TEXT   34


Guest on 23rd November 2022 01:55:44 AM

  1. ============
  2. Fiemap Ioctl
  3. ============
  5. The fiemap ioctl is an efficient method for userspace to get file
  6. extent mappings. Instead of block-by-block mapping (such as bmap), fiemap
  7. returns a list of extents.
  10. Request Basics
  11. --------------
  13. A fiemap request is encoded within struct fiemap:
  15. struct fiemap {
  16.         __u64   fm_start;        /* logical offset (inclusive) at
  17.                                   * which to start mapping (in) */
  18.         __u64   fm_length;       /* logical length of mapping which
  19.                                   * userspace cares about (in) */
  20.         __u32   fm_flags;        /* FIEMAP_FLAG_* flags for request (in/out) */
  21.         __u32   fm_mapped_extents; /* number of extents that were
  22.                                     * mapped (out) */
  23.         __u32   fm_extent_count; /* size of fm_extents array (in) */
  24.         __u32   fm_reserved;
  25.         struct fiemap_extent fm_extents[0]; /* array of mapped extents (out) */
  26. };
  29. fm_start, and fm_length specify the logical range within the file
  30. which the process would like mappings for. Extents returned mirror
  31. those on disk - that is, the logical offset of the 1st returned extent
  32. may start before fm_start, and the range covered by the last returned
  33. extent may end after fm_length. All offsets and lengths are in bytes.
  35. Certain flags to modify the way in which mappings are looked up can be
  36. set in fm_flags. If the kernel doesn't understand some particular
  37. flags, it will return EBADR and the contents of fm_flags will contain
  38. the set of flags which caused the error. If the kernel is compatible
  39. with all flags passed, the contents of fm_flags will be unmodified.
  40. It is up to userspace to determine whether rejection of a particular
  41. flag is fatal to its operation. This scheme is intended to allow the
  42. fiemap interface to grow in the future but without losing
  43. compatibility with old software.
  45. fm_extent_count specifies the number of elements in the fm_extents[] array
  46. that can be used to return extents.  If fm_extent_count is zero, then the
  47. fm_extents[] array is ignored (no extents will be returned), and the
  48. fm_mapped_extents count will hold the number of extents needed in
  49. fm_extents[] to hold the file's current mapping.  Note that there is
  50. nothing to prevent the file from changing between calls to FIEMAP.
  52. The following flags can be set in fm_flags:
  55. If this flag is set, the kernel will sync the file before mapping extents.
  58. If this flag is set, the extents returned will describe the inodes
  59. extended attribute lookup tree, instead of its data tree.
  62. Extent Mapping
  63. --------------
  65. Extent information is returned within the embedded fm_extents array
  66. which userspace must allocate along with the fiemap structure. The
  67. number of elements in the fiemap_extents[] array should be passed via
  68. fm_extent_count. The number of extents mapped by kernel will be
  69. returned via fm_mapped_extents. If the number of fiemap_extents
  70. allocated is less than would be required to map the requested range,
  71. the maximum number of extents that can be mapped in the fm_extent[]
  72. array will be returned and fm_mapped_extents will be equal to
  73. fm_extent_count. In that case, the last extent in the array will not
  74. complete the requested range and will not have the FIEMAP_EXTENT_LAST
  75. flag set (see the next section on extent flags).
  77. Each extent is described by a single fiemap_extent structure as
  78. returned in fm_extents.
  80. struct fiemap_extent {
  81.         __u64   fe_logical;  /* logical offset in bytes for the start of
  82.                               * the extent */
  83.         __u64   fe_physical; /* physical offset in bytes for the start
  84.                               * of the extent */
  85.         __u64   fe_length;   /* length in bytes for the extent */
  86.         __u64   fe_reserved64[2];
  87.         __u32   fe_flags;    /* FIEMAP_EXTENT_* flags for this extent */
  88.         __u32   fe_reserved[3];
  89. };
  91. All offsets and lengths are in bytes and mirror those on disk.  It is valid
  92. for an extents logical offset to start before the request or its logical
  93. length to extend past the request.  Unless FIEMAP_EXTENT_NOT_ALIGNED is
  94. returned, fe_logical, fe_physical, and fe_length will be aligned to the
  95. block size of the file system.  With the exception of extents flagged as
  96. FIEMAP_EXTENT_MERGED, adjacent extents will not be merged.
  98. The fe_flags field contains flags which describe the extent returned.
  99. A special flag, FIEMAP_EXTENT_LAST is always set on the last extent in
  100. the file so that the process making fiemap calls can determine when no
  101. more extents are available, without having to call the ioctl again.
  103. Some flags are intentionally vague and will always be set in the
  104. presence of other more specific flags. This way a program looking for
  105. a general property does not have to know all existing and future flags
  106. which imply that property.
  109. are set, FIEMAP_EXTENT_NOT_ALIGNED will also be set. A program looking
