Re: [PATCH v9 04/10] mm: thp: Support allocation of anonymous multi-size THP

[David Hildenbrand <david@redhat.com>]

On 12.12.23 16:38, Ryan Roberts wrote:
> On 12/12/2023 15:02, David Hildenbrand wrote:
>> On 07.12.23 17:12, Ryan Roberts wrote:
>>> Introduce the logic to allow THP to be configured (through the new sysfs
>>> interface we just added) to allocate large folios to back anonymous
>>> memory, which are larger than the base page size but smaller than
>>> PMD-size. We call this new THP extension "multi-size THP" (mTHP).
>>>
>>> mTHP continues to be PTE-mapped, but in many cases can still provide
>>> similar benefits to traditional PMD-sized THP: Page faults are
>>> significantly reduced (by a factor of e.g. 4, 8, 16, etc. depending on
>>> the configured order), but latency spikes are much less prominent
>>> because the size of each page isn't as huge as the PMD-sized variant and
>>> there is less memory to clear in each page fault. The number of per-page
>>> operations (e.g. ref counting, rmap management, lru list management) are
>>> also significantly reduced since those ops now become per-folio.
>>
>> I'll note that with always-pte-mapped-thp it will be much easier to support
>> incremental page clearing (e.g., zero only parts of the folio and map the
>> remainder in a pro-non-like fashion whereby we'll zero on the next page fault).
>> With a PMD-sized thp, you have to eventually place/rip out page tables to
>> achieve that.
> 
> But then you lose the benefits of reduced number of page faults; reducing page
> faults gives a big speed up for workloads with lots of short lived processes
> like compiling.

Well, you can do interesting things like "allocate order-5", but zero in 
order-3 chunks. You get less page faults and pay for alloc/rmap only once.

But yes, all has pros and cons.

[...]

>>
>>>
>>> Some architectures also employ TLB compression mechanisms to squeeze
>>> more entries in when a set of PTEs are virtually and physically
>>> contiguous and approporiately aligned. In this case, TLB misses will
>>> occur less often.
>>>
>>> The new behaviour is disabled by default, but can be enabled at runtime
>>> by writing to /sys/kernel/mm/transparent_hugepage/hugepage-XXkb/enabled
>>> (see documentation in previous commit). The long term aim is to change
>>> the default to include suitable lower orders, but there are some risks
>>> around internal fragmentation that need to be better understood first.
>>>
>>> Tested-by: Kefeng Wang <wangkefeng.wang@huawei.com>
>>> Tested-by: John Hubbard <jhubbard@nvidia.com>
>>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>>> ---
>>>    include/linux/huge_mm.h |   6 ++-
>>>    mm/memory.c             | 111 ++++++++++++++++++++++++++++++++++++----
>>>    2 files changed, 106 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
>>> index 609c153bae57..fa7a38a30fc6 100644
>>> --- a/include/linux/huge_mm.h
>>> +++ b/include/linux/huge_mm.h
>>> @@ -68,9 +68,11 @@ extern struct kobj_attribute shmem_enabled_attr;
>>>    #define HPAGE_PMD_NR (1<<HPAGE_PMD_ORDER)
>>
>> [...]
>>
>>> +
>>> +#ifdef CONFIG_TRANSPARENT_HUGEPAGE
>>> +static struct folio *alloc_anon_folio(struct vm_fault *vmf)
>>> +{
>>> +    struct vm_area_struct *vma = vmf->vma;
>>> +    unsigned long orders;
>>> +    struct folio *folio;
>>> +    unsigned long addr;
>>> +    pte_t *pte;
>>> +    gfp_t gfp;
>>> +    int order;
>>> +
>>> +    /*
>>> +     * If uffd is active for the vma we need per-page fault fidelity to
>>> +     * maintain the uffd semantics.
>>> +     */
>>> +    if (unlikely(userfaultfd_armed(vma)))
>>> +        goto fallback;
>>> +
>>> +    /*
>>> +     * Get a list of all the (large) orders below PMD_ORDER that are enabled
>>> +     * for this vma. Then filter out the orders that can't be allocated over
>>> +     * the faulting address and still be fully contained in the vma.
>>> +     */
>>> +    orders = thp_vma_allowable_orders(vma, vma->vm_flags, false, true, true,
>>> +                      BIT(PMD_ORDER) - 1);
>>> +    orders = thp_vma_suitable_orders(vma, vmf->address, orders);
>>> +
>>> +    if (!orders)
>>> +        goto fallback;
>>> +
>>> +    pte = pte_offset_map(vmf->pmd, vmf->address & PMD_MASK);
>>> +    if (!pte)
>>> +        return ERR_PTR(-EAGAIN);
>>> +
>>> +    /*
>>> +     * Find the highest order where the aligned range is completely
>>> +     * pte_none(). Note that all remaining orders will be completely
>>> +     * pte_none().
>>> +     */
>>> +    order = highest_order(orders);
>>> +    while (orders) {
>>> +        addr = ALIGN_DOWN(vmf->address, PAGE_SIZE << order);
>>> +        if (pte_range_none(pte + pte_index(addr), 1 << order))
>>> +            break;
>>> +        order = next_order(&orders, order);
>>> +    }
>>> +
>>> +    pte_unmap(pte);
>>> +
>>> +    /* Try allocating the highest of the remaining orders. */
>>> +    gfp = vma_thp_gfp_mask(vma);
>>> +    while (orders) {
>>> +        addr = ALIGN_DOWN(vmf->address, PAGE_SIZE << order);
>>> +        folio = vma_alloc_folio(gfp, order, vma, addr, true);
>>> +        if (folio) {
>>> +            clear_huge_page(&folio->page, vmf->address, 1 << order);
>>> +            return folio;
>>> +        }
>>> +        order = next_order(&orders, order);
>>> +    }
>>> +
>>> +fallback:
>>> +    return vma_alloc_zeroed_movable_folio(vma, vmf->address);
>>> +}
>>> +#else
>>> +#define alloc_anon_folio(vmf) \
>>> +        vma_alloc_zeroed_movable_folio((vmf)->vma, (vmf)->address)
>>> +#endif
>>
>> A neater alternative might be
>>
>> static struct folio *alloc_anon_folio(struct vm_fault *vmf)
>> {
>> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>>      /* magic */
>> fallback:
>> #endif
>>      return vma_alloc_zeroed_movable_folio((vmf)->vma, (vmf)->address):
>> }
> 
> I guess beauty lies in the eye of the beholder... I don't find it much neater
> personally :). But happy to make the change if you insist; what's the process
> now that its in mm-unstable? Just send a patch to Andrew for squashing?

That way it is clear that the fallback for thp is just what !thp does.

But either is fine for me; no need to change if you disagree.

-- 
Cheers,

David / dhildenb