minix/servers/mfs/super.c

369 lines
12 KiB
C
Raw Normal View History

2005-08-29 18:47:18 +02:00
/* This file manages the super block table and the related data structures,
* namely, the bit maps that keep track of which zones and which inodes are
2005-04-21 16:53:53 +02:00
* allocated and which are free. When a new inode or zone is needed, the
* appropriate bit map is searched for a free entry.
*
* The entry points into this file are
* alloc_bit: somebody wants to allocate a zone or inode; find one
* free_bit: indicate that a zone or inode is available for allocation
* get_super: search the 'superblock' table for a device
* mounted: tells if file inode is on mounted (or ROOT) file system
* read_super: read a superblock
*/
#include "fs.h"
#include <string.h>
#include <assert.h>
2005-04-21 16:53:53 +02:00
#include <minix/com.h>
#include <minix/u64.h>
#include <minix/bdev.h>
2005-04-21 16:53:53 +02:00
#include "buf.h"
#include "inode.h"
#include "super.h"
#include "const.h"
2005-04-21 16:53:53 +02:00
/*===========================================================================*
* alloc_bit *
*===========================================================================*/
PUBLIC bit_t alloc_bit(sp, map, origin)
struct super_block *sp; /* the filesystem to allocate from */
int map; /* IMAP (inode map) or ZMAP (zone map) */
bit_t origin; /* number of bit to start searching at */
{
/* Allocate a bit from a bit map and return its bit number. */
block_t start_block; /* first bit block */
block_t block;
2005-04-21 16:53:53 +02:00
bit_t map_bits; /* how many bits are there in the bit map? */
short bit_blocks; /* how many blocks are there in the bit map? */
unsigned word, bcount;
2005-04-21 16:53:53 +02:00
struct buf *bp;
bitchunk_t *wptr, *wlim, k;
bit_t i, b;
if (sp->s_rd_only)
panic("can't allocate bit on read-only filesys");
2005-04-21 16:53:53 +02:00
if (map == IMAP) {
start_block = START_BLOCK;
map_bits = (bit_t) (sp->s_ninodes + 1);
2005-04-21 16:53:53 +02:00
bit_blocks = sp->s_imap_blocks;
} else {
start_block = START_BLOCK + sp->s_imap_blocks;
map_bits = (bit_t) (sp->s_zones - (sp->s_firstdatazone - 1));
2005-04-21 16:53:53 +02:00
bit_blocks = sp->s_zmap_blocks;
}
/* Figure out where to start the bit search (depends on 'origin'). */
if (origin >= map_bits) origin = 0; /* for robustness */
/* Locate the starting place. */
block = (block_t) (origin / FS_BITS_PER_BLOCK(sp->s_block_size));
word = (origin % FS_BITS_PER_BLOCK(sp->s_block_size)) / FS_BITCHUNK_BITS;
2005-04-21 16:53:53 +02:00
/* Iterate over all blocks plus one, because we start in the middle. */
bcount = bit_blocks + 1;
do {
bp = get_block(sp->s_dev, start_block + block, NORMAL);
wlim = &bp->b_bitmap[FS_BITMAP_CHUNKS(sp->s_block_size)];
2005-04-21 16:53:53 +02:00
/* Iterate over the words in block. */
for (wptr = &bp->b_bitmap[word]; wptr < wlim; wptr++) {
/* Does this word contain a free bit? */
if (*wptr == (bitchunk_t) ~0) continue;
/* Find and allocate the free bit. */
k = (bitchunk_t) conv4(sp->s_native, (int) *wptr);
2005-04-21 16:53:53 +02:00
for (i = 0; (k & (1 << i)) != 0; ++i) {}
/* Bit number from the start of the bit map. */
b = ((bit_t) block * FS_BITS_PER_BLOCK(sp->s_block_size))
+ (wptr - &bp->b_bitmap[0]) * FS_BITCHUNK_BITS
2005-04-21 16:53:53 +02:00
+ i;
/* Don't allocate bits beyond the end of the map. */
