Re: [PATCH V3] mm/thp: Allocate transparent hugepages on local node

From: Aneesh Kumar K.V
Date: Tue Jan 20 2015 - 00:57:58 EST


Vlastimil Babka <vbabka@xxxxxxx> writes:

> On 01/17/2015 01:02 AM, Andrew Morton wrote:
>> On Fri, 16 Jan 2015 12:56:36 +0530 "Aneesh Kumar K.V" <aneesh.kumar@xxxxxxxxxxxxxxxxxx> wrote:
>>
>>> This make sure that we try to allocate hugepages from local node if
>>> allowed by mempolicy. If we can't, we fallback to small page allocation
>>> based on mempolicy. This is based on the observation that allocating pages
>>> on local node is more beneficial than allocating hugepages on remote node.
>>
>> The changelog is a bit incomplete. It doesn't describe the current
>> behaviour, nor what is wrong with it. What are the before-and-after
>> effects of this change?
>>
>> And what might be the user-visible effects?
>>
>>> --- a/mm/mempolicy.c
>>> +++ b/mm/mempolicy.c
>>> @@ -2030,6 +2030,46 @@ retry_cpuset:
>>> return page;
>>> }
>>>
>>> +struct page *alloc_hugepage_vma(gfp_t gfp, struct vm_area_struct *vma,
>>> + unsigned long addr, int order)
>>
>> alloc_pages_vma() is nicely documented. alloc_hugepage_vma() is not
>> documented at all. This makes it a bit had for readers to work out the
>> difference!
>>
>> Is it possible to scrunch them both into the same function? Probably
>> too messy?
>
> Hm that could work, alloc_pages_vma already has an if (MPOL_INTERLEAVE) part, so
> just put the THP specialities into an "else if (huge_page)" part there?
>
> You could probably test for GFP_TRANSHUGE the same way as __alloc_pages_slowpath
> does. There might be false positives theoretically, but is there anything else
> that would use these flags and not be a THP?
>

is that check correct ? ie,

if ((gfp & GFP_TRANSHUGE) == GFP_TRANSHUGE)

may not always indicate transparent hugepage if defrag = 0 . With defrag
cleared, we remove __GFP_WAIT from GFP_TRANSHUGE.

static inline gfp_t alloc_hugepage_gfpmask(int defrag, gfp_t extra_gfp)
{
return (GFP_TRANSHUGE & ~(defrag ? 0 : __GFP_WAIT)) | extra_gfp;
}

-aneesh

--
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/