On Wed, Feb 24, 2021 at 03:35:44AM -0500, Luo Longjun wrote: > @@ -2912,17 +2922,66 @@ static int locks_show(struct seq_file *f, void *v) > struct file_lock *fl, *bfl; > struct pid_namespace *proc_pidns = proc_pid_ns(file_inode(f->file)->i_sb); > > + struct list_head root; > + struct list_head *tail = &root; > + struct list_head *pos, *tmp; > + struct locks_traverse_list *node, *node_child; > + > + int ret = 0; > + > fl = hlist_entry(v, struct file_lock, fl_link); > > if (locks_translate_pid(fl, proc_pidns) == 0) > - return 0; > + return ret; > + > + INIT_LIST_HEAD(&root); > > - lock_get_status(f, fl, iter->li_pos, ""); > + node = kmalloc(sizeof(struct locks_traverse_list), GFP_KERNEL); Is it safe to allocate here? I thought this was under the blocked_lock_lock spinlock. And I still don't think you need a stack. Have you tried the suggestion in my previous mail? --b.
