releases/2.6.29.5/jbd2-update-locking-coments.patch

   1 From stable-bounces@linux.kernel.org  Tue Jun  9 02:28:16 2009
   2 From: Jan Kara <jack@suse.cz>
   3 Date: Tue,  2 Jun 2009 08:07:52 -0400
   4 Subject: jbd2: Update locking coments
   5 To: stable@kernel.org
   6 Cc: "Theodore Ts'o" <tytso@mit.edu>, linux-ext4@vger.kernel.org, Jan Kara <jack@suse.cz>
   7 Message-ID: <1243944479-20574-11-git-send-email-tytso@mit.edu>
   8
   9
  10 From: Jan Kara <jack@suse.cz>
  11
  12 (cherry picked from commit 86db97c87f744364d5889ca8a4134ca2048b8f83)
  13
  14 Update information about locking in JBD2 revoke code. Inconsistency in
  15 comments found by Lin Tan <tammy000@gmail.com>
  16
  17 CC: Lin Tan <tammy000@gmail.com>
  18 Signed-off-by: Jan Kara <jack@suse.cz>
  19 Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
  20 Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
  21 ---
  22  fs/jbd2/revoke.c |   24 +++++++++++++++++++-----
  23  1 file changed, 19 insertions(+), 5 deletions(-)
  24
  25 --- a/fs/jbd2/revoke.c
  26 +++ b/fs/jbd2/revoke.c
  27 @@ -55,6 +55,25 @@
  28   *                     need do nothing.
  29   * RevokeValid set, Revoked set:
  30   *                     buffer has been revoked.
  31 + *
  32 + * Locking rules:
  33 + * We keep two hash tables of revoke records. One hashtable belongs to the
  34 + * running transaction (is pointed to by journal->j_revoke), the other one
  35 + * belongs to the committing transaction. Accesses to the second hash table
  36 + * happen only from the kjournald and no other thread touches this table.  Also
  37 + * journal_switch_revoke_table() which switches which hashtable belongs to the
  38 + * running and which to the committing transaction is called only from
  39 + * kjournald. Therefore we need no locks when accessing the hashtable belonging
  40 + * to the committing transaction.
  41 + *
  42 + * All users operating on the hash table belonging to the running transaction
  43 + * have a handle to the transaction. Therefore they are safe from kjournald
  44 + * switching hash tables under them. For operations on the lists of entries in
  45 + * the hash table j_revoke_lock is used.
  46 + *
  47 + * Finally, also replay code uses the hash tables but at this moment noone else
  48 + * can touch them (filesystem isn't mounted yet) and hence no locking is
  49 + * needed.
  50   */
  51
  52  #ifndef __KERNEL__
  53 @@ -401,8 +420,6 @@ int jbd2_journal_revoke(handle_t *handle
  54   * the second time we would still have a pending revoke to cancel.  So,
  55   * do not trust the Revoked bit on buffers unless RevokeValid is also
  56   * set.
  57 - *
  58 - * The caller must have the journal locked.
  59   */
  60  int jbd2_journal_cancel_revoke(handle_t *handle, struct journal_head *jh)
  61  {
  62 @@ -480,10 +497,7 @@ void jbd2_journal_switch_revoke_table(jo
  63  /*
  64   * Write revoke records to the journal for all entries in the current
  65   * revoke hash, deleting the entries as we go.
  66 - *
  67 - * Called with the journal lock held.
  68   */
  69 -
  70  void jbd2_journal_write_revoke_records(journal_t *journal,
  71                                   transaction_t *transaction)
  72  {