Re: [RFC] perf: mmap2 not covering VM_CLONE regions

[Kees Cook]

On Thu, Oct 3, 2013 at 2:13 AM, Stephane Eranian <eranian@xxxxxxxxxx> wrote:
> On Thu, Oct 3, 2013 at 11:03 AM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>> On Thu, Oct 03, 2013 at 10:55:28AM +0200, Stephane Eranian wrote:
>>> I don't know the MM code but I assume that that vm_mm struct is
>>> allocated dynamically
>>> and maybe you already grabbing a lock while doing this. Could we
>>> leverage that lock
>>> to increment a global generation number?
>>
>> Sure; something like so.. I just don't like global state nor adding to
>> mm_struct for just this.
>>
> I understand, I don't like global state either. I see you already
> have uprobes state in there. Thus to me, the perf situation is
> not much worse in terms of usefulness (frequency of use).
>
> Thanks.
>
>> ---
>>  include/linux/mm_types.h | 1 +
>>  kernel/fork.c            | 6 ++++++
>>  2 files changed, 7 insertions(+)
>>
>> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
>> index d9851eeb6e1d..3877b1e72a5b 100644
>> --- a/include/linux/mm_types.h
>> +++ b/include/linux/mm_types.h
>> @@ -436,6 +436,7 @@ struct mm_struct {
>>         int first_nid;
>>  #endif
>>         struct uprobes_state uprobes_state;
>> +       u64 mm_id;
>>  };
>>
>>  /* first nid will either be a valid NID or one of these values */
>> diff --git a/kernel/fork.c b/kernel/fork.c
>> index 086fe73ad6bd..b315f6227629 100644
>> --- a/kernel/fork.c
>> +++ b/kernel/fork.c
>> @@ -523,6 +523,8 @@ static void mm_init_aio(struct mm_struct *mm)
>>  #endif
>>  }
>>
>> +static u64 global_mm_id;
>> +
>>  static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p)
>>  {
>>         atomic_set(&mm->mm_users, 1);
>> @@ -537,6 +539,7 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p)
>>         spin_lock_init(&mm->page_table_lock);
>>         mm_init_aio(mm);
>>         mm_init_owner(mm, p);
>> +       mm->mm_id = 0;
>>
>>         if (likely(!mm_alloc_pgd(mm))) {
>>                 mm->def_flags = 0;
>> @@ -1422,6 +1425,9 @@ static struct task_struct *copy_process(unsigned long clone_flags,
>>          */
>>         write_lock_irq(&tasklist_lock);
>>
>> +       if (p->mm && !p->mm->mm_id)
>> +               p->mm->mm_id = ++global_mm_id;
>> +
>>         /* CLONE_PARENT re-uses the old parent */
>>         if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) {
>>                 p->real_parent = current->real_parent;

FWIW, I like this approach.

-Kees

-- 
Kees Cook
Chrome OS Security
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/