[PATCH v4 00/15] virtio-mem: paravirtualized memory

[PATCH v4 00/15] virtio-mem: paravirtualized memory

[David Hildenbrand]

This series is based on v5.7-rc4. The patches are located at:
    https://github.com/davidhildenbrand/linux.git virtio-mem-v4

This is basically a resend of v3 [1], now based on v5.7-rc4 and restested.
One patch was reshuffled and two ACKs I missed to add were added. The
rebase did not require any modifications to patches.

Details about virtio-mem can be found in the cover letter of v2 [2]. A
basic QEMU implementation was posted yesterday [3].

[1] https://lkml.kernel.org/r/20200507103119.11219-1-david@redhat.com
[2] https://lkml.kernel.org/r/20200311171422.10484-1-david@redhat.com
[3] https://lkml.kernel.org/r/20200506094948.76388-1-david@redhat.com

v3 -> v4:
- Move "MAINTAINERS: Add myself as virtio-mem maintainer" to #2
- Add two ACKs from Andrew (in reply to v2)
-- "mm: Allow to offline unmovable PageOffline() pages via ..."
-- "mm/memory_hotplug: Introduce offline_and_remove_memory()"

v2 -> v3:
- "virtio-mem: Paravirtualized memory hotplug"
-- Include "linux/slab.h" to fix build issues
-- Remember the "region_size", helpful for patch #11
-- Minor simplifaction in virtio_mem_overlaps_range()
-- Use notifier_from_errno() instead of notifier_to_errno() in notifier
-- More reliable check for added memory when unloading the driver
- "virtio-mem: Allow to specify an ACPI PXM as nid"
-- Also print the nid
- Added patch #11-#15

David Hildenbrand (15):
  virtio-mem: Paravirtualized memory hotplug
  MAINTAINERS: Add myself as virtio-mem maintainer
  virtio-mem: Allow to specify an ACPI PXM as nid
  virtio-mem: Paravirtualized memory hotunplug part 1
  virtio-mem: Paravirtualized memory hotunplug part 2
  mm: Allow to offline unmovable PageOffline() pages via
    MEM_GOING_OFFLINE
  virtio-mem: Allow to offline partially unplugged memory blocks
  mm/memory_hotplug: Introduce offline_and_remove_memory()
  virtio-mem: Offline and remove completely unplugged memory blocks
  virtio-mem: Better retry handling
  virtio-mem: Add parent resource for all added "System RAM"
  virtio-mem: Drop manual check for already present memory
  virtio-mem: Unplug subblocks right-to-left
  virtio-mem: Use -ETXTBSY as error code if the device is busy
  virtio-mem: Try to unplug the complete online memory block first

 MAINTAINERS                     |    7 +
 drivers/acpi/numa/srat.c        |    1 +
 drivers/virtio/Kconfig          |   17 +
 drivers/virtio/Makefile         |    1 +
 drivers/virtio/virtio_mem.c     | 1962 +++++++++++++++++++++++++++++++
 include/linux/memory_hotplug.h  |    1 +
 include/linux/page-flags.h      |   10 +
 include/uapi/linux/virtio_ids.h |    1 +
 include/uapi/linux/virtio_mem.h |  208 ++++
 mm/memory_hotplug.c             |   81 +-
 mm/page_alloc.c                 |   26 +
 mm/page_isolation.c             |    9 +
 12 files changed, 2314 insertions(+), 10 deletions(-)
 create mode 100644 drivers/virtio/virtio_mem.c
 create mode 100644 include/uapi/linux/virtio_mem.h

-- 
2.25.3

[PATCH v4 01/15] virtio-mem: Paravirtualized memory hotplug

[David Hildenbrand]

Each virtio-mem device owns exactly one memory region. It is responsible
for adding/removing memory from that memory region on request.

When the device driver starts up, the requested amount of memory is
queried and then plugged to Linux. On request, further memory can be
plugged or unplugged. This patch only implements the plugging part.

On x86-64, memory can currently be plugged in 4MB ("subblock") granularity.
When required, a new memory block will be added (e.g., usually 128MB on
x86-64) in order to plug more subblocks. Only x86-64 was tested for now.

The online_page callback is used to keep unplugged subblocks offline
when onlining memory - similar to the Hyper-V balloon driver. Unplugged
pages are marked PG_offline, to tell dump tools (e.g., makedumpfile) to
skip them.

User space is usually responsible for onlining the added memory. The
memory hotplug notifier is used to synchronize virtio-mem activity
against memory onlining/offlining.

Each virtio-mem device can belong to a NUMA node, which allows us to
easily add/remove small chunks of memory to/from a specific NUMA node by
using multiple virtio-mem devices. Something that works even when the
guest has no idea about the NUMA topology.

One way to view virtio-mem is as a "resizable DIMM" or a DIMM with many
"sub-DIMMS".

This patch directly introduces the basic infrastructure to implement memory
unplug. Especially the memory block states and subblock bitmaps will be
heavily used there.

Notes:
- In case memory is to be onlined by user space, we limit the amount of
  offline memory blocks, to not run out of memory. This is esp. an
  issue if memory is added faster than it is getting onlined.
- Suspend/Hibernate is not supported due to the way virtio-mem devices
  behave. Limited support might be possible in the future.
- Reloading the device driver is not supported.

Reviewed-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Len Brown <lenb@kernel.org>
Cc: linux-acpi@vger.kernel.org
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/Kconfig          |   16 +
 drivers/virtio/Makefile         |    1 +
 drivers/virtio/virtio_mem.c     | 1533 +++++++++++++++++++++++++++++++
 include/uapi/linux/virtio_ids.h |    1 +
 include/uapi/linux/virtio_mem.h |  200 ++++
 5 files changed, 1751 insertions(+)
 create mode 100644 drivers/virtio/virtio_mem.c
 create mode 100644 include/uapi/linux/virtio_mem.h

diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index 69a32dfc318a..d6dde7d2cf76 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -78,6 +78,22 @@ config VIRTIO_BALLOON
 
 	 If unsure, say M.
 
