1        Locking scheme used for directory operations is based on two
   2kinds of locks - per-inode (->i_mutex) and per-filesystem
   5        For our purposes all operations fall in 5 classes:
   71) read access.  Locking rules: caller locks directory we are accessing.
   92) object creation.  Locking rules: same as above.
  113) object removal.  Locking rules: caller locks parent, finds victim,
  12locks victim and calls the method.
  144) rename() that is _not_ cross-directory.  Locking rules: caller locks
  15the parent, finds source and target, if target already exists - locks it
  16and then calls the method.
  185) link creation.  Locking rules:
  19        * lock parent
  20        * check that source is not a directory
  21        * lock source
  22        * call the method.
  246) cross-directory rename.  The trickiest in the whole bunch.  Locking
  26        * lock the filesystem
  27        * lock parents in "ancestors first" order.
  28        * find source and target.
  29        * if old parent is equal to or is a descendent of target
  30                fail with -ENOTEMPTY
  31        * if new parent is equal to or is a descendent of source
  32                fail with -ELOOP
  33        * if target exists - lock it.
  34        * call the method.
  37The rules above obviously guarantee that all directories that are going to be
  38read, modified or removed by method will be locked by caller.
  41If no directory is its own ancestor, the scheme above is deadlock-free.
  44        First of all, at any moment we have a partial ordering of the
  45objects - A < B iff A is an ancestor of B.
  47        That ordering can change.  However, the following is true:
  49(1) if object removal or non-cross-directory rename holds lock on A and
  50    attempts to acquire lock on B, A will remain the parent of B until we
  51    acquire the lock on B.  (Proof: only cross-directory rename can change
  52    the parent of object and it would have to lock the parent).
  54(2) if cross-directory rename holds the lock on filesystem, order will not
  55    change until rename acquires all locks.  (Proof: other cross-directory
  56    renames will be blocked on filesystem lock and we don't start changing
  57    the order until we had acquired all locks).
  59(3) any operation holds at most one lock on non-directory object and
  60    that lock is acquired after all other locks.  (Proof: see descriptions
  61    of operations).
  63        Now consider the minimal deadlock.  Each process is blocked on
  64attempt to acquire some lock and already holds at least one lock.  Let's
  65consider the set of contended locks.  First of all, filesystem lock is
  66not contended, since any process blocked on it is not holding any locks.
  67Thus all processes are blocked on ->i_mutex.
  69        Non-directory objects are not contended due to (3).  Thus link
  70creation can't be a part of deadlock - it can't be blocked on source
  71and it means that it doesn't hold any locks.
  73        Any contended object is either held by cross-directory rename or
  74has a child that is also contended.  Indeed, suppose that it is held by
  75operation other than cross-directory rename.  Then the lock this operation
  76is blocked on belongs to child of that object due to (1).
  78        It means that one of the operations is cross-directory rename.
  79Otherwise the set of contended objects would be infinite - each of them
  80would have a contended child and we had assumed that no object is its
  81own descendent.  Moreover, there is exactly one cross-directory rename
  82(see above).
  84        Consider the object blocking the cross-directory rename.  One
  85of its descendents is locked by cross-directory rename (otherwise we
  86would again have an infinite set of contended objects).  But that
  87means that cross-directory rename is taking locks out of order.  Due
  88to (2) the order hadn't changed since we had acquired filesystem lock.
  89But locking rules for cross-directory rename guarantee that we do not
  90try to acquire lock on descendent before the lock on ancestor.
  91Contradiction.  I.e.  deadlock is impossible.  Q.E.D.
  94        These operations are guaranteed to avoid loop creation.  Indeed,
  95the only operation that could introduce loops is cross-directory rename.
  96Since the only new (parent, child) pair added by rename() is (new parent,
  97source), such loop would have to contain these objects and the rest of it
  98would have to exist before rename().  I.e. at the moment of loop creation
  99rename() responsible for that would be holding filesystem lock and new parent
 100would have to be equal to or a descendent of source.  But that means that
 101new parent had been equal to or a descendent of source since the moment when
 102we had acquired filesystem lock and rename() would fail with -ELOOP in that
 105        While this locking scheme works for arbitrary DAGs, it relies on
 106ability to check that directory is a descendent of another object.  Current
 107implementation assumes that directory graph is a tree.  This assumption is
 108also preserved by all operations (cross-directory rename on a tree that would
 109not introduce a cycle will leave it a tree and link() fails for directories).
 111        Notice that "directory" in the above == "anything that might have
 112children", so if we are going to introduce hybrid objects we will need
 113either to make sure that link(2) doesn't work for them or to make changes
 114in is_subdir() that would make it work even in presence of such beasts.
 115 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.