[PATCH v3 0/2] mm/swapfile.c and swap.h cleanup

[PATCH v3 0/2] mm/swapfile.c and swap.h cleanup

[Chris Li]

This patch series, which builds on Kairui's swap improve cluster scan series.
https://lore.kernel.org/linux-mm/20250806161748.76651-1-ryncsn@gmail.com/

It introduces a new function, alloc_swap_scan_list(), for swapfile.c.

It also cleans up swap.h by removing comments that reference fields that
have been deleted.

There are no functional changes in this two-patch series.

Signed-off-by: Chris Li <chrisl@kernel.org>
---
Changes in v3:
- Fix a few scan_all argument to match having while loop or not.
- Update comment message.
- Add a ptach to clean up the comments in swap.h
- Link to v2: https://lore.kernel.org/r/20250808-swap-scan-list-v2-1-d50e4758ecee@kernel.org

Changes in v2:
- Adjust change base on Andrew's feedback on int type and break out of
  loop.
- Link to v1: https://lore.kernel.org/r/20250806-swap-scan-list-v1-1-a5fe2d9340a2@kernel.org

---
Chris Li (2):
      mm/swapfile.c: introduce function alloc_swap_scan_list()
      mm: swap.h: Remove deleted field from comments

 include/linux/swap.h |  7 ++---
 mm/swapfile.c        | 86 ++++++++++++++++++++++++++++------------------------
 2 files changed, 49 insertions(+), 44 deletions(-)
---
base-commit: f89484324d5876ee10765fa61da0332899fa1a6a
change-id: 20250806-swap-scan-list-2b89e3424b0a

Best regards,
-- 
Chris Li <chrisl@kernel.org>

[PATCH v3 1/2] mm/swapfile.c: introduce function alloc_swap_scan_list()

[Chris Li]

alloc_swap_scan_list() will scan the whole list or the first cluster.

This reduces the repeat patterns of isolating a cluster then scanning that
cluster. As a result, cluster_alloc_swap_entry() is shorter and shallower.

No functional change.

Signed-off-by: Chris Li <chrisl@kernel.org>
---
 mm/swapfile.c | 86 ++++++++++++++++++++++++++++++++---------------------------
 1 file changed, 47 insertions(+), 39 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 4a0cf4fb348d..a7ffabbe65ef 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -820,6 +820,29 @@ static unsigned int alloc_swap_scan_cluster(struct swap_info_struct *si,
 	return found;
 }
 
+static unsigned int alloc_swap_scan_list(struct swap_info_struct *si,
+					 struct list_head *list,
+					 unsigned int order,
+					 unsigned char usage,
+					 bool scan_all)
+{
+	unsigned int found = SWAP_ENTRY_INVALID;
+
+	do {
+		struct swap_cluster_info *ci = isolate_lock_cluster(si, list);
+		unsigned long offset;
+
+		if (!ci)
+			break;
+		offset = cluster_offset(si, ci);
+		found = alloc_swap_scan_cluster(si, ci, offset, order, usage);
+		if (found)
+			break;
+	} while (scan_all);
+
+	return found;
+}
+
 static void swap_reclaim_full_clusters(struct swap_info_struct *si, bool force)
 {
 	long to_scan = 1;
@@ -913,32 +936,24 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si, int o
 	 * to spread out the writes.
 	 */
 	if (si->flags & SWP_PAGE_DISCARD) {
-		ci = isolate_lock_cluster(si, &si->free_clusters);
-		if (ci) {
-			found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
-							order, usage);
-			if (found)
-				goto done;
-		}
+		found = alloc_swap_scan_list(si, &si->free_clusters, order, usage,
+					     false);
+		if (found)
+			goto done;
 	}
 
 	if (order < PMD_ORDER) {
-		while ((ci = isolate_lock_cluster(si, &si->nonfull_clusters[order]))) {
-			found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
-							order, usage);
-			if (found)
-				goto done;
-		}
+		found = alloc_swap_scan_list(si, &si->nonfull_clusters[order],
+					     order, usage, true);
+		if (found)
+			goto done;
 	}
 
 	if (!(si->flags & SWP_PAGE_DISCARD)) {
-		ci = isolate_lock_cluster(si, &si->free_clusters);
-		if (ci) {
-			found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
-							order, usage);
-			if (found)
-				goto done;
-		}
+		found = alloc_swap_scan_list(si, &si->free_clusters, order, usage,
+					     false);
+		if (found)
+			goto done;
 	}
 
