Re: [PATCH] mm/gup: don't check page lru flag before draining it

[yangge1116]

在 2024/6/17 下午5:52, David Hildenbrand 写道:
> Why would we want to make folio_maybe_dma_pinned() detection that worse

Just want to fix it using the existing function, seems a little 
unreasonable. I will prepare the V2 using folio_test_lru(folio) to check.

static unsigned long collect_longterm_unpinnable_pages(...)
{
...
    if (!folio_test_lru(folio) && drain_allow) {
        lru_add_drain_all();
        drain_allow = false;
    }
...
}

void folio_mark_lazyfree(struct folio *folio)
{
    if (folio_test_anon(folio) && folio_test_swapbacked(folio) &&
       !folio_test_swapcache(folio) && !folio_test_unevictable(folio)) {
        struct folio_batch *fbatch;

        folio_get(folio);
        if (!folio_test_clear_lru(folio)) {
            folio_put(folio);
            return;
        }

        local_lock(&cpu_fbatches.lock);
        fbatch = this_cpu_ptr(&cpu_fbatches.lru_lazyfree);
        folio_batch_add_and_move(fbatch, folio, lru_lazyfree_fn);
        local_unlock(&cpu_fbatches.lock);
    }
}