+config VIRTIO_MEM
+	tristate "Virtio mem driver"
+	default m
+	depends on X86_64
+	depends on VIRTIO
+	depends on MEMORY_HOTPLUG_SPARSE
+	depends on MEMORY_HOTREMOVE
+	help
+	 This driver provides access to virtio-mem paravirtualized memory
+	 devices, allowing to hotplug and hotunplug memory.
+
+	 This driver was only tested under x86-64, but should theoretically
+	 work on all architectures that support memory hotplug and hotremove.
+
+	 If unsure, say M.
+
 config VIRTIO_INPUT
 	tristate "Virtio input driver"
 	depends on VIRTIO
diff --git a/drivers/virtio/Makefile b/drivers/virtio/Makefile
index 29a1386ecc03..4d993791f2d7 100644
--- a/drivers/virtio/Makefile
+++ b/drivers/virtio/Makefile
@@ -7,3 +7,4 @@ virtio_pci-$(CONFIG_VIRTIO_PCI_LEGACY) += virtio_pci_legacy.o
 obj-$(CONFIG_VIRTIO_BALLOON) += virtio_balloon.o
 obj-$(CONFIG_VIRTIO_INPUT) += virtio_input.o
 obj-$(CONFIG_VIRTIO_VDPA) += virtio_vdpa.o
+obj-$(CONFIG_VIRTIO_MEM) += virtio_mem.o
diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
new file mode 100644
index 000000000000..5d1dcaa6fc42
--- /dev/null
+++ b/drivers/virtio/virtio_mem.c
@@ -0,0 +1,1533 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Virtio-mem device driver.
+ *
+ * Copyright Red Hat, Inc. 2020
+ *
+ * Author(s): David Hildenbrand <david@redhat.com>
+ */
+
+#include <linux/virtio.h>
+#include <linux/virtio_mem.h>
+#include <linux/workqueue.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/memory_hotplug.h>
+#include <linux/memory.h>
+#include <linux/hrtimer.h>
+#include <linux/crash_dump.h>
+#include <linux/mutex.h>
+#include <linux/bitmap.h>
+#include <linux/lockdep.h>
+
+enum virtio_mem_mb_state {
+	/* Unplugged, not added to Linux. Can be reused later. */
+	VIRTIO_MEM_MB_STATE_UNUSED = 0,
+	/* (Partially) plugged, not added to Linux. Error on add_memory(). */
+	VIRTIO_MEM_MB_STATE_PLUGGED,
+	/* Fully plugged, fully added to Linux, offline. */
+	VIRTIO_MEM_MB_STATE_OFFLINE,
+	/* Partially plugged, fully added to Linux, offline. */
+	VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL,
+	/* Fully plugged, fully added to Linux, online (!ZONE_MOVABLE). */
+	VIRTIO_MEM_MB_STATE_ONLINE,
+	/* Partially plugged, fully added to Linux, online (!ZONE_MOVABLE). */
+	VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL,
+	/*
+	 * Fully plugged, fully added to Linux, online (ZONE_MOVABLE).
+	 * We are not allowed to allocate (unplug) parts of this block that
+	 * are not movable (similar to gigantic pages). We will never allow
+	 * to online OFFLINE_PARTIAL to ZONE_MOVABLE (as they would contain
+	 * unmovable parts).
+	 */
+	VIRTIO_MEM_MB_STATE_ONLINE_MOVABLE,
+	VIRTIO_MEM_MB_STATE_COUNT
+};
+
+struct virtio_mem {
+	struct virtio_device *vdev;
+
+	/* We might first have to unplug all memory when starting up. */
+	bool unplug_all_required;
+
+	/* Workqueue that processes the plug/unplug requests. */
+	struct work_struct wq;
+	atomic_t config_changed;
+
+	/* Virtqueue for guest->host requests. */
+	struct virtqueue *vq;
+
+	/* Wait for a host response to a guest request. */
+	wait_queue_head_t host_resp;
+
+	/* Space for one guest request and the host response. */
+	struct virtio_mem_req req;
+	struct virtio_mem_resp resp;
+
+	/* The current size of the device. */
+	uint64_t plugged_size;
+	/* The requested size of the device. */
+	uint64_t requested_size;
+
+	/* The device block size (for communicating with the device). */
+	uint32_t device_block_size;
+	/* Physical start address of the memory region. */
+	uint64_t addr;
+	/* Maximum region size in bytes. */
+	uint64_t region_size;
+
+	/* The subblock size. */
+	uint32_t subblock_size;
+	/* The number of subblocks per memory block. */
+	uint32_t nb_sb_per_mb;
+
+	/* Id of the first memory block of this device. */
+	unsigned long first_mb_id;
+	/* Id of the last memory block of this device. */
+	unsigned long last_mb_id;
+	/* Id of the last usable memory block of this device. */
+	unsigned long last_usable_mb_id;
+	/* Id of the next memory bock to prepare when needed. */
+	unsigned long next_mb_id;
+
+	/* Summary of all memory block states. */
+	unsigned long nb_mb_state[VIRTIO_MEM_MB_STATE_COUNT];
+#define VIRTIO_MEM_NB_OFFLINE_THRESHOLD		10
+
+	/*
+	 * One byte state per memory block.
+	 *
+	 * Allocated via vmalloc(). When preparing new blocks, resized
+	 * (alloc+copy+free) when needed (crossing pages with the next mb).
+	 * (when crossing pages).
+	 *
+	 * With 128MB memory blocks, we have states for 512GB of memory in one
+	 * page.
+	 */
+	uint8_t *mb_state;
+
+	/*
+	 * $nb_sb_per_mb bit per memory block. Handled similar to mb_state.
+	 *
+	 * With 4MB subblocks, we manage 128GB of memory in one page.
+	 */
+	unsigned long *sb_bitmap;
+
+	/*
+	 * Mutex that protects the nb_mb_state, mb_state, and sb_bitmap.
+	 *
+	 * When this lock is held the pointers can't change, ONLINE and
+	 * OFFLINE blocks can't change the state and no subblocks will get
+	 * plugged.
+	 */
+	struct mutex hotplug_mutex;
+	bool hotplug_active;
+
+	/* An error occurred we cannot handle - stop processing requests. */
+	bool broken;
+
+	/* The driver is being removed. */
+	spinlock_t removal_lock;
+	bool removing;
+
+	/* Timer for retrying to plug/unplug memory. */
+	struct hrtimer retry_timer;
+#define VIRTIO_MEM_RETRY_TIMER_MS		30000
+
+	/* Memory notifier (online/offline events). */
+	struct notifier_block memory_notifier;
+
+	/* Next device in the list of virtio-mem devices. */
+	struct list_head next;
+};
+
+/*
+ * We have to share a single online_page callback among all virtio-mem
+ * devices. We use RCU to iterate the list in the callback.
+ */
+static DEFINE_MUTEX(virtio_mem_mutex);
+static LIST_HEAD(virtio_mem_devices);
+
+static void virtio_mem_online_page_cb(struct page *page, unsigned int order);
+
+/*
+ * Register a virtio-mem device so it will be considered for the online_page
+ * callback.
+ */
+static int register_virtio_mem_device(struct virtio_mem *vm)
+{
+	int rc = 0;
+
+	/* First device registers the callback. */
+	mutex_lock(&virtio_mem_mutex);
+	if (list_empty(&virtio_mem_devices))
+		rc = set_online_page_callback(&virtio_mem_online_page_cb);
+	if (!rc)
+		list_add_rcu(&vm->next, &virtio_mem_devices);
+	mutex_unlock(&virtio_mem_mutex);
+
+	return rc;
+}
+
+/*
+ * Unregister a virtio-mem device so it will no longer be considered for the
+ * online_page callback.
+ */
+static void unregister_virtio_mem_device(struct virtio_mem *vm)
+{
+	/* Last device unregisters the callback. */
+	mutex_lock(&virtio_mem_mutex);
+	list_del_rcu(&vm->next);
+	if (list_empty(&virtio_mem_devices))
+		restore_online_page_callback(&virtio_mem_online_page_cb);
+	mutex_unlock(&virtio_mem_mutex);
+
+	synchronize_rcu();
+}
+
+/*
+ * Calculate the memory block id of a given address.
+ */
+static unsigned long virtio_mem_phys_to_mb_id(unsigned long addr)
+{
+	return addr / memory_block_size_bytes();
+}
+
+/*
+ * Calculate the physical start address of a given memory block id.
+ */
+static unsigned long virtio_mem_mb_id_to_phys(unsigned long mb_id)
+{
+	return mb_id * memory_block_size_bytes();
+}
+
+/*
+ * Calculate the subblock id of a given address.
+ */
+static unsigned long virtio_mem_phys_to_sb_id(struct virtio_mem *vm,
+					      unsigned long addr)
+{
+	const unsigned long mb_id = virtio_mem_phys_to_mb_id(addr);
+	const unsigned long mb_addr = virtio_mem_mb_id_to_phys(mb_id);
+
+	return (addr - mb_addr) / vm->subblock_size;
+}
+
+/*
+ * Set the state of a memory block, taking care of the state counter.
+ */
+static void virtio_mem_mb_set_state(struct virtio_mem *vm, unsigned long mb_id,
+				    enum virtio_mem_mb_state state)
+{
+	const unsigned long idx = mb_id - vm->first_mb_id;
+	enum virtio_mem_mb_state old_state;
+
+	old_state = vm->mb_state[idx];
+	vm->mb_state[idx] = state;
+
+	BUG_ON(vm->nb_mb_state[old_state] == 0);
+	vm->nb_mb_state[old_state]--;
+	vm->nb_mb_state[state]++;
+}
+
+/*
+ * Get the state of a memory block.
+ */
+static enum virtio_mem_mb_state virtio_mem_mb_get_state(struct virtio_mem *vm,
+							unsigned long mb_id)
+{
+	const unsigned long idx = mb_id - vm->first_mb_id;
+
+	return vm->mb_state[idx];
+}
+
+/*
+ * Prepare the state array for the next memory block.
+ */
+static int virtio_mem_mb_state_prepare_next_mb(struct virtio_mem *vm)
+{
+	unsigned long old_bytes = vm->next_mb_id - vm->first_mb_id + 1;
+	unsigned long new_bytes = vm->next_mb_id - vm->first_mb_id + 2;
+	int old_pages = PFN_UP(old_bytes);
+	int new_pages = PFN_UP(new_bytes);
