Messages in this thread | | | Date | Mon, 3 Jun 2013 21:06:40 +0200 | From | Oleg Nesterov <> | Subject | [PATCH v2 0/4] proc: first_tid() fix/cleanup |
| |
Hello.
next_thread() should be avoided, probably next_tid() is the only "valid" user.
But now we have another reason to avoid (and probably even kill) it, we are going to replace or fix while_each_thread(), almost every lockless usage is wrong.
Changes:
1/4: Update the changelog, do not move the comment.
2/4: No changes.
3/4: Update the comment following the explanations from Eric.
Eric pointed that get_proc_task() without rcu lock can trigger the (bogus) warning. Extract the similar check from pid_delete_dentry() into the new helper and use it instead.
I didn't dare to preserve his ack, but the only change is the new proc_inode_is_dead() helper and
- if (pid_task(proc_pid(inode)) + if (proc_inode_is_dead(inode))
in proc_task_readdir().
4/4: New.
Oleg.
| |