process_inode_switch_wbs_work() can be switching over 100 inodes to a
different cgroup. Since switching an inode requires counting all dirty &
under-writeback pages in the address space of each inode, this can take
a significant amount of time. Add a possibility to reschedule after
processing each inode to avoid softlockups.
Acked-by: Tejun Heo <tj@xxxxxxxxxx>
Signed-off-by: Jan Kara <jack@xxxxxxx>
---
fs/fs-writeback.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index f2265aa9b4c2..40b42c385b55 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -500,6 +500,7 @@ static void process_inode_switch_wbs_work(struct bdi_writeback *new_wb,
*/
down_read(&bdi->wb_switch_rwsem);
+ inodep = isw->inodes;
/*
* By the time control reaches here, RCU grace period has passed
* since I_WB_SWITCH assertion and all wb stat update transactions
@@ -510,6 +511,7 @@ static void process_inode_switch_wbs_work(struct bdi_writeback *new_wb,
* gives us exclusion against all wb related operations on @inode
* including IO list manipulations and stat updates.
*/
+relock:
if (old_wb < new_wb) {
spin_lock(&old_wb->list_lock);
spin_lock_nested(&new_wb->list_lock, SINGLE_DEPTH_NESTING);
@@ -518,10 +520,17 @@ static void process_inode_switch_wbs_work(struct bdi_writeback *new_wb,
spin_lock_nested(&old_wb->list_lock, SINGLE_DEPTH_NESTING);
}
- for (inodep = isw->inodes; *inodep; inodep++) {
+ while (*inodep) {
WARN_ON_ONCE((*inodep)->i_wb != old_wb);
if (inode_do_switch_wbs(*inodep, old_wb, new_wb))
nr_switched++;
+ inodep++;
+ if (*inodep && need_resched()) {
+ spin_unlock(&new_wb->list_lock);
+ spin_unlock(&old_wb->list_lock);
+ cond_resched();
+ goto relock;
+ }
}
spin_unlock(&new_wb->list_lock);
--
2.51.0
