Merge branch 'branch-1.2' into debian

1 files changed, 1651 insertions, 0 deletions

diff --git a/src/drivers/ao_fat.c b/src/drivers/ao_fat.c
new file mode 100644
index 00000000..1a1b8eb0
--- /dev/null
+++ b/src/drivers/ao_fat.c
@@ -0,0 +1,1651 @@
+/*
+ * Copyright © 2013 Keith Packard <keithp@keithp.com>
+ *
+ * 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; version 2 of the License.
+ *
+ * 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, write to the Free Software Foundation, Inc.,
+ * 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
+ */
+
+#ifndef AO_FAT_TEST
+#include "ao.h"
+#endif
+
+#include "ao_fat.h"
+#include "ao_bufio.h"
+
+/* Include FAT commands */
+#ifndef AO_FAT_TEST
+#define FAT_COMMANDS	1
+#endif
+
+/* Spew FAT tracing */
+#define FAT_TRACE	0
+ 
+#ifdef DBG
+#undef DBG
+#endif
+
+#if FAT_TRACE
+#define DBG(...) printf(__VA_ARGS__)
+#else
+#define DBG(...)
+#endif
+
+/*
+ * Basic file system types
+ */
+
+typedef ao_fat_offset_t		offset_t;
+typedef ao_fat_sector_t		sector_t;
+typedef ao_fat_cluster_t	cluster_t;
+typedef ao_fat_dirent_t		dirent_t;
+typedef ao_fat_cluster_offset_t	cluster_offset_t;
+
+/* Global FAT lock */
+static uint8_t ao_fat_mutex;
+
+/* Partition information, sector numbers */
+
+static uint8_t	partition_type;
+static sector_t	partition_start, partition_end;
+
+#define AO_FAT_BAD_CLUSTER		0xffffff7
+#define AO_FAT_LAST_CLUSTER		0xfffffff
+#define AO_FAT_IS_LAST_CLUSTER(c)		(((c) & 0xffffff8) == 0xffffff8)
+#define AO_FAT_IS_LAST_CLUSTER16(c)	(((c) & 0xfff8) == 0xfff8)
+
+
+#define SECTOR_SIZE	512
+#define SECTOR_MASK	(SECTOR_SIZE - 1)
+#define SECTOR_SHIFT	9
+
+#define DIRENT_SIZE	32
+
+/* File system parameters */
+static uint8_t		sectors_per_cluster;
+static uint32_t		bytes_per_cluster;
+static sector_t		reserved_sector_count;
+static uint8_t		number_fat;
+static dirent_t		root_entries;
+static sector_t 	sectors_per_fat;
+static cluster_t	number_cluster;
+static sector_t		fat_start;
+static sector_t 	root_start;
+static sector_t 	data_start;
+static cluster_t	next_free;
+static uint8_t		filesystem_full;
+
+/* FAT32 extra data */
+static uint8_t		fat32;
+static uint8_t		fsinfo_dirty;
+static cluster_t	root_cluster;
+static sector_t		fsinfo_sector;
+static cluster_t	free_count;
+
+/*
+ * Deal with LSB FAT data structures
+ */
+static uint32_t
+get_u32(uint8_t *base)
+{
+	return ((uint32_t) base[0] |
+		((uint32_t) base[1] << 8) |
+		((uint32_t) base[2] << 16) |
+		((uint32_t) base[3] << 24));
+}
+
+static void
+put_u32(uint8_t *base, uint32_t value)
+{
+	base[0] = value;
+	base[1] = value >> 8;
+	base[2] = value >> 16;
+	base[3] = value >> 24;
+}
+
+static uint16_t
+get_u16(uint8_t *base)
+{
+	return ((uint16_t) base[0] | ((uint16_t) base[1] << 8));
+}
+
+static void
+put_u16(uint8_t *base, uint16_t value)
+{
+	base[0] = value;
+	base[1] = value >> 8;
+}
+
+static uint8_t
+_ao_fat_cluster_valid(cluster_t cluster)
+{
+	return (2 <= cluster && cluster < number_cluster);
+}
+
+/* Start using a sector */
+static uint8_t *
+_ao_fat_sector_get(sector_t sector)
+{
+	sector += partition_start;
+	if (sector >= partition_end)
+		return NULL;
+	return ao_bufio_get(sector);
+}
+
+/* Finish using a sector, 'w' is 1 if modified */
+#define _ao_fat_sector_put(b,w) ao_bufio_put(b,w)
+
+/* Get the next cluster entry in the chain */
+static cluster_t
+_ao_fat_entry_read(cluster_t cluster)
+{
+	sector_t	sector;
+	cluster_t	offset;
+	uint8_t		*buf;
+	cluster_t	ret;
+
+	if (!_ao_fat_cluster_valid(cluster))
+		return 0xfffffff7;
+
+	if (fat32)
+		cluster <<= 2;
+	else
+		cluster <<= 1;
+	sector = cluster >> (SECTOR_SHIFT);
+	offset = cluster & SECTOR_MASK;
+	buf = _ao_fat_sector_get(fat_start + sector);
+	if (!buf)
+		return 0;
+
+	if (fat32) {
+		ret = get_u32(buf + offset);
+		ret &= 0xfffffff;
+	} else {
+		ret = get_u16(buf + offset);
+		if (AO_FAT_IS_LAST_CLUSTER16(ret))
+			ret |= 0xfff0000;
+	}
+	_ao_fat_sector_put(buf, 0);
+	return ret;
+}
+
+/* Replace the referenced cluster entry in the chain with
+ * 'new_value'. Return the previous value.
