diff mbox series

[v3,2/3] mm: Call arch_swap_restore() from unuse_pte()

Message ID 20230517022115.3033604-3-pcc@google.com
State New
Headers
Series mm: Fix bug affecting swapping in MTE tagged pages |

Commit Message

Peter Collingbourne May 17, 2023, 2:21 a.m. UTC 
  We would like to move away from requiring architectures to restore
metadata from swap in the set_pte_at() implementation, as this is not only
error-prone but adds complexity to the arch-specific code. This requires
us to call arch_swap_restore() before calling swap_free() whenever pages
are restored from swap. We are currently doing so everywhere except in
unuse_pte(); do so there as well.

Signed-off-by: Peter Collingbourne <pcc@google.com>
Link: https://linux-review.googlesource.com/id/I68276653e612d64cde271ce1b5a99ae05d6bbc4f
---
 mm/swapfile.c | 7 +++++++
 1 file changed, 7 insertions(+)

Comments

David Hildenbrand May 17, 2023, 8:37 a.m. UTC | #1 
On 17.05.23 04:21, Peter Collingbourne wrote:
> We would like to move away from requiring architectures to restore
> metadata from swap in the set_pte_at() implementation, as this is not only
> error-prone but adds complexity to the arch-specific code. This requires
> us to call arch_swap_restore() before calling swap_free() whenever pages
> are restored from swap. We are currently doing so everywhere except in
> unuse_pte(); do so there as well.
> 
> Signed-off-by: Peter Collingbourne <pcc@google.com>
> Link: https://linux-review.googlesource.com/id/I68276653e612d64cde271ce1b5a99ae05d6bbc4f
> ---
>   mm/swapfile.c | 7 +++++++
>   1 file changed, 7 insertions(+)
> 
> diff --git a/mm/swapfile.c b/mm/swapfile.c
> index 274bbf797480..e9843fadecd6 100644
> --- a/mm/swapfile.c
> +++ b/mm/swapfile.c
> @@ -1794,6 +1794,13 @@ static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
>   		goto setpte;
>   	}
>   
> +	/*
> +	 * Some architectures may have to restore extra metadata to the page
> +	 * when reading from swap. This metadata may be indexed by swap entry
> +	 * so this must be called before swap_free().
> +	 */
> +	arch_swap_restore(entry, page_folio(page));
> +
>   	/* See do_swap_page() */
>   	BUG_ON(!PageAnon(page) && PageMappedToDisk(page));
>   	BUG_ON(PageAnon(page) && PageAnonExclusive(page));

Acked-by: David Hildenbrand <david@redhat.com>
Steven Price May 17, 2023, 2:58 p.m. UTC | #2 
On 17/05/2023 03:21, Peter Collingbourne wrote:
> We would like to move away from requiring architectures to restore
> metadata from swap in the set_pte_at() implementation, as this is not only
> error-prone but adds complexity to the arch-specific code. This requires
> us to call arch_swap_restore() before calling swap_free() whenever pages
> are restored from swap. We are currently doing so everywhere except in
> unuse_pte(); do so there as well.
> 
> Signed-off-by: Peter Collingbourne <pcc@google.com>
> Link: https://linux-review.googlesource.com/id/I68276653e612d64cde271ce1b5a99ae05d6bbc4f

Reviewed-by: Steven Price <steven.price@arm.com>

> ---
>  mm/swapfile.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/mm/swapfile.c b/mm/swapfile.c
> index 274bbf797480..e9843fadecd6 100644
> --- a/mm/swapfile.c
> +++ b/mm/swapfile.c
> @@ -1794,6 +1794,13 @@ static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
>  		goto setpte;
>  	}
>  
> +	/*
> +	 * Some architectures may have to restore extra metadata to the page
> +	 * when reading from swap. This metadata may be indexed by swap entry
> +	 * so this must be called before swap_free().
> +	 */
> +	arch_swap_restore(entry, page_folio(page));
> +
>  	/* See do_swap_page() */
>  	BUG_ON(!PageAnon(page) && PageMappedToDisk(page));
>  	BUG_ON(PageAnon(page) && PageAnonExclusive(page));
Catalin Marinas May 19, 2023, 4:42 p.m. UTC | #3 
On Tue, May 16, 2023 at 07:21:12PM -0700, Peter Collingbourne wrote:
> We would like to move away from requiring architectures to restore
> metadata from swap in the set_pte_at() implementation, as this is not only
> error-prone but adds complexity to the arch-specific code. This requires
> us to call arch_swap_restore() before calling swap_free() whenever pages
> are restored from swap. We are currently doing so everywhere except in
> unuse_pte(); do so there as well.
> 
> Signed-off-by: Peter Collingbourne <pcc@google.com>
> Link: https://linux-review.googlesource.com/id/I68276653e612d64cde271ce1b5a99ae05d6bbc4f

Acked-by: Catalin Marinas <catalin.marinas@arm.com>
diff mbox series

Patch

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 274bbf797480..e9843fadecd6 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -1794,6 +1794,13 @@  static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
 		goto setpte;
 	}
 
+	/*
+	 * Some architectures may have to restore extra metadata to the page
+	 * when reading from swap. This metadata may be indexed by swap entry
+	 * so this must be called before swap_free().
+	 */
+	arch_swap_restore(entry, page_folio(page));
+
 	/* See do_swap_page() */
 	BUG_ON(!PageAnon(page) && PageMappedToDisk(page));
 	BUG_ON(PageAnon(page) && PageAnonExclusive(page));