if (b >= map_bits) break;
/* Allocate and return bit number. */
k |= 1 << i;
*wptr = (bitchunk_t) conv4(sp->s_native, (int) k);
MARKDIRTY(bp);
2005-04-21 16:53:53 +02:00
put_block(bp, MAP_BLOCK);
return(b);
}
put_block(bp, MAP_BLOCK);
if (++block >= (unsigned int) bit_blocks) /* last block, wrap around */
block = 0;
2005-04-21 16:53:53 +02:00
word = 0;
} while (--bcount > 0);
return(NO_BIT); /* no bit could be allocated */
}
/*===========================================================================*
* free_bit *
*===========================================================================*/
PUBLIC void free_bit(sp, map, bit_returned)
struct super_block *sp; /* the filesystem to operate on */
int map; /* IMAP (inode map) or ZMAP (zone map) */
bit_t bit_returned; /* number of bit to insert into the map */
{
/* Return a zone or inode by turning off its bitmap bit. */
unsigned block, word, bit;
struct buf *bp;
bitchunk_t k, mask;
block_t start_block;
if (sp->s_rd_only)
panic("can't free bit on read-only filesys");
2005-04-21 16:53:53 +02:00
if (map == IMAP) {
start_block = START_BLOCK;
} else {
start_block = START_BLOCK + sp->s_imap_blocks;
}
block = bit_returned / FS_BITS_PER_BLOCK(sp->s_block_size);
2005-08-29 18:47:18 +02:00
word = (bit_returned % FS_BITS_PER_BLOCK(sp->s_block_size))
/ FS_BITCHUNK_BITS;
bit = bit_returned % FS_BITCHUNK_BITS;
2005-04-21 16:53:53 +02:00
mask = 1 << bit;
bp = get_block(sp->s_dev, start_block + block, NORMAL);
k = (bitchunk_t) conv4(sp->s_native, (int) bp->b_bitmap[word]);
2005-04-21 16:53:53 +02:00
if (!(k & mask)) {
if (map == IMAP) panic("tried to free unused inode");
else panic("tried to free unused block: %u", bit_returned);
2005-04-21 16:53:53 +02:00
}
k &= ~mask;
bp->b_bitmap[word] = (bitchunk_t) conv4(sp->s_native, (int) k);
MARKDIRTY(bp);
2005-04-21 16:53:53 +02:00
put_block(bp, MAP_BLOCK);
}
2005-04-21 16:53:53 +02:00
/*===========================================================================*
* get_super *
*===========================================================================*/
PUBLIC struct super_block *get_super(
dev_t dev /* device number whose super_block is sought */
)
2005-04-21 16:53:53 +02:00
{
2005-09-11 18:45:46 +02:00
if (dev == NO_DEV)
panic("request for super_block of NO_DEV");
2005-04-21 16:53:53 +02:00
if(superblock.s_dev != dev)
panic("wrong superblock: 0x%x", (int) dev);
2005-04-21 16:53:53 +02:00
return(&superblock);
2005-04-21 16:53:53 +02:00
}
2005-04-21 16:53:53 +02:00
/*===========================================================================*
* get_block_size *
*===========================================================================*/
PUBLIC unsigned int get_block_size(dev_t dev)
2005-04-21 16:53:53 +02:00
{
2005-09-11 18:45:46 +02:00
if (dev == NO_DEV)
panic("request for block size of NO_DEV");
2005-04-21 16:53:53 +02:00
return(fs_block_size);
2005-04-21 16:53:53 +02:00
}
/*===========================================================================*
* rw_super *
2005-04-21 16:53:53 +02:00
*===========================================================================*/
PRIVATE int rw_super(struct super_block *sp, int writing)
2005-04-21 16:53:53 +02:00
{
/* Read/write a superblock. */
int r;
static char *sbbuf;
dev_t save_dev = sp->s_dev;
/* To keep the 1kb on disk clean, only read/write up to and including
* this field.
