Re: [PATCH 6 of 8] x86/mm: Properly account for paged out pages

["Andres Lagar-Cavilla" <andres@lagarcavilla.org>]

> On Wed, Jan 25, Andres Lagar-Cavilla wrote:
>
>>  xen/arch/x86/mm/p2m.c |  3 ++-
>>  1 files changed, 2 insertions(+), 1 deletions(-)
>>
>>
>> If we hit the page after nominate but before paging it out, don't
>> decrement the
>> domain count of paged out pages.
>
> The meaning of paged_pages is to track released pages.
> In p2m_mem_paging_evict() it gets incremented once the page is removed
> from the domain, and in p2m_mem_paging_prep() it should be decremented
> right after the alloc_domheap_page() call.
> So the patch should move the atomic_dec() call after successful page
> allocation.
>
> Olaf

Hi Olaf,

there is a put_page in case things go wrong, after the alloc_domheap_page
call. So doing the decrement at alloc is a bit too soon.

Thanks
Andres
>
>> diff -r c41436e555cd -r d4336e35f0bc xen/arch/x86/mm/p2m.c
>> --- a/xen/arch/x86/mm/p2m.c
>> +++ b/xen/arch/x86/mm/p2m.c
>> @@ -1041,7 +1041,8 @@ int p2m_mem_paging_prep(struct domain *d
>>                      p2m_ram_rw, a);
>>      set_gpfn_from_mfn(mfn_x(mfn), gfn);
>>
>> -    atomic_dec(&d->paged_pages);
>> +    if ( !page_extant )
>> +        atomic_dec(&d->paged_pages);
>>
>>      ret = 0;
>>
>