Re: [PATCH v6 1/5] mm: rmap: support batched checks of the references for large folios

On 3/16/26 07:25, Baolin Wang wrote:

On 3/10/26 4:17 PM, David Hildenbrand (Arm) wrote:

quoted

On 3/10/26 02:37, Baolin Wang wrote:

quoted

I understand your point. I’m concerned that I can’t test this patch on
every architecture to validate the benefits. Anyway, let me try this on
my X86 machine first.

In any case, please make that a follow-up patch :)

Sure. However, after investigating RISC‑V and x86, I found that
ptep_clear_flush_young() does not flush the TLB on these architectures:

int ptep_clear_flush_young(struct vm_area_struct *vma,
               unsigned long address, pte_t *ptep)
{
    /*
     * On x86 CPUs, clearing the accessed bit without a TLB flush
     * doesn't cause data corruption. [ It could cause incorrect
     * page aging and the (mistaken) reclaim of hot pages, but the
     * chance of that should be relatively low. ]
     *
     * So as a performance optimization don't flush the TLB when
     * clearing the accessed bit, it will eventually be flushed by
     * a context switch or a VM operation anyway. [ In the rare
     * event of it not getting flushed for a long time the delay
     * shouldn't really matter because there's no real memory
     * pressure for swapout to react to. ]
     */
    return ptep_test_and_clear_young(vma, address, ptep);
}

You'd probably want an arch helper then, that tells you whether
a flush_tlb_range() after ptep_test_and_clear_young() is required.

Or some special flush_tlb_range() helper.

I agree that it requires more work.

-- 
Cheers,

David