+ */
+static cluster_t
+_ao_fat_entry_replace(cluster_t  cluster, cluster_t new_value)
+{
+	sector_t		sector;
+	cluster_offset_t	offset;
+	uint8_t			*buf;
+	cluster_t		ret;
+	cluster_t		old_value;
+	uint8_t			fat;
+
+	if (!_ao_fat_cluster_valid(cluster))
+		return 0xfffffff7;
+
+	/* Convert from cluster index to byte index */
+	if (fat32)
+		cluster <<= 2;
+	else
+		cluster <<= 1;
+	sector = cluster >> SECTOR_SHIFT;
+	offset = cluster & SECTOR_MASK;
+
+	new_value &= 0xfffffff;
+	for (fat = 0; fat < number_fat; fat++) {
+		buf = _ao_fat_sector_get(fat_start + fat * sectors_per_fat + sector);
+		if (!buf)
+			return 0;
+		if (fat32) {
+			old_value = get_u32(buf + offset);
+			put_u32(buf + offset, new_value | (old_value & 0xf0000000));
+			if (fat == 0) {
+				ret = old_value & 0xfffffff;
+
+				/* Track the free count if it wasn't marked
+				 * invalid when we mounted the file system
+				 */
+				if (free_count != 0xffffffff) {
+					if (new_value && !ret) {
+						--free_count;
+						fsinfo_dirty = 1;
+					} else if (!new_value && ret) {
+						++free_count;
+						fsinfo_dirty = 1;
+					}
+				}
+			}
+		} else {
+			if (fat == 0) {
+				ret = get_u16(buf + offset);
+				if (AO_FAT_IS_LAST_CLUSTER16(ret))
+					ret |= 0xfff0000;
+			}
+			put_u16(buf + offset, new_value);
+		}
+		_ao_fat_sector_put(buf, 1);
+	}
+	return ret;
+	
+}
+
+/*
+ * Walk a cluster chain and mark
+ * all of them as free
+ */
+static void
+_ao_fat_free_cluster_chain(cluster_t cluster)
+{
+	while (_ao_fat_cluster_valid(cluster)) {
+		if (cluster < next_free) {
+			next_free = cluster;
+			fsinfo_dirty = 1;
+		}
+		cluster = _ao_fat_entry_replace(cluster, 0x00000000);
+	}
+}
+
+/*
+ * _ao_fat_cluster_seek
+ * 
+ * The basic file system operation -- map a file cluster index to a
+ * partition cluster number. Done by computing the cluster number and
+ * then walking that many clusters from the first cluster. Returns
+ * 0xffff if we walk off the end of the file or the cluster chain
+ * is damaged somehow
+ */
+static cluster_t
+_ao_fat_cluster_seek(cluster_t cluster, cluster_t distance)
+{
+	while (distance) {
+		cluster = _ao_fat_entry_read(cluster);
+		if (!_ao_fat_cluster_valid(cluster))
+			break;
+		distance--;
+	}
+	return cluster;
+}
+
+/*
+ * _ao_fat_cluster_set_size
+ *
+ * Set the number of clusters in the specified chain,
+ * freeing extra ones or alocating new ones as needed
+ *
+ * Returns AO_FAT_BAD_CLUSTER on allocation failure
+ */
+
+static cluster_t
+_ao_fat_cluster_set_size(cluster_t first_cluster, cluster_t size)
+{
+	cluster_t	have;
+	cluster_t	last_cluster;
+	cluster_t	next_cluster;
+
+	/* Walk the cluster chain to the
+	 * spot where it needs to change. That
+	 * will either be the end of the chain (in case it needs to grow),
+	 * or after the desired number of clusters, in which case it needs to shrink
+	 */
+	next_cluster = first_cluster;
+	last_cluster = 0;
+	DBG("\tclusters:");
+	for (have = 0; have < size; have++) {
+		DBG(" %08x", next_cluster);
+		if (!_ao_fat_cluster_valid(next_cluster))
+			break;
+		last_cluster = next_cluster;
+		next_cluster = _ao_fat_entry_read(next_cluster);
+	}
+	DBG("\n");
+
+	/* At this point, last_cluster points to the last valid
+	 * cluster in the file, if any. That's the spot in the FAT
+	 * that needs to be rewritten, either to truncate the file by
+	 * writing an END marker, or to extend the file by writing
+	 * more clusters. next_cluster will contain the value of the
+	 * FAT at last_cluster.
+	 *
+	 * If this is at the head of the cluster chain, then
+	 * last_cluster will be zero and next_cluster will
+	 * be the first cluster in the chain.
+	 */
+	if (have == size) {
+		/* The file is large enough, truncate as needed */
+		if (_ao_fat_cluster_valid(next_cluster)) {
+			DBG("truncate between %08x and %08x\n", last_cluster, next_cluster);
+			if (last_cluster)
+				/*
+				 * Otherwise, rewrite the last cluster
+				 * in the chain with a LAST marker
+				 */
+				(void) _ao_fat_entry_replace(last_cluster,
+							    AO_FAT_LAST_CLUSTER);
+			else
+				/*
+				 * If the file is getting erased, then
+				 * rewrite the directory entry cluster
+				 * value
+				 */
+				first_cluster = 0;
+
+			/* Clear the remaining clusters in the chain */
+			_ao_fat_free_cluster_chain(next_cluster);
+
+			/* The file system is no longer full (if it was) */
+			filesystem_full = 0;
+		} else {
+			DBG("unchanged FAT chain\n");
+			/* The chain is already the right length, don't mess with it */
+			;
+		}
+	} else {
+		cluster_t	need;
+		cluster_t	free;
+
+		if (filesystem_full)
+			return AO_FAT_BAD_CLUSTER;
+
+		/* Set next_free if it has wrapped or wasn't set before */
+		if (next_free < 2 || number_cluster <= next_free) {
+			next_free = 2;
+			fsinfo_dirty = 1;
+		}
+
+		/* See if there are enough free clusters in the file system */
+		need = size - have;
+
+#define loop_cluster	for (free = next_free; need > 0;)
+#define next_cluster				\
+		if (++free == number_cluster)	\
+			free = 2;		\
+		if (free == next_free)		\
+			break;			\
+
+		loop_cluster {
+			if (!_ao_fat_entry_read(free))
+				need--;
+			next_cluster;
+		}
+
+		/* Still need some, tell the user that we've failed */
+		if (need) {
+			filesystem_full = 1;
+			return AO_FAT_BAD_CLUSTER;
+		}
+
+		/* Now go allocate those clusters and
+		 * thread them onto the chain
+		 */
