/* * linux/fs/fat/cache.c * * Written 1992,1993 by Werner Almesberger * * Mar 1999. AV. Changed cache, so that it uses the starting cluster instead * of inode number. * May 1999. AV. Fixed the bogosity with FAT32 (read "FAT28"). Fscking lusers. */ #include #include #include #include #include #include #if 0 # define PRINTK(x) printk x #else # define PRINTK(x) #endif static struct fat_cache *fat_cache,cache[FAT_CACHE]; static spinlock_t fat_cache_lock = SPIN_LOCK_UNLOCKED; /* Returns the this'th FAT entry, -1 if it is an end-of-file entry. If new_value is != -1, that FAT entry is replaced by it. */ int fat_access(struct super_block *sb,int nr,int new_value) { return MSDOS_SB(sb)->cvf_format->fat_access(sb,nr,new_value); } int fat_bmap(struct inode *inode,int sector) { return MSDOS_SB(inode->i_sb)->cvf_format->cvf_bmap(inode,sector); } int default_fat_access(struct super_block *sb,int nr,int new_value) { struct buffer_head *bh, *bh2, *c_bh, *c_bh2; unsigned char *p_first, *p_last; int copy, first, last, next, b; if ((unsigned) (nr-2) >= MSDOS_SB(sb)->clusters) return 0; if (MSDOS_SB(sb)->fat_bits == 32) { first = last = nr*4; } else if (MSDOS_SB(sb)->fat_bits == 16) { first = last = nr*2; } else { first = nr*3/2; last = first+1; } b = MSDOS_SB(sb)->fat_start + (first >> sb->s_blocksize_bits); if (!(bh = fat_bread(sb, b))) { printk("bread in fat_access failed\n"); return 0; } if ((first >> sb->s_blocksize_bits) == (last >> sb->s_blocksize_bits)) { bh2 = bh; } else { if (!(bh2 = fat_bread(sb, b+1))) { fat_brelse(sb, bh); printk("2nd bread in fat_access failed\n"); return 0; } } if (MSDOS_SB(sb)->fat_bits == 32) { p_first = p_last = NULL; /* GCC needs that stuff */ next = CF_LE_L(((__u32 *) bh->b_data)[(first & (sb->s_blocksize - 1)) >> 2]); /* Fscking Microsoft marketing department. Their "32" is 28. */ next &= 0xfffffff; if (next >= 0xffffff7) next = -1; PRINTK(("fat_bread: 0x%x, nr=0x%x, first=0x%x, next=0x%x\n", b, nr, first, next)); } else if (MSDOS_SB(sb)->fat_bits == 16) { p_first = p_last = NULL; /* GCC needs that stuff */ next = CF_LE_W(((__u16 *) bh->b_data)[(first & (sb->s_blocksize - 1)) >> 1]); if (next >= 0xfff7) next = -1; } else { p_first = &((__u8 *)bh->b_data)[first & (sb->s_blocksize - 1)]; p_last = &((__u8 *)bh2->b_data)[(first + 1) & (sb->s_blocksize - 1)]; if (nr & 1) next = ((*p_first >> 4) | (*p_last << 4)) & 0xfff; else next = (*p_first+(*p_last << 8)) & 0xfff; if (next >= 0xff7) next = -1; } if (new_value != -1) { if (MSDOS_SB(sb)->fat_bits == 32) { ((__u32 *)bh->b_data)[(first & (sb->s_blocksize - 1)) >> 2] = CT_LE_L(new_value); } else if (MSDOS_SB(sb)->fat_bits == 16) { ((__u16 *)bh->b_data)[(first & (sb->s_blocksize - 1)) >> 1] = CT_LE_W(new_value); } else { if (nr & 1) { *p_first = (*p_first & 0xf) | (new_value << 4); *p_last = new_value >> 4; } else { *p_first = new_value & 0xff; *p_last = (*p_last & 0xf0) | (new_value >> 8); } fat_mark_buffer_dirty(sb, bh2); } fat_mark_buffer_dirty(sb, bh); for (copy = 1; copy < MSDOS_SB(sb)->fats; copy++) { b = MSDOS_SB(sb)->fat_start + (first >> sb->s_blocksize_bits) + MSDOS_SB(sb)->fat_length * copy; if (!(c_bh = fat_bread(sb, b))) break; if (bh != bh2) { if (!(c_bh2 = fat_bread(sb, b+1))) { fat_brelse(sb, c_bh); break; } memcpy(c_bh2->b_data, bh2->b_data, sb->s_blocksize); fat_mark_buffer_dirty(sb, c_bh2); fat_brelse(sb, c_bh2); } memcpy(c_bh->b_data, bh->b_data, sb->s_blocksize); fat_mark_buffer_dirty(sb, c_bh); fat_brelse(sb, c_bh); } } fat_brelse(sb, bh); if (bh != bh2) fat_brelse(sb, bh2); return next; } void fat_cache_init(void) { static int initialized = 0; int count; spin_lock(&fat_cache_lock); if (initialized) { spin_unlock(&fat_cache_lock); return; } fat_cache = &cache[0]; for (count = 0; count < FAT_CACHE; count++) { cache[count].device = 0; cache[count].next = count == FAT_CACHE-1 ? NULL : &cache[count+1]; } initialized = 1; spin_unlock(&fat_cache_lock); } void fat_cache_lookup(struct inode *inode,int cluster,int *f_clu,int *d_clu) { struct fat_cache *walk; int first = MSDOS_I(inode)->i_start; if (!first) return; spin_lock(&fat_cache_lock); for (walk = fat_cache; walk; walk = walk->next) if (inode->i_dev == walk->device && walk->start_cluster == first && walk->file_cluster <= cluster && walk->file_cluster > *f_clu) { *d_clu = walk->disk_cluster; #ifdef DEBUG printk("cache hit: %d (%d)\n",walk->file_cluster,*d_clu); #endif if ((*f_clu = walk->file_cluster) == cluster) { spin_unlock(&fat_cache_lock); return; } } spin_unlock(&fat_cache_lock); #ifdef DEBUG printk("cache miss\n"); #endif } #ifdef DEBUG static void list_cache(void) { struct fat_cache *walk; for (walk = fat_cache; walk; walk = walk->next) { if (walk->device) printk("<%s,%d>(%d,%d) ", kdevname(walk->device), walk->start_cluster, walk->file_cluster, walk->disk_cluster); else printk("-- "); } printk("\n"); } #endif void fat_cache_add(struct inode *inode,int f_clu,int d_clu) { struct fat_cache *walk,*last; int first = MSDOS_I(inode)->i_start; last = NULL; spin_lock(&fat_cache_lock); for (walk = fat_cache; walk->next; walk = (last = walk)->next) if (inode->i_dev == walk->device && walk->start_cluster == first && walk->file_cluster == f_clu) { if (walk->disk_cluster != d_clu) { printk("FAT cache corruption inode=%ld\n", inode->i_ino); spin_unlock(&fat_cache_lock); fat_cache_inval_inode(inode); return; } /* update LRU */ if (last == NULL) { spin_unlock(&fat_cache_lock); return; } last->next = walk->next; walk->next = fat_cache; fat_cache = walk; #ifdef DEBUG list_cache(); #endif spin_unlock(&fat_cache_lock); return; } walk->device = inode->i_dev; walk->start_cluster = first; walk->file_cluster = f_clu; walk->disk_cluster = d_clu; last->next = NULL; walk->next = fat_cache; fat_cache = walk; spin_unlock(&fat_cache_lock); #ifdef DEBUG list_cache(); #endif } /* Cache invalidation occurs rarely, thus the LRU chain is not updated. It fixes itself after a while. */ void fat_cache_inval_inode(struct inode *inode) { struct fat_cache *walk; int first = MSDOS_I(inode)->i_start; spin_lock(&fat_cache_lock); for (walk = fat_cache; walk; walk = walk->next) if (walk->device == inode->i_dev && walk->start_cluster == first) walk->device = 0; spin_unlock(&fat_cache_lock); } void fat_cache_inval_dev(kdev_t device) { struct fat_cache *walk; spin_lock(&fat_cache_lock); for (walk = fat_cache; walk; walk = walk->next) if (walk->device == device) walk->device = 0; spin_unlock(&fat_cache_lock); } int fat_get_cluster(struct inode *inode,int cluster) { int nr,count; if (!(nr = MSDOS_I(inode)->i_start)) return 0; if (!cluster) return nr; count = 0; for (fat_cache_lookup(inode,cluster,&count,&nr); count < cluster; count++) { if ((nr = fat_access(inode->i_sb,nr,-1)) == -1) return 0; if (!nr) return 0; } fat_cache_add(inode,cluster,nr); return nr; } int default_fat_bmap(struct inode *inode,int sector) { struct super_block *sb = inode->i_sb; struct msdos_sb_info *sbi = MSDOS_SB(sb); int cluster, offset, last_block; if ((sbi->fat_bits != 32) && (inode->i_ino == MSDOS_ROOT_INO || (S_ISDIR(inode->i_mode) && !MSDOS_I(inode)->i_start))) { if (sector >= sbi->dir_entries >> sbi->dir_per_block_bits) return 0; return sector + sbi->dir_start; } last_block = (MSDOS_I(inode)->mmu_private + (sb->s_blocksize - 1)) >> sb->s_blocksize_bits; if (sector >= last_block) return 0; cluster = sector / sbi->cluster_size; offset = sector % sbi->cluster_size; if (!(cluster = fat_get_cluster(inode, cluster))) return 0; return (cluster - 2) * sbi->cluster_size + sbi->data_start + offset; } /* Free all clusters after the skip'th cluster. Doesn't use the cache, because this way we get an additional sanity check. */ int fat_free(struct inode *inode,int skip) { int nr,last; if (!(nr = MSDOS_I(inode)->i_start)) return 0; last = 0; while (skip--) { last = nr; if ((nr = fat_access(inode->i_sb,nr,-1)) == -1) return 0; if (!nr) { printk("fat_free: skipped EOF\n"); return -EIO; } } if (last) { fat_access(inode->i_sb,last,EOF_FAT(inode->i_sb)); fat_cache_inval_inode(inode); } else { fat_cache_inval_inode(inode); MSDOS_I(inode)->i_start = 0; MSDOS_I(inode)->i_logstart = 0; mark_inode_dirty(inode); } lock_fat(inode->i_sb); while (nr != -1) { if (!(nr = fat_access(inode->i_sb,nr,0))) { fat_fs_panic(inode->i_sb,"fat_free: deleting beyond EOF"); break; } if (MSDOS_SB(inode->i_sb)->free_clusters != -1) { MSDOS_SB(inode->i_sb)->free_clusters++; if (MSDOS_SB(inode->i_sb)->fat_bits == 32) { fat_clusters_flush(inode->i_sb); } } inode->i_blocks -= (1 << MSDOS_SB(inode->i_sb)->cluster_bits) / 512; } unlock_fat(inode->i_sb); return 0; }