*/
#define LAST_ONDISK_FIELD s_disk_version
int ondisk_bytes = (int) ((char *) &sp->LAST_ONDISK_FIELD - (char *) sp)
+ sizeof(sp->LAST_ONDISK_FIELD);
STATICINIT(sbbuf, _MIN_BLOCK_SIZE);
2005-04-21 16:53:53 +02:00
assert(ondisk_bytes > 0);
assert(ondisk_bytes < _MIN_BLOCK_SIZE);
assert(ondisk_bytes < sizeof(struct super_block));
if (sp->s_dev == NO_DEV)
panic("request for super_block of NO_DEV");
if(writing) {
memset(sbbuf, 0, _MIN_BLOCK_SIZE);
memcpy(sbbuf, sp, ondisk_bytes);
r = bdev_write(sp->s_dev, cvu64(SUPER_BLOCK_BYTES), sbbuf, _MIN_BLOCK_SIZE,
BDEV_NOFLAGS);
} else {
r = bdev_read(sp->s_dev, cvu64(SUPER_BLOCK_BYTES), sbbuf, _MIN_BLOCK_SIZE,
BDEV_NOFLAGS);
memset(sp, 0, sizeof(*sp));
memcpy(sp, sbbuf, ondisk_bytes);
sp->s_dev = save_dev;
}
Split block/character protocols and libdriver This patch separates the character and block driver communication protocols. The old character protocol remains the same, but a new block protocol is introduced. The libdriver library is replaced by two new libraries: libchardriver and libblockdriver. Their exposed API, and drivers that use them, have been updated accordingly. Together, libbdev and libblockdriver now completely abstract away the message format used by the block protocol. As the memory driver is both a character and a block device driver, it now implements its own message loop. The most important semantic change made to the block protocol is that it is no longer possible to return both partial results and an error for a single transfer. This simplifies the interaction between the caller and the driver, as the I/O vector no longer needs to be copied back. Also, drivers are now no longer supposed to decide based on the layout of the I/O vector when a transfer should be cut short. Put simply, transfers are now supposed to either succeed completely, or result in an error. After this patch, the state of the various pieces is as follows: - block protocol: stable - libbdev API: stable for synchronous communication - libblockdriver API: needs slight revision (the drvlib/partition API in particular; the threading API will also change shortly) - character protocol: needs cleanup - libchardriver API: needs cleanup accordingly - driver restarts: largely unsupported until endpoint changes are reintroduced As a side effect, this patch eliminates several bugs, hacks, and gcc -Wall and -W warnings all over the place. It probably introduces a few new ones, too. Update warning: this patch changes the protocol between MFS and disk drivers, so in order to use old/new images, the MFS from the ramdisk must be used to mount all file systems.
2011-11-22 13:27:53 +01:00
if (r != _MIN_BLOCK_SIZE)
return(EINVAL);
return OK;
}
/*===========================================================================*
* read_super *
*===========================================================================*/
PUBLIC int read_super(struct super_block *sp)
{
unsigned int magic;
block_t offset;
int version, native, r;
if((r=rw_super(sp, 0)) != OK)
return r;
2005-04-21 16:53:53 +02:00
magic = sp->s_magic; /* determines file system type */
/* Get file system version and type. */
if (magic == SUPER_MAGIC || magic == conv2(BYTE_SWAP, SUPER_MAGIC)) {
version = V1;
native = (magic == SUPER_MAGIC);
} else if (magic == SUPER_V2 || magic == conv2(BYTE_SWAP, SUPER_V2)) {
version = V2;
native = (magic == SUPER_V2);
} else if (magic == SUPER_V3) {
version = V3;
native = 1;
} else {
return(EINVAL);
}
/* If the super block has the wrong byte order, swap the fields; the magic
* number doesn't need conversion. */
sp->s_ninodes = (ino_t) conv4(native, (int) sp->s_ninodes);
sp->s_nzones = (zone1_t) conv2(native, (int) sp->s_nzones);
sp->s_imap_blocks = (short) conv2(native, (int) sp->s_imap_blocks);
sp->s_zmap_blocks = (short) conv2(native, (int) sp->s_zmap_blocks);
sp->s_firstdatazone_old =(zone1_t)conv2(native,(int)sp->s_firstdatazone_old);
sp->s_log_zone_size = (short) conv2(native, (int) sp->s_log_zone_size);
sp->s_max_size = (off_t) conv4(native, sp->s_max_size);
sp->s_zones = (zone_t)conv4(native, sp->s_zones);
2005-04-21 16:53:53 +02:00
/* In V1, the device size was kept in a short, s_nzones, which limited
* devices to 32K zones. For V2, it was decided to keep the size as a
* long. However, just changing s_nzones to a long would not work, since
* then the position of s_magic in the super block would not be the same
* in V1 and V2 file systems, and there would be no way to tell whether
* a newly mounted file system was V1 or V2. The solution was to introduce
* a new variable, s_zones, and copy the size there.
