qnx6.rst (7339B)
1.. SPDX-License-Identifier: GPL-2.0 2 3=================== 4The QNX6 Filesystem 5=================== 6 7The qnx6fs is used by newer QNX operating system versions. (e.g. Neutrino) 8It got introduced in QNX 6.4.0 and is used default since 6.4.1. 9 10Option 11====== 12 13mmi_fs Mount filesystem as used for example by Audi MMI 3G system 14 15Specification 16============= 17 18qnx6fs shares many properties with traditional Unix filesystems. It has the 19concepts of blocks, inodes and directories. 20 21On QNX it is possible to create little endian and big endian qnx6 filesystems. 22This feature makes it possible to create and use a different endianness fs 23for the target (QNX is used on quite a range of embedded systems) platform 24running on a different endianness. 25 26The Linux driver handles endianness transparently. (LE and BE) 27 28Blocks 29------ 30 31The space in the device or file is split up into blocks. These are a fixed 32size of 512, 1024, 2048 or 4096, which is decided when the filesystem is 33created. 34 35Blockpointers are 32bit, so the maximum space that can be addressed is 362^32 * 4096 bytes or 16TB 37 38The superblocks 39--------------- 40 41The superblock contains all global information about the filesystem. 42Each qnx6fs got two superblocks, each one having a 64bit serial number. 43That serial number is used to identify the "active" superblock. 44In write mode with reach new snapshot (after each synchronous write), the 45serial of the new master superblock is increased (old superblock serial + 1) 46 47So basically the snapshot functionality is realized by an atomic final 48update of the serial number. Before updating that serial, all modifications 49are done by copying all modified blocks during that specific write request 50(or period) and building up a new (stable) filesystem structure under the 51inactive superblock. 52 53Each superblock holds a set of root inodes for the different filesystem 54parts. (Inode, Bitmap and Longfilenames) 55Each of these root nodes holds information like total size of the stored 56data and the addressing levels in that specific tree. 57If the level value is 0, up to 16 direct blocks can be addressed by each 58node. 59 60Level 1 adds an additional indirect addressing level where each indirect 61addressing block holds up to blocksize / 4 bytes pointers to data blocks. 62Level 2 adds an additional indirect addressing block level (so, already up 63to 16 * 256 * 256 = 1048576 blocks that can be addressed by such a tree). 64 65Unused block pointers are always set to ~0 - regardless of root node, 66indirect addressing blocks or inodes. 67 68Data leaves are always on the lowest level. So no data is stored on upper 69tree levels. 70 71The first Superblock is located at 0x2000. (0x2000 is the bootblock size) 72The Audi MMI 3G first superblock directly starts at byte 0. 73 74Second superblock position can either be calculated from the superblock 75information (total number of filesystem blocks) or by taking the highest 76device address, zeroing the last 3 bytes and then subtracting 0x1000 from 77that address. 78 790x1000 is the size reserved for each superblock - regardless of the 80blocksize of the filesystem. 81 82Inodes 83------ 84 85Each object in the filesystem is represented by an inode. (index node) 86The inode structure contains pointers to the filesystem blocks which contain 87the data held in the object and all of the metadata about an object except 88its longname. (filenames longer than 27 characters) 89The metadata about an object includes the permissions, owner, group, flags, 90size, number of blocks used, access time, change time and modification time. 91 92Object mode field is POSIX format. (which makes things easier) 93 94There are also pointers to the first 16 blocks, if the object data can be 95addressed with 16 direct blocks. 96 97For more than 16 blocks an indirect addressing in form of another tree is 98used. (scheme is the same as the one used for the superblock root nodes) 99 100The filesize is stored 64bit. Inode counting starts with 1. (while long 101filename inodes start with 0) 102 103Directories 104----------- 105 106A directory is a filesystem object and has an inode just like a file. 107It is a specially formatted file containing records which associate each 108name with an inode number. 109 110'.' inode number points to the directory inode 111 112'..' inode number points to the parent directory inode 113 114Eeach filename record additionally got a filename length field. 115 116One special case are long filenames or subdirectory names. 117 118These got set a filename length field of 0xff in the corresponding directory 119record plus the longfile inode number also stored in that record. 120 121With that longfilename inode number, the longfilename tree can be walked 122starting with the superblock longfilename root node pointers. 123 124Special files 125------------- 126 127Symbolic links are also filesystem objects with inodes. They got a specific 128bit in the inode mode field identifying them as symbolic link. 129 130The directory entry file inode pointer points to the target file inode. 131 132Hard links got an inode, a directory entry, but a specific mode bit set, 133no block pointers and the directory file record pointing to the target file 134inode. 135 136Character and block special devices do not exist in QNX as those files 137are handled by the QNX kernel/drivers and created in /dev independent of the 138underlaying filesystem. 139 140Long filenames 141-------------- 142 143Long filenames are stored in a separate addressing tree. The staring point 144is the longfilename root node in the active superblock. 145 146Each data block (tree leaves) holds one long filename. That filename is 147limited to 510 bytes. The first two starting bytes are used as length field 148for the actual filename. 149 150If that structure shall fit for all allowed blocksizes, it is clear why there 151is a limit of 510 bytes for the actual filename stored. 152 153Bitmap 154------ 155 156The qnx6fs filesystem allocation bitmap is stored in a tree under bitmap 157root node in the superblock and each bit in the bitmap represents one 158filesystem block. 159 160The first block is block 0, which starts 0x1000 after superblock start. 161So for a normal qnx6fs 0x3000 (bootblock + superblock) is the physical 162address at which block 0 is located. 163 164Bits at the end of the last bitmap block are set to 1, if the device is 165smaller than addressing space in the bitmap. 166 167Bitmap system area 168------------------ 169 170The bitmap itself is divided into three parts. 171 172First the system area, that is split into two halves. 173 174Then userspace. 175 176The requirement for a static, fixed preallocated system area comes from how 177qnx6fs deals with writes. 178 179Each superblock got it's own half of the system area. So superblock #1 180always uses blocks from the lower half while superblock #2 just writes to 181blocks represented by the upper half bitmap system area bits. 182 183Bitmap blocks, Inode blocks and indirect addressing blocks for those two 184tree structures are treated as system blocks. 185 186The rational behind that is that a write request can work on a new snapshot 187(system area of the inactive - resp. lower serial numbered superblock) while 188at the same time there is still a complete stable filesystem structure in the 189other half of the system area. 190 191When finished with writing (a sync write is completed, the maximum sync leap 192time or a filesystem sync is requested), serial of the previously inactive 193superblock atomically is increased and the fs switches over to that - then 194stable declared - superblock. 195 196For all data outside the system area, blocks are just copied while writing.