linux/Documentation/filesystems/porting
<<
>>
Prefs
   1Changes since 2.5.0:
   2
   3---
   4[recommended]
   5
   6New helpers: sb_bread(), sb_getblk(), sb_find_get_block(), set_bh(),
   7        sb_set_blocksize() and sb_min_blocksize().
   8
   9Use them.
  10
  11(sb_find_get_block() replaces 2.4's get_hash_table())
  12
  13---
  14[recommended]
  15
  16New methods: ->alloc_inode() and ->destroy_inode().
  17
  18Remove inode->u.foo_inode_i
  19Declare
  20        struct foo_inode_info {
  21                /* fs-private stuff */
  22                struct inode vfs_inode;
  23        };
  24        static inline struct foo_inode_info *FOO_I(struct inode *inode)
  25        {
  26                return list_entry(inode, struct foo_inode_info, vfs_inode);
  27        }
  28
  29Use FOO_I(inode) instead of &inode->u.foo_inode_i;
  30
  31Add foo_alloc_inode() and foo_destroy_inode() - the former should allocate
  32foo_inode_info and return the address of ->vfs_inode, the latter should free
  33FOO_I(inode) (see in-tree filesystems for examples).
  34
  35Make them ->alloc_inode and ->destroy_inode in your super_operations.
  36
  37Keep in mind that now you need explicit initialization of private data
  38typically between calling iget_locked() and unlocking the inode.
  39
  40At some point that will become mandatory.
  41
  42---
  43[mandatory]
  44
  45Change of file_system_type method (->read_super to ->get_sb)
  46
  47->read_super() is no more.  Ditto for DECLARE_FSTYPE and DECLARE_FSTYPE_DEV.
  48
  49Turn your foo_read_super() into a function that would return 0 in case of
  50success and negative number in case of error (-EINVAL unless you have more
  51informative error value to report).  Call it foo_fill_super().  Now declare
  52
  53int foo_get_sb(struct file_system_type *fs_type,
  54        int flags, const char *dev_name, void *data, struct vfsmount *mnt)
  55{
  56        return get_sb_bdev(fs_type, flags, dev_name, data, foo_fill_super,
  57                           mnt);
  58}
  59
  60(or similar with s/bdev/nodev/ or s/bdev/single/, depending on the kind of
  61filesystem).
  62
  63Replace DECLARE_FSTYPE... with explicit initializer and have ->get_sb set as
  64foo_get_sb.
  65
  66---
  67[mandatory]
  68
  69Locking change: ->s_vfs_rename_sem is taken only by cross-directory renames.
  70Most likely there is no need to change anything, but if you relied on
  71global exclusion between renames for some internal purpose - you need to
  72change your internal locking.  Otherwise exclusion warranties remain the
  73same (i.e. parents and victim are locked, etc.).
  74
  75---
  76[informational]
  77
  78Now we have the exclusion between ->lookup() and directory removal (by
  79->rmdir() and ->rename()).  If you used to need that exclusion and do
  80it by internal locking (most of filesystems couldn't care less) - you
  81can relax your locking.
  82
  83---
  84[mandatory]
  85
  86->lookup(), ->truncate(), ->create(), ->unlink(), ->mknod(), ->mkdir(),
  87->rmdir(), ->link(), ->lseek(), ->symlink(), ->rename()
  88and ->readdir() are called without BKL now.  Grab it on entry, drop upon return
  89- that will guarantee the same locking you used to have.  If your method or its
  90parts do not need BKL - better yet, now you can shift lock_kernel() and
  91unlock_kernel() so that they would protect exactly what needs to be
  92protected.
  93
  94---
  95[mandatory]
  96
  97BKL is also moved from around sb operations. BKL should have been shifted into
  98individual fs sb_op functions.  If you don't need it, remove it.
  99
 100---
 101[informational]
 102
 103check for ->link() target not being a directory is done by callers.  Feel
 104free to drop it...
 105
 106---
 107[informational]
 108
 109->link() callers hold ->i_mutex on the object we are linking to.  Some of your
 110problems might be over...
 111
 112---
 113[mandatory]
 114
 115new file_system_type method - kill_sb(superblock).  If you are converting
 116an existing filesystem, set it according to ->fs_flags:
 117        FS_REQUIRES_DEV         -       kill_block_super
 118        FS_LITTER               -       kill_litter_super
 119        neither                 -       kill_anon_super
 120FS_LITTER is gone - just remove it from fs_flags.
 121
 122---
 123[mandatory]
 124
 125        FS_SINGLE is gone (actually, that had happened back when ->get_sb()
 126went in - and hadn't been documented ;-/).  Just remove it from fs_flags
 127(and see ->get_sb() entry for other actions).