 	/* Try reclaim full clusters if free and nonfull lists are drained */
@@ -952,13 +967,10 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si, int o
 		 * failure is not critical. Scanning one cluster still
 		 * keeps the list rotated and reclaimed (for HAS_CACHE).
 		 */
-		ci = isolate_lock_cluster(si, &si->frag_clusters[order]);
-		if (ci) {
-			found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
-							order, usage);
-			if (found)
-				goto done;
-		}
+		found = alloc_swap_scan_list(si, &si->frag_clusters[order], order,
+					     usage, false);
+		if (found)
+			goto done;
 	}
 
 	/*
@@ -977,19 +989,15 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si, int o
 		 * Clusters here have at least one usable slots and can't fail order 0
 		 * allocation, but reclaim may drop si->lock and race with another user.
 		 */
-		while ((ci = isolate_lock_cluster(si, &si->frag_clusters[o]))) {
-			found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
-							0, usage);
-			if (found)
-				goto done;
-		}
+		found = alloc_swap_scan_list(si, &si->frag_clusters[o],
+					     0, usage, true);
+		if (found)
+			goto done;
 
-		while ((ci = isolate_lock_cluster(si, &si->nonfull_clusters[o]))) {
-			found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
-							0, usage);
-			if (found)
-				goto done;
-		}
+		found = alloc_swap_scan_list(si, &si->nonfull_clusters[o],
+					     0, usage, true);
+		if (found)
+			goto done;
 	}
 done:
 	if (!(si->flags & SWP_SOLIDSTATE))

-- 
2.43.0

Re: [PATCH v3 1/2] mm/swapfile.c: introduce function alloc_swap_scan_list()

[Kairui Song]

