[PATCH v2 00/12] mm, swap: extendable swap devices (xswap)

[PATCH v2 00/12] mm, swap: extendable swap devices (xswap)

[Baoquan He]

Xswap is a swap device with no backing storage. Swap out pages live
in zswap. Its cluster_info[] array lives in a VM_SPARSE vmalloc area,
and the area is grown and shrunk on demand as swap usage changes.

The problem being solved is the static size of compressed swap. Both
zram and zswap need the size fixed in advance, and neither gives memory
back when the workload shrinks. The solution should be a device whose
size can scale up/down as per usage. Xswap does that by mapping the
metadata lazily instead of reserving it for the whole range.

Design
------
- si->cluster_info[] stays a plain array. Access is still
  &si->cluster_info[offset / SWAPFILE_CLUSTER]: no per-access branch, no
  RCU discipline, no tear-down state machine, no NULL return.
- Only an initial chunk is mapped at creation. The rest of the address
  space is reserved, not allocated, so an idle device costs nothing.
- Growth is driven by allocation. When no free cluster is left and the
  address space has room, the next chunk is mapped and added to the free
  list. No userspace involvement.
- Shrink is driven by frees. The free tail is counted, and whole chunks
  are unmapped once the mapped range is at most half in use and several
  chunks can go. One chunk is left mapped as slack, so the next
  allocation does not map it straight back.

Size
----
A device starts at 1xRAM, rounded down to the cluster. That costs
nothing, because the mapping is lazy. An optional per-device cap,
/sys/kernel/mm/xswap/type<N>/limit, lets an admin lower the ceiling.
Grow and shrink both work without it.

Interface
---------
  /sys/kernel/mm/xswap/create           write an optional priority
  /sys/kernel/mm/xswap/destroy          write a swap type
  /sys/kernel/mm/xswap/type<N>/limit    read/write, in pages
The device shows up in /proc/swaps as xswap<N>.

Note:
--------------------------
Writeback, rmap lookup, etc. Those are consumers of this base. I have
a writeback prototype on top of this base and will post it as a reference.

Testing
-------
x86_64, 8 GiB VM, lockdep and PROVE_RCU on, 4 and 8 GiB MADV_PAGEOUT
workloads against a zswap-backed device:
- grow from 36 clusters up to the 3974-cluster limit, and stop there
- shrink on workload exit, then grow again
- fill the device to the limit and shrink again
- create, swapoff, destroy, recreate
Throughput is within ~2-3% of plain swap+zswap on a 64G/64-thread
swapout, so it is dominated by zswap compression, not the cluster table.

Changelog
=========
- Patch 1 (mm: zswap: return -ENOENT when the swap device is gone) is not
  part of this series; it was posted separately.

- There is only one size knob now. The runtime ceiling and the debugfs
  per-device limit are gone. All that is left is the optional per-device
  cap, /sys/kernel/mm/xswap/type<N>/limit. Grow and shrink work without
  it.

- The shrink no longer keeps its own count of the free tail. It scans the
  tail instead, and dropping the counter also removes a call from the
  cluster allocation path. 

- The priority is no longer a patch of its own. The create attribute
  takes it:
    echo 100 > /sys/kernel/mm/xswap/create

RFC v3 -> v1
- Add patch 16 to support setting xswap device priority at creation.
  The create sysfs interface (/sys/kernel/mm/xswap/create) previously
  hardcoded every new device's priority to DEF_SWAP_PRIO, it now
  accepts an optional priority:

    echo "<percent> [<prio>]" > /sys/kernel/mm/xswap/create

- Bug fix: xswap_lock init ordering. mutex_init(&si->xswap_lock) was called
  after xswap_map_clusters() (which locks it), i.e. locking an uninitialized
  mutex. Init now before the first xswap_map_clusters() call. Thanks to Klara.

- Bug fix: Fixes a compile error in !CONFIG_XSWAP builds. xswap_debugfs_root
  is declared inside CONFIG_XSWAP ifdeffery scope, so the ungarded use
  caused error when CONFIG_XSWAP is off.

RFC v2-> RFC v3:
- Replace the "header-only swap file + swapon" creation hack with a
  proper file-less device created and destroyed via sysfs
  (/sys/kernel/mm/xswap/{create,destroy}).  This required the
  __swapoff() refactor and the free_swap_cluster_info() signature
  change (patches 4, 6, 14).

- Require zswap: refuse to create an xswap device when zswap is
  unavailable (patch 15).

- Split the unrelated zswap -ENOENT fix out of the series into a
  standalone patch (patch 1).

- Fix nr_free_tail over-counting on concurrent grow, shrink leaking
  detached clusters on early bail-out, a re-init race on cluster
  spinlocks in xswap_map_clusters(), the nr_clusters_mapped update
  ordering, and swapoff accessing the shrinker-unmapped cluster tail.

- Minor cleanups (checkpatch, /proc/swaps alignment, commit messages).

RFC v1-> RFC v2:
- Added __GFP_HIGH | __GFP_NOMEMALLOC to alloc_page() and kmalloc_array()
  in the grow path, plus memalloc_noreclaim_save/restore() wrapping,
  to prevent the grow path from consuming emergency memory reserves
  or recursing into swap under PF_MEMALLOC. This is folded into patch 3.
  This was pointed out by Nhat.

- Folded the mutex serialization fix into the cluster grow patch (patch
  3). This is suggested by Nhat.

- Fixed coding style issues: corrected indentation of declarations in
  xswap_unmap_clusters(), removed unnecessary block scope around the
  err variable in xswap_map_clusters().

- Rebased onto mm-unstable

Baoquan He (11):
  mm, swap: add CONFIG_XSWAP and xswap fields to swap_info_struct
  mm, swap: refactor free_swap_cluster_info to take swap_info_struct
  mm, swap: add xswap cluster grow via VM_SPARSE vmalloc
  mm, swap: add sysfs create interface for xswap
  mm, swap: add xswap grow trigger on cluster allocation
  mm, swap: add xswap_try_shrink and shrink trigger on cluster free
  mm, swap: free backing pages in xswap_unmap_clusters
  mm, swap: defer xswap shrink to workqueue to avoid lock recursion
  mm, swap: refactor swapoff + add xswap_destroy
  mm, swap: require zswap for xswap devices
  mm, swap: add sysfs per-device size limit for xswap

Chris Li (1):
  mm: xswap support for zswap

 include/linux/swap.h |   13 +
 mm/Kconfig           |    9 +
 mm/page_io.c         |   16 +
 mm/swap_state.c      |    7 +
 mm/swapfile.c        | 1170 ++++++++++++++++++++++++++++++++++++++----
 mm/zswap.c           |    7 +-
 6 files changed, 1118 insertions(+), 104 deletions(-)

-- 
2.54.0

[PATCH v2 01/12] mm: xswap support for zswap

[Baoquan He]

From: Chris Li <chrisl@kernel.org>

Introduce extendable swap device support - xswap.

An xswap device has no backing storage and no swap data section, so
it wastes no disk space. Creation is via a sysfs interface added in a
later patch.

Zswap writeback is gated on whether a real (non-xswap) swap device is
active. nr_real_swapfiles counts such devices and is maintained at
swapon/swapoff only, so the gate reflects "a device exists to write
back to" rather than "a device currently has free slots". This keeps
writeback working even when the real swap device is full, and avoids
a double decrement when a full device is swapped off.

Co-developed-by: Baoquan He <hebaoquan@kylinos.cn>
Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
Signed-off-by: Chris Li <chrisl@kernel.org>
---
 include/linux/swap.h |  2 ++
 mm/page_io.c         | 16 ++++++++++++++++
 mm/swap_state.c      |  7 +++++++
 mm/swapfile.c        | 38 +++++++++++++++++++++++++++++++++++---
 mm/zswap.c           |  7 ++++++-
 5 files changed, 66 insertions(+), 4 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index 5658a1634b85..787fe463dcbb 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -207,6 +207,7 @@ enum {
 	SWP_STABLE_WRITES = (1 << 11),	/* no overwrite PG_writeback pages */
 	SWP_SYNCHRONOUS_IO = (1 << 12),	/* synchronous IO is efficient */
 	SWP_HIBERNATION = (1 << 13),	/* pinned for hibernation */
+	SWP_XSWAP	= (1 << 14),	/* extendable swap device */
 					/* add others here before... */
 };
 
@@ -356,6 +357,7 @@ void free_folio_and_swap_cache(struct folio *folio);
 void free_pages_and_swap_cache(struct encoded_page **, int);
 /* linux/mm/swapfile.c */
 extern atomic_long_t nr_swap_pages;
+extern atomic_t nr_real_swapfiles;
 extern long total_swap_pages;
 extern atomic_t nr_rotate_swap;
 
diff --git a/mm/page_io.c b/mm/page_io.c
index 88962571cb93..5483c943e3e3 100644
--- a/mm/page_io.c
+++ b/mm/page_io.c
@@ -248,6 +248,17 @@ int swap_writeout(struct swap_io_ctx *ctx, struct folio *folio)
 	}
 	rcu_read_unlock();
 
+	/*
+	 * ctx->sis is set by swap_add_folio() which is called from
+	 * __swap_writepage() below.  Since we must avoid the writepage
+	 * path for xswap devices, use the swap_info from the folio's
+	 * swap entry directly instead of going through ctx.
+	 */
+	if (unlikely(__swap_entry_to_info(folio->swap)->flags & SWP_XSWAP)) {
+		folio_mark_dirty(folio);
+		return AOP_WRITEPAGE_ACTIVATE;
+	}
+
 	__swap_writepage(ctx, folio);
 	return 0;
 out_unlock:
@@ -480,6 +491,11 @@ void swap_read_folio(struct swap_io_ctx *ctx, struct folio *folio)
 	if (zswap_load(folio) != -ENOENT)
 		goto finish;
 
+	if (unlikely(sis->flags & SWP_XSWAP)) {
+		folio_unlock(folio);
+		goto finish;
+	}
+
 	/* We have to read from slower devices. Increase zswap protection. */
 	zswap_folio_swapin(folio);
 	swap_add_folio(ctx, folio, READ);
diff --git a/mm/swap_state.c b/mm/swap_state.c
index b76eb3d876fd..2eedb7a3d7bb 100644
--- a/mm/swap_state.c
+++ b/mm/swap_state.c
@@ -830,6 +830,13 @@ struct folio *swap_cluster_readahead(swp_entry_t entry, gfp_t gfp_mask,
 	struct blk_plug plug;
 	swp_entry_t ra_entry;
 
+	/*
+	 * The entry may have been freed by another task. Avoid swap_info_get()
+	 * which will print error message if the race happens.
+	 */
+	if (si->flags & SWP_XSWAP)
+		goto skip;
+
 	mask = swapin_nr_pages(offset) - 1;
 	if (!mask)
 		goto skip;
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 53bf01d5f7f1..193b08a54908 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -66,6 +66,7 @@ static void move_cluster(struct swap_info_struct *si,
 static DEFINE_SPINLOCK(swap_lock);
 static unsigned int nr_swapfiles;
 atomic_long_t nr_swap_pages;
+atomic_t nr_real_swapfiles;
 /*
  * Some modules use swappable objects and may try to swap them out under
  * memory pressure (via the shrinker). Before doing so, they may wish to
@@ -1208,6 +1209,9 @@ static void del_from_avail_list(struct swap_info_struct *si, bool swapoff)
 		 */
 		lockdep_assert_held(&si->lock);
 		si->flags &= ~SWP_WRITEOK;
+		/* Count active devices, not merely those on the avail list. */
+		if (!(si->flags & SWP_XSWAP))
+			atomic_sub(1, &nr_real_swapfiles);
 		atomic_long_or(SWAP_USAGE_OFFLIST_BIT, &si->inuse_pages);
 	} else {
 		/*
@@ -1265,6 +1269,8 @@ static void add_to_avail_list(struct swap_info_struct *si, bool swapon)
 	}
 
 	plist_add(&si->avail_list, &swap_avail_head);
+	if (swapon && !(si->flags & SWP_XSWAP))
+		atomic_add(1, &nr_real_swapfiles);
 
 skip:
 	spin_unlock(&swap_avail_lock);
@@ -2959,6 +2965,19 @@ static int setup_swap_extents(struct swap_info_struct *sis,
 	struct inode *inode = mapping->host;
 	int ret;
 
+	if (sis->flags & SWP_XSWAP) {
+		*span = 0;
+		/*
+		 * xswap devices have no backing block device and
+		 * physical writeout is skipped in swap_writeout(),
+		 * but sis->ops must still be set so that callers
+		 * like shrink_folio_list() can safely dereference
+		 * ops->flags.
+		 */
+		sis->ops = &swap_bdev_ops;
+		return 0;
+	}
+
 	ret = sio_pool_init();
 	if (ret)
 		return ret;
@@ -3167,7 +3186,8 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 
 	destroy_swap_extents(p, p->swap_file);
 
-	if (!(p->flags & SWP_SOLIDSTATE))
+	if (!(p->flags & SWP_XSWAP) &&
+	    !(p->flags & SWP_SOLIDSTATE))
 		atomic_dec(&nr_rotate_swap);
 
 	mutex_lock(&swapon_mutex);
@@ -3277,6 +3297,19 @@ static void swap_stop(struct seq_file *swap, void *v)
 	mutex_unlock(&swapon_mutex);
 }
 
