Re: [PATCH] kvm: Faults which trigger IO release the mmap_sem

[Andres Lagar-Cavilla]

On Wed, Sep 17, 2014 at 10:08 AM, Gleb Natapov <gleb@xxxxxxxxxx> wrote:
> On Wed, Sep 17, 2014 at 10:00:32AM -0700, Andres Lagar-Cavilla wrote:
>> On Wed, Sep 17, 2014 at 4:42 AM, Gleb Natapov <gleb@xxxxxxxxxx> wrote:
>> > On Wed, Sep 17, 2014 at 01:27:14PM +0200, Radim KrÄmÃÅ wrote:
>> >> 2014-09-17 13:26+0300, Gleb Natapov:
>> >> > For async_pf_execute() you do not need to even retry. Next guest's page fault
>> >> > will retry it for you.
>> >>
>> >> Wouldn't that be a waste of vmentries?
>> > This is how it will work with or without this second gup. Page is not
>> > mapped into a shadow page table on this path, it happens on a next fault.
>>
>> The point is that the gup in the async pf completion from the work
>> queue will not relinquish the mmap semaphore. And it most definitely
>> should, given that we are likely looking at swap/filemap.
>>
> I get this point and the patch looks good in general, but my point is
> that when _retry() is called from async_pf_execute() second gup is not
> needed. In the original code gup is called to do IO and nothing else.
> In your patch this is accomplished by the first gup already, so you
> can skip second gup if pagep == nullptr.

I see. However, if this function were to be used elsewhere in the
future, then the "if pagep == NULL don't retry" semantics may not
match the new caller's intention. Would you prefer an explicit flag?

Andres

>
> --
>                         Gleb.



-- 
Andres Lagar-Cavilla | Google Kernel Team | andreslc@xxxxxxxxxx
--
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/