Re: [PATCH][V3]Make get_user_pages interruptible

[Ying Han]

thanks Pekka and i think one example of the case you mentioned is in
access_process_vm() which is calling
get_user_pages(tsk, mm, addr, 1, write, 1, &pages, &vma). However, it
is allocating only one page here which
much less likely to be stuck under memory pressure. Like you said, in
order to make it more flexible for future
changes, i might make the change like:
>>>>                         */
>>>> -                       if (unlikely(test_tsk_thread_flag(tsk, TIF_MEMDIE)))
>>>> -                               return i ? i : -ENOMEM;
>>>> +                       if (unlikely(sigkill_pending(current) | | sigkill_pending(tsk)))
>>>> +                               return i ? i : -ERESTARTSYS;

is this something acceptable?



On Mon, Nov 24, 2008 at 1:13 PM, Pekka Enberg <penberg@xxxxxxxxxxxxxx> wrote:
> On Fri, Nov 21, 2008 at 5:31 PM, Ying Han <yinghan@xxxxxxxxxx> wrote:
>>>>                         */
>>>> -                       if (unlikely(test_tsk_thread_flag(tsk, TIF_MEMDIE)))
>>>> -                               return i ? i : -ENOMEM;
>>>> +                       if (unlikely(sigkill_pending(tsk)))
>>>> +                               return i ? i : -ERESTARTSYS;
>
> On Mon, Nov 24, 2008 at 12:02 PM, Paul Menage <menage@xxxxxxxxxx> wrote:
>>> You've changed the check from sigkill_pending(current) to sigkill_pending(tsk).
>>>
>>> I originally made that sigkill_pending(current) since we want to avoid
>>> tasks entering an unkillable state just because they're doing
>>> get_user_pages() on a system that's short of memory. Admittedly for
>>> the main case that we care about, mlock() (or an mmap() with
>>> MCL_FUTURE set) then tsk==current, but philosophically it seems to me
>>> to be more correct to do the check against current than tsk, since
>>> current is the thing that's actually allocating the memory. But maybe
>>> it would be better to check both?
>
> On Mon, Nov 24, 2008 at 11:02 PM, Ying Han <yinghan@xxxxxxxxxx> wrote:
>> In most of cases, tsk==current in get_user_pages(), that is why i
>> change current to tsk since
>> tsk is a superset of current, no? If that is right, why we need to check both?
>
> I'm not sure if it's strictly necessary but as I pointed out in the
> other mail, there can be callers that are doing get_user_pages() on
> behalf of other tasks and you probably want to be able to kill the
> task that's actually _calling_ get_user_pages() as well.
>
>                        Pekka
>
--
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/