[PATCH] use pdflush for unused inode writeback

This is pdflush's first application!  The writeback of
the unused inodes list by keventd is removed, and a
pdflush thread is dispatched instead.

There is a need for exclusion - to prevent all the
pdflush threads from working against the same request
queue.  This is implemented locally.  And this is a
problem, because other pdflush threads can be dispatched
to writeback other filesystem objects, and they don't
know that there's already a pdflush thread working that
request queue.

So moving the exclusion into the request queue itself
is on my things-to-do-list.  But the code as-is works
OK - under a `dbench 100' load the number of pdflush
instances can grow as high as four or five.  Some fine
tuning is needed...

diff --git a/fs/inode.c b/fs/inode.c
index 29ba87e29414a7e5ffd6bf56f39d6b33bdbb4a77..c6d3a3b864c05f957af9870cabb458c95ddc88c9 100644 (file)
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -433,7 +433,7 @@ void sync_inodes(void)
        }
 }
 
-static void try_to_sync_unused_inodes(void * arg)
+static void try_to_sync_unused_inodes(unsigned long pexclusive)
 {
        struct super_block * sb;
        int nr_inodes = inodes_stat.nr_unused;
@@ -450,10 +450,9 @@ static void try_to_sync_unused_inodes(void * arg)
        }
        spin_unlock(&sb_lock);
        spin_unlock(&inode_lock);
+       clear_bit(0, (unsigned long *)pexclusive);
 }
 
-static struct tq_struct unused_inodes_flush_task;
-
 /**
  *     write_inode_now -       write an inode to disk
  *     @inode: inode to write to disk
@@ -746,8 +745,15 @@ void prune_icache(int goal)
         * from here or we're either synchronously dogslow
         * or we deadlock with oom.
         */
-       if (goal)
-               schedule_task(&unused_inodes_flush_task);
+       if (goal) {
+               static unsigned long exclusive;
+
+               if (!test_and_set_bit(0, &exclusive)) {
+                       if (pdflush_operation(try_to_sync_unused_inodes,
+                                               (unsigned long)&exclusive))
+                               clear_bit(0, &exclusive);
+               }
+       }
 }
 /*
  * This is called from kswapd when we think we need some
@@ -1173,8 +1179,6 @@ void __init inode_init(unsigned long mempages)
                                         NULL);
        if (!inode_cachep)
                panic("cannot create inode slab cache");
-
-       unused_inodes_flush_task.routine = try_to_sync_unused_inodes;
 }
 
 static inline void do_atime_update(struct inode *inode)