+		need = size - have;
+		loop_cluster {
+			if (_ao_fat_entry_read(free) == 0) {
+				next_free = free + 1;
+				if (next_free >= number_cluster)
+					next_free = 2;
+				fsinfo_dirty = 1;
+				DBG("\tadd cluster. old %08x new %08x\n", last_cluster, free);
+				if (last_cluster)
+					_ao_fat_entry_replace(last_cluster, free);
+				else
+					first_cluster = free;
+				last_cluster = free;
+				need--;
+			}
+			next_cluster;
+		}
+#undef loop_cluster
+#undef next_cluster
+		DBG("\tlast cluster %08x\n", last_cluster);
+		/* Mark the new end of the chain */
+		_ao_fat_entry_replace(last_cluster, AO_FAT_LAST_CLUSTER);
+	}
+
+	DBG("\tfirst cluster %08x\n", first_cluster);
+	return first_cluster;
+}
+
+/* Start using a root directory entry */
+static uint8_t *
+_ao_fat_root_get(dirent_t e)
+{
+	offset_t		byte = e * DIRENT_SIZE;
+	sector_t		sector = byte >> SECTOR_SHIFT;
+	cluster_offset_t	offset = byte & SECTOR_MASK;
+	uint8_t			*buf;
+
+	if (fat32) {
+		cluster_t	cluster_distance = sector / sectors_per_cluster;
+		sector_t	sector_index = sector % sectors_per_cluster;
+		cluster_t	cluster = _ao_fat_cluster_seek(root_cluster, cluster_distance);
+
+		if (_ao_fat_cluster_valid(cluster))
+			sector = data_start + (cluster-2) * sectors_per_cluster + sector_index;
+		else
+			return NULL;
+	} else {
+		if (e >= root_entries)
+			return NULL;
+		sector = root_start + sector;
+	}
+
+	buf = _ao_fat_sector_get(sector);
+	if (!buf)
+		return NULL;
+	return buf + offset;
+}
+
+/* Finish using a root directory entry, 'w' is 1 if modified */
+static void
+_ao_fat_root_put(uint8_t *root, dirent_t e, uint8_t write)
+{
+	cluster_offset_t	offset = ((e * DIRENT_SIZE) & SECTOR_MASK);
+	uint8_t			*buf = root - offset;
+
+	_ao_fat_sector_put(buf, write);
+}
+
+/*
+ * _ao_fat_root_extend
+ *
+ * On FAT32, make the root directory at least 'ents' entries long
+ */
+static int8_t
+_ao_fat_root_extend(dirent_t ents)
+{
+	offset_t	byte_size;
+	cluster_t	cluster_size;
+	if (!fat32)
+		return 0;
+	
+	byte_size = (ents + 1) * 0x20;
+	cluster_size = (byte_size + bytes_per_cluster - 1) / bytes_per_cluster;
+	if (_ao_fat_cluster_set_size(root_cluster, cluster_size) != AO_FAT_BAD_CLUSTER)
+		return 1;
+	return 0;
+}
+		
+/*
+ * _ao_fat_setup_partition
+ * 
+ * Load the boot block and find the first partition
+ */
+static uint8_t
+_ao_fat_setup_partition(void)
+{
+	uint8_t *mbr;
+	uint8_t	*partition;
+	uint32_t partition_size;
+
+	mbr = ao_bufio_get(0);
+	if (!mbr)
+		return AO_FAT_FILESYSTEM_MBR_READ_FAILURE;
+
+	/* Check the signature */
+	if (mbr[0x1fe] != 0x55 || mbr[0x1ff] != 0xaa) {
+		DBG ("Invalid MBR signature %02x %02x\n",
+			mbr[0x1fe], mbr[0x1ff]);
+		ao_bufio_put(mbr, 0);
+		return AO_FAT_FILESYSTEM_INVALID_MBR_SIGNATURE;
+	}
+
+	/* Check to see if it's actually a boot block, in which
+	 * case it's presumably not a paritioned device
+	 */
+
+	if (mbr[0] == 0xeb) {
+		partition_start = 0;
+		partition_size = get_u16(mbr + 0x13);
+		if (partition_size == 0)
+			partition_size = get_u32(mbr + 0x20);
+	} else {
+		/* Just use the first partition */
+		partition = &mbr[0x1be];
+	
+		partition_type = partition[4];
+		switch (partition_type) {
+		case 4:		/* FAT16 up to 32M */
+		case 6:		/* FAT16 over 32M */
+			break;
+		case 0x0b:	/* FAT32 up to 2047GB */
+		case 0x0c:	/* FAT32 LBA */
+			break;
+		default:
+			DBG ("Invalid partition type %02x\n", partition_type);
+			ao_bufio_put(mbr, 0);
+			return AO_FAT_FILESYSTEM_INVALID_PARTITION_TYPE;
+		}
+
+		partition_start = get_u32(partition+8);
+		partition_size = get_u32(partition+12);
+		if (partition_size == 0) {
+			DBG ("Zero-sized partition\n");
+			ao_bufio_put(mbr, 0);
+			return AO_FAT_FILESYSTEM_ZERO_SIZED_PARTITION;
+		}
+	}
+	partition_end = partition_start + partition_size;
+	ao_bufio_put(mbr, 0);
+	return AO_FAT_FILESYSTEM_SUCCESS;
+}
+	
+static uint8_t
+_ao_fat_setup_fs(void)
+{
+	uint8_t		*boot = _ao_fat_sector_get(0);
+	uint32_t	data_sectors;
+
+	if (!boot)
+		return AO_FAT_FILESYSTEM_BOOT_READ_FAILURE;
+
+	/* Check the signature */
+	if (boot[0x1fe] != 0x55 || boot[0x1ff] != 0xaa) {
+		DBG ("Invalid BOOT signature %02x %02x\n",
+			boot[0x1fe], boot[0x1ff]);
+		_ao_fat_sector_put(boot, 0);
+		return AO_FAT_FILESYSTEM_INVALID_BOOT_SIGNATURE;
+	}
+
+	/* Check the sector size */
+	if (get_u16(boot + 0xb) != SECTOR_SIZE) {
+		DBG ("Invalid sector size %d\n",
+			get_u16(boot + 0xb));
+		_ao_fat_sector_put(boot, 0);
+		return AO_FAT_FILESYSTEM_INVALID_SECTOR_SIZE;
+	}
+
+	sectors_per_cluster = boot[0xd];
+	bytes_per_cluster = sectors_per_cluster << SECTOR_SHIFT;
+	reserved_sector_count = get_u16(boot+0xe);
+	number_fat = boot[0x10];
+	root_entries = get_u16(boot + 0x11);
+	sectors_per_fat = get_u16(boot+0x16);
+	fat32 = 0;
+	if (sectors_per_fat == 0) {
+		fat32 = 1;
+		sectors_per_fat = get_u32(boot+0x24);
+		root_cluster = get_u32(boot+0x2c);
+		fsinfo_sector = get_u16(boot + 0x30);
+	}
+	_ao_fat_sector_put(boot, 0);
+
+	free_count = 0xffffffff;
+	next_free = 0;
+	if (fat32 && fsinfo_sector) {
+		uint8_t	*fsinfo = _ao_fat_sector_get(fsinfo_sector);
+
+		if (fsinfo) {
+			free_count = get_u32(fsinfo + 0x1e8);
+			next_free = get_u32(fsinfo + 0x1ec);
+			_ao_fat_sector_put(fsinfo, 0);
+		}
+	}
+
+	fat_start = reserved_sector_count;;
+	root_start = fat_start + number_fat * sectors_per_fat;
+	data_start = root_start + ((root_entries * DIRENT_SIZE + SECTOR_MASK) >> SECTOR_SHIFT);
