Re: [PATCH v12 1/8] mm/gup: Introduce unpin_folio/unpin_folios helpers

[David Hildenbrand <david@redhat.com>]

On 02.04.24 15:52, David Hildenbrand wrote:
> On 25.02.24 08:56, Vivek Kasireddy wrote:
>> These helpers are the folio versions of unpin_user_page/unpin_user_pages.
>> They are currently only useful for unpinning folios pinned by
>> memfd_pin_folios() or other associated routines. However, they could
>> find new uses in the future, when more and more folio-only helpers
>> are added to GUP.
>>
>> Cc: David Hildenbrand <david@redhat.com>
>> Cc: Matthew Wilcox <willy@infradead.org>
>> Cc: Christoph Hellwig <hch@infradead.org>
>> Cc: Jason Gunthorpe <jgg@nvidia.com>
>> Cc: Peter Xu <peterx@redhat.com>
>> Suggested-by: David Hildenbrand <david@redhat.com>
>> Signed-off-by: Vivek Kasireddy <vivek.kasireddy@intel.com>
>> ---
>>    include/linux/mm.h |  2 ++
>>    mm/gup.c           | 81 ++++++++++++++++++++++++++++++++++++++++------
>>    2 files changed, 74 insertions(+), 9 deletions(-)
>>
>> diff --git a/include/linux/mm.h b/include/linux/mm.h
>> index 6f4825d82965..36e4c2b22600 100644
>> --- a/include/linux/mm.h
>> +++ b/include/linux/mm.h
>> @@ -1601,11 +1601,13 @@ static inline void put_page(struct page *page)
>>    #define GUP_PIN_COUNTING_BIAS (1U << 10)
>>    
>>    void unpin_user_page(struct page *page);
>> +void unpin_folio(struct folio *folio);
>>    void unpin_user_pages_dirty_lock(struct page **pages, unsigned long npages,
>>    				 bool make_dirty);
>>    void unpin_user_page_range_dirty_lock(struct page *page, unsigned long npages,
>>    				      bool make_dirty);
>>    void unpin_user_pages(struct page **pages, unsigned long npages);
>> +void unpin_folios(struct folio **folios, unsigned long nfolios);
>>    
>>    static inline bool is_cow_mapping(vm_flags_t flags)
>>    {
>> diff --git a/mm/gup.c b/mm/gup.c
>> index df83182ec72d..0a45eda6aaeb 100644
>> --- a/mm/gup.c
>> +++ b/mm/gup.c
>> @@ -30,6 +30,23 @@ struct follow_page_context {
>>    	unsigned int page_mask;
>>    };
>>    
>> +static inline void sanity_check_pinned_folios(struct folio **folios,
>> +					      unsigned long nfolios)
>> +{
>> +	if (!IS_ENABLED(CONFIG_DEBUG_VM))
>> +		return;
>> +
>> +	for (; nfolios; nfolios--, folios++) {
>> +		struct folio *folio = *folios;
>> +
>> +		if (is_zero_folio(folio) ||
>> +		    !folio_test_anon(folio))
>> +			continue;
>> +
>> +		VM_BUG_ON_FOLIO(!PageAnonExclusive(&folio->page), folio);
> 
> That change is wrong (and the split makes the check confusing).
> 
> It could be that the first subpage is no longer exclusive, but the given
> (sanity_check_pinned_pages() ) subpage is exclusive for large folios.
> 
> I suggest dropping that change, and instead, in
> unpin_folio()/unpin_folios(), reject any anon folios for now.
> 
> So, replace the sanity_check_pinned_folios() in unpin_folio() /
> unpin_folios() by a VM_WARN_ON(folio_test_anon(folio));

After reading patch #2: drop both the sanity check and VM_WARN_ON() from 
unpin_folio()/unpin_folios(), and add a comment to the patch description 
that we cannot do the sanity checking without the subpage, and that we 
can reintroduce it once we have a single per-folio AnonExclusive bit.

-- 
Cheers,

David / dhildenb