Add callbacks from the JBD (journal) to allow async notification of when

a handle has been committed to disk.

diff --git a/lustre/patches/patch-2.4.18 b/lustre/patches/patch-2.4.18
index c604d12..93e371a 100644 (file)
--- a/lustre/patches/patch-2.4.18
+++ b/lustre/patches/patch-2.4.18
@@ -126,3 +126,209 @@
        if (IDE_CONTROL_REG)
                OUT_BYTE(drive->ctl,IDE_CONTROL_REG);
        OUT_BYTE(0x00, IDE_FEATURE_REG);
+--- lum/include/linux/jbd.h.orig       Tue Apr 16 14:27:25 2002
++++ lum/include/linux/jbd.h    Wed Apr 24 05:19:00 2002
+@@ -249,6 +249,12 @@
+       return bh->b_private;
+ }
+ 
++struct journal_callback {
++      struct list_head jcb_list;
++      void (*jcb_func)(void *);
++      /* user data goes here */
++};
++
+ struct jbd_revoke_table_s;
+ 
+ /* The handle_t type represents a single atomic update being performed
+@@ -279,6 +285,12 @@
+          operations */
+       int                     h_err;
+ 
++      /* List of application registered callbacks for this handle.
++       * The function(s) will be called after the transaction that
++       * this handle is part of has been committed to disk.
++       */
++      struct list_head        h_jcb;
++
+       /* Flags */
+       unsigned int    h_sync:         1;      /* sync-on-close */
+       unsigned int    h_jdata:        1;      /* force data journaling */
+@@ -398,6 +410,10 @@
+ 
+       /* How many handles used this transaction? */
+       int t_handle_count;
++
++      /* List of registered callback functions for this transaction.
++       * Called when the transaction is committed. */
++      struct list_head        t_jcb;
+ };
+ 
+ 
+@@ -547,7 +563,7 @@
+ #define JFS_FLUSHED   0x008   /* The journal superblock has been flushed */
+ #define JFS_LOADED    0x010   /* The journal superblock has been loaded */
+ 
+-/* 
++/*
+  * Function declarations for the journaling transaction and buffer
+  * management
+  */
+@@ -646,6 +662,8 @@
+ extern int     journal_try_to_free_buffers(journal_t *, struct page *, int);
+ extern int     journal_stop(handle_t *);
+ extern int     journal_flush (journal_t *);
++extern void    journal_callback_set(handle_t *handle, void (*func)(void *),
++                                    void *cb_data);
+ 
+ extern void    journal_lock_updates (journal_t *);
+ extern void    journal_unlock_updates (journal_t *);
+--- lum/fs/jbd/commit.c.orig   Fri Apr 12 10:27:52 2002
++++ lum/fs/jbd/commit.c        Wed Apr 24 05:18:07 2002
+@@ -475,7 +475,7 @@
+            transaction's t_log_list queue, and metadata buffers are on
+            the t_iobuf_list queue.
+ 
+-         Wait for the transactions in reverse order.  That way we are
++         Wait for the buffers in reverse order.  That way we are
+          less likely to be woken up until all IOs have completed, and
+          so we incur less scheduling load.
+       */
+@@ -602,6 +602,21 @@
+       }
+       lock_journal(journal);
+ 
++      /* Call any callbacks that had been registered for handles in this
++       * transaction.  It is up to the callback to free any allocated
++       * memory.