 128
 129---
 130[mandatory]
 131
 132->setattr() is called without BKL now.  Caller _always_ holds ->i_mutex, so
 133watch for ->i_mutex-grabbing code that might be used by your ->setattr().
 134Callers of notify_change() need ->i_mutex now.
 135
 136---
 137[recommended]
 138
 139New super_block field "struct export_operations *s_export_op" for
 140explicit support for exporting, e.g. via NFS.  The structure is fully
 141documented at its declaration in include/linux/fs.h, and in
 142Documentation/filesystems/nfs/Exporting.
 143
 144Briefly it allows for the definition of decode_fh and encode_fh operations
 145to encode and decode filehandles, and allows the filesystem to use
 146a standard helper function for decode_fh, and provide file-system specific
 147support for this helper, particularly get_parent.
 148
 149It is planned that this will be required for exporting once the code
 150settles down a bit.
 151
 152[mandatory]
 153
 154s_export_op is now required for exporting a filesystem.
 155isofs, ext2, ext3, resierfs, fat
 156can be used as examples of very different filesystems.
 157
 158---
 159[mandatory]
 160
 161iget4() and the read_inode2 callback have been superseded by iget5_locked()
 162which has the following prototype,
 163
 164    struct inode *iget5_locked(struct super_block *sb, unsigned long ino,
 165                                int (*test)(struct inode *, void *),
 166                                int (*set)(struct inode *, void *),
 167                                void *data);
 168
 169'test' is an additional function that can be used when the inode
 170number is not sufficient to identify the actual file object. 'set'
 171should be a non-blocking function that initializes those parts of a
 172newly created inode to allow the test function to succeed. 'data' is
 173passed as an opaque value to both test and set functions.
 174
 175When the inode has been created by iget5_locked(), it will be returned with the
 176I_NEW flag set and will still be locked.  The filesystem then needs to finalize
 177the initialization. Once the inode is initialized it must be unlocked by
 178calling unlock_new_inode().
 179
 180The filesystem is responsible for setting (and possibly testing) i_ino
 181when appropriate. There is also a simpler iget_locked function that
 182just takes the superblock and inode number as arguments and does the
 183test and set for you.
 184
 185e.g.
 186        inode = iget_locked(sb, ino);
 187        if (inode->i_state & I_NEW) {
 188                err = read_inode_from_disk(inode);
 189                if (err < 0) {
 190                        iget_failed(inode);
 191                        return err;
 192                }
 193                unlock_new_inode(inode);
 194        }
 195
 196Note that if the process of setting up a new inode fails, then iget_failed()
 197should be called on the inode to render it dead, and an appropriate error
 198should be passed back to the caller.
 199
 200---
 201[recommended]
 202
 203->getattr() finally getting used.  See instances in nfs, minix, etc.
 204
 205---
 206[mandatory]
 207
 208->revalidate() is gone.  If your filesystem had it - provide ->getattr()
 209and let it call whatever you had as ->revlidate() + (for symlinks that
 210had ->revalidate()) add calls in ->follow_link()/->readlink().
 211
 212---
 213[mandatory]
 214
 215->d_parent changes are not protected by BKL anymore.  Read access is safe
 216if at least one of the following is true:
 217        * filesystem has no cross-directory rename()
 218        * we know that parent had been locked (e.g. we are looking at
 219->d_parent of ->lookup() argument).
 220        * we are called from ->rename().
 221        * the child's ->d_lock is held
 222Audit your code and add locking if needed.  Notice that any place that is
 223not protected by the conditions above is risky even in the old tree - you
 224had been relying on BKL and that's prone to screwups.  Old tree had quite
 225a few holes of that kind - unprotected access to ->d_parent leading to
 226anything from oops to silent memory corruption.
 227
 228---
 229[mandatory]
 230
 231        FS_NOMOUNT is gone.  If you use it - just set MS_NOUSER in flags
 232(see rootfs for one kind of solution and bdev/socket/pipe for another).
 233
 234---
 235[recommended]
 236
 237        Use bdev_read_only(bdev) instead of is_read_only(kdev).  The latter
 238is still alive, but only because of the mess in drivers/s390/block/dasd.c.
 239As soon as it gets fixed is_read_only() will die.
 240
 241---
 242[mandatory]
 243
 244->permission() is called without BKL now. Grab it on entry, drop upon
 245return - that will guarantee the same locking you used to have.  If
 246your method or its parts do not need BKL - better yet, now you can
 247shift lock_kernel() and unlock_kernel() so that they would protect
 248exactly what needs to be protected.