+static const char *swap_type_str(struct swap_info_struct *si)
+{
+	struct file *file = si->swap_file;
+
+	if (si->flags & SWP_XSWAP)
+		return "xswap\t";
+
+	if (S_ISBLK(file_inode(file)->i_mode))
+		return "partition";
+
+	return "file\t";
+}
+
 static int swap_show(struct seq_file *swap, void *v)
 {
 	struct swap_info_struct *si = v;
@@ -3296,8 +3329,7 @@ static int swap_show(struct seq_file *swap, void *v)
 	len = seq_file_path(swap, file, " \t\n\\");
 	seq_printf(swap, "%*s%s\t%lu\t%s%lu\t%s%d\n",
 			len < 40 ? 40 - len : 1, " ",
-			S_ISBLK(file_inode(file)->i_mode) ?
-				"partition" : "file\t",
+			swap_type_str(si),
 			bytes, bytes < 10000000 ? "\t" : "",
 			inuse, inuse < 10000000 ? "\t" : "",
 			si->prio);
diff --git a/mm/zswap.c b/mm/zswap.c
index b9948d4657d2..064970a4393f 100644
--- a/mm/zswap.c
+++ b/mm/zswap.c
@@ -1000,6 +1000,11 @@ static int zswap_writeback_entry(struct zswap_entry *entry,
 	if (!si)
 		return -ENOENT;
 
+	if (si->flags & SWP_XSWAP) {
+		put_swap_device(si);
+		return -EINVAL;
+	}
+
 	mpol = get_task_policy(current);
 	folio = swap_cache_alloc_folio(swpentry, GFP_KERNEL, BIT(0), NULL, mpol,
 				       NO_INTERLEAVE_INDEX);
@@ -1545,7 +1550,7 @@ bool zswap_store(struct folio *folio)
 	zswap_pool_put(pool);
 put_objcg:
 	obj_cgroup_put(objcg);
-	if (!ret && zswap_pool_reached_full)
+	if (!ret && zswap_pool_reached_full && atomic_read(&nr_real_swapfiles))
 		queue_work(shrink_wq, &zswap_shrink_work);
 check_old:
 	/*
-- 
2.54.0

[PATCH v2 02/12] mm, swap: add CONFIG_XSWAP and xswap fields to swap_info_struct

[Baoquan He]

Add CONFIG_XSWAP Kconfig option for extendable swap device support.
It depends on SWAP && 64BIT && ZSWAP, since xswap devices are backed by
zswap, and on SYSFS, which is currently the only way to create one.

Add three fields to struct swap_info_struct under CONFIG_XSWAP:
- cluster_vm: the VM_SPARSE vm_struct backing the cluster_info array
- nr_clusters_max: total number of clusters in the xswap address space
- nr_clusters_mapped: number of clusters currently mapped (lazy grow)
These fields enable lazy vmalloc-based dynamic cluster management.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 include/linux/swap.h | 5 +++++
 mm/Kconfig           | 9 +++++++++
 2 files changed, 14 insertions(+)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index 787fe463dcbb..09ecef1b5e90 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -248,6 +248,11 @@ struct swap_info_struct {
 	signed char	type;		/* strange name for an index */
 	unsigned int	max;		/* size of this swap device */
 	struct swap_cluster_info *cluster_info; /* cluster info. Only for SSD */
+#ifdef CONFIG_XSWAP
+	struct vm_struct	*cluster_vm;	/* VM_SPARSE area for cluster_info */
+	unsigned long		nr_clusters_max;/* total clusters in the xswap address space */
+	unsigned long		nr_clusters_mapped; /* currently mapped cluster count */
+#endif
 	struct list_head free_clusters; /* free clusters list */
 	struct list_head full_clusters; /* full clusters list */
 	struct list_head nonfull_clusters[SWAP_NR_ORDERS];
diff --git a/mm/Kconfig b/mm/Kconfig
index 604c58199acb..e9837c0f5d8a 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -122,6 +122,15 @@ config ZSWAP_COMPRESSOR_DEFAULT
        default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD
        default ""
 
+config XSWAP
+	bool "Extendable (virtual) swap device"
+	depends on SWAP && 64BIT && ZSWAP && SYSFS
+	help
+	  Adds support for extendable swap devices (xswap) that decouple
+	  PTE swap entries from physical backing storage. The cluster_info
+	  array is backed by a sparse vmalloc area that grows and shrinks
+	  on demand, avoiding static pre-allocation overhead.
+
 config ZSMALLOC
 	tristate
 
-- 
2.54.0

[PATCH v2 03/12] mm, swap: refactor free_swap_cluster_info to take swap_info_struct

[Baoquan He]

Change free_swap_cluster_info() to accept struct swap_info_struct*
instead of (cluster_info, maxpages) directly. Extract cluster_info
and maxpages from si inside the function, and clear si->cluster_info
after freeing so a caller's error path cannot free it again (a
double free on the swapon() error path). Also clean up swapoff:
remove the snapshot locals (maxpages/cluster_info) and move the
p->max/p->cluster_info clearing after free_swap_cluster_info().

The new signature will allow the xswap path (added in the next patch)
to access si->flags and call xswap_unmap_clusters() from within
free_swap_cluster_info().

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 193b08a54908..25bdb2f2fe0a 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -3055,14 +3055,17 @@ static void wait_for_allocation(struct swap_info_struct *si)
 	}
 }
 
-static void free_swap_cluster_info(struct swap_cluster_info *cluster_info,
-				   unsigned long maxpages)
+static void free_swap_cluster_info(struct swap_info_struct *si)
 {
+	struct swap_cluster_info *cluster_info = si->cluster_info;
+	unsigned long maxpages = si->max;
 	struct swap_cluster_info *ci;
-	int i, nr_clusters = DIV_ROUND_UP(maxpages, SWAPFILE_CLUSTER);
+	int i, nr_clusters;
 
 	if (!cluster_info)
 		return;
+
+	nr_clusters = DIV_ROUND_UP(maxpages, SWAPFILE_CLUSTER);
 	for (i = 0; i < nr_clusters; i++) {
 		ci = cluster_info + i;
 		/* Cluster with bad marks count will have a remaining table */
@@ -3074,6 +3077,7 @@ static void free_swap_cluster_info(struct swap_cluster_info *cluster_info,
 		spin_unlock(&ci->lock);
 	}
 	kvfree(cluster_info);
+	si->cluster_info = NULL;
 }
 
 /*
@@ -3101,11 +3105,9 @@ static void flush_percpu_swap_cluster(struct swap_info_struct *si)
 SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 {
 	struct swap_info_struct *p = NULL;
-	struct swap_cluster_info *cluster_info;
 	struct file *swap_file, *victim;
 	struct address_space *mapping;
 	struct inode *inode;
-	unsigned int maxpages;
 	int err, found = 0;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -3197,10 +3199,6 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 
 	swap_file = p->swap_file;
 	p->swap_file = NULL;
-	maxpages = p->max;
-	cluster_info = p->cluster_info;
-	p->max = 0;
-	p->cluster_info = NULL;
 	spin_unlock(&p->lock);
 	spin_unlock(&swap_lock);
 	arch_swap_invalidate_area(p->type);
@@ -3208,7 +3206,9 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 	mutex_unlock(&swapon_mutex);
 	kfree(p->global_cluster);
 	p->global_cluster = NULL;
-	free_swap_cluster_info(cluster_info, maxpages);
+	free_swap_cluster_info(p);
+	p->max = 0;
+	p->cluster_info = NULL;
 
 	inode = mapping->host;
 
@@ -3575,6 +3575,8 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 	if (!cluster_info)
 		goto err;
 
+	si->cluster_info = cluster_info;
+
 	for (i = 0; i < nr_clusters; i++)
 		spin_lock_init(&cluster_info[i].lock);
 
@@ -3638,7 +3640,7 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 	si->cluster_info = cluster_info;
 	return 0;
 err:
-	free_swap_cluster_info(cluster_info, maxpages);
+	free_swap_cluster_info(si);
 	return err;
 }
 
@@ -3857,7 +3859,7 @@ SYSCALL_DEFINE2(swapon, const char __user *, specialfile, int, swap_flags)
 	si->global_cluster = NULL;
 	inode = NULL;
 	destroy_swap_extents(si, swap_file);
-	free_swap_cluster_info(si->cluster_info, si->max);
+	free_swap_cluster_info(si);
 	si->cluster_info = NULL;
 	/*
 	 * Clear the SWP_USED flag after all resources are freed so
-- 
2.54.0

[PATCH v2 04/12] mm, swap: add xswap cluster grow via VM_SPARSE vmalloc

[Baoquan He]

Implement dynamic cluster_info array growth for xswap devices using a
VM_SPARSE vmalloc area. The cluster_info array is backed by a sparse
vmalloc region that is populated lazily in chunks: physical pages are
allocated and mapped on demand as swap usage grows, and the mapped
tail is unmapped again on the error/teardown paths. Only an initial
chunk is mapped at device setup, and every access is bounded to the
currently mapped range.

The mapping is hardened against concurrency and failure: partial map
failures clear their PTEs before freeing, stale grow ranges are rejected.
The grow path avoids emergency reserves via __GFP_HIGH|__GFP_NOMEMALLOC
and wraps allocations with memalloc_noreclaim_save(). A per-device
mutex (xswap_lock) serializes concurrent map/unmap page table modifications.
find_next_to_unuse() bounds its scan by the mapped range and wraps with >=
so a shrunk bound cannot spin forever.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 include/linux/swap.h |   1 +
 mm/swapfile.c        | 286 ++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 285 insertions(+), 2 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index 09ecef1b5e90..ad2311310032 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -252,6 +252,7 @@ struct swap_info_struct {
 	struct vm_struct	*cluster_vm;	/* VM_SPARSE area for cluster_info */
 	unsigned long		nr_clusters_max;/* total clusters in the xswap address space */
 	unsigned long		nr_clusters_mapped; /* currently mapped cluster count */
+	struct mutex		xswap_lock;	/* serialize map/unmap operations */
 #endif
 	struct list_head free_clusters; /* free clusters list */
 	struct list_head full_clusters; /* full clusters list */
diff --git a/mm/swapfile.c b/mm/swapfile.c
index 25bdb2f2fe0a..5140b3f846b2 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -49,6 +49,25 @@
 #include "internal.h"
 #include "swap.h"
 
+#ifdef CONFIG_XSWAP
+/*
+ * xswap: dynamically grow the cluster_info array via a VM_SPARSE area.
+ *
+ * XSWAP_GROW_CLUSTERS is the number of clusters to map in one grow
+ * operation.  It is set to the number of cluster_info structs that
+ * fit in a single page (at least 16), so that the vmalloc page table
+ * overhead is proportional to the number of clusters mapped.
+ */
+#define XSWAP_GROW_CLUSTERS \
+	max_t(unsigned long, PAGE_SIZE / sizeof(struct swap_cluster_info), 16)
+
+static int xswap_map_clusters(struct swap_info_struct *si,
+			      unsigned long start_idx, unsigned long nr);
+static void xswap_unmap_clusters(struct swap_info_struct *si,
+				 unsigned long start_idx, unsigned long nr);
+static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data);
+#endif
+
 static void swap_range_alloc(struct swap_info_struct *si,
 			     unsigned int nr_entries);
 static bool folio_swapcache_freeable(struct folio *folio);
@@ -2716,15 +2735,27 @@ static unsigned int find_next_to_unuse(struct swap_info_struct *si,
 					unsigned int prev)
 {
 	unsigned int i;
+	unsigned int end = si->max;
 	unsigned long swp_tb;
 
+#ifdef CONFIG_XSWAP
+	/* xswap may have shrunk and unmapped the cluster_info tail. */
+	if (si->flags & SWP_XSWAP) {
+		unsigned long mapped_end;
+
+		mapped_end = READ_ONCE(si->nr_clusters_mapped) * SWAPFILE_CLUSTER;
+		if (mapped_end < end)
+			end = mapped_end;
+	}
+#endif
+
 	/*
 	 * No need for swap_lock here: we're just looking
 	 * for whether an entry is in use, not modifying it; false
 	 * hits are okay, and sys_swapoff() has already prevented new
 	 * allocations from this area (while holding swap_lock).
 	 */
-	for (i = prev + 1; i < si->max; i++) {
+	for (i = prev + 1; i < end; i++) {
 		swp_tb = swap_table_get(__swap_offset_to_cluster(si, i),
 					i % SWAPFILE_CLUSTER);
 		if (!swp_tb_is_null(swp_tb) && !swp_tb_is_bad(swp_tb))
@@ -2733,7 +2764,7 @@ static unsigned int find_next_to_unuse(struct swap_info_struct *si,
 			cond_resched();
 	}
 
-	if (i == si->max)
+	if (i >= end)
 		i = 0;
 
 	return i;
@@ -3049,6 +3080,13 @@ static void wait_for_allocation(struct swap_info_struct *si)
 
 	BUG_ON(si->flags & SWP_WRITEOK);
 
+#ifdef CONFIG_XSWAP
+	/* Skip shrinker-unmapped cluster tail. */
+	if (si->flags & SWP_XSWAP)
+		end = min(end, READ_ONCE(si->nr_clusters_mapped) *
+			  SWAPFILE_CLUSTER);
+#endif
+
 	for (offset = 0; offset < end; offset += SWAPFILE_CLUSTER) {
 		ci = swap_cluster_lock(si, offset);
 		swap_cluster_unlock(ci);
@@ -3065,6 +3103,18 @@ static void free_swap_cluster_info(struct swap_info_struct *si)
 	if (!cluster_info)
 		return;
 
