Re: [PATCH v6 5/9] mm: thp: Extend THP to allocate anonymous large folios

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Oct 05, 2023 at 01:49:30PM +0100, Ryan Roberts wrote:
> On 05/10/2023 13:05, Daniel Gomez wrote:
> > On Fri, Sep 29, 2023 at 12:44:16PM +0100, Ryan Roberts wrote:
> >
> > Hi Ryan,
> >> Introduce the logic to allow THP to be configured (through the new
> >> anon_orders interface we just added) to allocate large folios to back
> >> anonymous memory, which are smaller than PMD-size (for example order-2,
> >> order-3, order-4, etc).
> >>
> >> These THPs continue 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.
> >>
> >> 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 because the anon_orders
> >> defaults to only enabling PMD-order, but can be enabled at runtime by
> >> writing to anon_orders (see documentation in previous commit). The long
> >> term aim is to default anon_orders to include suitable lower orders, but
> >> there are some risks around internal fragmentation that need to be
> >> better understood first.
> >>
> >> Signed-off-by: Ryan Roberts <ryan.roberts@xxxxxxx>
> >> ---
> >>  Documentation/admin-guide/mm/transhuge.rst |   9 +-
> >>  include/linux/huge_mm.h                    |   6 +-
> >>  mm/memory.c                                | 108 +++++++++++++++++++--
> >>  3 files changed, 111 insertions(+), 12 deletions(-)
> >>
> >> diff --git a/Documentation/admin-guide/mm/transhuge.rst b/Documentation/admin-guide/mm/transhuge.rst
> >> index 9f954e73a4ca..732c3b2f4ba8 100644
> >> --- a/Documentation/admin-guide/mm/transhuge.rst
> >> +++ b/Documentation/admin-guide/mm/transhuge.rst
> >> @@ -353,7 +353,9 @@ anonymous transparent huge pages, it is necessary to read
> >>  ``/proc/PID/smaps`` and count the AnonHugePages and AnonHugePteMap
> >>  fields for each mapping. Note that in both cases, AnonHugePages refers
> >>  only to PMD-mapped THPs. AnonHugePteMap refers to THPs that are mapped
> >> -using PTEs.
> >> +using PTEs. This includes all THPs whose order is smaller than
> >> +PMD-order, as well as any PMD-order THPs that happen to be PTE-mapped
> >> +for other reasons.
> >>
> >>  The number of file transparent huge pages mapped to userspace is available
> >>  by reading ShmemPmdMapped and ShmemHugePages fields in ``/proc/meminfo``.
> >> @@ -367,6 +369,11 @@ frequently will incur overhead.
> >>  There are a number of counters in ``/proc/vmstat`` that may be used to
> >>  monitor how successfully the system is providing huge pages for use.
> >>
> >> +.. note::
> >> +   Currently the below counters only record events relating to
> >> +   PMD-order THPs. Events relating to smaller order THPs are not
> >> +   included.
> >> +
> >>  thp_fault_alloc
> >>  	is incremented every time a huge page is successfully
> >>  	allocated to handle a page fault.
> >> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> >> index 2e7c338229a6..c4860476a1f5 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)
> >>
> >>  /*
> >> - * Mask of all large folio orders supported for anonymous THP.
> >> + * Mask of all large folio orders supported for anonymous THP; all orders up to
> >> + * and including PMD_ORDER, except order-0 (which is not "huge") and order-1
> >> + * (which is a limitation of the THP implementation).
> >>   */
> >> -#define THP_ORDERS_ALL_ANON	BIT(PMD_ORDER)
> >> +#define THP_ORDERS_ALL_ANON	((BIT(PMD_ORDER + 1) - 1) & ~(BIT(0) | BIT(1)))
> >>
> >>  /*
> >>   * Mask of all large folio orders supported for file THP.
