Messages in this thread |  | | | Subject | Re: [PATCH] Update comment on find_task_by_pid_ns | | From | Tetsuo Handa <> | | Date | Thu, 11 Feb 2010 10:21:15 +0900 |
| |
Andrew Morton wrote: > > What should we do? Adding rcu_read_lock()/rcu_read_unlock() to each > > callers? Or adding rcu_read_lock()/rcu_read_unlock() inside > > find_task_by_pid_ns()? > > Putting rcu_read_lock() in the callee isn't a complete solution. > Because the function would still be returning a task_struct* without > any locking held and without taking a reference against it. So that > pointer is useless to the caller! > > We could add a new function which looks up the task and then takes a > reference on it, insde suitable locks. The caller would then use the > task_struct and then remember to call put_task_struct() to unpin it. > This prevents the task_struct from getting freed while it's being > manipulated, but it doesn't prevent fields within it from being altered > - that's up to the caller to sort out.
Code for "struct task_struct" is too complicated for me to understand, but my understanding is that
(1) tasklist_lock is acquired for writing.
(2) "struct task_struct" (to exit()) is removed from task's list.
(3) tasklist_lock is released.
(4) Wait for RCU grace period.
(5) kfree() members of "struct task_struct".
(6) kfree() "struct task_struct" itself.
If above sequence is correct, I think
rcu_read_lock(); task = find_task_by_pid_ns(); if (task) do_something(task); rcu_read_unlock(); do_something() can safely access all members of task without read_lock(&tasklist_lock), except task->prev (I don't know the exact member) and task->usage, because do_something() finishes its work before (5). I think we need to call find_task_by_pid_ns() with both read_lock(&tasklist_lock) and rcu_read_lock()
read_lock(&tasklist_lock); rcu_read_lock(); task = find_task_by_pid_ns(); if (task) atomido_something(task); rcu_read_unlock(); read_unlock(&tasklist_lock); only when do_something() wants to access task->prev or task->usage .
> > One fix is to go through all those callsites and add the rcu_read_lock. > That kinda sucks. Perhaps writing the new function which returns a > pinned task_struct is better?
|  |