++       */
++      {
++              struct list_head *p, *n;
++              list_for_each_safe(p, n, &commit_transaction->t_jcb) {
++                      struct journal_callback *jcb;
++
++                      jcb = list_entry(p, struct journal_callback, jcb_list);
++                      list_del(p);
++                      jcb->jcb_func(jcb);
++              }
++      }
++
+       /* End of a transaction!  Finally, we can do checkpoint
+            processing: any buffers committed as a result of this
+            transaction can be removed from any checkpoint list it was on
+--- lum/fs/jbd/journal.c.orig  Fri Apr 12 10:27:52 2002
++++ lum/fs/jbd/journal.c       Wed Apr 24 05:18:49 2002
+@@ -58,6 +58,7 @@
+ #endif
+ EXPORT_SYMBOL(journal_flush);
+ EXPORT_SYMBOL(journal_revoke);
++EXPORT_SYMBOL(journal_callback_set);
+ 
+ EXPORT_SYMBOL(journal_init_dev);
+ EXPORT_SYMBOL(journal_init_inode);
+--- lum/fs/jbd/transaction.c.orig      Fri Apr 12 10:27:52 2002
++++ lum/fs/jbd/transaction.c   Wed Apr 24 05:31:14 2002
+@@ -57,6 +57,7 @@
+       transaction->t_state = T_RUNNING;
+       transaction->t_tid = journal->j_transaction_sequence++;
+       transaction->t_expires = jiffies + journal->j_commit_interval;
++      INIT_LIST_HEAD(&transaction->t_jcb);
+ 
+       /* Set up the commit timer for the new transaction. */
+       J_ASSERT (!journal->j_commit_timer_active);
+@@ -201,6 +202,20 @@
+       return 0;
+ }
+ 
++/* Allocate a new handle.  This should probably be in a slab... */
++static handle_t *get_handle(int nblocks)
++{
++      handle_t *handle = jbd_kmalloc(sizeof (handle_t), GFP_NOFS);
++      if (!handle)
++              return NULL;
++      memset(handle, 0, sizeof (handle_t));
++      handle->h_buffer_credits = nblocks;
++      handle->h_ref = 1;
++      INIT_LIST_HEAD(&handle->h_jcb);
++
++      return handle;
++}
++
+ /*
+  * Obtain a new handle.  
+  *
+@@ -227,14 +242,11 @@
+               handle->h_ref++;
+               return handle;
+       }
+-      
+-      handle = jbd_kmalloc(sizeof (handle_t), GFP_NOFS);
++
++      handle = get_handle(nblocks);
+       if (!handle)
+               return ERR_PTR(-ENOMEM);
+-      memset (handle, 0, sizeof (handle_t));
+ 
+-      handle->h_buffer_credits = nblocks;
+-      handle->h_ref = 1;
+       current->journal_info = handle;
+ 
+       err = start_this_handle(journal, handle);
+@@ -333,14 +345,11 @@
+       
+       if (is_journal_aborted(journal))
+               return ERR_PTR(-EIO);
+-      
+-      handle = jbd_kmalloc(sizeof (handle_t), GFP_NOFS);
++
++      handle = get_handle(nblocks);
+       if (!handle)
+               return ERR_PTR(-ENOMEM);
+-      memset (handle, 0, sizeof (handle_t));
+ 
+-      handle->h_buffer_credits = nblocks;
+-      handle->h_ref = 1;
+       current->journal_info = handle;
+ 
+       err = try_start_this_handle(journal, handle);
+@@ -1328,6 +1337,27 @@
+ #endif
+ 
+ /*
++ * Register a callback function for this handle.  The function will be
++ * called when the transaction that this handle is part of has been
++ * committed to disk.  There is no guarantee of ordering between handles
++ * within a single transaction, nor between callbacks registered on the
++ * same handle.
++ *
++ * The caller is responsible for allocating the journal_callback struct.
++ * This is to allow the caller to add as much extra data to the callback
++ * as needed, but reduce the overhead of multiple allocations.  The caller
++ * should use the journal_callback_size() function to reserve enough space
++ * for the JBD-private part of the callback struct.
++ */
++void journal_callback_set(handle_t *handle, void (*func)(void *), void *cb_data)
++{
++      struct journal_callback *jcb = cb_data;
++
++      list_add(&jcb->jcb_list, &handle->h_jcb);
++      jcb->jcb_func = func;
++}
++
++/*
+  * All done for a particular handle.
+  *
+  * There is not much action needed here.  We just return any remaining
+@@ -1393,7 +1423,11 @@
+                       wake_up(&journal->j_wait_transaction_locked);
+       }
+ 
+-      /* 
++      /* Move callbacks from the handle to the transaction. */
++      list_splice(&handle->h_jcb, &transaction->t_jcb);
++      INIT_LIST_HEAD(&handle->h_jcb);
++
++      /*
+        * If the handle is marked SYNC, we need to set another commit
+        * going!  We also want to force a commit if the current
+        * transaction is occupying too much of the log, or if the