+#ifdef CONFIG_XSWAP
+	if (si->flags & SWP_XSWAP) {
+		/* Unmap all mapped clusters and free the VM_SPARSE area */
+		if (si->nr_clusters_mapped > 0)
+			xswap_unmap_clusters(si, 0, si->nr_clusters_mapped);
+		free_vm_area(si->cluster_vm);
+		si->cluster_vm = NULL;
+		si->nr_clusters_mapped = 0;
+		return;
+	}
+#endif
+
 	nr_clusters = DIV_ROUND_UP(maxpages, SWAPFILE_CLUSTER);
 	for (i = 0; i < nr_clusters; i++) {
 		ci = cluster_info + i;
@@ -3562,6 +3612,174 @@ static unsigned long read_swap_header(struct swap_info_struct *si,
 	return maxpages;
 }
 
+#ifdef CONFIG_XSWAP
+static int xswap_map_clusters(struct swap_info_struct *si,
+			      unsigned long start_idx, unsigned long nr)
+{
+	unsigned long start_addr = (unsigned long)si->cluster_info +
+				   (size_t)start_idx * sizeof(struct swap_cluster_info);
+	unsigned long end_addr = start_addr + (size_t)nr * sizeof(struct swap_cluster_info);
+	/* Round to page boundaries for vm_area_map_pages(). */
+	unsigned long vm_start = PAGE_ALIGN(start_addr);
+	unsigned long vm_end = PAGE_ALIGN(end_addr);
+	unsigned int noreclaim_flags;
+	unsigned long mapped_end;
+	unsigned long npages;
+	struct page **pages;
+	unsigned long i;
+	int err;
+
+	mutex_lock(&si->xswap_lock);
+
+	/* Refuse a stale range: the boundary moved since the caller read it. */
+	if (start_idx != READ_ONCE(si->nr_clusters_mapped)) {
+		mutex_unlock(&si->xswap_lock);
+		return -EAGAIN;
+	}
+	if (start_idx + nr > si->nr_clusters_max) {
+		mutex_unlock(&si->xswap_lock);
+		return -EAGAIN;
+	}
+
+	/*
+	 * Mapping is page granular, so an earlier chunk can have mapped
+	 * past its own clusters up to the end of a page.  Find how much
+	 * of this range is already there and map only the rest; a range
+	 * that is covered entirely was mapped by a concurrent grower.
+	 */
+	mapped_end = vm_start;
+	if (vm_start < vm_end)
+		apply_to_existing_page_range(&init_mm, vm_start,
+					     vm_end - vm_start,
+					     xswap_mapped_end, &mapped_end);
+	if (vm_start >= vm_end || mapped_end == vm_end)
+		goto mapped;
+	vm_start = mapped_end;
+
+	npages = (vm_end - vm_start) >> PAGE_SHIFT;
+
+	/* Prevent recursive reclaim during vmap page table allocation. */
+	noreclaim_flags = memalloc_noreclaim_save();
+
+	pages = kmalloc_array(npages, sizeof(*pages),
+			      __GFP_HIGH | __GFP_NOMEMALLOC | GFP_KERNEL);
+	if (!pages) {
+		memalloc_noreclaim_restore(noreclaim_flags);
+		mutex_unlock(&si->xswap_lock);
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < npages; i++) {
+		/* __GFP_ZERO: cluster_info pointer fields must start NULL. */
+		pages[i] = alloc_page(__GFP_HIGH | __GFP_NOMEMALLOC |
+				      GFP_KERNEL | __GFP_ZERO);
+		if (!pages[i])
+			goto fail;
+	}
+
+	err = vm_area_map_pages(si->cluster_vm, vm_start, vm_end, pages);
+	if (err) {
+		/* -EBUSY: defensive, a concurrent grower got there first. */
+		if (err == -EBUSY) {
+			i = npages;
+			goto fail_nounmap;
+		}
+		i = npages;
+		goto fail;
+	}
+
+	/* Initialize spinlocks for newly mapped clusters */
+	for (i = start_idx; i < start_idx + nr; i++)
+		spin_lock_init(&si->cluster_info[i].lock);
+
+	kfree(pages);
+	memalloc_noreclaim_restore(noreclaim_flags);
+
+	/*
+	 * Pairs with READ_ONCE() in shrink/grow paths.
+	 */
+	smp_store_release(&si->nr_clusters_mapped, start_idx + nr);
+	mutex_unlock(&si->xswap_lock);
+	return 0;
+
+mapped:
+	/* Every page of the range is already mapped; account for it. */
+	for (i = start_idx; i < start_idx + nr; i++)
+		spin_lock_init(&si->cluster_info[i].lock);
+
+	/* Pairs with READ_ONCE() in shrink/grow paths. */
+	smp_store_release(&si->nr_clusters_mapped, start_idx + nr);
+	mutex_unlock(&si->xswap_lock);
+	return 0;
+
+fail_nounmap:
+	/*
+	 * The concurrent grower already mapped the range, initialized the
+	 * cluster spinlocks and advanced nr_clusters_mapped.  It may still
+	 * be holding those locks while adding clusters to the free list, so
+	 * do not touch them here; just free our unused pages.
+	 */
+	while (i > 0) {
+		i--;
+		if (pages[i])
+			__free_page(pages[i]);
+	}
+	kfree(pages);
+	memalloc_noreclaim_restore(noreclaim_flags);
+	mutex_unlock(&si->xswap_lock);
+	return 0;
+
+fail:
+	/* Clear PTEs vm_area_map_pages() may have left before freeing pages. */
+	vm_area_unmap_pages(si->cluster_vm, vm_start, vm_end);
+	while (i > 0) {
+		i--;
+		if (pages[i])
+			__free_page(pages[i]);
+	}
+	memalloc_noreclaim_restore(noreclaim_flags);
+	kfree(pages);
+	mutex_unlock(&si->xswap_lock);
+	return -ENOMEM;
+}
+
+static void xswap_unmap_clusters(struct swap_info_struct *si,
+				 unsigned long start_idx, unsigned long nr)
+{
+	unsigned long start_addr = (unsigned long)si->cluster_info +
+				   (size_t)start_idx * sizeof(struct swap_cluster_info);
+	unsigned long end_addr = start_addr + (size_t)nr * sizeof(struct swap_cluster_info);
+	/* Round to page boundaries for vm_area_unmap_pages(). */
+	unsigned long vm_start = PAGE_ALIGN(start_addr);
+	unsigned long vm_end = PAGE_ALIGN(end_addr);
+
+	mutex_lock(&si->xswap_lock);
+
+	if (vm_start >= vm_end) {
+		WRITE_ONCE(si->nr_clusters_mapped, start_idx);
+		mutex_unlock(&si->xswap_lock);
+		return;
+	}
+
+	vm_area_unmap_pages(si->cluster_vm, vm_start, vm_end);
+	/* vm_area_unmap_pages() clears PTEs but does not free pages. */
+	/* TODO: free backing pages via page table walk or tracking bitmap */
+
+	/* Pairs with READ_ONCE() in shrink/grow paths. */
+	WRITE_ONCE(si->nr_clusters_mapped, start_idx);
+	mutex_unlock(&si->xswap_lock);
+}
+
+/* Track the end of the run of pages that is already mapped. */
+static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data)
+{
+	unsigned long *mapped_end = data;
+
+	*mapped_end = addr + PAGE_SIZE;
+	return 0;
+}
+#endif /* CONFIG_XSWAP */
+
 static int setup_swap_clusters_info(struct swap_info_struct *si,
 				    union swap_header *swap_header,
 				    unsigned long maxpages)
@@ -3571,6 +3789,70 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 	int err = -ENOMEM;
 	unsigned long i;
 
+#ifdef CONFIG_XSWAP
+	if (si->flags & SWP_XSWAP) {
+		unsigned long size = PAGE_ALIGN(nr_clusters * sizeof(*cluster_info));
+		struct vm_struct *vm;
+
+		vm = get_vm_area(size, VM_SPARSE);
+		if (!vm)
+			goto err;
+
+		cluster_info = vm->addr;
+		si->cluster_vm = vm;
+		si->nr_clusters_max = nr_clusters;
+		si->cluster_info = cluster_info;
+
+		/* Must be initialized before xswap_map_clusters() locks it. */
+		mutex_init(&si->xswap_lock);
+
+		/* Map the initial chunk (at least cluster 0) */
+		if (xswap_map_clusters(si, 0, min_t(unsigned long,
+					XSWAP_GROW_CLUSTERS, nr_clusters)))
+			goto err_free_vm;
+
+		/* xswap: only cluster 0 slot 0 is bad */
+		err = swap_cluster_setup_bad_slot(si, cluster_info, 0, false);
+		if (err)
+			goto err_unmap;
+
+		INIT_LIST_HEAD(&si->free_clusters);
+		INIT_LIST_HEAD(&si->full_clusters);
+		INIT_LIST_HEAD(&si->discard_clusters);
+		for (i = 0; i < SWAP_NR_ORDERS; i++) {
+			INIT_LIST_HEAD(&si->nonfull_clusters[i]);
+			INIT_LIST_HEAD(&si->frag_clusters[i]);
+		}
+
+		/*
+		 * Cluster 0 holds the header slot and the last one holds the
+		 * holes past si->max; both have slots marked bad, so they are
+		 * not entirely free.  The clusters in between are.
+		 */
+		for (i = 0; i < si->nr_clusters_mapped; i++) {
+			struct swap_cluster_info *ci = &cluster_info[i];
+
+			if (ci->count) {
+				ci->flags = CLUSTER_FLAG_NONFULL;
+				list_add_tail(&ci->list, &si->nonfull_clusters[0]);
+			} else {
+				ci->flags = CLUSTER_FLAG_FREE;
+				list_add_tail(&ci->list, &si->free_clusters);
+			}
+		}
+
+		return 0;
+
+err_unmap:
+		xswap_unmap_clusters(si, 0, si->nr_clusters_mapped);
+err_free_vm:
+		free_vm_area(si->cluster_vm);
+		si->cluster_vm = NULL;
+		si->cluster_info = NULL;
+		return err;
+	}
+#endif /* CONFIG_XSWAP */
+
 	cluster_info = kvzalloc_objs(*cluster_info, nr_clusters);
 	if (!cluster_info)
 		goto err;
-- 
2.54.0

[PATCH v2 05/12] mm, swap: add sysfs create interface for xswap

[Baoquan He]

xswap devices have no backing storage, so there is no file to swapon.
Add a sysfs interface to create them directly:
  /sys/kernel/mm/xswap/create    write an optional priority to create
                                 a device (empty = DEF_SWAP_PRIO)

A new device is created with si->max and nr_clusters_max both equal to
full RAM: the cluster_info array is a sparse VM_SPARSE area mapped
lazily, so an idle device costs nothing. The runtime size can be
lowered per device afterwards via
/sys/kernel/mm/xswap/type<N>/limit (added later in this series).
The optional priority follows swapon(2)'s -p semantics (default
DEF_SWAP_PRIO, valid range -1..SWAP_FLAG_PRIO_MASK). The device shows
up in /proc/swaps as "xswap<N>".

Hibernation device discovery skips xswap devices: they have no bdev
to carry a resume image.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 165 ++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 160 insertions(+), 5 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 5140b3f846b2..4861e2d49e36 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -16,6 +16,8 @@
 #include <linux/kernel_stat.h>
 #include <linux/swap.h>
 #include <linux/vmalloc.h>
+#include <linux/kobject.h>
+#include <linux/sysfs.h>
 #include <linux/pagemap.h>
 #include <linux/namei.h>
 #include <linux/shmem_fs.h>
@@ -48,6 +50,7 @@
 #include "swap_table.h"
 #include "internal.h"
 #include "swap.h"
+#define DEF_SWAP_PRIO  -1
 