+
+	data_sectors = (partition_end - partition_start) - data_start;
+
+	number_cluster = data_sectors / sectors_per_cluster;
+
+	return AO_FAT_FILESYSTEM_SUCCESS;
+}
+
+/*
+ * State for an open file
+ */
+
+struct ao_file {
+	struct ao_fat_dirent	*dirent;
+	offset_t		offset;
+	offset_t		cluster_offset;
+	cluster_t		cluster;
+	uint8_t			busy;
+};
+
+#define AO_FAT_NFILE	8
+
+static struct ao_fat_dirent	ao_file_dirent[AO_FAT_NFILE];
+
+static struct ao_fat_dirent *
+_ao_fat_file_dirent_alloc(struct ao_fat_dirent *want)
+{
+	int8_t	d;
+	struct ao_fat_dirent *free = NULL, *dirent;
+
+	for (d = 0; d < AO_FAT_NFILE; d++) {
+
+		dirent = &ao_file_dirent[d];
+		/* See if there's another user of this file already */
+		if (want && dirent->name[0] != 0) {
+			if (dirent->entry == want->entry)
+				return dirent;
+		} else {
+			if (!free) {
+				free = dirent;
+				if (!want)
+					break;
+			}
+		}
+	}
+	if (free && want)
+		*free = *want;
+	return free;
+}
+
+static struct ao_file 		ao_file_table[AO_FAT_NFILE];
+
+static int8_t
+_ao_fat_fd_alloc(struct ao_fat_dirent *dirent)
+{
+	int8_t	fd;
+
+	for (fd = 0; fd < AO_FAT_NFILE; fd++)
+		if (!ao_file_table[fd].busy) {
+			ao_file_table[fd].dirent = _ao_fat_file_dirent_alloc(dirent);
+			ao_file_table[fd].busy = 1;
+			ao_file_table[fd].offset = 0;
+			ao_file_table[fd].cluster_offset = 0;
+			ao_file_table[fd].cluster = ao_file_table[fd].dirent->cluster;
+				
+			return fd;
+		}
+	return -AO_FAT_EMFILE;
+}
+
+static void
+_ao_fat_fd_free(int8_t fd)
+{
+	struct ao_file *file = &ao_file_table[fd];
+	struct ao_fat_dirent *dirent = file->dirent;
+	memset(&ao_file_table[fd], '\0', sizeof (struct ao_file));
+
+	/* Check and see if another ao_file references the same dirent */
+	for (fd = 0; fd < AO_FAT_NFILE; fd++)
+		if (ao_file_table[fd].dirent == dirent)
+			return;
+	memset(dirent, '\0', sizeof (struct ao_fat_dirent));
+}
+
+static struct ao_file *
+_ao_fat_fd_to_file(int8_t fd)
+{
+	struct ao_file *file;
+	if (fd < 0 || AO_FAT_NFILE <= fd)
+		return NULL;
+
+	file = &ao_file_table[fd];
+	if (!file->busy)
+		return NULL;
+	return file;
+}
+
+static uint8_t			ao_filesystem_setup;
+static uint8_t			ao_filesystem_status;
+
+static uint8_t
+_ao_fat_setup(void)
+{
+	if (!ao_filesystem_setup) {
+
+		ao_filesystem_setup = 1;
+		ao_bufio_setup();
+	
+		/* Re-initialize all global state; this will help to allow the
+		 * file system to get swapped someday
+		 */
+		partition_type = partition_start = partition_end = 0;
+		sectors_per_cluster = bytes_per_cluster = reserved_sector_count = 0;
+		number_fat = root_entries = sectors_per_fat = 0;
+		number_cluster = fat_start = root_start = data_start = 0;
+		next_free = filesystem_full = 0;
+		fat32 = fsinfo_dirty = root_cluster = fsinfo_sector = free_count = 0;
+
+		/* Reset open file table */
+		memset(&ao_file_table, '\0', sizeof (ao_file_table));
+
+		ao_filesystem_status = _ao_fat_setup_partition();
+		if (ao_filesystem_status != AO_FAT_FILESYSTEM_SUCCESS)
+			return ao_filesystem_status;
+		ao_filesystem_status = _ao_fat_setup_fs();
+		if (ao_filesystem_status != AO_FAT_FILESYSTEM_SUCCESS)
+			return ao_filesystem_status;
+	}
+	return ao_filesystem_status;
+}
+
+void
+ao_fat_unmount(void)
+{
+	ao_filesystem_setup = 0;
+}
+
+/*
+ * Basic file operations
+ */
+
+static uint32_t
+_ao_fat_current_sector(struct ao_file *file)
+{
+	cluster_t	cluster_offset;
+	uint32_t	sector_offset;
+	uint16_t	sector_index;
+	cluster_t	cluster;
+
+	DBG("current sector offset %d size %d\n",
+	    file->offset, file->dirent->size);
+
+	if (file->offset > file->dirent->size) {
+		printf ("file offset %d larger than size %d\n",
+			file->offset, file->dirent->size);
+		return 0xffffffff;
+	}
+
+	sector_offset = file->offset >> SECTOR_SHIFT;
+
+	if (!file->cluster || file->offset < file->cluster_offset) {
+		file->cluster = file->dirent->cluster;
+		file->cluster_offset = 0;
+		DBG("\treset to start of file %08x\n", file->cluster);
+	}
+
+	if (file->cluster_offset + bytes_per_cluster <= file->offset) {
+		cluster_t	cluster_distance;
+
+		cluster_offset = sector_offset / sectors_per_cluster;
+
+		cluster_distance = cluster_offset - file->cluster_offset / bytes_per_cluster;
+
+		DBG("\tseek forward %d clusters\n", cluster_distance);
+		cluster = _ao_fat_cluster_seek(file->cluster, cluster_distance);
+
+		if (!_ao_fat_cluster_valid(cluster)) {
+			printf ("invalid cluster %08x\n", cluster);
+			return 0xffffffff;
+		}
+		file->cluster = cluster;
+		file->cluster_offset = cluster_offset * bytes_per_cluster;
+	}
+
+	sector_index = sector_offset % sectors_per_cluster;
+	DBG("current cluster %08x sector_index %d sector %d\n",
+	    file->cluster, sector_index,
+	    data_start + (uint32_t) (file->cluster-2) * sectors_per_cluster + sector_index);
+	return data_start + (uint32_t) (file->cluster-2) * sectors_per_cluster + sector_index;
+}
+
+/*
+ * _ao_fat_invaldate_cluster_offset
+ *
+ * When the file size gets shrunk, invalidate
+ * any file structures referencing clusters beyond that point
+ */
+
+static void
+_ao_fat_invalidate_cluster_offset(struct ao_fat_dirent *dirent)
+{
+	int8_t		fd;
+	struct ao_file	*file;
+
+	for (fd = 0; fd < AO_FAT_NFILE; fd++) {
+		file = &ao_file_table[fd];
+		if (!file->busy)
+			continue;
+		if (file->dirent == dirent) {
+			if (file->cluster_offset >= dirent->size) {