 249
 250---
 251[mandatory]
 252
 253->statfs() is now called without BKL held.  BKL should have been
 254shifted into individual fs sb_op functions where it's not clear that
 255it's safe to remove it.  If you don't need it, remove it.
 256
 257---
 258[mandatory]
 259
 260        is_read_only() is gone; use bdev_read_only() instead.
 261
 262---
 263[mandatory]
 264
 265        destroy_buffers() is gone; use invalidate_bdev().
 266
 267---
 268[mandatory]
 269
 270        fsync_dev() is gone; use fsync_bdev().  NOTE: lvm breakage is
 271deliberate; as soon as struct block_device * is propagated in a reasonable
 272way by that code fixing will become trivial; until then nothing can be
 273done.
 274
 275[mandatory]
 276
 277        block truncatation on error exit from ->write_begin, and ->direct_IO
 278moved from generic methods (block_write_begin, cont_write_begin,
 279nobh_write_begin, blockdev_direct_IO*) to callers.  Take a look at
 280ext2_write_failed and callers for an example.
 281
 282[mandatory]
 283
 284        ->truncate is gone.  The whole truncate sequence needs to be
 285implemented in ->setattr, which is now mandatory for filesystems
 286implementing on-disk size changes.  Start with a copy of the old inode_setattr
 287and vmtruncate, and the reorder the vmtruncate + foofs_vmtruncate sequence to
 288be in order of zeroing blocks using block_truncate_page or similar helpers,
 289size update and on finally on-disk truncation which should not fail.
 290inode_change_ok now includes the size checks for ATTR_SIZE and must be called
 291in the beginning of ->setattr unconditionally.
 292
 293[mandatory]
 294
 295        ->clear_inode() and ->delete_inode() are gone; ->evict_inode() should
 296be used instead.  It gets called whenever the inode is evicted, whether it has
 297remaining links or not.  Caller does *not* evict the pagecache or inode-associated
 298metadata buffers; getting rid of those is responsibility of method, as it had
 299been for ->delete_inode(). Caller makes sure async writeback cannot be running
 300for the inode while (or after) ->evict_inode() is called.
 301
 302        ->drop_inode() returns int now; it's called on final iput() with
 303inode->i_lock held and it returns true if filesystems wants the inode to be
 304dropped.  As before, generic_drop_inode() is still the default and it's been
 305updated appropriately.  generic_delete_inode() is also alive and it consists
 306simply of return 1.  Note that all actual eviction work is done by caller after
 307->drop_inode() returns.
 308
 309        As before, clear_inode() must be called exactly once on each call of
 310->evict_inode() (as it used to be for each call of ->delete_inode()).  Unlike
 311before, if you are using inode-associated metadata buffers (i.e.
 312mark_buffer_dirty_inode()), it's your responsibility to call
 313invalidate_inode_buffers() before clear_inode().
 314
 315        NOTE: checking i_nlink in the beginning of ->write_inode() and bailing out
 316if it's zero is not *and* *never* *had* *been* enough.  Final unlink() and iput()
 317may happen while the inode is in the middle of ->write_inode(); e.g. if you blindly
 318free the on-disk inode, you may end up doing that while ->write_inode() is writing
 319to it.
 320
 321---
 322[mandatory]
 323
 324        .d_delete() now only advises the dcache as to whether or not to cache
 325unreferenced dentries, and is now only called when the dentry refcount goes to
 3260. Even on 0 refcount transition, it must be able to tolerate being called 0,
 3271, or more times (eg. constant, idempotent).
 328
 329---
 330[mandatory]
 331
 332        .d_compare() calling convention and locking rules are significantly
 333changed. Read updated documentation in Documentation/filesystems/vfs.txt (and
 334look at examples of other filesystems) for guidance.
 335
 336---
 337[mandatory]
 338
 339        .d_hash() calling convention and locking rules are significantly
 340changed. Read updated documentation in Documentation/filesystems/vfs.txt (and
 341look at examples of other filesystems) for guidance.
 342
 343---
 344[mandatory]
 345        dcache_lock is gone, replaced by fine grained locks. See fs/dcache.c
 346for details of what locks to replace dcache_lock with in order to protect
 347particular things. Most of the time, a filesystem only needs ->d_lock, which
 348protects *all* the dcache state of a given dentry.
 349
 350--
 351[mandatory]
 352
 353        Filesystems must RCU-free their inodes, if they can have been accessed
 354via rcu-walk path walk (basically, if the file can have had a path name in the
 355vfs namespace).
 356
 357        Even though i_dentry and i_rcu share storage in a union, we will
 358initialize the former in inode_init_always(), so just leave it alone in
 359the callback.  It used to be necessary to clean it there, but not anymore
 360(starting at 3.2).
