It needs to give a chance to be rescheduled while shrinking slab entries. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
@@ -569,6 +569,7 @@ unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink)
if (node_cnt + tree_cnt >= nr_shrink)
goto unlock_out;
+ cond_resched();
}
up_write(&sbi->extent_tree_lock);