Re: [PATCH 1/7] mm: remove cgroup_mm_owner_callbacks

[Balbir Singh <balbir@linux.vnet.ibm.com>]

Hugh Dickins wrote:
> On Wed, 19 Nov 2008, Paul Menage wrote:
>> On Wed, Nov 19, 2008 at 5:11 PM, Hugh Dickins <hugh@veritas.com> wrote:
>>>  assign_new_owner:
>>>        BUG_ON(c == p);
>>>        get_task_struct(c);
>>> -       read_unlock(&tasklist_lock);
>>> -       down_write(&mm->mmap_sem);
>>>        /*
>>>         * The task_lock protects c->mm from changing.
>>>         * We always want mm->owner->mm == mm
>>>         */
>>>        task_lock(c);
>>> +       /*
>>> +        * Delay read_unlock() till we have the task_lock()
>>> +        * to ensure that c does not slip away underneath us
>>> +        */
>>> +       read_unlock(&tasklist_lock);
>> How can c slip away when we've done get_task_struct(c) earlier?
> 

"c" cannot slip away, but c->mm can change, but see below.

> I don't know, I did vaguely wonder the same myself: just putting
> this back to how it was before (including that comment),
> maybe Balbir can enlighten us.

Looking at the patch, we do handle a c->mm != mm case. The code seems to keep
the task on the global tasklist and protects task->mm, which might not be
necessary here. It would seem reasonable to allow the read_unlock() to occur
prior to task_lock().

-- 
	Balbir

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>