:
+                       if (tcl > 0x7fffffff)
+                               goto next;
+                       total_clust = tcl; // fits in uint32_t
+                       spf = ((total_clust + 2) * 4 + bytes_per_sect - 1) / bytes_per_sect;
+                       if (spf <= sect_per_fat) {
+                               // do not need to adjust sect_per_fat.
+                               // so, was total_clust too big after all?
+                               if (total_clust <= MAX_CLUST_32)
+                                       goto found_total_clust; // no
+                               // yes, total_clust is _a bit_ too big
+                               goto next;
+                       }
+                       // adjust sect_per_fat, go back and recalc total_clust
+                       sect_per_fat = spf;
+               }
+ next:
+               if (sect_per_clust == 128)
+                       bb_error_msg_and_die("can't make FAT32 with >128 sectors/cluster");
+               sect_per_clust *= 2;
+               sect_per_fat = (sect_per_fat / 2) | 1;
+       }
+ found_total_clust:
+
+       //
+       // Print info
+       //
+       if (opts & OPT_v) {
+               fprintf(stderr,
+                       "Device '%s':\n"
+                       "heads:%u, sectors/track:%u, bytes/sector:%u\n"
+                       "media descriptor:0x%02x\n"
+                       "total sectors:%"OFF_FMT"u, clusters:%u, sectors/cluster:%u\n"
+                       "FATs:2, sectors/FAT:%u\n"
+                       "volumeID:%08x, label:'%s'\n",
+                       device_name,
+                       heads, sect_per_track, bytes_per_sect,
+                       (int)media_byte,
+                       volume_size_sect, (int)total_clust, (int)sect_per_clust,
+                       sect_per_fat,
+                       (int)volume_id, volume_label
+               );
+       }
+
+       //
+       // Write filesystem image sequentially (no seeking)
+       //
+       {
+               unsigned bufsize = reserved_sect;
+               //bufsize |= sect_per_fat; // can be quite large
+               bufsize |= 2; // use this instead
+               bufsize |= sect_per_clust;
+               buf = xzalloc(bufsize * bytes_per_sect);
+       }
+
+       { // boot, fsinfo sectors and their copies
+               struct msdos_boot_sector *boot_blk = (void*)buf;
+               struct fat32_fsinfo *info = (void*)(buf + bytes_per_sect);
+
+               strcpy(boot_blk->boot_jump, "\xeb\x58\x90" "mkdosfs"); // system_id[8] included :)
+               STORE_LE(boot_blk->bytes_per_sect, bytes_per_sect);
+               STORE_LE(boot_blk->sect_per_clust, sect_per_clust);
+               STORE_LE(boot_blk->reserved_sect, reserved_sect);
+               STORE_LE(boot_blk->fats, 2);
+               //STORE_LE(boot_blk->dir_entries, 0); // for FAT32, stays 0
+               if (volume_size_sect <= 0xffff)
+                       STORE_LE(boot_blk->volume_size_sect, volume_size_sect);
+               STORE_LE(boot_blk->media_byte, media_byte);
+               // Bad: this would make Linux to think that it's fat12/16
+               //if (sect_per_fat <= 0xffff)
+               //      STORE_LE(boot_blk->sect_per_fat, sect_per_fat);
+               //STORE_LE(boot_blk->sect_per_fat, 0);
+               STORE_LE(boot_blk->sect_per_track, sect_per_track);
+               STORE_LE(boot_blk->heads, heads);
+               //STORE_LE(boot_blk->hidden, 0);
+               STORE_LE(boot_blk->fat32_volume_size_sect, volume_size_sect);
+               STORE_LE(boot_blk->fat32_sect_per_fat, sect_per_fat);
+               //STORE_LE(boot_blk->fat32_flags, 0);
+               //STORE_LE(boot_blk->fat32_version[2], 0,0);
+               STORE_LE(boot_blk->fat32_root_cluster, 2);
+               STORE_LE(boot_blk->fat32_info_sector, info_sector_number);
+               STORE_LE(boot_blk->fat32_backup_boot, backup_boot_sector);
+               //STORE_LE(boot_blk->reserved2[3], 0,0,0);
+               STORE_LE(boot_blk->vi.ext_boot_sign, 0x29);
+               STORE_LE(boot_blk->vi.volume_id32, volume_id);
+               strncpy(boot_blk->vi.fs_type, "FAT32   ", sizeof(boot_blk->vi.fs_type));
+               strncpy(boot_blk->vi.volume_label, volume_label, sizeof(boot_blk->vi.volume_label));
+               memcpy(boot_blk->boot_code, boot_code, sizeof(boot_code));
+               STORE_LE(boot_blk->boot_sign, BOOT_SIGN);
+
+               STORE_LE(info->signature1, FAT_FSINFO_SIG1);
+               STORE_LE(info->signature2, FAT_FSINFO_SIG2);
+               // we've allocated cluster 2 for the root dir
+               STORE_LE(info->free_clusters, (total_clust - 1));
+               STORE_LE(info->next_cluster, 2);
+               STORE_LE(info->boot_sign, BOOT_SIGN);
+
+               // 1st copy