*
* Calculate some other numbers that depend on the version here too, to
* hide some of the differences.
*/
if (version == V1) {
sp->s_block_size = _STATIC_BLOCK_SIZE;
sp->s_zones = (zone_t) sp->s_nzones; /* only V1 needs this copy */
2005-04-21 16:53:53 +02:00
sp->s_inodes_per_block = V1_INODES_PER_BLOCK;
sp->s_ndzones = V1_NR_DZONES;
sp->s_nindirs = V1_INDIRECTS;
} else {
if (version == V2) {
sp->s_block_size = _STATIC_BLOCK_SIZE;
} else {
sp->s_block_size = (unsigned short)
conv2(native,(int) sp->s_block_size);
}
if (sp->s_block_size < _MIN_BLOCK_SIZE) {
2005-04-21 16:53:53 +02:00
return EINVAL;
}
2005-04-21 16:53:53 +02:00
sp->s_inodes_per_block = V2_INODES_PER_BLOCK(sp->s_block_size);
sp->s_ndzones = V2_NR_DZONES;
sp->s_nindirs = V2_INDIRECTS(sp->s_block_size);
}
/* For even larger disks, a similar problem occurs with s_firstdatazone.
* If the on-disk field contains zero, we assume that the value was too
* large to fit, and compute it on the fly.
*/
if (sp->s_firstdatazone_old == 0) {
offset = START_BLOCK + sp->s_imap_blocks + sp->s_zmap_blocks;
offset += (sp->s_ninodes + sp->s_inodes_per_block - 1) /
sp->s_inodes_per_block;
sp->s_firstdatazone = (offset + (1 << sp->s_log_zone_size) - 1) >>
sp->s_log_zone_size;
} else {
sp->s_firstdatazone = (zone_t) sp->s_firstdatazone_old;
}
if (sp->s_block_size < _MIN_BLOCK_SIZE)
return(EINVAL);
if ((sp->s_block_size % 512) != 0)
return(EINVAL);
if (SUPER_SIZE > sp->s_block_size)
return(EINVAL);
2005-09-11 18:45:46 +02:00
if ((sp->s_block_size % V2_INODE_SIZE) != 0 ||
2005-04-21 16:53:53 +02:00
(sp->s_block_size % V1_INODE_SIZE) != 0) {
return(EINVAL);
2005-04-21 16:53:53 +02:00
}
/* Limit s_max_size to LONG_MAX */
if ((unsigned long)sp->s_max_size > LONG_MAX)
sp->s_max_size = LONG_MAX;
2005-04-21 16:53:53 +02:00
sp->s_isearch = 0; /* inode searches initially start at 0 */
sp->s_zsearch = 0; /* zone searches initially start at 0 */
sp->s_version = version;
sp->s_native = native;
/* Make a few basic checks to see if super block looks reasonable. */
if (sp->s_imap_blocks < 1 || sp->s_zmap_blocks < 1
|| sp->s_ninodes < 1 || sp->s_zones < 1
|| sp->s_firstdatazone <= 4
|| sp->s_firstdatazone >= sp->s_zones
2005-04-21 16:53:53 +02:00
|| (unsigned) sp->s_log_zone_size > 4) {
printf("not enough imap or zone map blocks, \n");
printf("or not enough inodes, or not enough zones, \n"
"or invalid first data zone, or zone size too large\n");
2005-04-21 16:53:53 +02:00
return(EINVAL);
}
/* Check any flags we don't understand but are required to. Currently
* these don't exist so all such unknown bits are fatal.
*/
if(sp->s_flags & MFSFLAG_MANDATORY_MASK) {
printf("MFS: unsupported feature flags on this FS.\n"
"Please use a newer MFS to mount it.\n");
return(EINVAL);
}
2005-04-21 16:53:53 +02:00
return(OK);
}
/*===========================================================================*
* write_super *
*===========================================================================*/
PUBLIC int write_super(struct super_block *sp)
{
if(sp->s_rd_only)
panic("can't write superblock of readonly filesystem");
return rw_super(sp, 1);
}