 #ifdef CONFIG_XSWAP
 /*
@@ -66,7 +69,70 @@ static int xswap_map_clusters(struct swap_info_struct *si,
 static void xswap_unmap_clusters(struct swap_info_struct *si,
 				 unsigned long start_idx, unsigned long nr);
 static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data);
-#endif
+
+#ifdef CONFIG_SYSFS
+static int xswap_create(int prio);
+
+static ssize_t xswap_create_store(struct kobject *kobj,
+				  struct kobj_attribute *attr,
+				  const char *buf, size_t count)
+{
+	int prio = DEF_SWAP_PRIO;
+	int err;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	/* "[<prio>]" is optional; empty means DEF_SWAP_PRIO, i.e. the same
+	 * default as swapon(2) without SWAP_FLAG_PREFER.
+	 */
+	if (*skip_spaces(buf)) {
+		err = kstrtoint(buf, 10, &prio);
+		if (err)
+			return err;
+	}
+
+	err = xswap_create(prio);
+	if (err < 0)
+		return err;
+
+	return count;
+}
+
+static struct kobj_attribute xswap_create_attr = __ATTR(create, 0200, NULL,
+							xswap_create_store);
+
+static struct attribute *xswap_attrs[] = {
+	&xswap_create_attr.attr,
+	NULL,
+};
+
+static const struct attribute_group xswap_attr_group = {
+	.attrs = xswap_attrs,
+};
+
+static struct kobject *xswap_kobj;
+
+static void xswap_sysfs_init(void)
+{
+	xswap_kobj = kobject_create_and_add("xswap", mm_kobj);
+	if (!xswap_kobj) {
+		pr_err("xswap: failed to create sysfs kobject\n");
+		return;
+	}
+	if (sysfs_create_group(xswap_kobj, &xswap_attr_group))
+		pr_err("xswap: failed to create sysfs group\n");
+}
+#else
+static inline void xswap_sysfs_init(void)
+{
+}
+#endif /* CONFIG_SYSFS */
+#else /* !CONFIG_XSWAP */
+static inline void xswap_sysfs_init(void)
+{
+}
+#endif /* CONFIG_XSWAP */
 
 static void swap_range_alloc(struct swap_info_struct *si,
 			     unsigned int nr_entries);
@@ -94,7 +160,6 @@ atomic_t nr_real_swapfiles;
 EXPORT_SYMBOL_GPL(nr_swap_pages);
 /* protected with swap_lock. reading in vm_swap_full() doesn't need lock */
 long total_swap_pages;
-#define DEF_SWAP_PRIO  -1
 unsigned long swapfile_maximum_size;
 #ifdef CONFIG_MIGRATION
 bool swap_migration_ad_supported;
@@ -2255,6 +2320,9 @@ static int __find_hibernation_swap_type(dev_t device, sector_t offset)
 
 		if (!(sis->flags & SWP_WRITEOK))
 			continue;
+		/* xswap has no bdev to match a resume device */
+		if (sis->flags & SWP_XSWAP)
+			continue;
 
 		if (device == sis->bdev->bd_dev) {
 			struct swap_extent *se = first_se(sis);
@@ -2381,6 +2449,8 @@ int find_first_swap(dev_t *device)
 
 		if (!(sis->flags & SWP_WRITEOK))
 			continue;
+		if (sis->flags & SWP_XSWAP)
+			continue;
 		*device = sis->bdev->bd_dev;
 		spin_unlock(&swap_lock);
 		return type;
@@ -3313,7 +3383,7 @@ static void *swap_start(struct seq_file *swap, loff_t *pos)
 		return SEQ_START_TOKEN;
 
 	for (type = 0; (si = swap_type_to_info(type)); type++) {
-		if (!(si->swap_file))
+		if (!(si->swap_file) && !(si->flags & SWP_XSWAP))
 			continue;
 		if (!--l)
 			return si;
@@ -3334,7 +3404,7 @@ static void *swap_next(struct seq_file *swap, void *v, loff_t *pos)
 
 	++(*pos);
 	for (; (si = swap_type_to_info(type)); type++) {
-		if (!(si->swap_file))
+		if (!(si->swap_file) && !(si->flags & SWP_XSWAP))
 			continue;
 		return si;
 	}
@@ -3376,7 +3446,14 @@ static int swap_show(struct seq_file *swap, void *v)
 	inuse = K(swap_usage_in_pages(si));
 
 	file = si->swap_file;
-	len = seq_file_path(swap, file, " \t\n\\");
+	if (file)
+		len = seq_file_path(swap, file, " \t\n\\");
+	else {
+		char name[16];
+
+		len = scnprintf(name, sizeof(name), "xswap%d", si->type);
+		seq_puts(swap, name);
+	}
 	seq_printf(swap, "%*s%s\t%lu\t%s%lu\t%s%d\n",
 			len < 40 ? 40 - len : 1, " ",
 			swap_type_str(si),
@@ -3926,6 +4003,82 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 	return err;
 }
 
+#ifdef CONFIG_XSWAP
+#ifdef CONFIG_SYSFS
+/* Create a file-less xswap device.  si->max and the initial nr_clusters
+ * ceiling both equal full RAM; the runtime size can be lowered afterwards
+ * via /sys/kernel/mm/xswap/type<N>/limit.
+ */
+static int xswap_create(int prio)
+{
+	struct swap_info_struct *si;
+	unsigned long ram, maxpages;
+	int error;
+
+	if (prio != DEF_SWAP_PRIO && (prio < 0 || prio > SWAP_FLAG_PRIO_MASK))
+		return -EINVAL;
+
+	si = alloc_swap_info();
+	if (IS_ERR(si))
+		return PTR_ERR(si);
+
+	INIT_WORK(&si->discard_work, swap_discard_work);
+	INIT_WORK(&si->reclaim_work, swap_reclaim_work);
+
+	ram = totalram_pages();
+	maxpages = min_t(unsigned long, ram, swapfile_maximum_size);
+	if ((unsigned int)maxpages == 0)
+		maxpages = UINT_MAX;
+	/* Cluster-aligned, so no cluster holds a slot past si->max. */
+	if (maxpages > SWAPFILE_CLUSTER)
+		maxpages = rounddown(maxpages, SWAPFILE_CLUSTER);
+	if (maxpages < 2)
+		maxpages = 2;
+
+	si->bdev = NULL;
+	si->flags |= SWP_XSWAP | SWP_SOLIDSTATE;
+	si->max = maxpages;
+	si->pages = maxpages - 1;
+	/* no backing file: mirror the xswap branch of setup_swap_extents() */
+	si->ops = &swap_bdev_ops;
+
+	error = setup_swap_clusters_info(si, NULL, maxpages);
+	if (error)
+		goto bad_swap;
+
+	error = zswap_swapon(si->type, si->max);
+	if (error)
+		goto bad_swap;
+
+	mutex_lock(&swapon_mutex);
+	si->prio = prio;
+	si->list.prio = -si->prio;
+	si->avail_list.prio = -si->prio;
+	/* si->swap_file stays NULL: this is a file-less device */
+	enable_swap_info(si);
+	mutex_unlock(&swapon_mutex);
+
+	pr_info("xswap: adding extendable swap type %d (prio %d, %u pages, max %lu)\n",
+		si->type, prio, si->pages, maxpages);
+	atomic_inc(&proc_poll_event);
+	wake_up_interruptible(&proc_poll_wait);
+
+	return si->type;
+
+bad_swap:
+	kfree(si->global_cluster);
+	si->global_cluster = NULL;
+	destroy_swap_extents(si, NULL);	/* safe: xswap never sets SWP_ACTIVATED */
+	free_swap_cluster_info(si);
+	si->cluster_info = NULL;
+	spin_lock(&swap_lock);
+	si->flags = 0;
+	spin_unlock(&swap_lock);
+	return error;
+}
+#endif /* CONFIG_SYSFS */
+#endif /* CONFIG_XSWAP */
+
 SYSCALL_DEFINE2(swapon, const char __user *, specialfile, int, swap_flags)
 {
 	struct swap_info_struct *si;
@@ -4269,6 +4422,8 @@ static int __init swapfile_init(void)
 		swap_migration_ad_supported = true;
 #endif	/* CONFIG_MIGRATION */
 
+	xswap_sysfs_init();
+
 	return 0;
 }
 subsys_initcall(swapfile_init);
-- 
2.54.0

[PATCH v2 06/12] mm, swap: add xswap grow trigger on cluster allocation

[Baoquan He]

When cluster_alloc_swap_entry() fails to find a free cluster and
the xswap device still has room to grow, expand the mapped range
by XSWAP_GROW_CLUSTERS clusters.

Since xswap is always SWP_SOLIDSTATE, global_cluster_lock is never
held on this path.

This makes the xswap cluster space grow transparently as swap usage
increases, without any userspace intervention.

Growing maps pages into the VM_SPARSE area, which can sleep. The caller
holds local_lock(&percpu_swap_cluster.lock) across the whole slow path,
so drop it around xswap_map_clusters() and take it again afterwards; it
only protects the per-cpu cluster cache, which this path does not touch.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 4861e2d49e36..f603f8c4b599 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -1270,6 +1270,56 @@ static unsigned long cluster_alloc_swap_entry(struct swap_info_struct *si,
 		if (found)
 			goto done;
 	}
+
+#ifdef CONFIG_XSWAP
+	/*
+	 * For xswap: if no free cluster was found and more clusters
+	 * can be mapped, grow the cluster_info array and retry.
+	 */
+	if (!found && (si->flags & SWP_XSWAP) &&
+	    READ_ONCE(si->nr_clusters_mapped) < READ_ONCE(si->nr_clusters_max) &&
+	    list_empty(&si->free_clusters)) {
+		unsigned long nr_new = min(READ_ONCE(si->nr_clusters_max) -
+					  READ_ONCE(si->nr_clusters_mapped),
+					  XSWAP_GROW_CLUSTERS);
+		unsigned long start = READ_ONCE(si->nr_clusters_mapped);
+		unsigned long i;
+		int ret;
+
+		/*
+		 * Mapping pages into the VM_SPARSE area can sleep, which is
+		 * not allowed under local_lock.  The lock only protects the
+		 * per-cpu cluster cache, which this path does not touch, so
+		 * it can be dropped across the call.
+		 */
+		local_unlock(&percpu_swap_cluster.lock);
+		ret = xswap_map_clusters(si, start, nr_new);
+		local_lock(&percpu_swap_cluster.lock);
+
+		if (!ret) {
+			for (i = start; i < start + nr_new; i++) {
+				struct swap_cluster_info *ci = &si->cluster_info[i];
+
+				/*
+				 * A concurrent grower may have already added
+				 * these clusters to the free list.  Only add
+				 * clusters that are still off-list (NONE).
+				 * Lock ci->lock first: move_cluster() takes
+				 * si->lock internally.
+				 */
+				spin_lock(&ci->lock);
+				if (ci->flags == CLUSTER_FLAG_NONE)
+					move_cluster(si, ci, &si->free_clusters,
+						     CLUSTER_FLAG_FREE);
+				spin_unlock(&ci->lock);
+			}
+
+			/* Retry allocation from the free list */
+			found = alloc_swap_scan_list(si, &si->free_clusters,
+						    folio, false);
+		}
+	}
+#endif
 done:
 	if (!(si->flags & SWP_SOLIDSTATE))
 		spin_unlock(&si->global_cluster_lock);
-- 
2.54.0

[PATCH v2 07/12] mm, swap: add xswap_try_shrink and shrink trigger on cluster free

[Baoquan He]

Add xswap_try_shrink(). It scans backwards from the end of the mapped
range and counts the contiguous free clusters at the tail. It then
unmaps one or more chunks of that tail, in units of XSWAP_GROW_CLUSTERS.

The function only reclaims when the mapped range is at most half in use.
Growth is demand driven. Reclaiming on a smaller dip would only map the
same clusters again, and every unmap costs an RCU grace period. So it
leaves one chunk of free tail mapped as slack for the next allocation,
and waits until several chunks can go before it unmaps anything.

Wire the trigger in __free_cluster(). After a cluster is released to
the free list, call xswap_try_shrink() to attempt the tail shrink.

Also update the XSWAP_GROW_CLUSTERS comment. The constant is now used
for shrink as well as grow.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 95 +++++++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 89 insertions(+), 6 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index f603f8c4b599..abe3d0ee5aba 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -54,12 +54,13 @@
 
 #ifdef CONFIG_XSWAP
 /*
- * xswap: dynamically grow the cluster_info array via a VM_SPARSE area.
+ * xswap: dynamically grow and shrink the cluster_info array via a
+ * VM_SPARSE area.
  *
- * XSWAP_GROW_CLUSTERS is the number of clusters to map in one grow
- * operation.  It is set to the number of cluster_info structs that
- * fit in a single page (at least 16), so that the vmalloc page table
- * overhead is proportional to the number of clusters mapped.
+ * XSWAP_GROW_CLUSTERS is the number of clusters to map/unmap in one
+ * grow/shrink operation: the number of cluster_info structs that fit in
+ * a single page (at least 16), so that the vmalloc page table overhead
+ * is proportional to the number of clusters mapped.
  */
 #define XSWAP_GROW_CLUSTERS \
 	max_t(unsigned long, PAGE_SIZE / sizeof(struct swap_cluster_info), 16)
@@ -69,6 +70,7 @@ static int xswap_map_clusters(struct swap_info_struct *si,
 static void xswap_unmap_clusters(struct swap_info_struct *si,
 				 unsigned long start_idx, unsigned long nr);
 static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data);
+static void xswap_try_shrink(struct swap_info_struct *si);
 
 #ifdef CONFIG_SYSFS
 static int xswap_create(int prio);
@@ -694,6 +696,9 @@ static void __free_cluster(struct swap_info_struct *si, struct swap_cluster_info
 	swap_cluster_free_table(ci);
 	move_cluster(si, ci, &si->free_clusters, CLUSTER_FLAG_FREE);
 	ci->order = 0;
+#ifdef CONFIG_XSWAP
+	xswap_try_shrink(si);
+#endif
 }
 
 /*
@@ -1514,11 +1519,19 @@ static bool swap_alloc_fast(struct folio *folio)
 	/*
 	 * Once allocated, swap_info_struct will never be completely freed,
 	 * so checking it's liveness by get_swap_device_info is enough.
+	 *
+	 * The cached cluster points into si->cluster_info, which xswap
+	 * maps and unmaps at runtime.  The RCU section has to cover both
+	 * reading the cached offset and dereferencing it, so that an
+	 * unmap can wait for the readers already in flight.
 	 */
+	rcu_read_lock();
 	si = this_cpu_read(percpu_swap_cluster.si[order]);
 	offset = this_cpu_read(percpu_swap_cluster.offset[order]);
-	if (!si || !offset || !get_swap_device_info(si))
+	if (!si || !offset || !get_swap_device_info(si)) {
+		rcu_read_unlock();
 		return false;
+	}
 
 	ci = swap_cluster_lock(si, offset);
 	if (cluster_is_usable(ci, order)) {
@@ -1530,6 +1543,7 @@ static bool swap_alloc_fast(struct folio *folio)
 	}
 
 	put_swap_device(si);
+	rcu_read_unlock();
 	return folio_test_swapcache(folio);
 }
 