+	uint8_t *new_mb_state;
+
+	if (vm->mb_state && old_pages == new_pages)
+		return 0;
+
+	new_mb_state = vzalloc(new_pages * PAGE_SIZE);
+	if (!new_mb_state)
+		return -ENOMEM;
+
+	mutex_lock(&vm->hotplug_mutex);
+	if (vm->mb_state)
+		memcpy(new_mb_state, vm->mb_state, old_pages * PAGE_SIZE);
+	vfree(vm->mb_state);
+	vm->mb_state = new_mb_state;
+	mutex_unlock(&vm->hotplug_mutex);
+
+	return 0;
+}
+
+#define virtio_mem_for_each_mb_state(_vm, _mb_id, _state) \
+	for (_mb_id = _vm->first_mb_id; \
+	     _mb_id < _vm->next_mb_id && _vm->nb_mb_state[_state]; \
+	     _mb_id++) \
+		if (virtio_mem_mb_get_state(_vm, _mb_id) == _state)
+
+/*
+ * Mark all selected subblocks plugged.
+ *
+ * Will not modify the state of the memory block.
+ */
+static void virtio_mem_mb_set_sb_plugged(struct virtio_mem *vm,
+					 unsigned long mb_id, int sb_id,
+					 int count)
+{
+	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb + sb_id;
+
+	__bitmap_set(vm->sb_bitmap, bit, count);
+}
+
+/*
+ * Mark all selected subblocks unplugged.
+ *
+ * Will not modify the state of the memory block.
+ */
+static void virtio_mem_mb_set_sb_unplugged(struct virtio_mem *vm,
+					   unsigned long mb_id, int sb_id,
+					   int count)
+{
+	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb + sb_id;
+
+	__bitmap_clear(vm->sb_bitmap, bit, count);
+}
+
+/*
+ * Test if all selected subblocks are plugged.
+ */
+static bool virtio_mem_mb_test_sb_plugged(struct virtio_mem *vm,
+					  unsigned long mb_id, int sb_id,
+					  int count)
+{
+	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb + sb_id;
+
+	if (count == 1)
+		return test_bit(bit, vm->sb_bitmap);
+
+	/* TODO: Helper similar to bitmap_set() */
+	return find_next_zero_bit(vm->sb_bitmap, bit + count, bit) >=
+	       bit + count;
+}
+
+/*
+ * Find the first plugged subblock. Returns vm->nb_sb_per_mb in case there is
+ * none.
+ */
+static int virtio_mem_mb_first_plugged_sb(struct virtio_mem *vm,
+					  unsigned long mb_id)
+{
+	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb;
+
+	return find_next_bit(vm->sb_bitmap, bit + vm->nb_sb_per_mb, bit) - bit;
+}
+
+/*
+ * Find the first unplugged subblock. Returns vm->nb_sb_per_mb in case there is
+ * none.
+ */
+static int virtio_mem_mb_first_unplugged_sb(struct virtio_mem *vm,
+					    unsigned long mb_id)
+{
+	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb;
+
+	return find_next_zero_bit(vm->sb_bitmap, bit + vm->nb_sb_per_mb, bit) -
+	       bit;
+}
+
+/*
+ * Prepare the subblock bitmap for the next memory block.
+ */
+static int virtio_mem_sb_bitmap_prepare_next_mb(struct virtio_mem *vm)
+{
+	const unsigned long old_nb_mb = vm->next_mb_id - vm->first_mb_id;
+	const unsigned long old_nb_bits = old_nb_mb * vm->nb_sb_per_mb;
+	const unsigned long new_nb_bits = (old_nb_mb + 1) * vm->nb_sb_per_mb;
+	int old_pages = PFN_UP(BITS_TO_LONGS(old_nb_bits) * sizeof(long));
+	int new_pages = PFN_UP(BITS_TO_LONGS(new_nb_bits) * sizeof(long));
+	unsigned long *new_sb_bitmap, *old_sb_bitmap;
+
+	if (vm->sb_bitmap && old_pages == new_pages)
+		return 0;
+
+	new_sb_bitmap = vzalloc(new_pages * PAGE_SIZE);
+	if (!new_sb_bitmap)
+		return -ENOMEM;
+
+	mutex_lock(&vm->hotplug_mutex);
+	if (new_sb_bitmap)
+		memcpy(new_sb_bitmap, vm->sb_bitmap, old_pages * PAGE_SIZE);
+
+	old_sb_bitmap = vm->sb_bitmap;
+	vm->sb_bitmap = new_sb_bitmap;
+	mutex_unlock(&vm->hotplug_mutex);
+
+	vfree(old_sb_bitmap);
+	return 0;
+}
+
+/*
+ * Try to add a memory block to Linux. This will usually only fail
+ * if out of memory.
+ *
+ * Must not be called with the vm->hotplug_mutex held (possible deadlock with
+ * onlining code).
+ *
+ * Will not modify the state of the memory block.
+ */
+static int virtio_mem_mb_add(struct virtio_mem *vm, unsigned long mb_id)
+{
+	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id);
+	int nid = memory_add_physaddr_to_nid(addr);
+
+	dev_dbg(&vm->vdev->dev, "adding memory block: %lu\n", mb_id);
+	return add_memory(nid, addr, memory_block_size_bytes());
+}
+
+/*
+ * Try to remove a memory block from Linux. Will only fail if the memory block
+ * is not offline.
+ *
+ * Must not be called with the vm->hotplug_mutex held (possible deadlock with
+ * onlining code).
+ *
+ * Will not modify the state of the memory block.
+ */
+static int virtio_mem_mb_remove(struct virtio_mem *vm, unsigned long mb_id)
+{
+	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id);
+	int nid = memory_add_physaddr_to_nid(addr);
+
+	dev_dbg(&vm->vdev->dev, "removing memory block: %lu\n", mb_id);
+	return remove_memory(nid, addr, memory_block_size_bytes());
+}
+
+/*
+ * Trigger the workqueue so the device can perform its magic.
+ */
+static void virtio_mem_retry(struct virtio_mem *vm)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&vm->removal_lock, flags);
+	if (!vm->removing)
+		queue_work(system_freezable_wq, &vm->wq);
+	spin_unlock_irqrestore(&vm->removal_lock, flags);
+}
+
+/*
+ * Test if a virtio-mem device overlaps with the given range. Can be called
+ * from (notifier) callbacks lockless.
+ */
+static bool virtio_mem_overlaps_range(struct virtio_mem *vm,
+				      unsigned long start, unsigned long size)
+{
+	unsigned long dev_start = virtio_mem_mb_id_to_phys(vm->first_mb_id);
+	unsigned long dev_end = virtio_mem_mb_id_to_phys(vm->last_mb_id) +
+				memory_block_size_bytes();
+
+	return start < dev_end && dev_start < start + size;
+}
+
+/*
+ * Test if a virtio-mem device owns a memory block. Can be called from
+ * (notifier) callbacks lockless.
+ */
+static bool virtio_mem_owned_mb(struct virtio_mem *vm, unsigned long mb_id)
+{
+	return mb_id >= vm->first_mb_id && mb_id <= vm->last_mb_id;
+}
+
+static int virtio_mem_notify_going_online(struct virtio_mem *vm,
+					  unsigned long mb_id,
+					  enum zone_type zone)
+{
+	switch (virtio_mem_mb_get_state(vm, mb_id)) {
+	case VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL:
+		/*
+		 * We won't allow to online a partially plugged memory block
+		 * to the MOVABLE zone - it would contain unmovable parts.
+		 */
+		if (zone == ZONE_MOVABLE) {
+			dev_warn_ratelimited(&vm->vdev->dev,
+					     "memory block has holes, MOVABLE not supported\n");
+			return NOTIFY_BAD;
+		}
+		return NOTIFY_OK;
+	case VIRTIO_MEM_MB_STATE_OFFLINE:
+		return NOTIFY_OK;
+	default:
+		break;
+	}
+	dev_warn_ratelimited(&vm->vdev->dev,
+			     "memory block onlining denied\n");
+	return NOTIFY_BAD;
+}
+
+static void virtio_mem_notify_offline(struct virtio_mem *vm,
+				      unsigned long mb_id)
+{
+	switch (virtio_mem_mb_get_state(vm, mb_id)) {
+	case VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL:
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL);
+		break;
+	case VIRTIO_MEM_MB_STATE_ONLINE:
+	case VIRTIO_MEM_MB_STATE_ONLINE_MOVABLE:
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_OFFLINE);
+		break;
+	default:
+		BUG();
+		break;
+	}
+}
+
+static void virtio_mem_notify_online(struct virtio_mem *vm, unsigned long mb_id,
+				     enum zone_type zone)
+{
+	unsigned long nb_offline;
+
+	switch (virtio_mem_mb_get_state(vm, mb_id)) {
+	case VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL:
+		BUG_ON(zone == ZONE_MOVABLE);
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL);
+		break;
+	case VIRTIO_MEM_MB_STATE_OFFLINE:
+		if (zone == ZONE_MOVABLE)
+			virtio_mem_mb_set_state(vm, mb_id,
+					    VIRTIO_MEM_MB_STATE_ONLINE_MOVABLE);
+		else
+			virtio_mem_mb_set_state(vm, mb_id,
+						VIRTIO_MEM_MB_STATE_ONLINE);
+		break;
+	default:
+		BUG();
+		break;
+	}
+	nb_offline = vm->nb_mb_state[VIRTIO_MEM_MB_STATE_OFFLINE] +
+		     vm->nb_mb_state[VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL];
+
+	/* see if we can add new blocks now that we onlined one block */
+	if (nb_offline == VIRTIO_MEM_NB_OFFLINE_THRESHOLD - 1)
+		virtio_mem_retry(vm);
+}
+
+/*
+ * This callback will either be called synchronously from add_memory() or
+ * asynchronously (e.g., triggered via user space). We have to be careful
+ * with locking when calling add_memory().
+ */
+static int virtio_mem_memory_notifier_cb(struct notifier_block *nb,
+					 unsigned long action, void *arg)
+{
+	struct virtio_mem *vm = container_of(nb, struct virtio_mem,
+					     memory_notifier);
+	struct memory_notify *mhp = arg;
+	const unsigned long start = PFN_PHYS(mhp->start_pfn);
+	const unsigned long size = PFN_PHYS(mhp->nr_pages);
+	const unsigned long mb_id = virtio_mem_phys_to_mb_id(start);
+	enum zone_type zone;
