1The QNX6 Filesystem
   4The qnx6fs is used by newer QNX operating system versions. (e.g. Neutrino)
   5It got introduced in QNX 6.4.0 and is used default since 6.4.1.
  10mmi_fs          Mount filesystem as used for example by Audi MMI 3G system
  15qnx6fs shares many properties with traditional Unix filesystems. It has the
  16concepts of blocks, inodes and directories.
  17On QNX it is possible to create little endian and big endian qnx6 filesystems.
  18This feature makes it possible to create and use a different endianness fs
  19for the target (QNX is used on quite a range of embedded systems) plattform
  20running on a different endianness.
  21The Linux driver handles endianness transparently. (LE and BE)
  26The space in the device or file is split up into blocks. These are a fixed
  27size of 512, 1024, 2048 or 4096, which is decided when the filesystem is
  29Blockpointers are 32bit, so the maximum space that can be addressed is
  302^32 * 4096 bytes or 16TB
  32The superblocks
  35The superblock contains all global information about the filesystem.
  36Each qnx6fs got two superblocks, each one having a 64bit serial number.
  37That serial number is used to identify the "active" superblock.
  38In write mode with reach new snapshot (after each synchronous write), the
  39serial of the new master superblock is increased (old superblock serial + 1)
  41So basically the snapshot functionality is realized by an atomic final
  42update of the serial number. Before updating that serial, all modifications
  43are done by copying all modified blocks during that specific write request
  44(or period) and building up a new (stable) filesystem structure under the
  45inactive superblock.
  47Each superblock holds a set of root inodes for the different filesystem
  48parts. (Inode, Bitmap and Longfilenames)
  49Each of these root nodes holds information like total size of the stored
  50data and the addressing levels in that specific tree.
  51If the level value is 0, up to 16 direct blocks can be addressed by each
  53Level 1 adds an additional indirect addressing level where each indirect
  54addressing block holds up to blocksize / 4 bytes pointers to data blocks.
  55Level 2 adds an additional indirect addressing block level (so, already up
  56to 16 * 256 * 256 = 1048576 blocks that can be addressed by such a tree).
  58Unused block pointers are always set to ~0 - regardless of root node,
  59indirect addressing blocks or inodes.
  60Data leaves are always on the lowest level. So no data is stored on upper
  61tree levels.
  63The first Superblock is located at 0x2000. (0x2000 is the bootblock size)
  64The Audi MMI 3G first superblock directly starts at byte 0.
  65Second superblock position can either be calculated from the superblock
  66information (total number of filesystem blocks) or by taking the highest
  67device address, zeroing the last 3 bytes and then subtracting 0x1000 from
  68that address.
  700x1000 is the size reserved for each superblock - regardless of the
  71blocksize of the filesystem.
  76Each object in the filesystem is represented by an inode. (index node)
  77The inode structure contains pointers to the filesystem blocks which contain
  78the data held in the object and all of the metadata about an object except
  79its longname. (filenames longer than 27 characters)
  80The metadata about an object includes the permissions, owner, group, flags,
  81size, number of blocks used, access time, change time and modification time.
  83Object mode field is POSIX format. (which makes things easier)
  85There are also pointers to the first 16 blocks, if the object data can be
  86addressed with 16 direct blocks.
  87For more than 16 blocks an indirect addressing in form of another tree is
  88used. (scheme is the same as the one used for the superblock root nodes)
  90The filesize is stored 64bit. Inode counting starts with 1. (whilst long
  91filename inodes start with 0)
  96A directory is a filesystem object and has an inode just like a file.
  97It is a specially formatted file containing records which associate each
  98name with an inode number.
  99'.' inode number points to the directory inode
 100'..' inode number points to the parent directory inode
 101Eeach filename record additionally got a filename length field.
 103One special case are long filenames or subdirectory names.
 104These got set a filename length field of 0xff in the corresponding directory
 105record plus the longfile inode number also stored in that record.
 106With that longfilename inode number, the longfilename tree can be walked
 107starting with the superblock longfilename root node pointers.
 109Special files
 112Symbolic links are also filesystem objects with inodes. They got a specific
 113bit in the inode mode field identifying them as symbolic link.
 114The directory entry file inode pointer points to the target file inode.
 116Hard links got an inode, a directory entry, but a specific mode bit set,
 117no block pointers and the directory file record pointing to the target file
 120Character and block special devices do not exist in QNX as those files
 121are handled by the QNX kernel/drivers and created in /dev independent of the
 122underlaying filesystem.
 124Long filenames
 127Long filenames are stored in a separate addressing tree. The staring point
 128is the longfilename root node in the active superblock.
 129Each data block (tree leaves) holds one long filename. That filename is
 130limited to 510 bytes. The first two starting bytes are used as length field
 131for the actual filename.
 132If that structure shall fit for all allowed blocksizes, it is clear why there
 133is a limit of 510 bytes for the actual filename stored.
 138The qnx6fs filesystem allocation bitmap is stored in a tree under bitmap
 139root node in the superblock and each bit in the bitmap represents one
 140filesystem block.
 141The first block is block 0, which starts 0x1000 after superblock start.
 142So for a normal qnx6fs 0x3000 (bootblock + superblock) is the physical
 143address at which block 0 is located.
 145Bits at the end of the last bitmap block are set to 1, if the device is
 146smaller than addressing space in the bitmap.
 148Bitmap system area
 151The bitmap itself is devided into three parts.
 152First the system area, that is split into two halfs.
 153Then userspace.
 155The requirement for a static, fixed preallocated system area comes from how
 156qnx6fs deals with writes.
 157Each superblock got it's own half of the system area. So superblock #1
 158always uses blocks from the lower half whilst superblock #2 just writes to
 159blocks represented by the upper half bitmap system area bits.
 161Bitmap blocks, Inode blocks and indirect addressing blocks for those two
 162tree structures are treated as system blocks.
 164The rational behind that is that a write request can work on a new snapshot
 165(system area of the inactive - resp. lower serial numbered superblock) while
 166at the same time there is still a complete stable filesystem structer in the
 167other half of the system area.
 169When finished with writing (a sync write is completed, the maximum sync leap
 170time or a filesystem sync is requested), serial of the previously inactive
 171superblock atomically is increased and the fs switches over to that - then
 172stable declared - superblock.
 174For all data outside the system area, blocks are just copied while writing.
 175 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.