@@ -2307,8 +2321,12 @@ swp_entry_t swap_alloc_hibernation_slot(int type)
 	/*
 	 * Try the local cluster first if it matches the device. If
 	 * not, try grab a new cluster and override local cluster.
+	 *
+	 * Same RCU requirement as swap_alloc_fast(): the cached offset
+	 * indexes si->cluster_info, which xswap can unmap.
 	 */
 	local_lock(&percpu_swap_cluster.lock);
+	rcu_read_lock();
 	pcp_si = this_cpu_read(percpu_swap_cluster.si[0]);
 	pcp_offset = this_cpu_read(percpu_swap_cluster.offset[0]);
 	if (pcp_si == si && pcp_offset) {
@@ -2318,6 +2336,7 @@ swp_entry_t swap_alloc_hibernation_slot(int type)
 		else
 			swap_cluster_unlock(ci);
 	}
+	rcu_read_unlock();
 	if (!offset)
 		offset = cluster_alloc_swap_entry(si, NULL);
 	local_unlock(&percpu_swap_cluster.lock);
@@ -3888,6 +3907,15 @@ static void xswap_unmap_clusters(struct swap_info_struct *si,
 		return;
 	}
 
+	/*
+	 * A per-cpu cluster cache can still hold an offset in this range.
+	 * Invalidate those references, then wait out the readers that have
+	 * already loaded one, so that nobody can dereference cluster_info
+	 * past this point.  swapoff() needs the same before it releases.
+	 */
+	flush_percpu_swap_cluster(si);
+	synchronize_rcu();
+
 	vm_area_unmap_pages(si->cluster_vm, vm_start, vm_end);
 	/* vm_area_unmap_pages() clears PTEs but does not free pages. */
 	/* TODO: free backing pages via page table walk or tracking bitmap */
@@ -3905,6 +3933,61 @@ static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data)
 	*mapped_end = addr + PAGE_SIZE;
 	return 0;
 }
+
+/*
+ * Automatic reclaim: leave one chunk of the free tail mapped as slack, so
+ * that the next allocation does not grow the range straight back, and only
+ * unmap once several chunks can go, so the unmap is worth the RCU grace
+ * period it costs.
+ */
+#define XSWAP_SHRINK_SLACK	XSWAP_GROW_CLUSTERS
+#define XSWAP_SHRINK_MIN	(XSWAP_GROW_CLUSTERS * 4)
+
+/*
+ * Try to shrink the cluster_info tail: unmap contiguous free clusters
+ * at the end of the mapped range.
+ */
+static void xswap_try_shrink(struct swap_info_struct *si)
+{
+	struct swap_cluster_info *ci;
+	unsigned long nr_mapped, last, idx;
+
+	if (!(si->flags & SWP_XSWAP))
+		return;
+
+	nr_mapped = READ_ONCE(si->nr_clusters_mapped);
+	if (nr_mapped <= 1) /* keep cluster 0 */
+		return;
+
+	/*
+	 * Reclaim on our own, but only once the mapped range is at most
+	 * half in use: growth is demand driven, so reclaiming on a smaller
+	 * dip would only map the same clusters again, and every unmap costs
+	 * an RCU grace period.
+	 */
+	if (atomic_long_read(&si->inuse_pages) * 2 > nr_mapped * SWAPFILE_CLUSTER)
+		return;
+
+	/* Find the last non-free cluster from the tail */
+	last = nr_mapped;
+	while (last > 1) {
+		idx = last - 1;
+		ci = &si->cluster_info[idx];
+		if (ci->count || ci->flags != CLUSTER_FLAG_FREE)
+			break;
+		last = idx;
+	}
+
+	if (last == nr_mapped)
+		return; /* nothing to shrink */
+
+	if (nr_mapped - last < XSWAP_SHRINK_SLACK + XSWAP_SHRINK_MIN)
+		return;
+
+	last += XSWAP_SHRINK_SLACK;
+
+	xswap_unmap_clusters(si, last, nr_mapped - last);
+}
 #endif /* CONFIG_XSWAP */
 
 static int setup_swap_clusters_info(struct swap_info_struct *si,
-- 
2.54.0

[PATCH v2 08/12] mm, swap: free backing pages in xswap_unmap_clusters

[Baoquan He]

vm_area_unmap_pages() clears PTEs and frees intermediate page table
pages. It does not free the backing physical pages that
xswap_map_clusters() allocated. Those pages were leaked on every
unmap.

Collect the backing pages before the unmap. Walk the page table with
apply_to_existing_page_range() and a small callback that stores
pte_page() of every present PTE into a dynamically allocated array.
After vm_area_unmap_pages() clears the PTEs, free the collected pages
with __free_page().

The array allocation runs under memalloc_noreclaim_save(). Reclaim
must not recurse into the caller, which holds xswap_lock. If the
allocation fails, return -ENOMEM without unmapping, so that no backing
page is leaked.

xswap_unmap_clusters() now returns an error code.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 57 ++++++++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 50 insertions(+), 7 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index abe3d0ee5aba..e10bb6d245fe 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -67,8 +67,8 @@
 
 static int xswap_map_clusters(struct swap_info_struct *si,
 			      unsigned long start_idx, unsigned long nr);
-static void xswap_unmap_clusters(struct swap_info_struct *si,
-				 unsigned long start_idx, unsigned long nr);
+static int xswap_unmap_clusters(struct swap_info_struct *si,
+				unsigned long start_idx, unsigned long nr);
 static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data);
 static void xswap_try_shrink(struct swap_info_struct *si);
 
@@ -3889,8 +3889,26 @@ static int xswap_map_clusters(struct swap_info_struct *si,
 	return -ENOMEM;
 }
 
-static void xswap_unmap_clusters(struct swap_info_struct *si,
-				 unsigned long start_idx, unsigned long nr)
+struct xswap_page_data {
+	struct page **pages;
+	int nr;
+	int max;
+};
+
+static int xswap_collect_page(pte_t *pte, unsigned long addr, void *data)
+{
+	struct xswap_page_data *xpd = data;
+	pte_t pteval = ptep_get(pte);
+
+	if (!pte_present(pteval))
+		return 0;
+	if (xpd->nr < xpd->max)
+		xpd->pages[xpd->nr++] = pte_page(pteval);
+	return 0;
+}
+
+static int xswap_unmap_clusters(struct swap_info_struct *si,
+				unsigned long start_idx, unsigned long nr)
 {
 	unsigned long start_addr = (unsigned long)si->cluster_info +
 				   (size_t)start_idx * sizeof(struct swap_cluster_info);
@@ -3898,13 +3916,18 @@ static void xswap_unmap_clusters(struct swap_info_struct *si,
 	/* Round to page boundaries for vm_area_unmap_pages(). */
 	unsigned long vm_start = PAGE_ALIGN(start_addr);
 	unsigned long vm_end = PAGE_ALIGN(end_addr);
+	unsigned long size;
+	unsigned long npages;
+	struct xswap_page_data xpd;
+	unsigned int noreclaim_flags;
+	int i;
 
 	mutex_lock(&si->xswap_lock);
 
 	if (vm_start >= vm_end) {
 		WRITE_ONCE(si->nr_clusters_mapped, start_idx);
 		mutex_unlock(&si->xswap_lock);
-		return;
+		return 0;
 	}
 
 	/*
@@ -3916,13 +3939,33 @@ static void xswap_unmap_clusters(struct swap_info_struct *si,
 	flush_percpu_swap_cluster(si);
 	synchronize_rcu();
 
+	size = vm_end - vm_start;
+	npages = size >> PAGE_SHIFT;
+
+	noreclaim_flags = memalloc_noreclaim_save();
+	xpd.pages = kmalloc_array(npages, sizeof(*xpd.pages),
+				  __GFP_HIGH | __GFP_NOMEMALLOC | GFP_KERNEL);
+	memalloc_noreclaim_restore(noreclaim_flags);
+	if (!xpd.pages) {
+		mutex_unlock(&si->xswap_lock);
+		return -ENOMEM;
+	}
+
+	xpd.nr = 0;
+	xpd.max = npages;
+	apply_to_existing_page_range(&init_mm, vm_start, size,
+				     xswap_collect_page, &xpd);
+
 	vm_area_unmap_pages(si->cluster_vm, vm_start, vm_end);
-	/* vm_area_unmap_pages() clears PTEs but does not free pages. */
-	/* TODO: free backing pages via page table walk or tracking bitmap */
+
+	for (i = 0; i < xpd.nr; i++)
+		__free_page(xpd.pages[i]);
+	kfree(xpd.pages);
 
 	/* Pairs with READ_ONCE() in shrink/grow paths. */
 	WRITE_ONCE(si->nr_clusters_mapped, start_idx);
 	mutex_unlock(&si->xswap_lock);
+	return 0;
 }
 
 /* Track the end of the run of pages that is already mapped. */
-- 
2.54.0

[PATCH v2 09/12] mm, swap: defer xswap shrink to workqueue to avoid lock recursion

[Baoquan He]

__free_cluster() called xswap_try_shrink() directly, with ci->lock
held. The shrink path calls xswap_unmap_clusters(), which unmaps the
vmalloc pages backing cluster_info. On return, swap_cache_del_folio()
calls swap_cluster_unlock(ci) on that address. The address is no
longer mapped, so it faults on a not-present page.

Replace the direct call with schedule_work(). The shrink then runs in
an independent workqueue context, where no cluster lock is held.

__free_cluster() schedules the work only for xswap devices, and only
while SWP_WRITEOK is set. The work function bails out once swapoff has
begun. __swapoff() also cancels any pending work before it walks the
clusters.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 include/linux/swap.h |   1 +
 mm/swapfile.c        | 124 +++++++++++++++++++++++++++++++++----------
 2 files changed, 98 insertions(+), 27 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index ad2311310032..6307d8e8f7ed 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -252,6 +252,7 @@ struct swap_info_struct {
 	struct vm_struct	*cluster_vm;	/* VM_SPARSE area for cluster_info */
 	unsigned long		nr_clusters_max;/* total clusters in the xswap address space */
 	unsigned long		nr_clusters_mapped; /* currently mapped cluster count */
+	struct work_struct	xswap_shrink_work; /* deferred shrink trigger */
 	struct mutex		xswap_lock;	/* serialize map/unmap operations */
 #endif
 	struct list_head free_clusters; /* free clusters list */
diff --git a/mm/swapfile.c b/mm/swapfile.c
index e10bb6d245fe..d2a3303b0572 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -697,7 +697,9 @@ static void __free_cluster(struct swap_info_struct *si, struct swap_cluster_info
 	move_cluster(si, ci, &si->free_clusters, CLUSTER_FLAG_FREE);
 	ci->order = 0;
 #ifdef CONFIG_XSWAP
-	xswap_try_shrink(si);
+	/* Only xswap devices, and not while the device is being torn down. */
+	if ((si->flags & SWP_XSWAP) && (si->flags & SWP_WRITEOK))
+		schedule_work(&si->xswap_shrink_work);
 #endif
 }
 
@@ -3244,6 +3246,7 @@ static void free_swap_cluster_info(struct swap_info_struct *si)
 
 #ifdef CONFIG_XSWAP
 	if (si->flags & SWP_XSWAP) {
+		cancel_work_sync(&si->xswap_shrink_work);
 		/* Unmap all mapped clusters and free the VM_SPARSE area */
 		if (si->nr_clusters_mapped > 0)
 			xswap_unmap_clusters(si, 0, si->nr_clusters_mapped);
@@ -3347,6 +3350,11 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 	spin_unlock(&p->lock);
 	spin_unlock(&swap_lock);
 
+#ifdef CONFIG_XSWAP
+	if (p->flags & SWP_XSWAP)
+		cancel_work_sync(&p->xswap_shrink_work);
+#endif
+
 	wait_for_allocation(p);
 
 	set_current_oom_origin();
@@ -3907,8 +3915,9 @@ static int xswap_collect_page(pte_t *pte, unsigned long addr, void *data)
 	return 0;
 }
 
-static int xswap_unmap_clusters(struct swap_info_struct *si,
-				unsigned long start_idx, unsigned long nr)
+/* Caller must hold si->xswap_lock; -ENOMEM leaves the mapping intact. */
+static int xswap_unmap_clusters_locked(struct swap_info_struct *si,
+				       unsigned long start_idx, unsigned long nr)
 {
 	unsigned long start_addr = (unsigned long)si->cluster_info +
 				   (size_t)start_idx * sizeof(struct swap_cluster_info);
@@ -3922,11 +3931,8 @@ static int xswap_unmap_clusters(struct swap_info_struct *si,
 	unsigned int noreclaim_flags;
 	int i;
 
-	mutex_lock(&si->xswap_lock);
-
 	if (vm_start >= vm_end) {
 		WRITE_ONCE(si->nr_clusters_mapped, start_idx);
-		mutex_unlock(&si->xswap_lock);
 		return 0;
 	}
 
@@ -3946,10 +3952,8 @@ static int xswap_unmap_clusters(struct swap_info_struct *si,
 	xpd.pages = kmalloc_array(npages, sizeof(*xpd.pages),
 				  __GFP_HIGH | __GFP_NOMEMALLOC | GFP_KERNEL);
 	memalloc_noreclaim_restore(noreclaim_flags);
-	if (!xpd.pages) {
-		mutex_unlock(&si->xswap_lock);
+	if (!xpd.pages)
 		return -ENOMEM;
-	}
 
 	xpd.nr = 0;
 	xpd.max = npages;
@@ -3964,10 +3968,20 @@ static int xswap_unmap_clusters(struct swap_info_struct *si,
 
