diff mbox series

mm/memcg: use get_page() for device private pages in mc_handle_swap_pte()

Message ID 20230715032802.2508163-1-linmiaohe@huawei.com
State New
Headers
Series mm/memcg: use get_page() for device private pages in mc_handle_swap_pte() |

Commit Message

Miaohe Lin July 15, 2023, 3:28 a.m. UTC 
  When page table locked is held, the page can't be freed from under us.
So use get_page() to get the extra page reference to simplify the code.
No functional change intended.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
---
 mm/memcontrol.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Comments

Matthew Wilcox July 15, 2023, 3:56 a.m. UTC | #1 
On Sat, Jul 15, 2023 at 11:28:02AM +0800, Miaohe Lin wrote:
> When page table locked is held, the page can't be freed from under us.

But the page isn't mapped into the page table ... there's a swap entry
in the page table, so I don't think your logic holds.

> So use get_page() to get the extra page reference to simplify the code.
> No functional change intended.
> 
> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
> ---
>  mm/memcontrol.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 93e3cc581b51..4ca382efb1ca 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -5670,8 +5670,9 @@ static struct page *mc_handle_swap_pte(struct vm_area_struct *vma,
>  	 */
>  	if (is_device_private_entry(ent)) {
>  		page = pfn_swap_entry_to_page(ent);
> -		if (!get_page_unless_zero(page))
> -			return NULL;
> +		/* Get a page reference while we know the page can't be freed. */
> +		get_page(page);
> +
>  		return page;
>  	}
>  
> -- 
> 2.33.0
> 
>
Miaohe Lin July 17, 2023, 2:28 a.m. UTC | #2 
On 2023/7/15 11:56, Matthew Wilcox wrote:
> On Sat, Jul 15, 2023 at 11:28:02AM +0800, Miaohe Lin wrote:
>> When page table locked is held, the page can't be freed from under us.
> 
> But the page isn't mapped into the page table ... there's a swap entry
> in the page table, so I don't think your logic holds.
> 

IIUC, device_private_entry will hold one page refcnt when it's set to page table.
And there's similar code in do_swap_page():

  vm_fault_t do_swap_page(struct vm_fault *vmf)
    if (unlikely(non_swap_entry(entry))) {
      if (is_device_private_entry(entry))
        /*
         * Get a page reference while we know the page can't be
         * freed.
         */
        get_page(vmf->page);
        pte_unmap_unlock(vmf->pte, vmf->ptl);
        ret = vmf->page->pgmap->ops->migrate_to_ram(vmf);
        put_page(vmf->page);
    ...

If my logic doesn't hold, do_swap_page() will need to fix the code. Or am I miss something?

Thanks Matthew.
Miaohe Lin July 23, 2023, 1:16 a.m. UTC | #3 
On 2023/7/17 10:28, Miaohe Lin wrote:
> On 2023/7/15 11:56, Matthew Wilcox wrote:
>> On Sat, Jul 15, 2023 at 11:28:02AM +0800, Miaohe Lin wrote:
>>> When page table locked is held, the page can't be freed from under us.
>>
>> But the page isn't mapped into the page table ... there's a swap entry
>> in the page table, so I don't think your logic holds.
>>
> 
> IIUC, device_private_entry will hold one page refcnt when it's set to page table.

Take remove_migration_pte() as example, it will hold extra one page refcnt when set device private entry:
  remove_migration_pte()
    ...
    folio_get(folio);
    ...
    if (unlikely(is_device_private_page(new))) {
      make_[writable|readable]_device_private_entry();
    }
    ...
    set_pte_at

> And there's similar code in do_swap_page():
> 
>   vm_fault_t do_swap_page(struct vm_fault *vmf)
>     if (unlikely(non_swap_entry(entry))) {
>       if (is_device_private_entry(entry))
>         /*
>          * Get a page reference while we know the page can't be
>          * freed.
>          */
>         get_page(vmf->page);
>         pte_unmap_unlock(vmf->pte, vmf->ptl);
>         ret = vmf->page->pgmap->ops->migrate_to_ram(vmf);
>         put_page(vmf->page);
>     ...
> 
> If my logic doesn't hold, do_swap_page() will need to fix the code. Or am I miss something?

Can I have your opinion?

Thanks.
diff mbox series

Patch

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 93e3cc581b51..4ca382efb1ca 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -5670,8 +5670,9 @@  static struct page *mc_handle_swap_pte(struct vm_area_struct *vma,
 	 */
 	if (is_device_private_entry(ent)) {
 		page = pfn_swap_entry_to_page(ent);
-		if (!get_page_unless_zero(page))
-			return NULL;
+		/* Get a page reference while we know the page can't be freed. */
+		get_page(page);
+
 		return page;
 	}