Re: [RFC PATCH v3 3/8] mm: Hot page tracking and promotion

[Bharata B Rao <bharata@amd.com>]

On 26-Nov-25 6:54 PM, Alok Rathore wrote:
>> +/*
>> + * Walks the PFNs of the zone, isolates and migrates them in batches.
>> + */
>> +static void kmigrated_walk_zone(unsigned long start_pfn, unsigned long end_pfn,
>> +                int src_nid)
>> +{
>> +    int cur_nid = NUMA_NO_NODE;
>> +    LIST_HEAD(migrate_list);
>> +    int batch_count = 0;
>> +    struct folio *folio;
>> +    struct page *page;
>> +    unsigned long pfn;
>> +
>> +    pfn = start_pfn;
>> +    do {
>> +        unsigned long nid = NUMA_NO_NODE, freq = 0, time = 0, nr = 1;
>> +
>> +        if (!pfn_valid(pfn))
>> +            goto out_next;
>> +
>> +        page = pfn_to_online_page(pfn);
>> +        if (!page)
>> +            goto out_next;
>> +
>> +        folio = page_folio(page);
>> +        nr = folio_nr_pages(folio);
>> +        if (folio_nid(folio) != src_nid)
>> +            goto out_next;
>> +
>> +        if (!folio_test_lru(folio))
>> +            goto out_next;
>> +
>> +        if (pghot_get_hotness(pfn, &nid, &freq, &time))
> 
> Better to remove freq value, it’s not used later.

Yes, can be removed. I am leaving it here just for a future possibility
of having kmigrated being driven to migrate hottest pages on priority
compared to hot pages at which point frequency value may come in handy.

Regards,
Bharata.