 	/* Pairs with READ_ONCE() in shrink/grow paths. */
 	WRITE_ONCE(si->nr_clusters_mapped, start_idx);
-	mutex_unlock(&si->xswap_lock);
 	return 0;
 }
 
+static int xswap_unmap_clusters(struct swap_info_struct *si,
+				unsigned long start_idx, unsigned long nr)
+{
+	int ret;
+
+	mutex_lock(&si->xswap_lock);
+	ret = xswap_unmap_clusters_locked(si, start_idx, nr);
+	mutex_unlock(&si->xswap_lock);
+	return ret;
+}
+
 /* Track the end of the run of pages that is already mapped. */
 static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data)
 {
@@ -3986,6 +4000,16 @@ static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data)
 #define XSWAP_SHRINK_SLACK	XSWAP_GROW_CLUSTERS
 #define XSWAP_SHRINK_MIN	(XSWAP_GROW_CLUSTERS * 4)
 
+static void xswap_shrink_work_fn(struct work_struct *work)
+{
+	struct swap_info_struct *si = container_of(work,
+			struct swap_info_struct, xswap_shrink_work);
+
+	if (!(READ_ONCE(si->flags) & SWP_WRITEOK))
+		return;
+	xswap_try_shrink(si);
+}
+
 /*
  * Try to shrink the cluster_info tail: unmap contiguous free clusters
  * at the end of the mapped range.
@@ -3993,14 +4017,16 @@ static int xswap_mapped_end(pte_t *pte, unsigned long addr, void *data)
 static void xswap_try_shrink(struct swap_info_struct *si)
 {
 	struct swap_cluster_info *ci;
-	unsigned long nr_mapped, last, idx;
+	unsigned long nr_mapped, nr_tail, nr_unmap, start_idx, i;
 
 	if (!(si->flags & SWP_XSWAP))
 		return;
 
+	mutex_lock(&si->xswap_lock);
+
 	nr_mapped = READ_ONCE(si->nr_clusters_mapped);
-	if (nr_mapped <= 1) /* keep cluster 0 */
-		return;
+	if (nr_mapped <= 1)	/* keep cluster 0 */
+		goto out_unlock;
 
 	/*
 	 * Reclaim on our own, but only once the mapped range is at most
@@ -4009,27 +4035,70 @@ static void xswap_try_shrink(struct swap_info_struct *si)
 	 * an RCU grace period.
 	 */
 	if (atomic_long_read(&si->inuse_pages) * 2 > nr_mapped * SWAPFILE_CLUSTER)
-		return;
+		goto out_unlock;
 
-	/* Find the last non-free cluster from the tail */
-	last = nr_mapped;
-	while (last > 1) {
-		idx = last - 1;
-		ci = &si->cluster_info[idx];
-		if (ci->count || ci->flags != CLUSTER_FLAG_FREE)
+	/*
+	 * Count the free clusters at the tail of the mapped range.  Scanned,
+	 * not tracked: the count must be exact to size the unmap, and an
+	 * incremental count falls behind on out-of-order frees.
+	 */
+	nr_tail = 0;
+	while (nr_mapped - nr_tail > 1) {
+		ci = &si->cluster_info[nr_mapped - nr_tail - 1];
+		if (READ_ONCE(ci->count) ||
+		    READ_ONCE(ci->flags) != CLUSTER_FLAG_FREE)
 			break;
-		last = idx;
+		nr_tail++;
 	}
+	if (nr_tail < XSWAP_SHRINK_SLACK + XSWAP_SHRINK_MIN)
+		goto out_unlock;
 
-	if (last == nr_mapped)
-		return; /* nothing to shrink */
+	nr_unmap = rounddown(nr_tail - XSWAP_SHRINK_SLACK, XSWAP_GROW_CLUSTERS);
+	if (!nr_unmap)
+		goto out_unlock;
+	start_idx = nr_mapped - nr_unmap;
 
-	if (nr_mapped - last < XSWAP_SHRINK_SLACK + XSWAP_SHRINK_MIN)
-		return;
+	/*
+	 * Only shrink a run that reaches the mapped end; otherwise
+	 * truncating nr_clusters_mapped would orphan the active tail.
+	 */
+	spin_lock(&si->lock);
+	for (i = start_idx; i < nr_mapped; i++) {
+		ci = &si->cluster_info[i];
+		if (READ_ONCE(ci->flags) != CLUSTER_FLAG_FREE)
+			break;
+		/* Skip clusters whose lock is currently held. */
+		if (!spin_trylock(&ci->lock)) {
+			spin_unlock(&si->lock);
+			goto out_unlock;
+		}
+		spin_unlock(&ci->lock);
+	}
+	if (i != nr_mapped) {
+		spin_unlock(&si->lock);
+		goto out_unlock;
+	}
 
-	last += XSWAP_SHRINK_SLACK;
+	for (i = start_idx; i < nr_mapped; i++) {
+		ci = &si->cluster_info[i];
+		list_del_init(&ci->list);
+		WRITE_ONCE(ci->flags, CLUSTER_FLAG_NONE);
+	}
+	spin_unlock(&si->lock);
 
-	xswap_unmap_clusters(si, last, nr_mapped - last);
+	if (xswap_unmap_clusters_locked(si, start_idx, nr_unmap)) {
+		spin_lock(&si->lock);
+		for (i = start_idx; i < nr_mapped; i++) {
+			ci = &si->cluster_info[i];
+			WRITE_ONCE(ci->flags, CLUSTER_FLAG_FREE);
+			list_add_tail(&ci->list, &si->free_clusters);
+		}
+		spin_unlock(&si->lock);
+		goto out_unlock;
+	}
+
+out_unlock:
+	mutex_unlock(&si->xswap_lock);
 }
 #endif /* CONFIG_XSWAP */
 
@@ -4094,6 +4163,7 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 			}
 		}
 
+		INIT_WORK(&si->xswap_shrink_work, xswap_shrink_work_fn);
 		return 0;
 
 err_unmap:
-- 
2.54.0

[PATCH v2 10/12] mm, swap: refactor swapoff + add xswap_destroy

[Baoquan He]

1. Extract __swapoff() from sys_swapoff(): the core teardown logic
   now lives in __swapoff(), shared by sys_swapoff() and the new
   xswap_destroy(). swap_file operations are guarded with NULL check
   so __swapoff() works for file-less devices too. sys_swapoff()
   retains file-matching; a NULL guard on p->swap_file ensures xswap
   devices are never matched by the file path.
2. Add xswap_destroy(int type): tears down a file-less xswap device
   by its swap type. Validates SWP_XSWAP | SWP_WRITEOK, removes
   from lists, delegates to __swapoff().
3. Add /sys/kernel/mm/xswap/destroy: write a swap type to tear down
   that xswap device. Requires CAP_SYS_ADMIN. Out-of-range types
   are rejected before the implicit int truncation.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 199 ++++++++++++++++++++++++++++++++++----------------
 1 file changed, 138 insertions(+), 61 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index d2a3303b0572..0321b6b0da56 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -74,6 +74,7 @@ static void xswap_try_shrink(struct swap_info_struct *si);
 
 #ifdef CONFIG_SYSFS
 static int xswap_create(int prio);
+static int xswap_destroy(int type);
 
 static ssize_t xswap_create_store(struct kobject *kobj,
 				  struct kobj_attribute *attr,
@@ -104,8 +105,35 @@ static ssize_t xswap_create_store(struct kobject *kobj,
 static struct kobj_attribute xswap_create_attr = __ATTR(create, 0200, NULL,
 							xswap_create_store);
 
+static ssize_t xswap_destroy_store(struct kobject *kobj,
+				   struct kobj_attribute *attr,
+				   const char *buf, size_t count)
+{
+	unsigned long type;
+	int err;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	err = kstrtoul(buf, 0, &type);
+	if (err)
+		return err;
+	if (type >= MAX_SWAPFILES)
+		return -EINVAL;
+
+	err = xswap_destroy(type);
+	if (err)
+		return err;
+
+	return count;
+}
+
+static struct kobj_attribute xswap_destroy_attr = __ATTR(destroy, 0200, NULL,
+							 xswap_destroy_store);
+
 static struct attribute *xswap_attrs[] = {
 	&xswap_create_attr.attr,
+	&xswap_destroy_attr.attr,
 	NULL,
 };
 
@@ -3294,61 +3322,13 @@ static void flush_percpu_swap_cluster(struct swap_info_struct *si)
 }
 
 
-SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
+/* Common swap teardown after list removal; shared by sys_swapoff() and
+ * xswap_destroy().
+ */
+static int __swapoff(struct swap_info_struct *p)
 {
-	struct swap_info_struct *p = NULL;
-	struct file *swap_file, *victim;
-	struct address_space *mapping;
-	struct inode *inode;
-	int err, found = 0;
-
-	if (!capable(CAP_SYS_ADMIN))
-		return -EPERM;
-
-	BUG_ON(!current->mm);
-
-	CLASS(filename, pathname)(specialfile);
-	victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0);
-	if (IS_ERR(victim))
-		return PTR_ERR(victim);
-
-	mapping = victim->f_mapping;
-	spin_lock(&swap_lock);
-	plist_for_each_entry(p, &swap_active_head, list) {
-		if (p->flags & SWP_WRITEOK) {
-			if (p->swap_file->f_mapping == mapping) {
-				found = 1;
-				break;
-			}
-		}
-	}
-	if (!found) {
-		err = -EINVAL;
-		spin_unlock(&swap_lock);
-		goto out_dput;
-	}
-
-	/* Refuse swapoff while the device is pinned for hibernation */
-	if (p->flags & SWP_HIBERNATION) {
-		err = -EBUSY;
-		spin_unlock(&swap_lock);
-		goto out_dput;
-	}
-
-	if (!security_vm_enough_memory_mm(current->mm, p->pages))
-		vm_unacct_memory(p->pages);
-	else {
-		err = -ENOMEM;
-		spin_unlock(&swap_lock);
-		goto out_dput;
-	}
-	spin_lock(&p->lock);
-	del_from_avail_list(p, true);
-	plist_del(&p->list, &swap_active_head);
-	atomic_long_sub(p->pages, &nr_swap_pages);
-	total_swap_pages -= p->pages;
-	spin_unlock(&p->lock);
-	spin_unlock(&swap_lock);
+	struct file *swap_file = NULL;
+	int err;
 
 #ifdef CONFIG_XSWAP
 	if (p->flags & SWP_XSWAP)
@@ -3364,7 +3344,7 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 	if (err) {
 		/* re-insert swap space back into swap_list */
 		reinsert_swap_info(p);
-		goto out_dput;
+		return err;
 	}
 
 	/*
@@ -3407,12 +3387,14 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 	p->max = 0;
 	p->cluster_info = NULL;
 
-	inode = mapping->host;
+	if (swap_file) {
+		struct inode *inode = swap_file->f_mapping->host;
 
-	inode_lock(inode);
-	inode->i_flags &= ~S_SWAPFILE;
-	inode_unlock(inode);
-	filp_close(swap_file, NULL);
+		inode_lock(inode);
+		inode->i_flags &= ~S_SWAPFILE;
+		inode_unlock(inode);
+		filp_close(swap_file, NULL);
+	}
 
 	/*
 	 * Clear the SWP_USED flag after all resources are freed so that swapon
@@ -3423,10 +3405,69 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
 	p->flags = 0;
 	spin_unlock(&swap_lock);
 
-	err = 0;
 	atomic_inc(&proc_poll_event);
 	wake_up_interruptible(&proc_poll_wait);
 
+	return 0;
+}
+
+SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
+{
+	struct swap_info_struct *p = NULL;
+	struct file *victim;
+	struct address_space *mapping;
+	int err, found = 0;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	BUG_ON(!current->mm);
+
+	CLASS(filename, pathname)(specialfile);
+	victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0);
+	if (IS_ERR(victim))
+		return PTR_ERR(victim);
+
+	mapping = victim->f_mapping;
+	spin_lock(&swap_lock);
+	plist_for_each_entry(p, &swap_active_head, list) {
+		if (p->flags & SWP_WRITEOK) {
+			if (p->swap_file && p->swap_file->f_mapping == mapping) {
+				found = 1;
+				break;
+			}
+		}
+	}
+	if (!found) {
+		err = -EINVAL;
+		spin_unlock(&swap_lock);
+		goto out_dput;
+	}
+
+	/* Refuse swapoff while the device is pinned for hibernation */
+	if (p->flags & SWP_HIBERNATION) {
+		err = -EBUSY;
+		spin_unlock(&swap_lock);
+		goto out_dput;
+	}
+
+	if (!security_vm_enough_memory_mm(current->mm, p->pages))
+		vm_unacct_memory(p->pages);
+	else {
+		err = -ENOMEM;
+		spin_unlock(&swap_lock);
+		goto out_dput;
+	}
+	spin_lock(&p->lock);
+	del_from_avail_list(p, true);
+	plist_del(&p->list, &swap_active_head);
+	atomic_long_sub(p->pages, &nr_swap_pages);
+	total_swap_pages -= p->pages;
+	spin_unlock(&p->lock);
+	spin_unlock(&swap_lock);
+
+	err = __swapoff(p);
+
 out_dput:
 	filp_close(victim, NULL);
 	return err;
@@ -4322,6 +4363,42 @@ static int xswap_create(int prio)
 	spin_unlock(&swap_lock);
 	return error;
 }
+
+/* Tear down a file-less xswap device by its swap type. */
+static int xswap_destroy(int type)
+{
+	struct swap_info_struct *p;
+
+	p = swap_type_to_info(type);
+	if (!p)
+		return -EINVAL;
+
+	spin_lock(&swap_lock);
+	if (!(p->flags & SWP_WRITEOK) || !(p->flags & SWP_XSWAP)) {
+		spin_unlock(&swap_lock);
+		return -EINVAL;
+	}
+	/* Refuse swapoff while the device is pinned for hibernation */
+	if (p->flags & SWP_HIBERNATION) {
+		spin_unlock(&swap_lock);
+		return -EBUSY;
+	}
+	if (!security_vm_enough_memory_mm(current->mm, p->pages))
+		vm_unacct_memory(p->pages);
+	else {
+		spin_unlock(&swap_lock);
+		return -ENOMEM;
+	}
+	spin_lock(&p->lock);
+	del_from_avail_list(p, true);
+	plist_del(&p->list, &swap_active_head);
+	atomic_long_sub(p->pages, &nr_swap_pages);
+	total_swap_pages -= p->pages;
+	spin_unlock(&p->lock);
+	spin_unlock(&swap_lock);
+
+	return __swapoff(p);
+}
 #endif /* CONFIG_SYSFS */
 #endif /* CONFIG_XSWAP */
 
-- 
2.54.0

[PATCH v2 11/12] mm, swap: require zswap for xswap devices

[Baoquan He]

xswap has no backing storage: swapped-out pages live only in zswap.
Without zswap, swapout always bounces back, so the device would
consume swap entry space without ever freeing memory. Refuse to
create a device when zswap is unavailable.