 361
 362--
 363[recommended]
 364        vfs now tries to do path walking in "rcu-walk mode", which avoids
 365atomic operations and scalability hazards on dentries and inodes (see
 366Documentation/filesystems/path-lookup.txt). d_hash and d_compare changes
 367(above) are examples of the changes required to support this. For more complex
 368filesystem callbacks, the vfs drops out of rcu-walk mode before the fs call, so
 369no changes are required to the filesystem. However, this is costly and loses
 370the benefits of rcu-walk mode. We will begin to add filesystem callbacks that
 371are rcu-walk aware, shown below. Filesystems should take advantage of this
 372where possible.
 373
 374--
 375[mandatory]
 376        d_revalidate is a callback that is made on every path element (if
 377the filesystem provides it), which requires dropping out of rcu-walk mode. This
 378may now be called in rcu-walk mode (nd->flags & LOOKUP_RCU). -ECHILD should be
 379returned if the filesystem cannot handle rcu-walk. See
 380Documentation/filesystems/vfs.txt for more details.
 381
 382        permission and check_acl are inode permission checks that are called
 383on many or all directory inodes on the way down a path walk (to check for
 384exec permission). These must now be rcu-walk aware (flags & IPERM_FLAG_RCU).
 385See Documentation/filesystems/vfs.txt for more details.
 386 
 387--
 388[mandatory]
 389        In ->fallocate() you must check the mode option passed in.  If your
 390filesystem does not support hole punching (deallocating space in the middle of a
 391file) you must return -EOPNOTSUPP if FALLOC_FL_PUNCH_HOLE is set in mode.
 392Currently you can only have FALLOC_FL_PUNCH_HOLE with FALLOC_FL_KEEP_SIZE set,
 393so the i_size should not change when hole punching, even when puching the end of
 394a file off.
 395
 396--
 397[mandatory]
 398        ->get_sb() is gone.  Switch to use of ->mount().  Typically it's just
 399a matter of switching from calling get_sb_... to mount_... and changing the
 400function type.  If you were doing it manually, just switch from setting ->mnt_root
 401to some pointer to returning that pointer.  On errors return ERR_PTR(...).
 402
 403--
 404[mandatory]
 405        ->permission() and generic_permission()have lost flags
 406argument; instead of passing IPERM_FLAG_RCU we add MAY_NOT_BLOCK into mask.
 407        generic_permission() has also lost the check_acl argument; ACL checking
 408has been taken to VFS and filesystems need to provide a non-NULL ->i_op->get_acl
 409to read an ACL from disk.
 410
 411--
 412[mandatory]
 413        If you implement your own ->llseek() you must handle SEEK_HOLE and
 414SEEK_DATA.  You can hanle this by returning -EINVAL, but it would be nicer to
 415support it in some way.  The generic handler assumes that the entire file is
 416data and there is a virtual hole at the end of the file.  So if the provided
 417offset is less than i_size and SEEK_DATA is specified, return the same offset.
 418If the above is true for the offset and you are given SEEK_HOLE, return the end
 419of the file.  If the offset is i_size or greater return -ENXIO in either case.
 420
 421[mandatory]
 422        If you have your own ->fsync() you must make sure to call
 423filemap_write_and_wait_range() so that all dirty pages are synced out properly.
 424You must also keep in mind that ->fsync() is not called with i_mutex held
 425anymore, so if you require i_mutex locking you must make sure to take it and
 426release it yourself.
 427
 428--
 429[mandatory]
 430        d_alloc_root() is gone, along with a lot of bugs caused by code
 431misusing it.  Replacement: d_make_root(inode).  The difference is,
 432d_make_root() drops the reference to inode if dentry allocation fails.  
 433
 434--
 435[mandatory]
 436        The witch is dead!  Well, 2/3 of it, anyway.  ->d_revalidate() and
 437->lookup() do *not* take struct nameidata anymore; just the flags.
 438--
 439[mandatory]
 440        ->create() doesn't take struct nameidata *; unlike the previous
 441two, it gets "is it an O_EXCL or equivalent?" boolean argument.  Note that
 442local filesystems can ignore tha argument - they are guaranteed that the
 443object doesn't exist.  It's remote/distributed ones that might care...
 444--
 445[mandatory]
 446        FS_REVAL_DOT is gone; if you used to have it, add ->d_weak_revalidate()
 447in your dentry operations instead.
 448--
 449[mandatory]
 450        vfs_readdir() is gone; switch to iterate_dir() instead
 451--
 452[mandatory]
 453        ->readdir() is gone now; switch to ->iterate()
 454
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.