+               xwrite(dev, buf, bytes_per_sect * backup_boot_sector);
+               // 2nd copy and possibly zero sectors
+               xwrite(dev, buf, bytes_per_sect * (reserved_sect - backup_boot_sector));
+       }
+
+       { // file allocation tables
+               unsigned i,j;
+               unsigned char *fat = (void*)buf;
+
+               memset(buf, 0, bytes_per_sect * 2);
+               // initial FAT entries
+               mark_cluster(0, 0x0fffff00 | media_byte);
+               mark_cluster(1, 0xffffffff);
+               // mark cluster 2 as EOF (used for root dir)
+               mark_cluster(2, EOF_FAT32);
+               for (i = 0; i < NUM_FATS; i++) {
+                       xwrite(dev, buf, bytes_per_sect);
+                       for (j = 1; j < sect_per_fat; j++)
+                               xwrite(dev, buf + bytes_per_sect, bytes_per_sect);
+               }
+       }
+
+       // root directory
+       // empty directory is just a set of zero bytes
+       memset(buf, 0, sect_per_clust * bytes_per_sect);
+       if (volume_label[0]) {
+               // create dir entry for volume_label
+               struct msdos_dir_entry *de;
+#if 0
+               struct tm tm;
+               uint16_t t, d;
+#endif
+               de = (void*)buf;
+               strncpy(de->name, volume_label, sizeof(de->name));
+               STORE_LE(de->attr, ATTR_VOLUME);
+#if 0
+               localtime_r(&create_time, &tm);
+               t = (tm.tm_sec >> 1) + (tm.tm_min << 5) + (tm.tm_hour << 11);
+               d = tm.tm_mday + ((tm.tm_mon+1) << 5) + ((tm.tm_year-80) << 9);
+               STORE_LE(de->time, t);
+               STORE_LE(de->date, d);
+               //STORE_LE(de->ctime_cs, 0);
+               de->ctime = de->time;
+               de->cdate = de->date;
+               de->adate = de->date;
+#endif
+       }
+       xwrite(dev, buf, sect_per_clust * bytes_per_sect);
+
+#if 0
+       if (opts & OPT_c) {
+               unsigned start_data_sector;
+               unsigned start_data_block;
+               unsigned badblocks = 0;
+               int try, got;
+               off_t currently_testing;
+               char *blkbuf = xmalloc(BLOCK_SIZE * TEST_BUFFER_BLOCKS);
+
+               // N.B. the two following vars are in hard sectors, i.e. SECTOR_SIZE byte sectors!
+               start_data_sector = (reserved_sect + NUM_FATS * sect_per_fat) * (bytes_per_sect / SECTOR_SIZE);
+               start_data_block = (start_data_sector + SECTORS_PER_BLOCK - 1) / SECTORS_PER_BLOCK;
+
+               bb_info_msg("Searching for bad blocks ");
+               currently_testing = 0;
+               try = TEST_BUFFER_BLOCKS;
+               while (currently_testing < volume_size_blocks) {
+                       if (currently_testing + try > volume_size_blocks)
+                               try = volume_size_blocks - currently_testing;
+                       // perform a test on a block. return the number of blocks
+                       // that could be read successfully.
+                       // seek to the correct location
+                       xlseek(dev, currently_testing * BLOCK_SIZE, SEEK_SET);
+                       // try reading
+                       got = read(dev, blkbuf, try * BLOCK_SIZE);
+                       if (got < 0)
+                               got = 0;
+                       if (got & (BLOCK_SIZE - 1))
+                               bb_error_msg("Unexpected values in do_check: probably bugs");
+                       got /= BLOCK_SIZE;
+                       currently_testing += got;
+                       if (got == try) {
+                               try = TEST_BUFFER_BLOCKS;
+                               continue;
+                       }
+                       try = 1;
+                       if (currently_testing < start_data_block)
+                               bb_error_msg_and_die("bad blocks before data-area: cannot make fs");
+
+                       // mark all of the sectors in the block as bad
+                       for (i = 0; i < SECTORS_PER_BLOCK; i++) {
+                               int cluster = (currently_testing * SECTORS_PER_BLOCK + i - start_data_sector) / (int) (sect_per_clust) / (bytes_per_sect / SECTOR_SIZE);
+                               if (cluster < 0)
+                                       bb_error_msg_and_die("Invalid cluster number in mark_sector: probably bug!");
+                               mark_cluster(cluster, BAD_FAT32);
+                       }
+                       badblocks++;
+                       currently_testing++;
+               }
+               free(blkbuf);
+               if (badblocks)
+                       bb_info_msg("%d bad block(s)", badblocks);
+       }
+#endif
+
+       // cleanup
+       if (ENABLE_FEATURE_CLEAN_UP) {
+               free(buf);
+               close(dev);
+       }
+
+       return 0;
+}