1$NetBSD: README,v 1.4 2009/01/03 08:25:35 lukem Exp $ 2 3makefs - build a file system image from a directory tree 4 5NOTES: 6 7 * This tool uses modified local copies of source found in other 8 parts of the tree. This is intentional. 9 10 * makefs is a work in progress, and subject to change. 11 12 13user overview: 14-------------- 15 16makefs creates a file system image from a given directory tree. 17the following file system types can be built: 18 ffs BSD fast file system 19 cd9660 ISO 9660 file system 20 21Support for the following file systems maybe be added in the future 22 ext2fs Linux EXT2 file system 23 fat MS-DOS `FAT' file system (FAT12, FAT16, FAT32) 24 25Various file system independent parameters and contraints can be 26specified, such as: 27 - minimum file system size (in KB) 28 - maximum file system size (in KB) 29 - free inodes 30 - free blocks (in KB) 31 - mtree(8) specification file containing permissions and ownership 32 to use in image, overridding the settings in the directory tree 33 - file containing list of files to specifically exclude or include 34 - fnmatch(3) pattern of filenames to exclude or include 35 - endianness of target file system 36 37File system specific parameters can be given as well, with a command 38line option such as "-o fsspeccific-options,comma-separated". 39For example, ffs would allow tuning of: 40 - block & fragment size 41 - cylinder groups 42 - number of blocks per inode 43 - minimum free space 44 45Other file systems might have controls on how to "munge" file names to 46fit within the constraints of the target file system. 47 48Exit codes: 49 0 all ok 50 1 fatal error 51 2 some files couldn't be added during image creation 52 (bad perms, missing file, etc). image will continue 53 to be made 54 55 56Implementation overview: 57------------------------ 58 59The implementation must allow for easy addition of extra file systems 60with minimal changes to the file system independent sections. 61 62The main program will: 63 - parse the options, including calling fs-specific routines to 64 validate fs-specific options 65 - walk the tree, building up a data structure which represents 66 the tree to stuff into the image. The structure will 67 probably be a similar tree to what mtree(8) uses internally; 68 a linked list of entries per directory with a child pointer 69 to children of directories. ".." won't be stored in the list; 70 the fs-specific tree walker should add this if required by the fs. 71 this builder have the smarts to handle hard links correctly. 72 - (optionally) Change the permissions in the tree according to 73 the mtree(8) specfile 74 - Call an fs-specific routine to build the image based on the 75 data structures. 76 77Each fs-specific module should have the following external interfaces: 78 79 prepare_options optional file system specific defaults that need to be 80 setup before parsing fs-specific options. 81 82 parse_options parse the string for fs-specific options, feeding 83 errors back to the user as appropriate 84 85 cleanup_options optional file system specific data that need to be 86 cleaned up when done with this filesystem. 87 88 make_fs take the data structures representing the 89 directory tree and fs parameters, 90 validate that the parameters are valid 91 (e.g, the requested image will be large enough), 92 create the image, and 93 populate the image 94 95prepare_options and cleanup_options are optional and can be NULL. 96 97NOTE: All file system specific options are referenced via the fs_specific 98pointer from the fsinfo_t strucutre. It is up to the filesystem to allocate 99and free any data needed for this via the prepare and cleanup callbacks. 100 101Each fs-specific module will need to add it's routines to the dispatch array 102in makefs.c and add prototypes for these to makefs.h 103 104All other implementation details should not need to change any of the 105generic code. 106 107ffs implementation 108------------------ 109 110In the ffs case, we can leverage off sbin/newfs/mkfs.c to actually build 111the image. When building and populating the image, the implementation 112can be greatly simplified if some assumptions are made: 113 - the total required size (in blocks and inodes) is determined 114 as part of the validation phase 115 - a "file" (including a directory) has a known size, so 116 support for growing a file is not necessary 117 118Two underlying primitives are provided: 119 make_inode create an inode, returning the inode number 120 121 write_file write file (from memory if DIR, file descriptor 122 if FILE or SYMLINK), referencing given inode. 123 it is smart enough to know if a short symlink 124 can be stuffed into the inode, etc. 125 126When creating a directory, the directory entries in the previously 127built tree data structure is scanned and built in memory so it can 128be written entirely as a single write_file() operation. 129