+	int rc = NOTIFY_OK;
+
+	if (!virtio_mem_overlaps_range(vm, start, size))
+		return NOTIFY_DONE;
+
+	/*
+	 * Memory is onlined/offlined in memory block granularity. We cannot
+	 * cross virtio-mem device boundaries and memory block boundaries. Bail
+	 * out if this ever changes.
+	 */
+	if (WARN_ON_ONCE(size != memory_block_size_bytes() ||
+			 !IS_ALIGNED(start, memory_block_size_bytes())))
+		return NOTIFY_BAD;
+
+	/*
+	 * Avoid circular locking lockdep warnings. We lock the mutex
+	 * e.g., in MEM_GOING_ONLINE and unlock it in MEM_ONLINE. The
+	 * blocking_notifier_call_chain() has it's own lock, which gets unlocked
+	 * between both notifier calls and will bail out. False positive.
+	 */
+	lockdep_off();
+
+	switch (action) {
+	case MEM_GOING_OFFLINE:
+		mutex_lock(&vm->hotplug_mutex);
+		if (vm->removing) {
+			rc = notifier_from_errno(-EBUSY);
+			mutex_unlock(&vm->hotplug_mutex);
+			break;
+		}
+		vm->hotplug_active = true;
+		break;
+	case MEM_GOING_ONLINE:
+		mutex_lock(&vm->hotplug_mutex);
+		if (vm->removing) {
+			rc = notifier_from_errno(-EBUSY);
+			mutex_unlock(&vm->hotplug_mutex);
+			break;
+		}
+		vm->hotplug_active = true;
+		zone = page_zonenum(pfn_to_page(mhp->start_pfn));
+		rc = virtio_mem_notify_going_online(vm, mb_id, zone);
+		break;
+	case MEM_OFFLINE:
+		virtio_mem_notify_offline(vm, mb_id);
+		vm->hotplug_active = false;
+		mutex_unlock(&vm->hotplug_mutex);
+		break;
+	case MEM_ONLINE:
+		zone = page_zonenum(pfn_to_page(mhp->start_pfn));
+		virtio_mem_notify_online(vm, mb_id, zone);
+		vm->hotplug_active = false;
+		mutex_unlock(&vm->hotplug_mutex);
+		break;
+	case MEM_CANCEL_OFFLINE:
+	case MEM_CANCEL_ONLINE:
+		if (!vm->hotplug_active)
+			break;
+		vm->hotplug_active = false;
+		mutex_unlock(&vm->hotplug_mutex);
+		break;
+	default:
+		break;
+	}
+
+	lockdep_on();
+
+	return rc;
+}
+
+/*
+ * Set a range of pages PG_offline.
+ */
+static void virtio_mem_set_fake_offline(unsigned long pfn,
+					unsigned int nr_pages)
+{
+	for (; nr_pages--; pfn++)
+		__SetPageOffline(pfn_to_page(pfn));
+}
+
+/*
+ * Clear PG_offline from a range of pages.
+ */
+static void virtio_mem_clear_fake_offline(unsigned long pfn,
+					  unsigned int nr_pages)
+{
+	for (; nr_pages--; pfn++)
+		__ClearPageOffline(pfn_to_page(pfn));
+}
+
+/*
+ * Release a range of fake-offline pages to the buddy, effectively
+ * fake-onlining them.
+ */
+static void virtio_mem_fake_online(unsigned long pfn, unsigned int nr_pages)
+{
+	const int order = MAX_ORDER - 1;
+	int i;
+
+	/*
+	 * We are always called with subblock granularity, which is at least
+	 * aligned to MAX_ORDER - 1.
+	 */
+	virtio_mem_clear_fake_offline(pfn, nr_pages);
+
+	for (i = 0; i < nr_pages; i += 1 << order)
+		generic_online_page(pfn_to_page(pfn + i), order);
+}
+
+static void virtio_mem_online_page_cb(struct page *page, unsigned int order)
+{
+	const unsigned long addr = page_to_phys(page);
+	const unsigned long mb_id = virtio_mem_phys_to_mb_id(addr);
+	struct virtio_mem *vm;
+	int sb_id;
+
+	/*
+	 * We exploit here that subblocks have at least MAX_ORDER - 1
+	 * size/alignment and that this callback is is called with such a
+	 * size/alignment. So we cannot cross subblocks and therefore
+	 * also not memory blocks.
+	 */
+	rcu_read_lock();
+	list_for_each_entry_rcu(vm, &virtio_mem_devices, next) {
+		if (!virtio_mem_owned_mb(vm, mb_id))
+			continue;
+
+		sb_id = virtio_mem_phys_to_sb_id(vm, addr);
+		/*
+		 * If plugged, online the pages, otherwise, set them fake
+		 * offline (PageOffline).
+		 */
+		if (virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id, 1))
+			generic_online_page(page, order);
+		else
+			virtio_mem_set_fake_offline(PFN_DOWN(addr), 1 << order);
+		rcu_read_unlock();
+		return;
+	}
+	rcu_read_unlock();
+
+	/* not virtio-mem memory, but e.g., a DIMM. online it */
+	generic_online_page(page, order);
+}
+
+static uint64_t virtio_mem_send_request(struct virtio_mem *vm,
+					const struct virtio_mem_req *req)
+{
+	struct scatterlist *sgs[2], sg_req, sg_resp;
+	unsigned int len;
+	int rc;
+
+	/* don't use the request residing on the stack (vaddr) */
+	vm->req = *req;
+
+	/* out: buffer for request */
+	sg_init_one(&sg_req, &vm->req, sizeof(vm->req));
+	sgs[0] = &sg_req;
+
+	/* in: buffer for response */
+	sg_init_one(&sg_resp, &vm->resp, sizeof(vm->resp));
+	sgs[1] = &sg_resp;
+
+	rc = virtqueue_add_sgs(vm->vq, sgs, 1, 1, vm, GFP_KERNEL);
+	if (rc < 0)
+		return rc;
+
+	virtqueue_kick(vm->vq);
+
+	/* wait for a response */
+	wait_event(vm->host_resp, virtqueue_get_buf(vm->vq, &len));
+
+	return virtio16_to_cpu(vm->vdev, vm->resp.type);
+}
+
+static int virtio_mem_send_plug_request(struct virtio_mem *vm, uint64_t addr,
+					uint64_t size)
+{
+	const uint64_t nb_vm_blocks = size / vm->device_block_size;
+	const struct virtio_mem_req req = {
+		.type = cpu_to_virtio16(vm->vdev, VIRTIO_MEM_REQ_PLUG),
+		.u.plug.addr = cpu_to_virtio64(vm->vdev, addr),
+		.u.plug.nb_blocks = cpu_to_virtio16(vm->vdev, nb_vm_blocks),
+	};
+
+	if (atomic_read(&vm->config_changed))
+		return -EAGAIN;
+
+	switch (virtio_mem_send_request(vm, &req)) {
+	case VIRTIO_MEM_RESP_ACK:
+		vm->plugged_size += size;
+		return 0;
+	case VIRTIO_MEM_RESP_NACK:
+		return -EAGAIN;
+	case VIRTIO_MEM_RESP_BUSY:
+		return -EBUSY;
+	case VIRTIO_MEM_RESP_ERROR:
+		return -EINVAL;
+	default:
+		return -ENOMEM;
+	}
+}
+
+static int virtio_mem_send_unplug_request(struct virtio_mem *vm, uint64_t addr,
+					  uint64_t size)
+{
+	const uint64_t nb_vm_blocks = size / vm->device_block_size;
+	const struct virtio_mem_req req = {
+		.type = cpu_to_virtio16(vm->vdev, VIRTIO_MEM_REQ_UNPLUG),
+		.u.unplug.addr = cpu_to_virtio64(vm->vdev, addr),
+		.u.unplug.nb_blocks = cpu_to_virtio16(vm->vdev, nb_vm_blocks),
+	};
+
+	if (atomic_read(&vm->config_changed))
+		return -EAGAIN;
+
+	switch (virtio_mem_send_request(vm, &req)) {
+	case VIRTIO_MEM_RESP_ACK:
+		vm->plugged_size -= size;
+		return 0;
+	case VIRTIO_MEM_RESP_BUSY:
+		return -EBUSY;
+	case VIRTIO_MEM_RESP_ERROR:
+		return -EINVAL;
+	default:
+		return -ENOMEM;
+	}
+}
+
+static int virtio_mem_send_unplug_all_request(struct virtio_mem *vm)
+{
+	const struct virtio_mem_req req = {
+		.type = cpu_to_virtio16(vm->vdev, VIRTIO_MEM_REQ_UNPLUG_ALL),
+	};
+
+	switch (virtio_mem_send_request(vm, &req)) {
+	case VIRTIO_MEM_RESP_ACK:
+		vm->unplug_all_required = false;
+		vm->plugged_size = 0;
+		/* usable region might have shrunk */
+		atomic_set(&vm->config_changed, 1);
+		return 0;
+	case VIRTIO_MEM_RESP_BUSY:
+		return -EBUSY;
+	default:
+		return -ENOMEM;
+	}
+}
+
+/*
+ * Plug selected subblocks. Updates the plugged state, but not the state
+ * of the memory block.
+ */
+static int virtio_mem_mb_plug_sb(struct virtio_mem *vm, unsigned long mb_id,
+				 int sb_id, int count)
+{
+	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id) +
+			      sb_id * vm->subblock_size;
+	const uint64_t size = count * vm->subblock_size;
+	int rc;
+
+	dev_dbg(&vm->vdev->dev, "plugging memory block: %lu : %i - %i\n", mb_id,
+		sb_id, sb_id + count - 1);
+
+	rc = virtio_mem_send_plug_request(vm, addr, size);
+	if (!rc)
+		virtio_mem_mb_set_sb_plugged(vm, mb_id, sb_id, count);
+	return rc;
+}
+
+/*
+ * Unplug selected subblocks. Updates the plugged state, but not the state
+ * of the memory block.
+ */
+static int virtio_mem_mb_unplug_sb(struct virtio_mem *vm, unsigned long mb_id,
+				   int sb_id, int count)
+{
+	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id) +
+			      sb_id * vm->subblock_size;
+	const uint64_t size = count * vm->subblock_size;
+	int rc;
+
+	dev_dbg(&vm->vdev->dev, "unplugging memory block: %lu : %i - %i\n",
+		mb_id, sb_id, sb_id + count - 1);
+
+	rc = virtio_mem_send_unplug_request(vm, addr, size);
+	if (!rc)
+		virtio_mem_mb_set_sb_unplugged(vm, mb_id, sb_id, count);
+	return rc;
+}
+
+/*
+ * Unplug the desired number of plugged subblocks of a offline or not-added
+ * memory block. Will fail if any subblock cannot get unplugged (instead of
+ * skipping it).
+ *
+ * Will not modify the state of the memory block.
+ *
+ * Note: can fail after some subblocks were unplugged.
+ */
+static int virtio_mem_mb_unplug_any_sb(struct virtio_mem *vm,
+				       unsigned long mb_id, uint64_t *nb_sb)
+{
+	int sb_id, count;
+	int rc;
+
+	while (*nb_sb) {
+		sb_id = virtio_mem_mb_first_plugged_sb(vm, mb_id);
+		if (sb_id >= vm->nb_sb_per_mb)
+			break;
+		count = 1;
+		while (count < *nb_sb &&
+		       sb_id + count  < vm->nb_sb_per_mb &&