+				file->cluster_offset = 0;
+				file->cluster = dirent->cluster;
+			}
+		}
+	}
+}
+
+ 
+/*
+ * _ao_fat_set_size
+ *
+ * Set the size of the current file, truncating or extending
+ * the cluster chain as needed
+ */
+static int8_t
+_ao_fat_set_size(struct ao_file *file, uint32_t size)
+{
+	uint8_t		*dent;
+	cluster_t	first_cluster;
+	cluster_t	have_clusters, need_clusters;
+
+	DBG ("Set size %d\n", size);
+	if (size == file->dirent->size) {
+		DBG("\tsize match\n");
+		return AO_FAT_SUCCESS;
+	}
+
+	first_cluster = file->dirent->cluster;
+	have_clusters = (file->dirent->size + bytes_per_cluster - 1) / bytes_per_cluster;
+	need_clusters = (size + bytes_per_cluster - 1) / bytes_per_cluster;
+
+	DBG ("\tfirst cluster %08x have %d need %d\n", first_cluster, have_clusters, need_clusters);
+	if (have_clusters != need_clusters) {
+		if (file->cluster && size > file->cluster_offset) {
+			cluster_t	offset_clusters = (file->cluster_offset + bytes_per_cluster) / bytes_per_cluster;
+			cluster_t	extra_clusters = need_clusters - offset_clusters;
+			cluster_t	next_cluster;
+
+			DBG ("\tset size relative offset_clusters %d extra_clusters %d\n",
+			     offset_clusters, extra_clusters);
+
+			/* Need one more to account for file->cluster, which we already have */
+			next_cluster = _ao_fat_cluster_set_size(file->cluster, extra_clusters + 1);
+			if (next_cluster == AO_FAT_BAD_CLUSTER)
+				return -AO_FAT_ENOSPC;
+		} else {
+			DBG ("\tset size absolute need_clusters %d\n", need_clusters);
+			first_cluster = _ao_fat_cluster_set_size(first_cluster, need_clusters);
+
+			if (first_cluster == AO_FAT_BAD_CLUSTER)
+				return -AO_FAT_ENOSPC;
+		}
+	}
+
+	DBG ("\tupdate directory size\n");
+	/* Update the directory entry */
+	dent = _ao_fat_root_get(file->dirent->entry);
+	if (!dent) {
+		printf ("dent update failed\n");
+		return -AO_FAT_EIO;
+	}
+	put_u32(dent + 0x1c, size);
+	put_u16(dent + 0x1a, first_cluster);
+	if (fat32)
+		put_u16(dent + 0x14, first_cluster >> 16);
+	_ao_fat_root_put(dent, file->dirent->entry, 1);
+
+	file->dirent->size = size;
+	file->dirent->cluster = first_cluster;
+	if (have_clusters > need_clusters)
+		_ao_fat_invalidate_cluster_offset(file->dirent);
+	DBG ("set size done\n");
+	return AO_FAT_SUCCESS;
+}
+
+/*
+ * _ao_fat_root_init
+ *
+ * Initialize a root directory entry
+ */
+static void
+_ao_fat_root_init(uint8_t *dent, char name[11], uint8_t attr)
+{
+	memset(dent, '\0', 0x20);
+	memmove(dent, name, 11);
+
+	dent[0x0b] = 0x00;
+	dent[0x0c] = 0x00;
+	dent[0x0d] = 0x00;
+
+	/* XXX fix time */
+	put_u16(dent + 0x0e, 0);
+	/* XXX fix date */
+	put_u16(dent + 0x10, 0);
+	/* XXX fix date */
+	put_u16(dent + 0x12, 0);
+
+	/* XXX fix time */
+	put_u16(dent + 0x16, 0);
+	/* XXX fix date */
+	put_u16(dent + 0x18, 0);
+
+	/* cluster number */
+	/* Low cluster bytes */
+	put_u16(dent + 0x1a, 0);
+	/* FAT32 high cluster bytes */
+	put_u16(dent + 0x14, 0);
+
+	/* size */
+	put_u32(dent + 0x1c, 0);
+}
+
+
+static void
+_ao_fat_dirent_init(struct ao_fat_dirent *dirent, uint8_t *dent, uint16_t entry)
+{
+	memcpy(dirent->name, dent + 0x00, 11);
+	dirent->attr = dent[0x0b];
+	dirent->size = get_u32(dent+0x1c);
+	dirent->cluster = get_u16(dent+0x1a);
+	if (fat32)
+		dirent->cluster |= (cluster_t) get_u16(dent + 0x14) << 16;
+	dirent->entry = entry;
+}
+
+/*
+ * _ao_fat_flush_fsinfo
+ *
+ * Write out any fsinfo changes to disk
+ */
+
+static void
+_ao_fat_flush_fsinfo(void)
+{
+	uint8_t	*fsinfo;
+
+	if (!fat32)
+		return;
+
+	if (!fsinfo_dirty)
+		return;
+	fsinfo_dirty = 0;
+	if (!fsinfo_sector)
+		return;
+
+	fsinfo = _ao_fat_sector_get(fsinfo_sector);
+	if (fsinfo) {
+		put_u32(fsinfo + 0x1e8, free_count);
+		put_u32(fsinfo + 0x1ec, next_free);
+		_ao_fat_sector_put(fsinfo, 1);
+	}
+}
+
+/*
+ * Public API
+ */
+
+/*
+ * ao_fat_sync
+ *
+ * Flush any pending I/O to storage
+ */
+
+static void
+_ao_fat_sync(void)
+{
+	if (_ao_fat_setup() != AO_FAT_FILESYSTEM_SUCCESS)
+		return;
+	_ao_fat_flush_fsinfo();
+	ao_bufio_flush();
+}
+
+void
+ao_fat_sync(void)
+{
+	ao_mutex_get(&ao_fat_mutex);
+	_ao_fat_sync();
+	ao_mutex_put(&ao_fat_mutex);
+}
+
+/*
+ * ao_fat_full
+ *
+ * Returns TRUE if the filesystem cannot take
+ * more data
+ */
+
+int8_t
+ao_fat_full(void)
+{
+	ao_mutex_get(&ao_fat_mutex);
+	if (_ao_fat_setup() != AO_FAT_FILESYSTEM_SUCCESS) {
+		ao_mutex_put(&ao_fat_mutex);
+		return 1;
+	}
+	ao_mutex_put(&ao_fat_mutex);
+	return filesystem_full;
+}
+
+static int8_t
+_ao_fat_readdir(uint16_t *entry, struct ao_fat_dirent *dirent)
+{
+	uint8_t	*dent;
+
+	if (_ao_fat_setup() != AO_FAT_FILESYSTEM_SUCCESS)
+		return -AO_FAT_EIO;
+
+	for (;;) {
+		dent = _ao_fat_root_get(*entry);
+		if (!dent)
+			return -AO_FAT_EDIREOF;
+
+		if (dent[0] == AO_FAT_DENT_END) {
+			_ao_fat_root_put(dent, *entry, 0);
+			return -AO_FAT_EDIREOF;
+		}
+		if (dent[0] != AO_FAT_DENT_EMPTY && (dent[0xb] & 0xf) != 0xf) {
+			_ao_fat_dirent_init(dirent, dent, *entry);
+			_ao_fat_root_put(dent, *entry, 0);
+			(*entry)++;
+			return AO_FAT_SUCCESS;
+		}
+		_ao_fat_root_put(dent, *entry, 0);
+		(*entry)++;
+	}
+}
+
+int8_t
+ao_fat_readdir(uint16_t *entry, struct ao_fat_dirent *dirent)
+{
+	int8_t status;
+
+	ao_mutex_get(&ao_fat_mutex);
+	status = _ao_fat_readdir(entry, dirent);
+	ao_mutex_put(&ao_fat_mutex);
+	return status;
+}
+
+/*
+ * ao_fat_open
+ *
+ * Open an existing file.
