Re: [PATCH RFC 07/11] mm/mremap: use ptep_get() for the destination PTE

Alexander Gordeev <[email protected]> Thu, 30 Jul 2026 13:29:24 +0200

Newsgroups	gmane.comp.freedesktop.xorg.drivers.intel,gmane.linux.kernel,gmane.comp.video.dri.devel,gmane.linux.ports.parisc,gmane.comp.emulators.xen.devel,gmane.linux.kernel.mm,gmane.linux.file-systems,gmane.linux.kernel.cross-arch,gmane.linux.kernel.bpf,gmane.linux.kernel.perf.user
Message-ID	<[email protected]>

On Mon, Jul 27, 2026 at 05:46:58PM +0100, Muhammad Usama Anjum wrote:
> move_ptes() directly dereferences new_ptep when checking that the
> destination slot is empty. new_ptep now points to hw_pte_t table storage,
> while pte_none() consumes a logical pte_t value.
> 
> Read the destination entry through the standard ptep_get() accessor before
> passing the logical value to pte_none().
> 
> Signed-off-by: Muhammad Usama Anjum <[email protected]>
> ---
>  mm/mremap.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/mremap.c b/mm/mremap.c
> index d966ee8248b6b..ecb90ccee6fdd 100644
> --- a/mm/mremap.c
> +++ b/mm/mremap.c
> @@ -264,7 +264,7 @@ static int move_ptes(struct pagetable_move_control *pmc,
>  
>  	for (; old_addr < old_end; old_ptep += nr_ptes, old_addr += nr_ptes * PAGE_SIZE,
>  		new_ptep += nr_ptes, new_addr += nr_ptes * PAGE_SIZE) {
> -		VM_WARN_ON_ONCE(!pte_none(*new_ptep));
> +		VM_WARN_ON_ONCE(!pte_none(ptep_get(new_ptep)));

Same as the previous patch, except that I did it already:
https://lore.kernel.org/linux-mm/[email protected]/

>  
>  		nr_ptes = 1;
>  		max_nr_ptes = (old_end - old_addr) >> PAGE_SHIFT;
> -- 
> 2.47.3
>