  110. for inline or tail-packed data can key on the specific flag. Software
  111. which simply cares not to try operating on non-aligned extents
  112. however, can just key on FIEMAP_EXTENT_NOT_ALIGNED, and not have to
  113. worry about all present and future flags which might imply unaligned
  114. data. Note that the opposite is not true - it would be valid for
  115. FIEMAP_EXTENT_NOT_ALIGNED to appear alone.
  118. This is the last extent in the file. A mapping attempt past this
  119. extent will return nothing.
  122. The location of this extent is currently unknown. This may indicate
  123. the data is stored on an inaccessible volume or that no storage has
  124. been allocated for the file yet.
  127.   - This will also set FIEMAP_EXTENT_UNKNOWN.
  128. Delayed allocation - while there is data for this extent, its
  129. physical location has not been allocated yet.
  132. This extent does not consist of plain filesystem blocks but is
  133. encoded (e.g. encrypted or compressed).  Reading the data in this
  134. extent via I/O to the block device will have undefined results.
  136. Note that it is *always* undefined to try to update the data
  137. in-place by writing to the indicated location without the
  138. assistance of the filesystem, or to access the data using the
  139. information returned by the FIEMAP interface while the filesystem
  140. is mounted.  In other words, user applications may only read the
  141. extent data via I/O to the block device while the filesystem is
  142. unmounted, and then only if the FIEMAP_EXTENT_ENCODED flag is
  143. clear; user applications must not try reading or writing to the
  144. filesystem via the block device under any other circumstances.
  147.   - This will also set FIEMAP_EXTENT_ENCODED
  148. The data in this extent has been encrypted by the file system.
  151. Extent offsets and length are not guaranteed to be block aligned.
  154.   This will also set FIEMAP_EXTENT_NOT_ALIGNED
  155. Data is located within a meta data block.
  158.   This will also set FIEMAP_EXTENT_NOT_ALIGNED
  159. Data is packed into a block with data from other files.
  162. Unwritten extent - the extent is allocated but its data has not been
  163. initialized.  This indicates the extent's data will be all zero if read
  164. through the filesystem but the contents are undefined if read directly from
  165. the device.
  168. This will be set when a file does not support extents, i.e., it uses a block
  169. based addressing scheme.  Since returning an extent for each block back to
  170. userspace would be highly inefficient, the kernel will try to merge most
  171. adjacent blocks into 'extents'.
  174. VFS -> File System Implementation
  175. ---------------------------------
  177. File systems wishing to support fiemap must implement a ->fiemap callback on
  178. their inode_operations structure. The fs ->fiemap call is responsible for
  179. defining its set of supported fiemap flags, and calling a helper function on
  180. each discovered extent:
  182. struct inode_operations {
  183.        ...
  185.        int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
  186.                      u64 len);
  188. ->fiemap is passed struct fiemap_extent_info which describes the
  189. fiemap request:
  191. struct fiemap_extent_info {
  192.         unsigned int fi_flags;          /* Flags as passed from user */
  193.         unsigned int fi_extents_mapped; /* Number of mapped extents */
  194.         unsigned int fi_extents_max;    /* Size of fiemap_extent array */
  195.         struct fiemap_extent *fi_extents_start; /* Start of fiemap_extent array */
  196. };
  198. It is intended that the file system should not need to access any of this
  199. structure directly. Filesystem handlers should be tolerant to signals and return
  200. EINTR once fatal signal received.
  203. Flag checking should be done at the beginning of the ->fiemap callback via the
  204. fiemap_check_flags() helper:
  206. int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
  208. The struct fieinfo should be passed in as received from ioctl_fiemap(). The
  209. set of fiemap flags which the fs understands should be passed via fs_flags. If
  210. fiemap_check_flags finds invalid user flags, it will place the bad values in
  211. fieinfo->fi_flags and return -EBADR. If the file system gets -EBADR, from
  212. fiemap_check_flags(), it should immediately exit, returning that error back to
  213. ioctl_fiemap().
  216. For each extent in the request range, the file system should call
  217. the helper function, fiemap_fill_next_extent():
  219. int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
  220.                             u64 phys, u64 len, u32 flags, u32 dev);
  222. fiemap_fill_next_extent() will use the passed values to populate the
  223. next free extent in the fm_extents array. 'General' extent flags will
  224. automatically be set from specific flags on behalf of the calling file
  225. system so that the userspace API is not broken.
  227. fiemap_fill_next_extent() returns 0 on success, and 1 when the
  228. user-supplied fm_extents array is full. If an error is encountered
  229. while copying the extent to user memory, -EFAULT will be returned.

Raw Paste

Login or Register to edit or fork this paste. It's free.