+ */
+static int8_t
+_ao_fat_open(char name[11], uint8_t mode)
+{
+	uint16_t		entry = 0;
+	static struct ao_fat_dirent	dirent;
+	int8_t			status;
+
+	if (_ao_fat_setup() != AO_FAT_FILESYSTEM_SUCCESS)
+		return -AO_FAT_EIO;
+
+	for (;;) {
+		status = _ao_fat_readdir(&entry, &dirent);
+		if (status < 0) {
+			if (status == -AO_FAT_EDIREOF)
+				return -AO_FAT_ENOENT;
+			return status;
+		}
+		if (!memcmp(name, dirent.name, 11)) {
+			if (AO_FAT_IS_DIR(dirent.attr))
+				return -AO_FAT_EISDIR;
+			if (!AO_FAT_IS_FILE(dirent.attr))
+				return -AO_FAT_EPERM;
+			if (mode > AO_FAT_OPEN_READ && (dirent.attr & AO_FAT_FILE_READ_ONLY))
+				return -AO_FAT_EACCESS;
+			return _ao_fat_fd_alloc(&dirent);
+		}
+	}
+	return -AO_FAT_ENOENT;
+}
+
+int8_t
+ao_fat_open(char name[11], uint8_t mode)
+{
+	int8_t status;
+
+	ao_mutex_get(&ao_fat_mutex);
+	status = _ao_fat_open(name, mode);
+	ao_mutex_put(&ao_fat_mutex);
+	return status;
+}
+
+/*
+ * ao_fat_close
+ *
+ * Close the currently open file
+ */
+static int8_t
+_ao_fat_close(int8_t fd)
+{
+	struct ao_file *file;
+
+	file = _ao_fat_fd_to_file(fd);
+	if (!file)
+		return -AO_FAT_EBADF;
+
+	_ao_fat_fd_free(fd);
+	_ao_fat_sync();
+	return AO_FAT_SUCCESS;
+}
+
+int8_t
+ao_fat_close(int8_t fd)
+{
+	int8_t status;
+
+	ao_mutex_get(&ao_fat_mutex);
+	status = _ao_fat_close(fd);
+	ao_mutex_put(&ao_fat_mutex);
+	return status;
+}
+
+/*
+ * ao_fat_creat
+ *
+ * Open and truncate an existing file or
+ * create a new file
+ */
+
+static int8_t
+_ao_fat_creat(char name[11])
+{
+	uint16_t		entry;
+	int8_t			fd;
+	int8_t			status;
+	uint8_t			*dent;
+	struct ao_file		*file;
+
+	if (_ao_fat_setup() != AO_FAT_FILESYSTEM_SUCCESS)
+		return -AO_FAT_EIO;
+
+	fd = _ao_fat_open(name, AO_FAT_OPEN_WRITE);
+	if (fd >= 0) {
+		file = &ao_file_table[fd];
+		status = _ao_fat_set_size(file, 0);
+		if (status < 0) {
+			_ao_fat_close(fd);
+			fd = status;
+		}
+	} else {
+		if (fd == -AO_FAT_ENOENT) {
+			entry = 0;
+			for (;;) {
+				dent = _ao_fat_root_get(entry);
+				if (!dent) {
+				
+					if (_ao_fat_root_extend(entry))
+						continue;
+					fd = -AO_FAT_ENOSPC;
+					break;
+				}
+				if (dent[0] == AO_FAT_DENT_EMPTY || dent[0] == AO_FAT_DENT_END) {
+					fd = _ao_fat_fd_alloc(NULL);
+					if (fd < 0) {
+						_ao_fat_root_put(dent, entry, 0);
+						break;
+					}
+
+					file = &ao_file_table[fd];
+					/* Initialize the dent */
+					_ao_fat_root_init(dent, name, AO_FAT_FILE_REGULAR);
+
+					/* Now initialize the dirent from the dent */
+					_ao_fat_dirent_init(file->dirent, dent, entry);
+
+					/* And write the dent to storage */
+					_ao_fat_root_put(dent, entry, 1);
+
+					status = -AO_FAT_SUCCESS;
+					break;
+				} else {
+					_ao_fat_root_put(dent, entry, 0);
+				}
+				entry++;
+			}
+		}
+	}
+	return fd;
+}
+
+int8_t
+ao_fat_creat(char name[11])
+{
+	int8_t	status;
+
+	ao_mutex_get(&ao_fat_mutex);
+	status = _ao_fat_creat(name);
+	ao_mutex_put(&ao_fat_mutex);
+	return status;
+}
+
+/*
+ * ao_fat_map_current
+ *
+ * Map the sector pointed at by the current file offset
+ */
+
+static void *
+ao_fat_map_current(struct ao_file *file, int len, cluster_offset_t *offsetp, cluster_offset_t *this_time)
+{
+	cluster_offset_t 	offset;
+	sector_t		sector;
+	void			*buf;
+
+	offset = file->offset & SECTOR_MASK;
+	sector = _ao_fat_current_sector(file);
+	if (sector == 0xffffffff) {
+		return NULL;
+	}
+	buf = _ao_fat_sector_get(sector);
+	if (!buf)
+		return NULL;
+	if (offset + len < SECTOR_SIZE)
+		*this_time = len;
+	else
+		*this_time = SECTOR_SIZE - offset;
+	*offsetp = offset;
+	return buf;
+}
+
+/*
+ * ao_fat_read
+ *
+ * Read from the file
+ */
+int
+ao_fat_read(int8_t fd, void *dst, int len)
+{
+	uint8_t			*dst_b = dst;
+	cluster_offset_t	this_time;
+	cluster_offset_t	offset;
+	uint8_t			*buf;
+	int			ret = 0;
+	struct ao_file		*file;
+
+	ao_mutex_get(&ao_fat_mutex);
+	file = _ao_fat_fd_to_file(fd);
+	if (!file) {
+		ret = -AO_FAT_EBADF;
+		goto done;
+	}
+
+	if (file->offset + len > file->dirent->size)
+		len = file->dirent->size - file->offset;
+
+	if (len < 0)
+		len = 0;
+
+	while (len) {
+		buf = ao_fat_map_current(file, len, &offset, &this_time);
+		if (!buf) {
+			ret = -AO_FAT_EIO;
+			break;
+		}
+		memcpy(dst_b, buf + offset, this_time);
