fs: move i_wb_list out from under inode_lock
Protect the inode writeback list with a new global lock
inode_wb_list_lock and use it to protect the list manipulations and
traversals. This lock replaces the inode_lock as the inodes on the
list can be validity checked while holding the inode->i_lock and
hence the inode_lock is no longer needed to protect the list.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 5de56a2..ed80065 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -176,6 +176,17 @@
}
/*
+ * Remove the inode from the writeback list it is on.
+ */
+void inode_wb_list_del(struct inode *inode)
+{
+ spin_lock(&inode_wb_list_lock);
+ list_del_init(&inode->i_wb_list);
+ spin_unlock(&inode_wb_list_lock);
+}
+
+
+/*
* Redirty an inode: set its when-it-was dirtied timestamp and move it to the
* furthest end of its superblock's dirty-inode list.
*
@@ -188,6 +199,7 @@
{
struct bdi_writeback *wb = &inode_to_bdi(inode)->wb;
+ assert_spin_locked(&inode_wb_list_lock);
if (!list_empty(&wb->b_dirty)) {
struct inode *tail;
@@ -205,14 +217,17 @@
{
struct bdi_writeback *wb = &inode_to_bdi(inode)->wb;
+ assert_spin_locked(&inode_wb_list_lock);
list_move(&inode->i_wb_list, &wb->b_more_io);
}
static void inode_sync_complete(struct inode *inode)
{
/*
- * Prevent speculative execution through spin_unlock(&inode_lock);
+ * Prevent speculative execution through
+ * spin_unlock(&inode_wb_list_lock);
*/
+
smp_mb();
wake_up_bit(&inode->i_state, __I_SYNC);
}
@@ -286,6 +301,7 @@
*/
static void queue_io(struct bdi_writeback *wb, unsigned long *older_than_this)
{
+ assert_spin_locked(&inode_wb_list_lock);
list_splice_init(&wb->b_more_io, &wb->b_io);
move_expired_inodes(&wb->b_dirty, &wb->b_io, older_than_this);
}
@@ -308,25 +324,23 @@
wqh = bit_waitqueue(&inode->i_state, __I_SYNC);
while (inode->i_state & I_SYNC) {
spin_unlock(&inode->i_lock);
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
__wait_on_bit(wqh, &wq, inode_wait, TASK_UNINTERRUPTIBLE);
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
spin_lock(&inode->i_lock);
}
}
/*
- * Write out an inode's dirty pages. Called under inode_lock. Either the
- * caller has ref on the inode (either via __iget or via syscall against an fd)
- * or the inode has I_WILL_FREE set (via generic_forget_inode)
+ * Write out an inode's dirty pages. Called under inode_wb_list_lock. Either
+ * the caller has an active reference on the inode or the inode has I_WILL_FREE
+ * set.
*
* If `wait' is set, wait on the writeout.
*
* The whole writeout design is quite complex and fragile. We want to avoid
* starvation of particular inodes when others are being redirtied, prevent
* livelocks, etc.
- *
- * Called under inode_lock.