+		       virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id + count,
+						     1))
+			count++;
+
+		rc = virtio_mem_mb_unplug_sb(vm, mb_id, sb_id, count);
+		if (rc)
+			return rc;
+		*nb_sb -= count;
+	}
+
+	return 0;
+}
+
+/*
+ * Unplug all plugged subblocks of an offline or not-added memory block.
+ *
+ * Will not modify the state of the memory block.
+ *
+ * Note: can fail after some subblocks were unplugged.
+ */
+static int virtio_mem_mb_unplug(struct virtio_mem *vm, unsigned long mb_id)
+{
+	uint64_t nb_sb = vm->nb_sb_per_mb;
+
+	return virtio_mem_mb_unplug_any_sb(vm, mb_id, &nb_sb);
+}
+
+/*
+ * Prepare tracking data for the next memory block.
+ */
+static int virtio_mem_prepare_next_mb(struct virtio_mem *vm,
+				      unsigned long *mb_id)
+{
+	int rc;
+
+	if (vm->next_mb_id > vm->last_usable_mb_id)
+		return -ENOSPC;
+
+	/* Resize the state array if required. */
+	rc = virtio_mem_mb_state_prepare_next_mb(vm);
+	if (rc)
+		return rc;
+
+	/* Resize the subblock bitmap if required. */
+	rc = virtio_mem_sb_bitmap_prepare_next_mb(vm);
+	if (rc)
+		return rc;
+
+	vm->nb_mb_state[VIRTIO_MEM_MB_STATE_UNUSED]++;
+	*mb_id = vm->next_mb_id++;
+	return 0;
+}
+
+/*
+ * Don't add too many blocks that are not onlined yet to avoid running OOM.
+ */
+static bool virtio_mem_too_many_mb_offline(struct virtio_mem *vm)
+{
+	unsigned long nb_offline;
+
+	nb_offline = vm->nb_mb_state[VIRTIO_MEM_MB_STATE_OFFLINE] +
+		     vm->nb_mb_state[VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL];
+	return nb_offline >= VIRTIO_MEM_NB_OFFLINE_THRESHOLD;
+}
+
+/*
+ * Try to plug the desired number of subblocks and add the memory block
+ * to Linux.
+ *
+ * Will modify the state of the memory block.
+ */
+static int virtio_mem_mb_plug_and_add(struct virtio_mem *vm,
+				      unsigned long mb_id,
+				      uint64_t *nb_sb)
+{
+	const int count = min_t(int, *nb_sb, vm->nb_sb_per_mb);
+	int rc, rc2;
+
+	if (WARN_ON_ONCE(!count))
+		return -EINVAL;
+
+	/*
+	 * Plug the requested number of subblocks before adding it to linux,
+	 * so that onlining will directly online all plugged subblocks.
+	 */
+	rc = virtio_mem_mb_plug_sb(vm, mb_id, 0, count);
+	if (rc)
+		return rc;
+
+	/*
+	 * Mark the block properly offline before adding it to Linux,
+	 * so the memory notifiers will find the block in the right state.
+	 */
+	if (count == vm->nb_sb_per_mb)
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_OFFLINE);
+	else
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL);
+
+	/* Add the memory block to linux - if that fails, try to unplug. */
+	rc = virtio_mem_mb_add(vm, mb_id);
+	if (rc) {
+		enum virtio_mem_mb_state new_state = VIRTIO_MEM_MB_STATE_UNUSED;
+
+		dev_err(&vm->vdev->dev,
+			"adding memory block %lu failed with %d\n", mb_id, rc);
+		rc2 = virtio_mem_mb_unplug_sb(vm, mb_id, 0, count);
+
+		/*
+		 * TODO: Linux MM does not properly clean up yet in all cases
+		 * where adding of memory failed - especially on -ENOMEM.
+		 */
+		if (rc2)
+			new_state = VIRTIO_MEM_MB_STATE_PLUGGED;
+		virtio_mem_mb_set_state(vm, mb_id, new_state);
+		return rc;
+	}
+
+	*nb_sb -= count;
+	return 0;
+}
+
+/*
+ * Try to plug the desired number of subblocks of a memory block that
+ * is already added to Linux.
+ *
+ * Will modify the state of the memory block.
+ *
+ * Note: Can fail after some subblocks were successfully plugged.
+ */
+static int virtio_mem_mb_plug_any_sb(struct virtio_mem *vm, unsigned long mb_id,
+				     uint64_t *nb_sb, bool online)
+{
+	unsigned long pfn, nr_pages;
+	int sb_id, count;
+	int rc;
+
+	if (WARN_ON_ONCE(!*nb_sb))
+		return -EINVAL;
+
+	while (*nb_sb) {
+		sb_id = virtio_mem_mb_first_unplugged_sb(vm, mb_id);
+		if (sb_id >= vm->nb_sb_per_mb)
+			break;
+		count = 1;
+		while (count < *nb_sb &&
+		       sb_id + count < vm->nb_sb_per_mb &&
+		       !virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id + count,
+						      1))
+			count++;
+
+		rc = virtio_mem_mb_plug_sb(vm, mb_id, sb_id, count);
+		if (rc)
+			return rc;
+		*nb_sb -= count;
+		if (!online)
+			continue;
+
+		/* fake-online the pages if the memory block is online */
+		pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
+			       sb_id * vm->subblock_size);
+		nr_pages = PFN_DOWN(count * vm->subblock_size);
+		virtio_mem_fake_online(pfn, nr_pages);
+	}
+
+	if (virtio_mem_mb_test_sb_plugged(vm, mb_id, 0, vm->nb_sb_per_mb)) {
+		if (online)
+			virtio_mem_mb_set_state(vm, mb_id,
+						VIRTIO_MEM_MB_STATE_ONLINE);
+		else
+			virtio_mem_mb_set_state(vm, mb_id,
+						VIRTIO_MEM_MB_STATE_OFFLINE);
+	}
+
+	return rc;
+}
+
+/*
+ * Try to plug the requested amount of memory.
+ */
+static int virtio_mem_plug_request(struct virtio_mem *vm, uint64_t diff)
+{
+	uint64_t nb_sb = diff / vm->subblock_size;
+	unsigned long mb_id;
+	int rc;
+
+	if (!nb_sb)
+		return 0;
+
+	/* Don't race with onlining/offlining */
+	mutex_lock(&vm->hotplug_mutex);
+
+	/* Try to plug subblocks of partially plugged online blocks. */
+	virtio_mem_for_each_mb_state(vm, mb_id,
+				     VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL) {
+		rc = virtio_mem_mb_plug_any_sb(vm, mb_id, &nb_sb, true);
+		if (rc || !nb_sb)
+			goto out_unlock;
+		cond_resched();
+	}
+
+	/* Try to plug subblocks of partially plugged offline blocks. */
+	virtio_mem_for_each_mb_state(vm, mb_id,
+				     VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL) {
+		rc = virtio_mem_mb_plug_any_sb(vm, mb_id, &nb_sb, false);
+		if (rc || !nb_sb)
+			goto out_unlock;
+		cond_resched();
+	}
+
+	/*
+	 * We won't be working on online/offline memory blocks from this point,
+	 * so we can't race with memory onlining/offlining. Drop the mutex.
+	 */
+	mutex_unlock(&vm->hotplug_mutex);
+
+	/* Try to plug and add unused blocks */
+	virtio_mem_for_each_mb_state(vm, mb_id, VIRTIO_MEM_MB_STATE_UNUSED) {
+		if (virtio_mem_too_many_mb_offline(vm))
+			return -ENOSPC;
+
+		rc = virtio_mem_mb_plug_and_add(vm, mb_id, &nb_sb);
+		if (rc || !nb_sb)
+			return rc;
+		cond_resched();
+	}
+
+	/* Try to prepare, plug and add new blocks */
+	while (nb_sb) {
+		if (virtio_mem_too_many_mb_offline(vm))
+			return -ENOSPC;
+
+		rc = virtio_mem_prepare_next_mb(vm, &mb_id);
+		if (rc)
+			return rc;
+		rc = virtio_mem_mb_plug_and_add(vm, mb_id, &nb_sb);
+		if (rc)
+			return rc;
+		cond_resched();
+	}
+
+	return 0;
+out_unlock:
+	mutex_unlock(&vm->hotplug_mutex);
+	return rc;
+}
+
+/*
+ * Try to unplug all blocks that couldn't be unplugged before, for example,
+ * because the hypervisor was busy.
+ */
+static int virtio_mem_unplug_pending_mb(struct virtio_mem *vm)
+{
+	unsigned long mb_id;
+	int rc;
+
+	virtio_mem_for_each_mb_state(vm, mb_id, VIRTIO_MEM_MB_STATE_PLUGGED) {
+		rc = virtio_mem_mb_unplug(vm, mb_id);
+		if (rc)
+			return rc;
+		virtio_mem_mb_set_state(vm, mb_id, VIRTIO_MEM_MB_STATE_UNUSED);
+	}
+
+	return 0;
+}
+
+/*
+ * Update all parts of the config that could have changed.
+ */
+static void virtio_mem_refresh_config(struct virtio_mem *vm)
+{
+	const uint64_t phys_limit = 1UL << MAX_PHYSMEM_BITS;
+	uint64_t new_plugged_size, usable_region_size, end_addr;
+
+	/* the plugged_size is just a reflection of what _we_ did previously */
+	virtio_cread(vm->vdev, struct virtio_mem_config, plugged_size,
+		     &new_plugged_size);
+	if (WARN_ON_ONCE(new_plugged_size != vm->plugged_size))
+		vm->plugged_size = new_plugged_size;
+
+	/* calculate the last usable memory block id */
+	virtio_cread(vm->vdev, struct virtio_mem_config,
+		     usable_region_size, &usable_region_size);
+	end_addr = vm->addr + usable_region_size;
+	end_addr = min(end_addr, phys_limit);
+	vm->last_usable_mb_id = virtio_mem_phys_to_mb_id(end_addr) - 1;
+
+	/* see if there is a request to change the size */
+	virtio_cread(vm->vdev, struct virtio_mem_config, requested_size,
+		     &vm->requested_size);
+
+	dev_info(&vm->vdev->dev, "plugged size: 0x%llx", vm->plugged_size);
+	dev_info(&vm->vdev->dev, "requested size: 0x%llx", vm->requested_size);
+}
+
+/*
+ * Workqueue function for handling plug/unplug requests and config updates.
+ */
+static void virtio_mem_run_wq(struct work_struct *work)
+{
+	struct virtio_mem *vm = container_of(work, struct virtio_mem, wq);
+	uint64_t diff;
+	int rc;
+
+	hrtimer_cancel(&vm->retry_timer);
+
+	if (vm->broken)
+		return;
+
+retry:
+	rc = 0;
+
+	/* Make sure we start with a clean state if there are leftovers. */
+	if (unlikely(vm->unplug_all_required))
+		rc = virtio_mem_send_unplug_all_request(vm);
+
+	if (atomic_read(&vm->config_changed)) {
+		atomic_set(&vm->config_changed, 0);
+		virtio_mem_refresh_config(vm);
+	}
+
+	/* Unplug any leftovers from previous runs */
+	if (!rc)
+		rc = virtio_mem_unplug_pending_mb(vm);
+
+	if (!rc && vm->requested_size != vm->plugged_size) {
+		if (vm->requested_size > vm->plugged_size) {
+			diff = vm->requested_size - vm->plugged_size;
+			rc = virtio_mem_plug_request(vm, diff);
+		}
+		/* TODO: try to unplug memory */
+	}
+
+	switch (rc) {
+	case 0:
+		break;
+	case -ENOSPC:
+		/*
+		 * We cannot add any more memory (alignment, physical limit)
+		 * or we have too many offline memory blocks.
+		 */
+		break;
+	case -EBUSY:
+		/*
+		 * The hypervisor cannot process our request right now
+		 * (e.g., out of memory, migrating).
+		 */
+	case -ENOMEM:
+		/* Out of memory, try again later. */
+		hrtimer_start(&vm->retry_timer,
+			      ms_to_ktime(VIRTIO_MEM_RETRY_TIMER_MS),
+			      HRTIMER_MODE_REL);
+		break;
+	case -EAGAIN:
+		/* Retry immediately (e.g., the config changed). */
+		goto retry;
+	default:
+		/* Unknown error, mark as broken */
+		dev_err(&vm->vdev->dev,
+			"unknown error, marking device broken: %d\n", rc);
+		vm->broken = true;
+	}
+}
+
+static enum hrtimer_restart virtio_mem_timer_expired(struct hrtimer *timer)
+{
+	struct virtio_mem *vm = container_of(timer, struct virtio_mem,
+					     retry_timer);
+
+	virtio_mem_retry(vm);
+	return HRTIMER_NORESTART;
+}
+
+static void virtio_mem_handle_response(struct virtqueue *vq)
+{
+	struct virtio_mem *vm = vq->vdev->priv;
+
+	wake_up(&vm->host_resp);
+}
+
+static int virtio_mem_init_vq(struct virtio_mem *vm)
+{
+	struct virtqueue *vq;
+
+	vq = virtio_find_single_vq(vm->vdev, virtio_mem_handle_response,
+				   "guest-request");
+	if (IS_ERR(vq))
+		return PTR_ERR(vq);
+	vm->vq = vq;
+
+	return 0;
+}
+
+/*
+ * Test if any memory in the range is present in Linux.
+ */
+static bool virtio_mem_any_memory_present(unsigned long start,
+					  unsigned long size)
+{
+	const unsigned long start_pfn = PFN_DOWN(start);
+	const unsigned long end_pfn = PFN_UP(start + size);
+	unsigned long pfn;
+
+	for (pfn = start_pfn; pfn != end_pfn; pfn++)
+		if (present_section_nr(pfn_to_section_nr(pfn)))
+			return true;
+
+	return false;
+}
+
+static int virtio_mem_init(struct virtio_mem *vm)
+{
+	const uint64_t phys_limit = 1UL << MAX_PHYSMEM_BITS;
+
+	if (!vm->vdev->config->get) {
+		dev_err(&vm->vdev->dev, "config access disabled\n");
+		return -EINVAL;
+	}
+
+	/*
+	 * We don't want to (un)plug or reuse any memory when in kdump. The
+	 * memory is still accessible (but not mapped).
+	 */
+	if (is_kdump_kernel()) {
+		dev_warn(&vm->vdev->dev, "disabled in kdump kernel\n");
+		return -EBUSY;
+	}
+
+	/* Fetch all properties that can't change. */
+	virtio_cread(vm->vdev, struct virtio_mem_config, plugged_size,
+		     &vm->plugged_size);
+	virtio_cread(vm->vdev, struct virtio_mem_config, block_size,
+		     &vm->device_block_size);
+	virtio_cread(vm->vdev, struct virtio_mem_config, addr, &vm->addr);
+	virtio_cread(vm->vdev, struct virtio_mem_config, region_size,
+		     &vm->region_size);
+
+	/*
+	 * If we still have memory plugged, we might have to unplug all
+	 * memory first. However, if somebody simply unloaded the driver
+	 * we would have to reinitialize the old state - something we don't
+	 * support yet. Detect if we have any memory in the area present.
+	 */
+	if (vm->plugged_size) {
+		uint64_t usable_region_size;
+
+		virtio_cread(vm->vdev, struct virtio_mem_config,
+			     usable_region_size, &usable_region_size);
+
+		if (virtio_mem_any_memory_present(vm->addr,
+						  usable_region_size)) {
+			dev_err(&vm->vdev->dev,
+				"reloading the driver is not supported\n");
+			return -EINVAL;
+		}
+		/*
+		 * Note: it might happen that the device is busy and
+		 * unplugging all memory might take some time.
+		 */
+		dev_info(&vm->vdev->dev, "unplugging all memory required\n");
+		vm->unplug_all_required = 1;
+	}
+
+	/*
+	 * We always hotplug memory in memory block granularity. This way,
+	 * we have to wait for exactly one memory block to online.
+	 */
+	if (vm->device_block_size > memory_block_size_bytes()) {
+		dev_err(&vm->vdev->dev,
+			"The block size is not supported (too big).\n");
+		return -EINVAL;
+	}
+
+	/* bad device setup - warn only */
+	if (!IS_ALIGNED(vm->addr, memory_block_size_bytes()))
+		dev_warn(&vm->vdev->dev,
+			 "The alignment of the physical start address can make some memory unusable.\n");
+	if (!IS_ALIGNED(vm->addr + vm->region_size, memory_block_size_bytes()))
+		dev_warn(&vm->vdev->dev,
+			 "The alignment of the physical end address can make some memory unusable.\n");
+	if (vm->addr + vm->region_size > phys_limit)
+		dev_warn(&vm->vdev->dev,
+			 "Some memory is not addressable. This can make some memory unusable.\n");
+
+	/*
+	 * Calculate the subblock size:
+	 * - At least MAX_ORDER - 1 / pageblock_order.
+	 * - At least the device block size.
+	 * In the worst case, a single subblock per memory block.
+	 */
+	vm->subblock_size = PAGE_SIZE * 1u << max_t(uint32_t, MAX_ORDER - 1,
+						    pageblock_order);
+	vm->subblock_size = max_t(uint32_t, vm->device_block_size,
+				  vm->subblock_size);
+	vm->nb_sb_per_mb = memory_block_size_bytes() / vm->subblock_size;
+
+	/* Round up to the next full memory block */
+	vm->first_mb_id = virtio_mem_phys_to_mb_id(vm->addr - 1 +
+						   memory_block_size_bytes());
+	vm->next_mb_id = vm->first_mb_id;
+	vm->last_mb_id = virtio_mem_phys_to_mb_id(vm->addr +
+			 vm->region_size) - 1;
+
+	dev_info(&vm->vdev->dev, "start address: 0x%llx", vm->addr);
+	dev_info(&vm->vdev->dev, "region size: 0x%llx", vm->region_size);
+	dev_info(&vm->vdev->dev, "device block size: 0x%x",
+		 vm->device_block_size);
+	dev_info(&vm->vdev->dev, "memory block size: 0x%lx",
+		 memory_block_size_bytes());
+	dev_info(&vm->vdev->dev, "subblock size: 0x%x",
+		 vm->subblock_size);
+
+	return 0;
+}
+
+static int virtio_mem_probe(struct virtio_device *vdev)
+{
+	struct virtio_mem *vm;
+	int rc = -EINVAL;
+
+	vdev->priv = vm = kzalloc(sizeof(*vm), GFP_KERNEL);
+	if (!vm)
+		return -ENOMEM;
+
+	init_waitqueue_head(&vm->host_resp);
+	vm->vdev = vdev;
+	INIT_WORK(&vm->wq, virtio_mem_run_wq);
+	mutex_init(&vm->hotplug_mutex);
+	INIT_LIST_HEAD(&vm->next);
+	spin_lock_init(&vm->removal_lock);
+	hrtimer_init(&vm->retry_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	vm->retry_timer.function = virtio_mem_timer_expired;
+
+	/* register the virtqueue */
+	rc = virtio_mem_init_vq(vm);
+	if (rc)
+		goto out_free_vm;
+
+	/* initialize the device by querying the config */
+	rc = virtio_mem_init(vm);
+	if (rc)
+		goto out_del_vq;
+
+	/* register callbacks */
+	vm->memory_notifier.notifier_call = virtio_mem_memory_notifier_cb;
+	rc = register_memory_notifier(&vm->memory_notifier);
+	if (rc)
+		goto out_del_vq;
+	rc = register_virtio_mem_device(vm);
+	if (rc)
+		goto out_unreg_mem;
+
+	virtio_device_ready(vdev);
+
+	/* trigger a config update to start processing the requested_size */
+	atomic_set(&vm->config_changed, 1);
+	queue_work(system_freezable_wq, &vm->wq);
+
+	return 0;
+out_unreg_mem:
+	unregister_memory_notifier(&vm->memory_notifier);
+out_del_vq:
+	vdev->config->del_vqs(vdev);
+out_free_vm:
+	kfree(vm);
+	vdev->priv = NULL;
+
+	return rc;
+}
+
+static void virtio_mem_remove(struct virtio_device *vdev)
+{
+	struct virtio_mem *vm = vdev->priv;
+	unsigned long mb_id;
+	int rc;
+
+	/*
+	 * Make sure the workqueue won't be triggered anymore and no memory
+	 * blocks can be onlined/offlined until we're finished here.
+	 */
+	mutex_lock(&vm->hotplug_mutex);
+	spin_lock_irq(&vm->removal_lock);
+	vm->removing = true;
+	spin_unlock_irq(&vm->removal_lock);
+	mutex_unlock(&vm->hotplug_mutex);
+
+	/* wait until the workqueue stopped */
+	cancel_work_sync(&vm->wq);
+	hrtimer_cancel(&vm->retry_timer);
+
+	/*
+	 * After we unregistered our callbacks, user space can online partially
+	 * plugged offline blocks. Make sure to remove them.
+	 */
+	virtio_mem_for_each_mb_state(vm, mb_id,
+				     VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL) {
+		rc = virtio_mem_mb_remove(vm, mb_id);
+		BUG_ON(rc);
+		virtio_mem_mb_set_state(vm, mb_id, VIRTIO_MEM_MB_STATE_UNUSED);
+	}
+
+	/* unregister callbacks */
+	unregister_virtio_mem_device(vm);
+	unregister_memory_notifier(&vm->memory_notifier);
+
+	/*
+	 * There is no way we could reliably remove all memory we have added to
+	 * the system. And there is no way to stop the driver/device from going
