Message-ID: <a6ff77ba-18f7-f3ed-3f9a-00f04edccdb3@intel.com>
Date:   Wed, 2 Aug 2023 21:12:07 +0800
From:   "Yin, Fengwei" <fengwei.yin@...el.com>
To:     Ryan Roberts <ryan.roberts@....com>, <linux-mm@...ck.org>,
        <linux-kernel@...r.kernel.org>, <akpm@...ux-foundation.org>,
        <yuzhao@...gle.com>, <willy@...radead.org>, <david@...hat.com>,
        <shy828301@...il.com>, <hughd@...gle.com>
Subject: Re: [PATCH 1/3] mm: add functions folio_in_range() and
 folio_within_vma()



On 8/2/2023 7:35 PM, Ryan Roberts wrote:
> On 02/08/2023 12:14, Ryan Roberts wrote:
>> On 28/07/2023 08:09, Yin Fengwei wrote:
>>> It will be used to check whether the folio is mapped to specific
>>> VMA and whether the mapping address of folio is in the range.
>>>
>>> Also a helper function folio_within_vma() to check whether folio
>>> is in the range of vma based on folio_in_range().
>>>
>>> Signed-off-by: Yin Fengwei <fengwei.yin@...el.com>
>>> ---
>>>  mm/internal.h | 69 +++++++++++++++++++++++++++++++++++++++++++++++++++
>>>  1 file changed, 69 insertions(+)
>>>
>>> diff --git a/mm/internal.h b/mm/internal.h
>>> index 5a03bc4782a2..63de32154a48 100644
>>> --- a/mm/internal.h
>>> +++ b/mm/internal.h
>>> @@ -585,6 +585,75 @@ extern long faultin_vma_page_range(struct vm_area_struct *vma,
>>>  				   bool write, int *locked);
>>>  extern bool mlock_future_ok(struct mm_struct *mm, unsigned long flags,
>>>  			       unsigned long bytes);
>>> +
>>> +/*
>>> + * Check whether the folio is in specific range
>>> + *
>>> + * First, check whether the folio is in the range of vma.
>>> + * Then, check whether the folio is mapped to the range of [start, end].
>>> + * In the end, check whether the folio is fully mapped to the range.
>>> + *
>>> + * @pte page table pointer will be checked whether the large folio
>>> + *      is fully mapped to. Currently, if mremap in the middle of
>>> + *      large folio, the large folio could be mapped to to different
>>> + *      VMA and address check can't identify this situation.
>>> + */
>>> +static inline bool
>>> +folio_in_range(struct folio *folio, struct vm_area_struct *vma,
>>> +		unsigned long start, unsigned long end, pte_t *pte)
>>
>> This api seems a bit redundant to me. Wouldn't it be better to remove the vma
>> parameter and instead fix up the start/end addresses in folio_within_vma()?
> 
> I have created a function as part of my "pte batch-zap" patch set [1], which
> counts the number of contiguously mapped pages of a folio
> (folio_nr_pages_cont_mapped()). I wonder if actually this should be the
> primitive, which can be shared for more cases. Then your folio_within_vma()
> function could just compare the nr_pages to folio_nr_pages() to decide if the
> folio is fully and contiguously mapped in the VMA.
That means we need to unify the parameters. But I don't care about the page and
you don't care about the VMA. :). Maybe we can share the PTE check part?

> 
> I also wonder if you should change the name of folio_within_vma() to something
> like folio_test_cont_in_vma() to disambiguate from the case where the folio may
> be fully mapped with a discontiguity (although perhaps that's not possible
> because a mremap would result in distinct vmas... would a new mmap in the hole
> cause a merge of all 3?).
I don't think it's possible as mremap reuse original pgoff of VMA to new VMA. I suppose
it will prevent VMA merging. But I didn't check detail.

I hate to add the PTE check as it makes folio_within_vma() much heavy and can only
be called with page table holding. But MREMAP_DONTUNMAP could create the VMA which
just has part of folio mapped. And the first version folio_within_vma() can't identify
it.


Regards
Yin, Fengwei

> 
> [1] https://lore.kernel.org/linux-mm/20230727141837.3386072-4-ryan.roberts@arm.com/
> 
>>
>>> +{
>>> +	pte_t ptent;
>>> +	unsigned long i, nr = folio_nr_pages(folio);
>>> +	pgoff_t pgoff, addr;
>>> +	unsigned long vma_pglen = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
>>> +
>>> +	VM_WARN_ON_FOLIO(folio_test_ksm(folio), folio);
>>> +
>>> +	if (start < vma->vm_start)
>>> +		start = vma->vm_start;
>>> +	if (end > vma->vm_end)
>>> +		end = vma->vm_end;
>>> +
>>> +	pgoff = folio_pgoff(folio);
>>> +	/* if folio start address is not in vma range */
>>> +	if (pgoff < vma->vm_pgoff || pgoff > vma->vm_pgoff + vma_pglen)
>>> +		return false;
>>> +
>>> +	addr = vma->vm_start + ((pgoff - vma->vm_pgoff) << PAGE_SHIFT);
>>> +	if (addr < start || end - addr < folio_size(folio))
>>> +		return false;
>>> +
>>> +	/* not necessary to check pte for none large folio */
>>> +	if (!folio_test_large(folio))
>>> +		return true;
>>> +
>>> +	if (!pte)
>>> +		return false;
>>> +
>>> +	/* check whether parameter pte is associated with folio */
>>> +	ptent = ptep_get(pte);
>>> +	if (pte_none(ptent) || !pte_present(ptent) ||
>>> +			pte_pfn(ptent) - folio_pfn(folio) >= nr)
>>> +		return false;
>>> +
>>> +	pte -= pte_pfn(ptent) - folio_pfn(folio);
>>> +	for (i = 0; i < nr; i++, pte++) {
>>> +		ptent = ptep_get(pte);
>>> +
>>> +		if (pte_none(ptent) || !pte_present(ptent) ||
>>> +				pte_pfn(ptent) - folio_pfn(folio) >= nr)
>>> +			return false;
>>> +	}
>>
>> I don't think I see anything to ensure you don't wander off the end (or start)
>> of the pgtable? If the folio is mremapped so that it straddles multiple tables
>> (or is bigger than a single table?) then I think pte can become invalid? Perhaps
>> you intended start/end to always be within the same pgtable, but that is not
>> guarranteed in the case that folio_within_vma() is making the call.
>>
>> Also I want to check that this function is definitely always called under the
>> PTL for the table that pte belongs to?
>>
>>> +
>>> +	return true;
>>> +}
>>> +
>>> +static inline bool
>>> +folio_within_vma(struct folio *folio, struct vm_area_struct *vma, pte_t *pte)
>>> +{
>>> +	return folio_in_range(folio, vma, vma->vm_start, vma->vm_end, pte);
>>> +}
>>> +
>>>  /*
>>>   * mlock_vma_folio() and munlock_vma_folio():
>>>   * should be called with vma's mmap_lock held for read or write,
>>
> 

Powered by blists - more mailing lists