On Tue, Aug 12, 2025 at 5:13 PM Chris Li <chrisl@kernel.org> wrote:
>
> alloc_swap_scan_list() will scan the whole list or the first cluster.
>
> This reduces the repeat patterns of isolating a cluster then scanning that
> cluster. As a result, cluster_alloc_swap_entry() is shorter and shallower.
>
> No functional change.
>
> Signed-off-by: Chris Li <chrisl@kernel.org>
> ---
>  mm/swapfile.c | 86 ++++++++++++++++++++++++++++++++---------------------------
>  1 file changed, 47 insertions(+), 39 deletions(-)
>
> diff --git a/mm/swapfile.c b/mm/swapfile.c
> index 4a0cf4fb348d..a7ffabbe65ef 100644
> --- a/mm/swapfile.c
> +++ b/mm/swapfile.c
> @@ -820,6 +820,29 @@ static unsigned int alloc_swap_scan_cluster(struct swap_info_struct *si,
>         return found;
>  }
>
> +static unsigned int alloc_swap_scan_list(struct swap_info_struct *si,
> +                                        struct list_head *list,
> +                                        unsigned int order,
> +                                        unsigned char usage,
> +                                        bool scan_all)
> +{
> +       unsigned int found = SWAP_ENTRY_INVALID;
> +
> +       do {
> +               struct swap_cluster_info *ci = isolate_lock_cluster(si, list);
> +               unsigned long offset;
> +
> +               if (!ci)
> +                       break;
> +               offset = cluster_offset(si, ci);
> +               found = alloc_swap_scan_cluster(si, ci, offset, order, usage);
> +               if (found)
> +                       break;
> +       } while (scan_all);
> +
> +       return found;
> +}
> +
>  static void swap_reclaim_full_clusters(struct swap_info_struct *si, bool force)
>  {
>         long to_scan = 1;
> @@ -913,32 +936,24 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si, int o
>          * to spread out the writes.
>          */
>         if (si->flags & SWP_PAGE_DISCARD) {
> -               ci = isolate_lock_cluster(si, &si->free_clusters);
> -               if (ci) {
> -                       found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
> -                                                       order, usage);
> -                       if (found)
> -                               goto done;
> -               }
> +               found = alloc_swap_scan_list(si, &si->free_clusters, order, usage,
> +                                            false);
> +               if (found)
> +                       goto done;
>         }
>
>         if (order < PMD_ORDER) {
> -               while ((ci = isolate_lock_cluster(si, &si->nonfull_clusters[order]))) {
> -                       found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
> -                                                       order, usage);
> -                       if (found)
> -                               goto done;
> -               }
> +               found = alloc_swap_scan_list(si, &si->nonfull_clusters[order],
> +                                            order, usage, true);
> +               if (found)
> +                       goto done;
>         }
>
>         if (!(si->flags & SWP_PAGE_DISCARD)) {
> -               ci = isolate_lock_cluster(si, &si->free_clusters);
> -               if (ci) {
> -                       found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
> -                                                       order, usage);
> -                       if (found)
> -                               goto done;
> -               }
> +               found = alloc_swap_scan_list(si, &si->free_clusters, order, usage,
> +                                            false);
> +               if (found)
> +                       goto done;
>         }
>
>         /* Try reclaim full clusters if free and nonfull lists are drained */
> @@ -952,13 +967,10 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si, int o
>                  * failure is not critical. Scanning one cluster still
>                  * keeps the list rotated and reclaimed (for HAS_CACHE).
>                  */
> -               ci = isolate_lock_cluster(si, &si->frag_clusters[order]);
> -               if (ci) {
> -                       found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
> -                                                       order, usage);
> -                       if (found)
> -                               goto done;
> -               }
> +               found = alloc_swap_scan_list(si, &si->frag_clusters[order], order,
> +                                            usage, false);
> +               if (found)
> +                       goto done;
>         }
>
>         /*
> @@ -977,19 +989,15 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si, int o
>                  * Clusters here have at least one usable slots and can't fail order 0
>                  * allocation, but reclaim may drop si->lock and race with another user.
>                  */
> -               while ((ci = isolate_lock_cluster(si, &si->frag_clusters[o]))) {
> -                       found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
> -                                                       0, usage);
> -                       if (found)
> -                               goto done;
> -               }
> +               found = alloc_swap_scan_list(si, &si->frag_clusters[o],
> +                                            0, usage, true);
> +               if (found)
> +                       goto done;
>
> -               while ((ci = isolate_lock_cluster(si, &si->nonfull_clusters[o]))) {
> -                       found = alloc_swap_scan_cluster(si, ci, cluster_offset(si, ci),
> -                                                       0, usage);
> -                       if (found)
> -                               goto done;
> -               }
> +               found = alloc_swap_scan_list(si, &si->nonfull_clusters[o],
> +                                            0, usage, true);
> +               if (found)
> +                       goto done;
>         }
>  done:
>         if (!(si->flags & SWP_SOLIDSTATE))
>
> --
> 2.43.0
>

I've been testing on top of a locally updated version of V2 for about
two days, and it's identical to this one. This looks great, thanks!

Reviewed-by: Kairui Song <kasong@tencent.com>

Re: [PATCH v3 1/2] mm/swapfile.c: introduce function alloc_swap_scan_list()

[Chris Li]

On Tue, Aug 12, 2025 at 9:57 AM Kairui Song <ryncsn@gmail.com> wrote:
>
> I've been testing on top of a locally updated version of V2 for about
> two days, and it's identical to this one. This looks great, thanks!

Thanks for the review.

I have been running the swap stress tests on kernel compile as well.
No issue so far.

Chris

Re: [PATCH v3 1/2] mm/swapfile.c: introduce function alloc_swap_scan_list()

[Nhat Pham]

On Tue, Aug 12, 2025 at 12:10 AM Chris Li <chrisl@kernel.org> wrote:
>
> alloc_swap_scan_list() will scan the whole list or the first cluster.
>
> This reduces the repeat patterns of isolating a cluster then scanning that
> cluster. As a result, cluster_alloc_swap_entry() is shorter and shallower.
>
> No functional change.
>
> Signed-off-by: Chris Li <chrisl@kernel.org>

Code seems more readable now, IMO. Thanks, Chris!