+		_ao_fat_sector_put(buf, 0);
+
+		ret += this_time;
+		len -= this_time;
+		dst_b += this_time;
+		file->offset = file->offset + this_time;
+	}
+done:
+	ao_mutex_put(&ao_fat_mutex);
+	return ret;
+}
+
+/*
+ * ao_fat_write
+ *
+ * Write to the file, extended as necessary
+ */
+int
+ao_fat_write(int8_t fd, void *src, int len)
+{
+	uint8_t			*src_b = src;
+	cluster_offset_t	this_time;
+	cluster_offset_t	offset;
+	uint8_t			*buf;
+	int			ret = 0;
+	struct ao_file		*file;
+
+	ao_mutex_get(&ao_fat_mutex);
+	file = _ao_fat_fd_to_file(fd);
+	if (!file) {
+		ret = -AO_FAT_EBADF;
+		goto done;
+	}
+
+	if (file->offset + len > file->dirent->size) {
+		ret = _ao_fat_set_size(file, file->offset + len);
+		if (ret < 0)
+			goto done;
+	}
+
+	while (len) {
+		buf = ao_fat_map_current(file, len, &offset, &this_time);
+		if (!buf) {
+			ret = -AO_FAT_EIO;
+			break;
+		}
+		memcpy(buf + offset, src_b, this_time);
+		_ao_fat_sector_put(buf, 1);
+
+		ret += this_time;
+		len -= this_time;
+		src_b += this_time;
+		file->offset = file->offset + this_time;
+	}
+done:
+	ao_mutex_put(&ao_fat_mutex);
+	return ret;
+}
+
+/*
+ * ao_fat_seek
+ *
+ * Set the position for the next I/O operation
+ * Note that this doesn't actually change the size
+ * of the file if the requested position is beyond
+ * the current file length, that would take a future
+ * write
+ */
+int32_t
+ao_fat_seek(int8_t fd, int32_t pos, uint8_t whence)
+{
+	offset_t	new_offset;
+	struct ao_file	*file;
+	int32_t		ret;
+
+	ao_mutex_get(&ao_fat_mutex);
+	file = _ao_fat_fd_to_file(fd);
+	if (!file) {
+		ret = -AO_FAT_EBADF;
+		goto done;
+	}
+
+	new_offset = file->offset;
+	switch (whence) {
+	case AO_FAT_SEEK_SET:
+		new_offset = pos;
+		break;
+	case AO_FAT_SEEK_CUR:
+		new_offset += pos;
+		break;
+	case AO_FAT_SEEK_END:
+		new_offset = file->dirent->size + pos;
+		break;
+	}
+	ret = file->offset = new_offset;
+done:
+	ao_mutex_put(&ao_fat_mutex);
+	return ret;
+}
+
+/*
+ * ao_fat_unlink
+ *
+ * Remove a file from the directory, marking
+ * all clusters as free
+ */
+int8_t
+ao_fat_unlink(char name[11])
+{
+	uint16_t		entry = 0;
+	static struct ao_fat_dirent	dirent;
+	int8_t			ret;
+
+	ao_mutex_get(&ao_fat_mutex);
+	if (_ao_fat_setup() != AO_FAT_FILESYSTEM_SUCCESS) {
+		ret = -AO_FAT_EIO;
+		goto done;
+	}
+
+	while (ao_fat_readdir(&entry, &dirent)) {
+		if (memcmp(name, dirent.name, 11) == 0) {
+			uint8_t	*next;
+			uint8_t	*ent;
+			uint8_t	delete;
+
+			if (AO_FAT_IS_DIR(dirent.attr)) {
+				ret = -AO_FAT_EISDIR;
+				goto done;
+			}
+			if (!AO_FAT_IS_FILE(dirent.attr)) {
+				ret = -AO_FAT_EPERM;
+				goto done;
+			}
+
+			_ao_fat_free_cluster_chain(dirent.cluster);
+			next = _ao_fat_root_get(dirent.entry + 1);
+			if (next && next[0] != AO_FAT_DENT_END)
+				delete = AO_FAT_DENT_EMPTY;
+			else
+				delete = AO_FAT_DENT_END;
+			if (next)
+				_ao_fat_root_put(next, dirent.entry + 1, 0);
+			ent = _ao_fat_root_get(dirent.entry);
+			if (ent) {
+				memset(ent, '\0', DIRENT_SIZE);
+				*ent = delete;
+				_ao_fat_root_put(ent, dirent.entry, 1);
+			}
+			ao_bufio_flush();
+			ret = AO_FAT_SUCCESS;
+			goto done;
+		}
+	}
+	ret = -AO_FAT_ENOENT;
+done:
+	ao_mutex_put(&ao_fat_mutex);
+	return ret;
+}
+
+int8_t
+ao_fat_rename(char old[11], char new[11])
+{
+	return -AO_FAT_EIO;
+}
+
+#if FAT_COMMANDS
+
+static const char *filesystem_errors[] = {
+	[AO_FAT_FILESYSTEM_SUCCESS] = "FAT file system operating normally",
+	[AO_FAT_FILESYSTEM_MBR_READ_FAILURE] = "MBR media read error",
+	[AO_FAT_FILESYSTEM_INVALID_MBR_SIGNATURE] = "MBR signature invalid",
+	[AO_FAT_FILESYSTEM_INVALID_PARTITION_TYPE] = "Unsupported paritition type",
+	[AO_FAT_FILESYSTEM_ZERO_SIZED_PARTITION] = "Partition has zero sectors",
+	[AO_FAT_FILESYSTEM_BOOT_READ_FAILURE] = "Boot block media read error",
+	[AO_FAT_FILESYSTEM_INVALID_BOOT_SIGNATURE] = "Boot block signature invalid",
+	[AO_FAT_FILESYSTEM_INVALID_SECTOR_SIZE] = "Sector size not 512",
+};
+	
+static void
+ao_fat_mbr_cmd(void)
+{
+	uint8_t	status;
+
+	ao_mutex_get(&ao_fat_mutex);
+	status = _ao_fat_setup();
+	if (status == AO_FAT_FILESYSTEM_SUCCESS) {
+		printf ("partition type: %02x\n", partition_type);
+		printf ("partition start: %08x\n", partition_start);
+
