writeback: queue work on stack in writeback_inodes_sb
authorChristoph Hellwig <hch@lst.de>
Tue, 8 Jun 2010 16:14:43 +0000 (18:14 +0200)
committerJens Axboe <jaxboe@fusionio.com>
Fri, 11 Jun 2010 10:58:07 +0000 (12:58 +0200)
If we want to rely on s_umount in the caller we need to wait for completion
of the I/O submission before returning to the caller.  Refactor
bdi_sync_writeback into a bdi_queue_work_onstack helper and use it for this
case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
fs/fs-writeback.c

index dbf6f108e8685f247977e67fb5f91bfb95606c9f..759666966c6dca572c5d304d322fe40b132a261f 100644 (file)
@@ -178,30 +178,22 @@ static void bdi_alloc_queue_work(struct backing_dev_info *bdi,
 }
 
 /**
- * bdi_sync_writeback - start and wait for writeback
- * @bdi: the backing device to write from
+ * bdi_queue_work_onstack - start and wait for writeback
  * @sb: write inodes from this super_block
  *
  * Description:
- *   This does WB_SYNC_ALL data integrity writeback and waits for the
- *   IO to complete. Callers must hold the sb s_umount semaphore for
+ *   This function initiates writeback and waits for the operation to
+ *   complete. Callers must hold the sb s_umount semaphore for
  *   reading, to avoid having the super disappear before we are done.
  */
-static void bdi_sync_writeback(struct backing_dev_info *bdi,
-                              struct super_block *sb)
+static void bdi_queue_work_onstack(struct wb_writeback_args *args)
 {
-       struct wb_writeback_args args = {
-               .sb             = sb,
-               .sync_mode      = WB_SYNC_ALL,
-               .nr_pages       = LONG_MAX,
-               .range_cyclic   = 0,
-       };
        struct bdi_work work;
 
-       bdi_work_init(&work, &args);
+       bdi_work_init(&work, args);
        __set_bit(WS_ONSTACK, &work.state);
 
-       bdi_queue_work(bdi, &work);
+       bdi_queue_work(args->sb->s_bdi, &work);
        bdi_wait_on_work_done(&work);
 }
 
@@ -944,7 +936,7 @@ int bdi_writeback_task(struct bdi_writeback *wb)
 
 /*
  * Schedule writeback for all backing devices. This does WB_SYNC_NONE
- * writeback, for integrity writeback see bdi_sync_writeback().
+ * writeback, for integrity writeback see bdi_queue_work_onstack().
  */
 static void bdi_writeback_all(struct super_block *sb, long nr_pages)
 {
@@ -1183,12 +1175,15 @@ void writeback_inodes_sb(struct super_block *sb)
 {
        unsigned long nr_dirty = global_page_state(NR_FILE_DIRTY);
        unsigned long nr_unstable = global_page_state(NR_UNSTABLE_NFS);
-       long nr_to_write;
+       struct wb_writeback_args args = {
+               .sb             = sb,
+               .sync_mode      = WB_SYNC_NONE,
+       };
 
-       nr_to_write = nr_dirty + nr_unstable +
+       args.nr_pages = nr_dirty + nr_unstable +
                        (inodes_stat.nr_inodes - inodes_stat.nr_unused);
 
-       bdi_start_writeback(sb->s_bdi, sb, nr_to_write);
+       bdi_queue_work_onstack(&args);
 }
 EXPORT_SYMBOL(writeback_inodes_sb);
 
@@ -1218,7 +1213,14 @@ EXPORT_SYMBOL(writeback_inodes_sb_if_idle);
  */
 void sync_inodes_sb(struct super_block *sb)
 {
-       bdi_sync_writeback(sb->s_bdi, sb);
+       struct wb_writeback_args args = {
+               .sb             = sb,
+               .sync_mode      = WB_SYNC_ALL,
+               .nr_pages       = LONG_MAX,
+               .range_cyclic   = 0,
+       };
+
+       bdi_queue_work_onstack(&args);
        wait_sb_inodes(sb);
 }
 EXPORT_SYMBOL(sync_inodes_sb);