/* * Copyright (C) 2010 The Android Open Source Project * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ #include <fcntl.h> #include <arpa/inet.h> #include <sys/ioctl.h> #include <sys/stat.h> #include <sys/types.h> #include <string.h> #if defined(__linux__) #include <linux/fs.h> #elif defined(__APPLE__) && defined(__MACH__) #include <sys/disk.h> #endif #include "ext4_utils.h" #include "output_file.h" #include "backed_block.h" #include "uuid.h" #include "allocate.h" #include "indirect.h" #include "extent.h" #include "ext4.h" #include "jbd2.h" int force = 0; struct fs_info info; struct fs_aux_info aux_info; /* returns 1 if a is a power of b */ static int is_power_of(int a, int b) { while (a > b) { if (a % b) return 0; a /= b; } return (a == b) ? 1 : 0; } /* Returns 1 if the bg contains a backup superblock. On filesystems with the sparse_super feature, only block groups 0, 1, and powers of 3, 5, and 7 have backup superblocks. Otherwise, all block groups have backup superblocks */ int ext4_bg_has_super_block(int bg) { /* Without sparse_super, every block group has a superblock */ if (!(info.feat_ro_compat & EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER)) return 1; if (bg == 0 || bg == 1) return 1; if (is_power_of(bg, 3) || is_power_of(bg, 5) || is_power_of(bg, 7)) return 1; return 0; } /* Write the filesystem image to a file */ void write_ext4_image(const char *filename, int gz, int sparse) { int ret = 0; struct output_file *out = open_output_file(filename, gz, sparse); off_t off; if (!out) return; /* The write_data* functions expect only block aligned calls. * This is not an issue, except when we write out the super * block on a system with a block size > 1K. So, we need to * deal with that here. */ if (info.block_size > 1024) { u8 buf[4096] = { 0 }; /* The larget supported ext4 block size */ memcpy(buf + 1024, (u8*)aux_info.sb, 1024); write_data_block(out, 0, buf, info.block_size); } else { write_data_block(out, 1024, (u8*)aux_info.sb, 1024); } write_data_block(out, (u64)(aux_info.first_data_block + 1) * info.block_size, (u8*)aux_info.bg_desc, aux_info.bg_desc_blocks * info.block_size); for_each_data_block(write_data_block, write_data_file, out); pad_output_file(out, info.len); close_output_file(out); } /* Compute the rest of the parameters of the filesystem from the basic info */ void ext4_create_fs_aux_info() { aux_info.first_data_block = (info.block_size > 1024) ? 0 : 1; aux_info.len_blocks = info.len / info.block_size; aux_info.inode_table_blocks = DIV_ROUND_UP(info.inodes_per_group * info.inode_size, info.block_size); aux_info.groups = DIV_ROUND_UP(aux_info.len_blocks - aux_info.first_data_block, info.blocks_per_group); aux_info.blocks_per_ind = info.block_size / sizeof(u32); aux_info.blocks_per_dind = aux_info.blocks_per_ind * aux_info.blocks_per_ind; aux_info.blocks_per_tind = aux_info.blocks_per_dind * aux_info.blocks_per_dind; aux_info.bg_desc_blocks = DIV_ROUND_UP(aux_info.groups * sizeof(struct ext2_group_desc), info.block_size); aux_info.bg_desc_reserve_blocks = DIV_ROUND_UP(aux_info.groups * 1024 * sizeof(struct ext2_group_desc), info.block_size) - aux_info.bg_desc_blocks; if (aux_info.bg_desc_reserve_blocks > aux_info.blocks_per_ind) aux_info.bg_desc_reserve_blocks = aux_info.blocks_per_ind; aux_info.default_i_flags = EXT4_NOATIME_FL; u32 last_group_size = aux_info.len_blocks % info.blocks_per_group; u32 last_header_size = 2 + aux_info.inode_table_blocks; if (ext4_bg_has_super_block(aux_info.groups - 1)) last_header_size += 1 + aux_info.bg_desc_blocks + aux_info.bg_desc_reserve_blocks; if (last_group_size > 0 && last_group_size < last_header_size) { aux_info.groups--; aux_info.len_blocks -= last_group_size; } aux_info.sb = calloc(info.block_size, 1); if (!aux_info.sb) critical_error_errno("calloc"); aux_info.bg_desc = calloc(info.block_size, aux_info.bg_desc_blocks); if (!aux_info.bg_desc) critical_error_errno("calloc"); } void ext4_free_fs_aux_info() { free(aux_info.sb); free(aux_info.bg_desc); } /* Fill in the superblock memory buffer based on the filesystem parameters */ void ext4_fill_in_sb() { unsigned int i; struct ext4_super_block *sb = aux_info.sb; sb->s_inodes_count = info.inodes_per_group * aux_info.groups; sb->s_blocks_count_lo = aux_info.len_blocks; sb->s_r_blocks_count_lo = 0; sb->s_free_blocks_count_lo = 0; sb->s_free_inodes_count = 0; sb->s_first_data_block = aux_info.first_data_block; sb->s_log_block_size = log_2(info.block_size / 1024); sb->s_obso_log_frag_size = log_2(info.block_size / 1024); sb->s_blocks_per_group = info.blocks_per_group; sb->s_obso_frags_per_group = info.blocks_per_group; sb->s_inodes_per_group = info.inodes_per_group; sb->s_mtime = 0; sb->s_wtime = 0; sb->s_mnt_count = 0; sb->s_max_mnt_count = 0xFFFF; sb->s_magic = EXT4_SUPER_MAGIC; sb->s_state = EXT4_VALID_FS; sb->s_errors = EXT4_ERRORS_RO; sb->s_minor_rev_level = 0; sb->s_lastcheck = 0; sb->s_checkinterval = 0; sb->s_creator_os = EXT4_OS_LINUX; sb->s_rev_level = EXT4_DYNAMIC_REV; sb->s_def_resuid = EXT4_DEF_RESUID; sb->s_def_resgid = EXT4_DEF_RESGID; sb->s_first_ino = EXT4_GOOD_OLD_FIRST_INO; sb->s_inode_size = info.inode_size; sb->s_block_group_nr = 0; sb->s_feature_compat = info.feat_compat; sb->s_feature_incompat = info.feat_incompat; sb->s_feature_ro_compat = info.feat_ro_compat; generate_uuid("extandroid/make_ext4fs", info.label, sb->s_uuid); memset(sb->s_volume_name, 0, sizeof(sb->s_volume_name)); strncpy(sb->s_volume_name, info.label, sizeof(sb->s_volume_name)); memset(sb->s_last_mounted, 0, sizeof(sb->s_last_mounted)); sb->s_algorithm_usage_bitmap = 0; sb->s_reserved_gdt_blocks = aux_info.bg_desc_reserve_blocks; sb->s_prealloc_blocks = 0; sb->s_prealloc_dir_blocks = 0; //memcpy(sb->s_journal_uuid, sb->s_uuid, sizeof(sb->s_journal_uuid)); if (info.feat_compat & EXT4_FEATURE_COMPAT_HAS_JOURNAL) sb->s_journal_inum = EXT4_JOURNAL_INO; sb->s_journal_dev = 0; sb->s_last_orphan = 0; sb->s_hash_seed[0] = 0; /* FIXME */ sb->s_def_hash_version = DX_HASH_TEA; sb->s_reserved_char_pad = EXT4_JNL_BACKUP_BLOCKS; sb->s_desc_size = sizeof(struct ext2_group_desc); sb->s_default_mount_opts = 0; /* FIXME */ sb->s_first_meta_bg = 0; sb->s_mkfs_time = 0; //sb->s_jnl_blocks[17]; /* FIXME */ sb->s_blocks_count_hi = aux_info.len_blocks >> 32; sb->s_r_blocks_count_hi = 0; sb->s_free_blocks_count_hi = 0; sb->s_min_extra_isize = sizeof(struct ext4_inode) - EXT4_GOOD_OLD_INODE_SIZE; sb->s_want_extra_isize = sizeof(struct ext4_inode) - EXT4_GOOD_OLD_INODE_SIZE; sb->s_flags = 0; sb->s_raid_stride = 0; sb->s_mmp_interval = 0; sb->s_mmp_block = 0; sb->s_raid_stripe_width = 0; sb->s_log_groups_per_flex = 0; sb->s_kbytes_written = 0; for (i = 0; i < aux_info.groups; i++) { u64 group_start_block = aux_info.first_data_block + i * info.blocks_per_group; u32 header_size = 0; if (ext4_bg_has_super_block(i)) { if (i != 0) { queue_data_block((u8 *)sb, info.block_size, group_start_block); queue_data_block((u8 *)aux_info.bg_desc, aux_info.bg_desc_blocks * info.block_size, group_start_block + 1); } header_size = 1 + aux_info.bg_desc_blocks + aux_info.bg_desc_reserve_blocks; } aux_info.bg_desc[i].bg_block_bitmap = group_start_block + header_size; aux_info.bg_desc[i].bg_inode_bitmap = group_start_block + header_size + 1; aux_info.bg_desc[i].bg_inode_table = group_start_block + header_size + 2; aux_info.bg_desc[i].bg_free_blocks_count = sb->s_blocks_per_group; aux_info.bg_desc[i].bg_free_inodes_count = sb->s_inodes_per_group; aux_info.bg_desc[i].bg_used_dirs_count = 0; } } void ext4_create_resize_inode() { struct block_allocation *reserve_inode_alloc = create_allocation(); u32 reserve_inode_len = 0; unsigned int i; struct ext4_inode *inode = get_inode(EXT4_RESIZE_INO); if (inode == NULL) { error("failed to get resize inode"); return; } for (i = 0; i < aux_info.groups; i++) { if (ext4_bg_has_super_block(i)) { u64 group_start_block = aux_info.first_data_block + i * info.blocks_per_group; u32 reserved_block_start = group_start_block + 1 + aux_info.bg_desc_blocks; u32 reserved_block_len = aux_info.bg_desc_reserve_blocks; append_region(reserve_inode_alloc, reserved_block_start, reserved_block_len, i); reserve_inode_len += reserved_block_len; } } inode_attach_resize(inode, reserve_inode_alloc); inode->i_mode = S_IFREG | S_IRUSR | S_IWUSR; inode->i_links_count = 1; free_alloc(reserve_inode_alloc); } /* Allocate the blocks to hold a journal inode and connect them to the reserved journal inode */ void ext4_create_journal_inode() { struct ext4_inode *inode = get_inode(EXT4_JOURNAL_INO); if (inode == NULL) { error("failed to get journal inode"); return; } u8 *journal_data = inode_allocate_data_extents(inode, info.journal_blocks * info.block_size, info.journal_blocks * info.block_size); if (!journal_data) { error("failed to allocate extents for journal data"); return; } inode->i_mode = S_IFREG | S_IRUSR | S_IWUSR; inode->i_links_count = 1; journal_superblock_t *jsb = (journal_superblock_t *)journal_data; jsb->s_header.h_magic = htonl(JBD2_MAGIC_NUMBER); jsb->s_header.h_blocktype = htonl(JBD2_SUPERBLOCK_V2); jsb->s_blocksize = htonl(info.block_size); jsb->s_maxlen = htonl(info.journal_blocks); jsb->s_nr_users = htonl(1); jsb->s_first = htonl(1); jsb->s_sequence = htonl(1); memcpy(aux_info.sb->s_jnl_blocks, &inode->i_block, sizeof(inode->i_block)); } /* Update the number of free blocks and inodes in the filesystem and in each block group */ void ext4_update_free() { unsigned int i; for (i = 0; i < aux_info.groups; i++) { u32 bg_free_blocks = get_free_blocks(i); u32 bg_free_inodes = get_free_inodes(i); aux_info.bg_desc[i].bg_free_blocks_count = bg_free_blocks; aux_info.sb->s_free_blocks_count_lo += bg_free_blocks; aux_info.bg_desc[i].bg_free_inodes_count = bg_free_inodes; aux_info.sb->s_free_inodes_count += bg_free_inodes; aux_info.bg_desc[i].bg_used_dirs_count += get_directories(i); } } static u64 get_block_device_size(const char *filename) { int fd = open(filename, O_RDONLY); u64 size = 0; int ret; if (fd < 0) return 0; #if defined(__linux__) ret = ioctl(fd, BLKGETSIZE64, &size); #elif defined(__APPLE__) && defined(__MACH__) ret = ioctl(fd, DKIOCGETBLOCKCOUNT, &size); #else return 0; #endif close(fd); if (ret) return 0; return size; } u64 get_file_size(const char *filename) { struct stat buf; int ret; ret = stat(filename, &buf); if (ret) return 0; if (S_ISREG(buf.st_mode)) return buf.st_size; else if (S_ISBLK(buf.st_mode)) return get_block_device_size(filename); else return 0; } u64 parse_num(const char *arg) { char *endptr; u64 num = strtoull(arg, &endptr, 10); if (*endptr == 'k' || *endptr == 'K') num *= 1024LL; else if (*endptr == 'm' || *endptr == 'M') num *= 1024LL * 1024LL; else if (*endptr == 'g' || *endptr == 'G') num *= 1024LL * 1024LL * 1024LL; return num; }