Acked-by: Nhat Pham <nphamcs@gmail.com>

[PATCH v3 2/2] mm: swap.h: Remove deleted field from comments

[Chris Li]

The comment for struct swap_info_struct.lock incorrectly mentions fields
that have already been deleted from the structure.

Updates the comments to accurately reflect the current struct
swap_info_struct.

There is no functional change.

Signed-off-by: Chris Li <chrisl@kernel.org>
---
 include/linux/swap.h | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index a060d102e0d1..c2da85cb7fe7 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -320,11 +320,8 @@ struct swap_info_struct {
 	struct completion comp;		/* seldom referenced */
 	spinlock_t lock;		/*
 					 * protect map scan related fields like
-					 * swap_map, lowest_bit, highest_bit,
-					 * inuse_pages, cluster_next,
-					 * cluster_nr, lowest_alloc,
-					 * highest_alloc, free/discard cluster
-					 * list. other fields are only changed
+					 * swap_map, inuse_pages and all cluster
+					 * lists. other fields are only changed
 					 * at swapon/swapoff, so are protected
 					 * by swap_lock. changing flags need
 					 * hold this lock and swap_lock. If

-- 
2.43.0

Re: [PATCH v3 2/2] mm: swap.h: Remove deleted field from comments

[Kairui Song]

On Tue, Aug 12, 2025 at 3:26 PM Chris Li <chrisl@kernel.org> wrote:
>
> The comment for struct swap_info_struct.lock incorrectly mentions fields
> that have already been deleted from the structure.
>
> Updates the comments to accurately reflect the current struct
> swap_info_struct.
>
> There is no functional change.
>
> Signed-off-by: Chris Li <chrisl@kernel.org>
> ---
>  include/linux/swap.h | 7 ++-----
>  1 file changed, 2 insertions(+), 5 deletions(-)
>
> diff --git a/include/linux/swap.h b/include/linux/swap.h
> index a060d102e0d1..c2da85cb7fe7 100644
> --- a/include/linux/swap.h
> +++ b/include/linux/swap.h
> @@ -320,11 +320,8 @@ struct swap_info_struct {
>         struct completion comp;         /* seldom referenced */
>         spinlock_t lock;                /*
>                                          * protect map scan related fields like
> -                                        * swap_map, lowest_bit, highest_bit,
> -                                        * inuse_pages, cluster_next,
> -                                        * cluster_nr, lowest_alloc,
> -                                        * highest_alloc, free/discard cluster
> -                                        * list. other fields are only changed
> +                                        * swap_map, inuse_pages and all cluster
> +                                        * lists. other fields are only changed
>                                          * at swapon/swapoff, so are protected
>                                          * by swap_lock. changing flags need
>                                          * hold this lock and swap_lock. If
>
> --
> 2.43.0
>

Thanks.

Reviewed-by: Kairui Song <kasong@tencent.com>

Re: [PATCH v3 2/2] mm: swap.h: Remove deleted field from comments

[Barry Song]

On Tue, Aug 12, 2025 at 3:11 PM Chris Li <chrisl@kernel.org> wrote:
>
> The comment for struct swap_info_struct.lock incorrectly mentions fields
> that have already been deleted from the structure.
>
> Updates the comments to accurately reflect the current struct
> swap_info_struct.
>
> There is no functional change.
>
> Signed-off-by: Chris Li <chrisl@kernel.org>

Reviewed-by: Barry Song <baohua@kernel.org>

Thanks
Barry

Re: [PATCH v3 2/2] mm: swap.h: Remove deleted field from comments

[Nhat Pham]

On Tue, Aug 12, 2025 at 12:11 AM Chris Li <chrisl@kernel.org> wrote:
>
> The comment for struct swap_info_struct.lock incorrectly mentions fields
> that have already been deleted from the structure.
>
> Updates the comments to accurately reflect the current struct
> swap_info_struct.
>
> There is no functional change.
>
> Signed-off-by: Chris Li <chrisl@kernel.org>

Acked-by: Nhat Pham <nphamcs@gmail.com>