Re: [PATCH v1 4/4] mm/mmu_gather: Store and process pages in contig ranges

[David Hildenbrand <david@redhat.com>]

On 25.08.23 06:09, Matthew Wilcox wrote:
> On Thu, Aug 10, 2023 at 11:33:32AM +0100, Ryan Roberts wrote:
>> +void folios_put_refs(struct folio_range *folios, int nr)
>> +{
>> +	int i;
>> +	LIST_HEAD(pages_to_free);
>> +	struct lruvec *lruvec = NULL;
>> +	unsigned long flags = 0;
>> +	unsigned int lock_batch;
>> +
>> +	for (i = 0; i < nr; i++) {
>> +		struct folio *folio = page_folio(folios[i].start);
>> +		int refs = folios[i].end - folios[i].start;
>> +
>> +		/*
>> +		 * Make sure the IRQ-safe lock-holding time does not get
>> +		 * excessive with a continuous string of pages from the
>> +		 * same lruvec. The lock is held only if lruvec != NULL.
>> +		 */
>> +		if (lruvec && ++lock_batch == SWAP_CLUSTER_MAX) {
>> +			unlock_page_lruvec_irqrestore(lruvec, flags);
>> +			lruvec = NULL;
>> +		}
>> +
>> +		if (is_huge_zero_page(&folio->page))
>> +			continue;
>> +
>> +		if (folio_is_zone_device(folio)) {
>> +			if (lruvec) {
>> +				unlock_page_lruvec_irqrestore(lruvec, flags);
>> +				lruvec = NULL;
>> +			}
>> +			if (put_devmap_managed_page(&folio->page))
>> +				continue;
>> +			if (folio_put_testzero(folio))
> 
> We're only putting one ref for the zone_device folios?  Surely
> this should be ref_sub_and_test like below?

I suspect put_devmap_managed_page() then also needs care?

-- 
Cheers,

David / dhildenb