+	 * away. Warn at least.
+	 */
+	if (vm->nb_mb_state[VIRTIO_MEM_MB_STATE_OFFLINE] ||
+	    vm->nb_mb_state[VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL] ||
+	    vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE] ||
+	    vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL] ||
+	    vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE_MOVABLE])
+		dev_warn(&vdev->dev, "device still has system memory added\n");
+
+	/* remove all tracking data - no locking needed */
+	vfree(vm->mb_state);
+	vfree(vm->sb_bitmap);
+
+	/* reset the device and cleanup the queues */
+	vdev->config->reset(vdev);
+	vdev->config->del_vqs(vdev);
+
+	kfree(vm);
+	vdev->priv = NULL;
+}
+
+static void virtio_mem_config_changed(struct virtio_device *vdev)
+{
+	struct virtio_mem *vm = vdev->priv;
+
+	atomic_set(&vm->config_changed, 1);
+	virtio_mem_retry(vm);
+}
+
+#ifdef CONFIG_PM_SLEEP
+static int virtio_mem_freeze(struct virtio_device *vdev)
+{
+	/*
+	 * When restarting the VM, all memory is usually unplugged. Don't
+	 * allow to suspend/hibernate.
+	 */
+	dev_err(&vdev->dev, "save/restore not supported.\n");
+	return -EPERM;
+}
+
+static int virtio_mem_restore(struct virtio_device *vdev)
+{
+	return -EPERM;
+}
+#endif
+
+static struct virtio_device_id virtio_mem_id_table[] = {
+	{ VIRTIO_ID_MEM, VIRTIO_DEV_ANY_ID },
+	{ 0 },
+};
+
+static struct virtio_driver virtio_mem_driver = {
+	.driver.name = KBUILD_MODNAME,
+	.driver.owner = THIS_MODULE,
+	.id_table = virtio_mem_id_table,
+	.probe = virtio_mem_probe,
+	.remove = virtio_mem_remove,
+	.config_changed = virtio_mem_config_changed,
+#ifdef CONFIG_PM_SLEEP
+	.freeze	=	virtio_mem_freeze,
+	.restore =	virtio_mem_restore,
+#endif
+};
+
+module_virtio_driver(virtio_mem_driver);
+MODULE_DEVICE_TABLE(virtio, virtio_mem_id_table);
+MODULE_AUTHOR("David Hildenbrand <david@redhat.com>");
+MODULE_DESCRIPTION("Virtio-mem driver");
+MODULE_LICENSE("GPL");
diff --git a/include/uapi/linux/virtio_ids.h b/include/uapi/linux/virtio_ids.h
index ecc27a17401a..b052355ac7a3 100644
--- a/include/uapi/linux/virtio_ids.h
+++ b/include/uapi/linux/virtio_ids.h
@@ -44,6 +44,7 @@
 #define VIRTIO_ID_VSOCK        19 /* virtio vsock transport */
 #define VIRTIO_ID_CRYPTO       20 /* virtio crypto */
 #define VIRTIO_ID_IOMMU        23 /* virtio IOMMU */
+#define VIRTIO_ID_MEM          24 /* virtio mem */
 #define VIRTIO_ID_FS           26 /* virtio filesystem */
 #define VIRTIO_ID_PMEM         27 /* virtio pmem */
 #define VIRTIO_ID_MAC80211_HWSIM 29 /* virtio mac80211-hwsim */
diff --git a/include/uapi/linux/virtio_mem.h b/include/uapi/linux/virtio_mem.h
new file mode 100644
index 000000000000..1bfade78bdfd
--- /dev/null
+++ b/include/uapi/linux/virtio_mem.h
@@ -0,0 +1,200 @@
+/* SPDX-License-Identifier: BSD-3-Clause */
+/*
+ * Virtio Mem Device
+ *
+ * Copyright Red Hat, Inc. 2020
+ *
+ * Authors:
+ *     David Hildenbrand <david@redhat.com>
+ *
+ * This header is BSD licensed so anyone can use the definitions
+ * to implement compatible drivers/servers:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of IBM nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL IBM OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
+ * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+ * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
+ * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#ifndef _LINUX_VIRTIO_MEM_H
+#define _LINUX_VIRTIO_MEM_H
+
+#include <linux/types.h>
+#include <linux/virtio_types.h>
+#include <linux/virtio_ids.h>
+#include <linux/virtio_config.h>
+
+/*
+ * Each virtio-mem device manages a dedicated region in physical address
+ * space. Each device can belong to a single NUMA node, multiple devices
+ * for a single NUMA node are possible. A virtio-mem device is like a
+ * "resizable DIMM" consisting of small memory blocks that can be plugged
+ * or unplugged. The device driver is responsible for (un)plugging memory
+ * blocks on demand.
+ *
+ * Virtio-mem devices can only operate on their assigned memory region in
+ * order to (un)plug memory. A device cannot (un)plug memory belonging to
+ * other devices.
+ *
+ * The "region_size" corresponds to the maximum amount of memory that can
+ * be provided by a device. The "size" corresponds to the amount of memory
+ * that is currently plugged. "requested_size" corresponds to a request
+ * from the device to the device driver to (un)plug blocks. The
+ * device driver should try to (un)plug blocks in order to reach the
+ * "requested_size". It is impossible to plug more memory than requested.
+ *
+ * The "usable_region_size" represents the memory region that can actually
+ * be used to (un)plug memory. It is always at least as big as the
+ * "requested_size" and will grow dynamically. It will only shrink when
+ * explicitly triggered (VIRTIO_MEM_REQ_UNPLUG).
+ *
+ * There are no guarantees what will happen if unplugged memory is
+ * read/written. Such memory should, in general, not be touched. E.g.,
+ * even writing might succeed, but the values will simply be discarded at
+ * random points in time.
+ *
+ * It can happen that the device cannot process a request, because it is
+ * busy. The device driver has to retry later.
+ *
+ * Usually, during system resets all memory will get unplugged, so the
+ * device driver can start with a clean state. However, in specific
+ * scenarios (if the device is busy) it can happen that the device still
+ * has memory plugged. The device driver can request to unplug all memory
+ * (VIRTIO_MEM_REQ_UNPLUG) - which might take a while to succeed if the
+ * device is busy.
+ */
+
+/* --- virtio-mem: guest -> host requests --- */
+
+/* request to plug memory blocks */
+#define VIRTIO_MEM_REQ_PLUG			0
+/* request to unplug memory blocks */
+#define VIRTIO_MEM_REQ_UNPLUG			1
+/* request to unplug all blocks and shrink the usable size */
+#define VIRTIO_MEM_REQ_UNPLUG_ALL		2
+/* request information about the plugged state of memory blocks */
+#define VIRTIO_MEM_REQ_STATE			3
+
+struct virtio_mem_req_plug {
+	__virtio64 addr;
+	__virtio16 nb_blocks;
+};
+
+struct virtio_mem_req_unplug {
+	__virtio64 addr;
+	__virtio16 nb_blocks;
+};
+
+struct virtio_mem_req_state {
+	__virtio64 addr;
+	__virtio16 nb_blocks;
+};
+
+struct virtio_mem_req {
+	__virtio16 type;
+	__virtio16 padding[3];
+
+	union {
+		struct virtio_mem_req_plug plug;
+		struct virtio_mem_req_unplug unplug;
+		struct virtio_mem_req_state state;
+	} u;
+};
+
+
+/* --- virtio-mem: host -> guest response --- */
+
+/*
+ * Request processed successfully, applicable for
+ * - VIRTIO_MEM_REQ_PLUG
+ * - VIRTIO_MEM_REQ_UNPLUG
+ * - VIRTIO_MEM_REQ_UNPLUG_ALL
+ * - VIRTIO_MEM_REQ_STATE
+ */
+#define VIRTIO_MEM_RESP_ACK			0
+/*
+ * Request denied - e.g. trying to plug more than requested, applicable for
+ * - VIRTIO_MEM_REQ_PLUG
+ */
+#define VIRTIO_MEM_RESP_NACK			1
+/*
+ * Request cannot be processed right now, try again later, applicable for
+ * - VIRTIO_MEM_REQ_PLUG
+ * - VIRTIO_MEM_REQ_UNPLUG
+ * - VIRTIO_MEM_REQ_UNPLUG_ALL
+ */
+#define VIRTIO_MEM_RESP_BUSY			2
+/*
+ * Error in request (e.g. addresses/alignment), applicable for
+ * - VIRTIO_MEM_REQ_PLUG
+ * - VIRTIO_MEM_REQ_UNPLUG
+ * - VIRTIO_MEM_REQ_STATE
+ */
+#define VIRTIO_MEM_RESP_ERROR			3
+
+
+/* State of memory blocks is "plugged" */
+#define VIRTIO_MEM_STATE_PLUGGED		0
+/* State of memory blocks is "unplugged" */
+#define VIRTIO_MEM_STATE_UNPLUGGED		1
+/* State of memory blocks is "mixed" */
+#define VIRTIO_MEM_STATE_MIXED			2
+
+struct virtio_mem_resp_state {
+	__virtio16 state;
+};
+
+struct virtio_mem_resp {
+	__virtio16 type;
+	__virtio16 padding[3];
+
+	union {
+		struct virtio_mem_resp_state state;
+	} u;
+};
+
+/* --- virtio-mem: configuration --- */
+
+struct virtio_mem_config {
+	/* Block size and alignment. Cannot change. */
+	__u32 block_size;
+	__u32 padding;
+	/* Start address of the memory region. Cannot change. */
+	__u64 addr;
+	/* Region size (maximum). Cannot change. */
+	__u64 region_size;
+	/*
+	 * Currently usable region size. Can grow up to region_size. Can
+	 * shrink due to VIRTIO_MEM_REQ_UNPLUG_ALL (in which case no config
+	 * update will be sent).
+	 */
+	__u64 usable_region_size;
+	/*
+	 * Currently used size. Changes due to plug/unplug requests, but no
+	 * config updates will be sent.
+	 */
+	__u64 plugged_size;
+	/* Requested size. New plug requests cannot exceed it. Can change. */
+	__u64 requested_size;
+};
+
+#endif /* _LINUX_VIRTIO_MEM_H */
-- 
2.25.3