Runtime disabling of zswap after creation is safe: existing entries
stay loadable (zswap_load() gates on zswap_never_enabled(), not the
runtime zswap_enabled flag) and new swapouts merely bounce back to
memory without freeing it.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 mm/swapfile.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 0321b6b0da56..c2ca0a9ac625 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -4305,6 +4305,10 @@ static int xswap_create(int prio)
 	if (prio != DEF_SWAP_PRIO && (prio < 0 || prio > SWAP_FLAG_PRIO_MASK))
 		return -EINVAL;
 
+	/* xswap has no backing store, it relies on zswap. */
+	if (!zswap_is_enabled())
+		return -EOPNOTSUPP;
+
 	si = alloc_swap_info();
 	if (IS_ERR(si))
 		return PTR_ERR(si);
-- 
2.54.0

[PATCH v2 12/12] mm, swap: add sysfs per-device size limit for xswap

[Baoquan He]

Add a per-device knob for the xswap size ceiling, si->nr_clusters.
Grow and shrink both work without it; a cap below the mapped range
only makes shrink target the cap directly.

  /sys/kernel/mm/xswap/type<N>/limit    read/write, in pages

Reading reports the usable size, si->pages, in 4 KiB pages. Writing
sets a new ceiling, rounded up to SWAPFILE_CLUSTER (the device's growth
granularity).

Each device gets its own kobject directory under /sys/kernel/mm/xswap/.

Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
---
 include/linux/swap.h |   4 +
 mm/swapfile.c        | 184 ++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 179 insertions(+), 9 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index 6307d8e8f7ed..c76c801c34ea 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -16,6 +16,8 @@
 #include <uapi/linux/mempolicy.h>
 #include <asm/page.h>
 
+struct kobject;
+
 #define SWAP_FLAG_PREFER	0x8000	/* set if swap priority specified */
 #define SWAP_FLAG_PRIO_MASK	0x7fff
 #define SWAP_FLAG_DISCARD	0x10000 /* enable discard for swap */
@@ -251,7 +253,9 @@ struct swap_info_struct {
 #ifdef CONFIG_XSWAP
 	struct vm_struct	*cluster_vm;	/* VM_SPARSE area for cluster_info */
 	unsigned long		nr_clusters_max;/* total clusters in the xswap address space */
+	unsigned long		nr_clusters;	/* growth ceiling, set by type<N>/limit */
 	unsigned long		nr_clusters_mapped; /* currently mapped cluster count */
+	struct kobject		*xswap_dev_kobj; /* sysfs: /sys/kernel/mm/xswap/type<N>/ */
 	struct work_struct	xswap_shrink_work; /* deferred shrink trigger */
 	struct mutex		xswap_lock;	/* serialize map/unmap operations */
 #endif
diff --git a/mm/swapfile.c b/mm/swapfile.c
index c2ca0a9ac625..fd3bc83f6644 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -50,8 +50,13 @@
 #include "swap_table.h"
 #include "internal.h"
 #include "swap.h"
+
 #define DEF_SWAP_PRIO  -1
 
+static DEFINE_SPINLOCK(swap_lock);
+
+static long swap_usage_in_pages(struct swap_info_struct *si);
+
 #ifdef CONFIG_XSWAP
 /*
  * xswap: dynamically grow and shrink the cluster_info array via a
@@ -158,6 +163,142 @@ static inline void xswap_sysfs_init(void)
 {
 }
 #endif /* CONFIG_SYSFS */
+
+#ifdef CONFIG_SYSFS
+/*
+ * Per-device directory: /sys/kernel/mm/xswap/type<N>/limit (rw).
+ * It is the runtime size ceiling of the device, in pages (rounded up to
+ * SWAPFILE_CLUSTER, the device's growth granularity).  Lowering it below
+ * the current usage only stops further growth down to the used size and
+ * triggers shrink of the free tail.
+ */
+struct xswap_sysfs_dev {
+	struct kobject kobj;
+	struct swap_info_struct *si;
+};
+
+static ssize_t xswap_limit_show(struct kobject *kobj,
+				struct kobj_attribute *attr, char *buf)
+{
+	struct swap_info_struct *si =
+		container_of(kobj, struct xswap_sysfs_dev, kobj)->si;
+
+	return sysfs_emit(buf, "%u\n", READ_ONCE(si->pages));
+}
+
+static ssize_t xswap_limit_store(struct kobject *kobj,
+				 struct kobj_attribute *attr,
+				 const char *buf, size_t count)
+{
+	struct swap_info_struct *si =
+		container_of(kobj, struct xswap_sysfs_dev, kobj)->si;
+	unsigned long val, clusters, new_pages;
+	int err;
+
+	err = kstrtoul(buf, 0, &val);
+	if (err)
+		return err;
+
+	spin_lock(&swap_lock);
+	/* Reject writes once swapoff has cleared SWP_WRITEOK. */
+	if (!(si->flags & SWP_WRITEOK)) {
+		spin_unlock(&swap_lock);
+		return -ENODEV;
+	}
+
+	clusters = DIV_ROUND_UP(val, SWAPFILE_CLUSTER);
+	if (clusters > si->nr_clusters_max)
+		clusters = si->nr_clusters_max;
+
+	spin_lock(&si->lock);
+	si->nr_clusters = clusters;
+	spin_unlock(&si->lock);
+
+	new_pages = min_t(unsigned long, clusters * SWAPFILE_CLUSTER, si->max);
+	if (new_pages)
+		new_pages--;
+
+	if (new_pages < swap_usage_in_pages(si))
+		new_pages = swap_usage_in_pages(si);
+	if (new_pages != si->pages) {
+		long delta = (long)new_pages - (long)si->pages;
+
+		si->pages = new_pages;
+		atomic_long_add(delta, &nr_swap_pages);
+		total_swap_pages += delta;
+	}
+	spin_unlock(&swap_lock);
+
+	/* Lowering the ceiling may free tail clusters. */
+	xswap_try_shrink(si);
+
+	return count;
+}
+
+static struct kobj_attribute xswap_limit_attr =
+	__ATTR(limit, 0644, xswap_limit_show, xswap_limit_store);
+
+static void xswap_dev_release(struct kobject *kobj)
+{
+	kfree(container_of(kobj, struct xswap_sysfs_dev, kobj));
+}
+
+static const struct kobj_type xswap_dev_ktype = {
+	.sysfs_ops = &kobj_sysfs_ops,
+	.release = xswap_dev_release,
+};
+
+static int xswap_dev_kobj_add(struct swap_info_struct *si)
+{
+	struct xswap_sysfs_dev *dev;
+	int err;
+
+	if (!xswap_kobj)
+		return 0;
+
+	dev = kzalloc_obj(*dev, GFP_KERNEL);
+	if (!dev)
+		return -ENOMEM;
+	dev->si = si;
+
+	err = kobject_init_and_add(&dev->kobj, &xswap_dev_ktype, xswap_kobj,
+				   "type%d", si->type);
+	if (err) {
+		kobject_put(&dev->kobj);
+		return err;
+	}
+
+	err = sysfs_create_file(&dev->kobj, &xswap_limit_attr.attr);
+	if (err) {
+		kobject_del(&dev->kobj);
+		kobject_put(&dev->kobj);
+		return err;
+	}
+	si->xswap_dev_kobj = &dev->kobj;
+	return 0;
+}
+
+static void xswap_dev_kobj_del(struct swap_info_struct *si)
+{
+	struct kobject *kobj = si->xswap_dev_kobj;
+
+	if (!kobj)
+		return;
+	si->xswap_dev_kobj = NULL;
+	sysfs_remove_file(kobj, &xswap_limit_attr.attr);
+	kobject_del(kobj);
+	kobject_put(kobj);
+}
+#else /* !CONFIG_SYSFS */
+static inline int xswap_dev_kobj_add(struct swap_info_struct *si)
+{
+	return 0;
+}
+
+static inline void xswap_dev_kobj_del(struct swap_info_struct *si)
+{
+}
+#endif /* CONFIG_SYSFS */
 #else /* !CONFIG_XSWAP */
 static inline void xswap_sysfs_init(void)
 {
@@ -178,7 +319,6 @@ static void move_cluster(struct swap_info_struct *si,
  *
  * Also protects swap_active_head total_swap_pages, and the SWP_WRITEOK flag.
  */
-static DEFINE_SPINLOCK(swap_lock);
 static unsigned int nr_swapfiles;
 atomic_long_t nr_swap_pages;
 atomic_t nr_real_swapfiles;
@@ -3274,6 +3414,7 @@ static void free_swap_cluster_info(struct swap_info_struct *si)
 
 #ifdef CONFIG_XSWAP
 	if (si->flags & SWP_XSWAP) {
+		xswap_dev_kobj_del(si);
 		cancel_work_sync(&si->xswap_shrink_work);
 		/* Unmap all mapped clusters and free the VM_SPARSE area */
 		if (si->nr_clusters_mapped > 0)
@@ -4057,8 +4198,9 @@ static void xswap_shrink_work_fn(struct work_struct *work)
  */
 static void xswap_try_shrink(struct swap_info_struct *si)
 {
+	unsigned long nr_mapped, nr_ceiling, nr_tail, nr_unmap;
+	unsigned long start_idx, slack, min, i;
 	struct swap_cluster_info *ci;
-	unsigned long nr_mapped, nr_tail, nr_unmap, start_idx, i;
 
 	if (!(si->flags & SWP_XSWAP))
 		return;
@@ -4066,18 +4208,34 @@ static void xswap_try_shrink(struct swap_info_struct *si)
 	mutex_lock(&si->xswap_lock);
 
 	nr_mapped = READ_ONCE(si->nr_clusters_mapped);
+	nr_ceiling = READ_ONCE(si->nr_clusters);
+
 	if (nr_mapped <= 1)	/* keep cluster 0 */
 		goto out_unlock;
 
 	/*
-	 * Reclaim on our own, but only once the mapped range is at most
-	 * half in use: growth is demand driven, so reclaiming on a smaller
-	 * dip would only map the same clusters again, and every unmap costs
-	 * an RCU grace period.
+	 * A cap below the mapped range is reason enough to reclaim on its
+	 * own; otherwise only once the mapped range is at most half in use,
+	 * because growth is demand driven and reclaiming on a smaller dip
+	 * would only map the same clusters again, at the price of an RCU
+	 * grace period per unmap.
 	 */
-	if (atomic_long_read(&si->inuse_pages) * 2 > nr_mapped * SWAPFILE_CLUSTER)
+	if (nr_ceiling >= nr_mapped &&
+	    atomic_long_read(&si->inuse_pages) * 2 > nr_mapped * SWAPFILE_CLUSTER)
 		goto out_unlock;
 
+	/*
+	 * Keep one chunk of free tail as slack so that the next allocation
+	 * does not grow the range back immediately - unless a cap asks for
+	 * the whole tail.
+	 */
+	slack = XSWAP_SHRINK_SLACK;
+	min = XSWAP_SHRINK_MIN;
+	if (nr_ceiling < nr_mapped) {
+		slack = 0;
+		min = XSWAP_GROW_CLUSTERS;
+	}
+
 	/*
 	 * Count the free clusters at the tail of the mapped range.  Scanned,
 	 * not tracked: the count must be exact to size the unmap, and an
@@ -4091,10 +4249,13 @@ static void xswap_try_shrink(struct swap_info_struct *si)
 			break;
 		nr_tail++;
 	}
-	if (nr_tail < XSWAP_SHRINK_SLACK + XSWAP_SHRINK_MIN)
+	if (nr_tail < slack + min)
 		goto out_unlock;
 