*/
static int
writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
@@ -368,7 +382,7 @@
inode->i_state |= I_SYNC;
inode->i_state &= ~I_DIRTY_PAGES;
spin_unlock(&inode->i_lock);
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
ret = do_writepages(mapping, wbc);
@@ -388,12 +402,10 @@
* due to delalloc, clear dirty metadata flags right before
* write_inode()
*/
- spin_lock(&inode_lock);
spin_lock(&inode->i_lock);
dirty = inode->i_state & I_DIRTY;
inode->i_state &= ~(I_DIRTY_SYNC | I_DIRTY_DATASYNC);
spin_unlock(&inode->i_lock);
- spin_unlock(&inode_lock);
/* Don't write the inode if only I_DIRTY_PAGES was set */
if (dirty & (I_DIRTY_SYNC | I_DIRTY_DATASYNC)) {
int err = write_inode(inode, wbc);
@@ -401,7 +413,7 @@
ret = err;
}
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
spin_lock(&inode->i_lock);
inode->i_state &= ~I_SYNC;
if (!(inode->i_state & I_FREEING)) {
@@ -543,10 +555,10 @@
*/
redirty_tail(inode);
}
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
iput(inode);
cond_resched();
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
if (wbc->nr_to_write <= 0) {
wbc->more_io = 1;
return 1;
@@ -565,7 +577,7 @@
if (!wbc->wb_start)
wbc->wb_start = jiffies; /* livelock avoidance */
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
if (!wbc->for_kupdate || list_empty(&wb->b_io))
queue_io(wb, wbc->older_than_this);
@@ -583,7 +595,7 @@
if (ret)
break;
}
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
/* Leave any unwritten inodes on b_io */
}
@@ -592,11 +604,11 @@
{
WARN_ON(!rwsem_is_locked(&sb->s_umount));
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
if (!wbc->for_kupdate || list_empty(&wb->b_io))
queue_io(wb, wbc->older_than_this);
writeback_sb_inodes(sb, wb, wbc, true);
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
}
/*
@@ -735,7 +747,7 @@
* become available for writeback. Otherwise
* we'll just busyloop.
*/
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
if (!list_empty(&wb->b_more_io)) {
inode = wb_inode(wb->b_more_io.prev);
trace_wbc_writeback_wait(&wbc, wb->bdi);
@@ -743,7 +755,7 @@
inode_wait_for_writeback(inode);
spin_unlock(&inode->i_lock);
}
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
}
return wrote;
@@ -1009,7 +1021,6 @@
{
struct super_block *sb = inode->i_sb;
struct backing_dev_info *bdi = NULL;
- bool wakeup_bdi = false;
/*
* Don't do this for I_DIRTY_PAGES - that doesn't actually
@@ -1033,7 +1044,6 @@
if (unlikely(block_dump))
block_dump___mark_inode_dirty(inode);
- spin_lock(&inode_lock);
spin_lock(&inode->i_lock);
if ((inode->i_state & flags) != flags) {
const int was_dirty = inode->i_state & I_DIRTY;
@@ -1059,12 +1069,12 @@
if (inode->i_state & I_FREEING)
goto out_unlock_inode;
- spin_unlock(&inode->i_lock);
/*
* If the inode was already on b_dirty/b_io/b_more_io, don't
* reposition it (that would break b_dirty time-ordering).
*/
if (!was_dirty) {
+ bool wakeup_bdi = false;
bdi = inode_to_bdi(inode);
if (bdi_cap_writeback_dirty(bdi)) {
@@ -1081,18 +1091,20 @@
wakeup_bdi = true;
}
+ spin_unlock(&inode->i_lock);
+ spin_lock(&inode_wb_list_lock);
inode->dirtied_when = jiffies;
list_move(&inode->i_wb_list, &bdi->wb.b_dirty);
+ spin_unlock(&inode_wb_list_lock);
+
+ if (wakeup_bdi)
+ bdi_wakeup_thread_delayed(bdi);
+ return;
}
- goto out;
}
out_unlock_inode:
spin_unlock(&inode->i_lock);
-out:
- spin_unlock(&inode_lock);
- if (wakeup_bdi)
- bdi_wakeup_thread_delayed(bdi);
}
EXPORT_SYMBOL(__mark_inode_dirty);
@@ -1296,9 +1308,9 @@
wbc.nr_to_write = 0;
might_sleep();
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
ret = writeback_single_inode(inode, &wbc);
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
if (sync)
inode_sync_wait(inode);
return ret;
@@ -1320,9 +1332,9 @@
{
int ret;
- spin_lock(&inode_lock);
+ spin_lock(&inode_wb_list_lock);
ret = writeback_single_inode(inode, wbc);
- spin_unlock(&inode_lock);
+ spin_unlock(&inode_wb_list_lock);
return ret;
}
EXPORT_SYMBOL(sync_inode);