> >> diff --git a/mm/memory.c b/mm/memory.c
> >> index b5b82fc8e164..92ed9c782dc9 100644
> >> --- a/mm/memory.c
> >> +++ b/mm/memory.c
> >> @@ -4059,6 +4059,87 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >>  	return ret;
> >>  }
> >>
> >> +static bool vmf_pte_range_changed(struct vm_fault *vmf, int nr_pages)
> >> +{
> >> +	int i;
> >> +
> >> +	if (nr_pages == 1)
> >> +		return vmf_pte_changed(vmf);
> >> +
> >> +	for (i = 0; i < nr_pages; i++) {
> >> +		if (!pte_none(ptep_get_lockless(vmf->pte + i)))
> >> +			return true;
> >> +	}
> >> +
> >> +	return false;
> >> +}
> >> +
> >> +#ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >> +static struct folio *alloc_anon_folio(struct vm_fault *vmf)
> >> +{
> >> +	gfp_t gfp;
> >> +	pte_t *pte;
> >> +	unsigned long addr;
> >> +	struct folio *folio;
> >> +	struct vm_area_struct *vma = vmf->vma;
> >> +	unsigned int orders;
> >> +	int order;
> >> +
> >> +	/*
> >> +	 * If uffd is active for the vma we need per-page fault fidelity to
> >> +	 * maintain the uffd semantics.
> >> +	 */
> >> +	if (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 = hugepage_vma_check(vma, vma->vm_flags, false, true, true,
> >> +				    BIT(PMD_ORDER) - 1);
> >> +	orders = transhuge_vma_suitable(vma, vmf->address, orders);
> >> +
> >> +	if (!orders)
> >> +		goto fallback;
> >> +
> >> +	pte = pte_offset_map(vmf->pmd, vmf->address & PMD_MASK);
> >> +	if (!pte)
> >> +		return ERR_PTR(-EAGAIN);
> >> +
> >> +	order = first_order(orders);
> >> +	while (orders) {
> >> +		addr = ALIGN_DOWN(vmf->address, PAGE_SIZE << order);
> >> +		vmf->pte = pte + pte_index(addr);
> >> +		if (!vmf_pte_range_changed(vmf, 1 << order))
> >> +			break;
> >> +		order = next_order(&orders, order);
> >> +	}
> >> +
> >> +	vmf->pte = NULL;
> >> +	pte_unmap(pte);
> >> +
> >> +	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);
> >
> > I was checking your series and noticed about the hugepage flag. I think
> > you've changed it from v1 -> v2 from being false to true when orders >=2
> > but I'm not sure about the reasoning. Is this because of your statement
> > in the cover letter [1]?
>
> That hugepage flags is spec'ed as follows:
>
>  * @hugepage: For hugepages try only the preferred node if possible.
>
> The intent of passing true for orders higher than 0, is that we would prefer to
> allocate a smaller order folio that is on the preferred node than a higher order
> folio that is not on the preferred node. The assumption is that the on-going
> cost of accessing the memory on the non-preferred node will outweigh the benefit
> of allocating it as a high order folio.
>
> Thanks,
> Ryan

I think I'm confused about the @hugepage name. I guess activating that
for any order >= 2 doesn't imply you are in fact allocating a huge page
isn't? I can see order is passed from vma_alloc_folio -> __folio_alloc
-> __alloc_pages but I assumed (before reading your patch) you always
want this disabled except for HPAGE_PMD_ORDER allocation cases. But
order is not a limitation for the preferred node here, regardless this
is a huge page or not.

I see the motivation, thanks for sharing.
>
>
> >
> > [1] cover letter snippet:
> >
> > "to implement variable order, large folios for anonymous memory.
> > (previously called ..., but now exposed as an extension to THP;
> > "small-order THP")"
> >
> > Thanks,
> > Daniel
> >
> >> +		if (folio) {
> >> +			clear_huge_page(&folio->page, addr, 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
> >> +
> >>  /*
> >>   * We enter with non-exclusive mmap_lock (to exclude vma changes,
> >>   * but allow concurrent faults), and pte mapped but not yet locked.