-	nr_unmap = rounddown(nr_tail - XSWAP_SHRINK_SLACK, XSWAP_GROW_CLUSTERS);
+	nr_unmap = rounddown(nr_tail - slack, XSWAP_GROW_CLUSTERS);
+	if (nr_ceiling < nr_mapped)
+		nr_unmap = min(nr_unmap, rounddown(nr_mapped - nr_ceiling,
+						   XSWAP_GROW_CLUSTERS));
 	if (!nr_unmap)
 		goto out_unlock;
 	start_idx = nr_mapped - nr_unmap;
@@ -4164,6 +4325,8 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 		cluster_info = vm->addr;
 		si->cluster_vm = vm;
 		si->nr_clusters_max = nr_clusters;
+		/* No cap unless type<N>/limit is written. */
+		si->nr_clusters = nr_clusters;
 		si->cluster_info = cluster_info;
 
 		/* Must be initialized before xswap_map_clusters() locks it. */
@@ -4205,6 +4368,9 @@ static int setup_swap_clusters_info(struct swap_info_struct *si,
 		}
 
 		INIT_WORK(&si->xswap_shrink_work, xswap_shrink_work_fn);
+		if (xswap_dev_kobj_add(si))
+			pr_warn("xswap: failed to add sysfs interface for type %d\n",
+				si->type);
 		return 0;
 
 err_unmap:
-- 
2.54.0

Re: [PATCH v2 01/12] mm: xswap support for zswap

[Kunwu Chan]

On Sun, 13 Sep 2026 15:50:03 +0800 Baoquan He <hebaoquan@kylinos.cn> wrote:

> From: Chris Li <chrisl@kernel.org>
> 
> Introduce extendable swap device support - xswap.
> 
> An xswap device has no backing storage and no swap data section, so
> it wastes no disk space. Creation is via a sysfs interface added in a
> later patch.
> 
> Zswap writeback is gated on whether a real (non-xswap) swap device is
> active. nr_real_swapfiles counts such devices and is maintained at
> swapon/swapoff only, so the gate reflects "a device exists to write
> back to" rather than "a device currently has free slots". This keeps
> writeback working even when the real swap device is full, and avoids
> a double decrement when a full device is swapped off.
> 
> Co-developed-by: Baoquan He <hebaoquan@kylinos.cn>
> Signed-off-by: Baoquan He <hebaoquan@kylinos.cn>
> Signed-off-by: Chris Li <chrisl@kernel.org>
> ---
>  include/linux/swap.h |  2 ++
>  mm/page_io.c         | 16 ++++++++++++++++
>  mm/swap_state.c      |  7 +++++++
>  mm/swapfile.c        | 38 +++++++++++++++++++++++++++++++++++---
>  mm/zswap.c           |  7 ++++++-
>  5 files changed, 66 insertions(+), 4 deletions(-)
> 
> diff --git a/include/linux/swap.h b/include/linux/swap.h
> index 5658a1634b85..787fe463dcbb 100644
> --- a/include/linux/swap.h
> +++ b/include/linux/swap.h
> @@ -207,6 +207,7 @@ enum {
>  	SWP_STABLE_WRITES = (1 << 11),	/* no overwrite PG_writeback pages */
>  	SWP_SYNCHRONOUS_IO = (1 << 12),	/* synchronous IO is efficient */
>  	SWP_HIBERNATION = (1 << 13),	/* pinned for hibernation */
> +	SWP_XSWAP	= (1 << 14),	/* extendable swap device */
>  					/* add others here before... */
>  };
>  
> @@ -356,6 +357,7 @@ void free_folio_and_swap_cache(struct folio *folio);
>  void free_pages_and_swap_cache(struct encoded_page **, int);
>  /* linux/mm/swapfile.c */
>  extern atomic_long_t nr_swap_pages;
> +extern atomic_t nr_real_swapfiles;
>  extern long total_swap_pages;
>  extern atomic_t nr_rotate_swap;
>  
> diff --git a/mm/page_io.c b/mm/page_io.c
> index 88962571cb93..5483c943e3e3 100644
> --- a/mm/page_io.c
> +++ b/mm/page_io.c
> @@ -248,6 +248,17 @@ int swap_writeout(struct swap_io_ctx *ctx, struct folio *folio)
>  	}
>  	rcu_read_unlock();
>  
> +	/*
> +	 * ctx->sis is set by swap_add_folio() which is called from
> +	 * __swap_writepage() below.  Since we must avoid the writepage
> +	 * path for xswap devices, use the swap_info from the folio's
> +	 * swap entry directly instead of going through ctx.
> +	 */
> +	if (unlikely(__swap_entry_to_info(folio->swap)->flags & SWP_XSWAP)) {
> +		folio_mark_dirty(folio);
> +		return AOP_WRITEPAGE_ACTIVATE;
> +	}
> +
>  	__swap_writepage(ctx, folio);
>  	return 0;
>  out_unlock:
> @@ -480,6 +491,11 @@ void swap_read_folio(struct swap_io_ctx *ctx, struct folio *folio)
>  	if (zswap_load(folio) != -ENOENT)
>  		goto finish;
>  
> +	if (unlikely(sis->flags & SWP_XSWAP)) {
> +		folio_unlock(folio);
> +		goto finish;
> +	}
> +
>  	/* We have to read from slower devices. Increase zswap protection. */
>  	zswap_folio_swapin(folio);
>  	swap_add_folio(ctx, folio, READ);
> diff --git a/mm/swap_state.c b/mm/swap_state.c
> index b76eb3d876fd..2eedb7a3d7bb 100644
> --- a/mm/swap_state.c
> +++ b/mm/swap_state.c
> @@ -830,6 +830,13 @@ struct folio *swap_cluster_readahead(swp_entry_t entry, gfp_t gfp_mask,
>  	struct blk_plug plug;
>  	swp_entry_t ra_entry;
>  
> +	/*
> +	 * The entry may have been freed by another task. Avoid swap_info_get()
> +	 * which will print error message if the race happens.
> +	 */
> +	if (si->flags & SWP_XSWAP)
> +		goto skip;
> +
>  	mask = swapin_nr_pages(offset) - 1;
>  	if (!mask)
>  		goto skip;
> diff --git a/mm/swapfile.c b/mm/swapfile.c
> index 53bf01d5f7f1..193b08a54908 100644
> --- a/mm/swapfile.c
> +++ b/mm/swapfile.c
> @@ -66,6 +66,7 @@ static void move_cluster(struct swap_info_struct *si,
>  static DEFINE_SPINLOCK(swap_lock);
>  static unsigned int nr_swapfiles;
>  atomic_long_t nr_swap_pages;
> +atomic_t nr_real_swapfiles;
>  /*
>   * Some modules use swappable objects and may try to swap them out under
>   * memory pressure (via the shrinker). Before doing so, they may wish to
> @@ -1208,6 +1209,9 @@ static void del_from_avail_list(struct swap_info_struct *si, bool swapoff)
>  		 */
>  		lockdep_assert_held(&si->lock);
>  		si->flags &= ~SWP_WRITEOK;
> +		/* Count active devices, not merely those on the avail list. */
> +		if (!(si->flags & SWP_XSWAP))
> +			atomic_sub(1, &nr_real_swapfiles);
>  		atomic_long_or(SWAP_USAGE_OFFLIST_BIT, &si->inuse_pages);
>  	} else {
>  		/*
> @@ -1265,6 +1269,8 @@ static void add_to_avail_list(struct swap_info_struct *si, bool swapon)
>  	}
>  
>  	plist_add(&si->avail_list, &swap_avail_head);
> +	if (swapon && !(si->flags & SWP_XSWAP))
> +		atomic_add(1, &nr_real_swapfiles);
>  
>  skip:
>  	spin_unlock(&swap_avail_lock);
> @@ -2959,6 +2965,19 @@ static int setup_swap_extents(struct swap_info_struct *sis,
>  	struct inode *inode = mapping->host;
>  	int ret;
>  
> +	if (sis->flags & SWP_XSWAP) {
> +		*span = 0;
> +		/*
> +		 * xswap devices have no backing block device and
> +		 * physical writeout is skipped in swap_writeout(),
> +		 * but sis->ops must still be set so that callers
> +		 * like shrink_folio_list() can safely dereference
> +		 * ops->flags.
> +		 */
> +		sis->ops = &swap_bdev_ops;
> +		return 0;
> +	}
> +
>  	ret = sio_pool_init();
>  	if (ret)
>  		return ret;
> @@ -3167,7 +3186,8 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
>  
>  	destroy_swap_extents(p, p->swap_file);
>  
> -	if (!(p->flags & SWP_SOLIDSTATE))
> +	if (!(p->flags & SWP_XSWAP) &&
> +	    !(p->flags & SWP_SOLIDSTATE))
>  		atomic_dec(&nr_rotate_swap);
>  
>  	mutex_lock(&swapon_mutex);
> @@ -3277,6 +3297,19 @@ static void swap_stop(struct seq_file *swap, void *v)
>  	mutex_unlock(&swapon_mutex);
>  }
>  
> +static const char *swap_type_str(struct swap_info_struct *si)
> +{
> +	struct file *file = si->swap_file;
> +
> +	if (si->flags & SWP_XSWAP)
> +		return "xswap\t";
> +
> +	if (S_ISBLK(file_inode(file)->i_mode))
> +		return "partition";
> +
> +	return "file\t";
> +}
> +
>  static int swap_show(struct seq_file *swap, void *v)
>  {
>  	struct swap_info_struct *si = v;
> @@ -3296,8 +3329,7 @@ static int swap_show(struct seq_file *swap, void *v)
>  	len = seq_file_path(swap, file, " \t\n\\");
>  	seq_printf(swap, "%*s%s\t%lu\t%s%lu\t%s%d\n",
>  			len < 40 ? 40 - len : 1, " ",
> -			S_ISBLK(file_inode(file)->i_mode) ?
> -				"partition" : "file\t",
> +			swap_type_str(si),
>  			bytes, bytes < 10000000 ? "\t" : "",
>  			inuse, inuse < 10000000 ? "\t" : "",
>  			si->prio);
> diff --git a/mm/zswap.c b/mm/zswap.c
> index b9948d4657d2..064970a4393f 100644
> --- a/mm/zswap.c
> +++ b/mm/zswap.c
> @@ -1000,6 +1000,11 @@ static int zswap_writeback_entry(struct zswap_entry *entry,
>  	if (!si)
>  		return -ENOENT;
>  
> +	if (si->flags & SWP_XSWAP) {
> +		put_swap_device(si);
> +		return -EINVAL;
> +	}
> +
>  	mpol = get_task_policy(current);
>  	folio = swap_cache_alloc_folio(swpentry, GFP_KERNEL, BIT(0), NULL, mpol,
>  				       NO_INTERLEAVE_INDEX);
> @@ -1545,7 +1550,7 @@ bool zswap_store(struct folio *folio)
>  	zswap_pool_put(pool);
>  put_objcg:
>  	obj_cgroup_put(objcg);
> -	if (!ret && zswap_pool_reached_full)
> +	if (!ret && zswap_pool_reached_full && atomic_read(&nr_real_swapfiles))
>  		queue_work(shrink_wq, &zswap_shrink_work);
>  check_old:
>  	/*
> -- 
> 2.54.0
> 
> 

The SWP_XSWAP flag and nr_real_swapfiles accounting look correct. The writeback guard 
in page_io.c (both swap_writeout and swap_read_folio paths) properly short-circuits 
xswap devices, and the zswap writeback gating on nr_real_swapfiles ensures writeback 
only fires when a real swap device is active. 

No issues found.

Reviewed-by: Kunwu Chan <chentao@kylinos.cn>

Thanks,
Kunwu

Re: [PATCH v2 01/12] mm: xswap support for zswap

[Baoquan He]

On 09/15/26 at 06:03pm, Kunwu Chan wrote:
> On Sun, 13 Sep 2026 15:50:03 +0800 Baoquan He <hebaoquan@kylinos.cn> wrote:
...snip... 
> 
> The SWP_XSWAP flag and nr_real_swapfiles accounting look correct. The writeback guard 
> in page_io.c (both swap_writeout and swap_read_folio paths) properly short-circuits 
> xswap devices, and the zswap writeback gating on nr_real_swapfiles ensures writeback 
> only fires when a real swap device is active. 
> 
> No issues found.
> 
> Reviewed-by: Kunwu Chan <chentao@kylinos.cn>

Thanks for careful reviewing. I will post v3 soon. I found this v2 is
based on a older commit of mm-unstable, and I forgot adding base-id, so
Sashiko reviewing is failed to launch. I have rebased this onto the
latest mm-new, hope Sashiko can be satisfied.

Re: [PATCH v2 01/12] mm: xswap support for zswap

[KunWu Chan]

On Wed, Sep 16, 2026 at 10:52 AM Baoquan He <baoquan.he@linux.dev> wrote:
>
> On 09/15/26 at 06:03pm, Kunwu Chan wrote:
> > On Sun, 13 Sep 2026 15:50:03 +0800 Baoquan He <hebaoquan@kylinos.cn> wrote:
> ...snip...
> >
> > The SWP_XSWAP flag and nr_real_swapfiles accounting look correct. The writeback guard
> > in page_io.c (both swap_writeout and swap_read_folio paths) properly short-circuits
> > xswap devices, and the zswap writeback gating on nr_real_swapfiles ensures writeback
> > only fires when a real swap device is active.
> >
> > No issues found.
> >
> > Reviewed-by: Kunwu Chan <chentao@kylinos.cn>
>
> Thanks for careful reviewing. I will post v3 soon. I found this v2 is
> based on a older commit of mm-unstable, and I forgot adding base-id, so
> Sashiko reviewing is failed to launch. I have rebased this onto the
> latest mm-new, hope Sashiko can be satisfied.

Thanks, Baoquan.

I’ll keep following the series and look forward to reviewing v3.