+		printf ("partition end:   %08x\n", partition_end);
+
+		printf ("fat32: %d\n", fat32);
+		printf ("sectors per cluster %d\n", sectors_per_cluster);
+		printf ("reserved sectors %d\n", reserved_sector_count);
+		printf ("number of FATs %d\n", number_fat);
+		printf ("root entries %d\n", root_entries);
+		printf ("sectors per fat %d\n", sectors_per_fat);
+
+		printf ("fat  start %d\n", fat_start);
+		printf ("root start %d\n", root_start);
+		printf ("data start %d\n", data_start);
+	} else {
+		printf ("FAT filesystem not available: %s\n", filesystem_errors[status]);
+	}
+	ao_mutex_put(&ao_fat_mutex);
+}
+
+struct ao_fat_attr {
+	uint8_t	bit;
+	char	label;
+};
+
+static const struct ao_fat_attr ao_fat_attr[] = {
+	{ .bit = AO_FAT_FILE_READ_ONLY, .label = 'R' },
+	{ .bit = AO_FAT_FILE_HIDDEN, .label = 'H' },
+	{ .bit = AO_FAT_FILE_SYSTEM, .label = 'S' },
+	{ .bit = AO_FAT_FILE_VOLUME_LABEL, .label = 'V' },
+	{ .bit = AO_FAT_FILE_DIRECTORY, .label = 'D' },
+	{ .bit = AO_FAT_FILE_ARCHIVE, .label = 'A' },
+};
+
+#define NUM_FAT_ATTR	(sizeof (ao_fat_attr) / sizeof (ao_fat_attr[0]))
+
+static void
+ao_fat_list_cmd(void)
+{
+	uint16_t		entry = 0;
+	static struct ao_fat_dirent	dirent;
+	int			i;
+	int8_t			status;
+
+	while ((status = ao_fat_readdir(&entry, &dirent)) == AO_FAT_SUCCESS) {
+		for (i = 0; i < 8; i++)
+			putchar(dirent.name[i]);
+		putchar('.');
+		for (; i < 11; i++)
+			putchar(dirent.name[i]);
+		for (i = 0; i < NUM_FAT_ATTR; i++)
+			putchar (dirent.attr & ao_fat_attr[i].bit ? ao_fat_attr[i].label : ' ');
+		printf (" @%08x %d\n", dirent.cluster, dirent.size);
+	}
+	if (status != -AO_FAT_EDIREOF)
+		printf ("readdir failed: %d\n", status);
+}
+
+static uint8_t
+ao_fat_parse_name(char name[11])
+{
+	uint8_t	c;
+
+	name[0] = '\0';
+	ao_cmd_white();
+	c = 0;
+	while (ao_cmd_lex_c != '\n') {
+		if (ao_cmd_lex_c == '.') {
+			for (; c < 8; c++)
+				name[c] = ' ';
+		} else {
+			if (c < 11)
+				name[c++] = ao_cmd_lex_c;
+		}
+		ao_cmd_lex();
+	}
+	while (c < 11)
+		name[c++] = ' ';
+}
+
+static void
+ao_fat_dump_cmd(void)
+{
+	static char	name[11];
+	int8_t		fd;
+	int		cnt, i;
+	static char	buf[32];
+
+	ao_fat_parse_name(name);
+	if (name[0] == '\0') {
+		ao_cmd_status = ao_cmd_syntax_error;
+		return;
+	}
+		
+	fd = ao_fat_open(name, AO_FAT_OPEN_READ);
+	if (fd < 0) {
+		printf ("Open failed: %d\n", fd);
+		return;
+	}
+	while ((cnt = ao_fat_read(fd, buf, sizeof(buf))) > 0) {
+		for (i = 0; i < cnt; i++)
+			putchar(buf[i]);
+	}
+	ao_fat_close(fd);
+}
+
+static void
+ao_fat_write_cmd(void)
+{
+	static char	name[11];
+	int8_t		fd;
+	int		cnt, i;
+	static char	buf[64];
+	char		c;
+	int		status;
+
+	ao_fat_parse_name(name);
+	if (name[0] == '\0') {
+		ao_cmd_status = ao_cmd_syntax_error;
+		return;
+	}
+		
+	fd = ao_fat_creat(name);
+	if (fd < 0) {
+		printf ("Open failed: %d\n", fd);
+		return;
+	}
+	flush();
+	while ((c = getchar()) != 4) {
+		if (c == '\r') c = '\n';
+		if (ao_echo()) {
+			if (c == '\n') putchar ('\r');
+			putchar(c); flush();
+		}
+		status = ao_fat_write(fd, &c, 1);
+		if (status != 1) {
+			printf ("Write failure %d\n", status);
+			break;
+		}
+	}
+	ao_fat_close(fd);
+}
+
+static void
+put32(uint32_t a)
+{
+	ao_cmd_put16(a >> 16);
+	ao_cmd_put16(a);
+}
+
+static void
+ao_fat_hexdump_cmd(void)
+{
+	char		name[11];
+	int8_t		fd;
+	int		cnt, i;
+	char		buf[8];
+	uint32_t	addr;
+
+	ao_fat_parse_name(name);
+	if (name[0] == '\0') {
+		ao_cmd_status = ao_cmd_syntax_error;
+		return;
+	}
+		
+	fd = ao_fat_open(name, AO_FAT_OPEN_READ);
+	if (fd < 0) {
+		printf ("Open failed: %d\n", fd);
+		return;
+	}
+	addr = 0;
+	while ((cnt = ao_fat_read(fd, buf, sizeof(buf))) > 0) {
+		put32(addr);
+		for (i = 0; i < cnt; i++) {
+			putchar(' ');
+			ao_cmd_put8(buf[i]);
+		}
+		putchar('\n');
+		addr += cnt;
+	}
+	ao_fat_close(fd);
+}
+
+static const struct ao_cmds ao_fat_cmds[] = {
+	{ ao_fat_mbr_cmd,	"M\0Show FAT MBR and other info" },
+	{ ao_fat_list_cmd,	"F\0List FAT directory" },
+	{ ao_fat_dump_cmd,	"D <name>\0Dump FAT file" },
+	{ ao_fat_write_cmd,	"W <name>\0Write FAT file (end with ^D)" },
+	{ ao_fat_hexdump_cmd,	"H <name>\0HEX dump FAT file" },
+	{ 0, NULL },
+};
+
+#endif
+
+void
+ao_fat_init(void)
+{
+	ao_bufio_init();
+	ao_cmd_register(&ao_fat_cmds[0]);
+}