> >> @@ -4066,6 +4147,9 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >>   */
> >>  static vm_fault_t do_anonymous_page(struct vm_fault *vmf)
> >>  {
> >> +	int i;
> >> +	int nr_pages = 1;
> >> +	unsigned long addr = vmf->address;
> >>  	bool uffd_wp = vmf_orig_pte_uffd_wp(vmf);
> >>  	struct vm_area_struct *vma = vmf->vma;
> >>  	struct folio *folio;
> >> @@ -4110,10 +4194,15 @@ static vm_fault_t do_anonymous_page(struct vm_fault *vmf)
> >>  	/* Allocate our own private page. */
> >>  	if (unlikely(anon_vma_prepare(vma)))
> >>  		goto oom;
> >> -	folio = vma_alloc_zeroed_movable_folio(vma, vmf->address);
> >> +	folio = alloc_anon_folio(vmf);
> >> +	if (IS_ERR(folio))
> >> +		return 0;
> >>  	if (!folio)
> >>  		goto oom;
> >>
> >> +	nr_pages = folio_nr_pages(folio);
> >> +	addr = ALIGN_DOWN(vmf->address, nr_pages * PAGE_SIZE);
> >> +
> >>  	if (mem_cgroup_charge(folio, vma->vm_mm, GFP_KERNEL))
> >>  		goto oom_free_page;
> >>  	folio_throttle_swaprate(folio, GFP_KERNEL);
> >> @@ -4130,12 +4219,12 @@ static vm_fault_t do_anonymous_page(struct vm_fault *vmf)
> >>  	if (vma->vm_flags & VM_WRITE)
> >>  		entry = pte_mkwrite(pte_mkdirty(entry), vma);
> >>
> >> -	vmf->pte = pte_offset_map_lock(vma->vm_mm, vmf->pmd, vmf->address,
> >> -			&vmf->ptl);
> >> +	vmf->pte = pte_offset_map_lock(vma->vm_mm, vmf->pmd, addr, &vmf->ptl);
> >>  	if (!vmf->pte)
> >>  		goto release;
> >> -	if (vmf_pte_changed(vmf)) {
> >> -		update_mmu_tlb(vma, vmf->address, vmf->pte);
> >> +	if (vmf_pte_range_changed(vmf, nr_pages)) {
> >> +		for (i = 0; i < nr_pages; i++)
> >> +			update_mmu_tlb(vma, addr + PAGE_SIZE * i, vmf->pte + i);
> >>  		goto release;
> >>  	}
> >>
> >> @@ -4150,16 +4239,17 @@ static vm_fault_t do_anonymous_page(struct vm_fault *vmf)
> >>  		return handle_userfault(vmf, VM_UFFD_MISSING);
> >>  	}
> >>
> >> -	inc_mm_counter(vma->vm_mm, MM_ANONPAGES);
> >> -	folio_add_new_anon_rmap(folio, vma, vmf->address);
> >> +	folio_ref_add(folio, nr_pages - 1);
> >> +	add_mm_counter(vma->vm_mm, MM_ANONPAGES, nr_pages);
> >> +	folio_add_new_anon_rmap(folio, vma, addr);
> >>  	folio_add_lru_vma(folio, vma);
> >>  setpte:
> >>  	if (uffd_wp)
> >>  		entry = pte_mkuffd_wp(entry);
> >> -	set_pte_at(vma->vm_mm, vmf->address, vmf->pte, entry);
> >> +	set_ptes(vma->vm_mm, addr, vmf->pte, entry, nr_pages);
> >>
> >>  	/* No need to invalidate - it was non-present before */
> >> -	update_mmu_cache_range(vmf, vma, vmf->address, vmf->pte, 1);
> >> +	update_mmu_cache_range(vmf, vma, addr, vmf->pte, nr_pages);
> >>  unlock:
> >>  	if (vmf->pte)
> >>  		pte_unmap_unlock(vmf->pte, vmf->ptl);
> >> --
> >> 2.25.1
> >>
>




[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux