Re: [PATCH] mm: pgtable: protect lockless kernel page table walks with RCU

[Matthew Wilcox]

On Fri, Jun 12, 2026 at 05:38:27AM +0100, David Carlier wrote:
> @@ -434,6 +434,14 @@ static void kernel_pgtable_work_func(struct work_struct *work)
>  	spin_unlock(&kernel_pgtable_work.lock);
>  
>  	iommu_sva_invalidate_kva_range(PAGE_OFFSET, TLB_FLUSH_ALL);
> +
> +	/*
> +	 * Lockless kernel page table walkers (ptdump, and any other user of
> +	 * walk_kernel_page_table_range_lockless()) dereference these pages
> +	 * under rcu_read_lock(). Wait for a grace period so no walker can
> +	 * still be reading a page we are about to free.
> +	 */
> +	synchronize_rcu();
>  	list_for_each_entry_safe(pt, next, &page_list, pt_list)
>  		__pagetable_free(pt);

synchronize_rcu() is rather expensive.  Can't you rcu-free the page
tables instead?  There's an rcu head in struct page.