[PATCH v4 02/15] MAINTAINERS: Add myself as virtio-mem maintainer

[David Hildenbrand]

Let's make sure patches/bug reports find the right person.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 MAINTAINERS | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 2926327e4976..014bbf5897c2 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -17972,6 +17972,13 @@ S:	Maintained
 F:	drivers/iommu/virtio-iommu.c
 F:	include/uapi/linux/virtio_iommu.h
 
+VIRTIO MEM DRIVER
+M:	David Hildenbrand <david@redhat.com>
+L:	virtualization@lists.linux-foundation.org
+S:	Maintained
+F:	drivers/virtio/virtio_mem.c
+F:	include/uapi/linux/virtio_mem.h
+
 VIRTUAL BOX GUEST DEVICE DRIVER
 M:	Hans de Goede <hdegoede@redhat.com>
 M:	Arnd Bergmann <arnd@arndb.de>
-- 
2.25.3

[PATCH v4 03/15] virtio-mem: Allow to specify an ACPI PXM as nid

[David Hildenbrand]

We want to allow to specify (similar as for a DIMM), to which node a
virtio-mem device (and, therefore, its memory) belongs. Add a new
virtio-mem feature flag and export pxm_to_node, so it can be used in kernel
module context.

Acked-by: Michal Hocko <mhocko@suse.com> # for the export
Acked-by: "Rafael J. Wysocki" <rafael@kernel.org> # for the export
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Len Brown <lenb@kernel.org>
Cc: linux-acpi@vger.kernel.org
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/acpi/numa/srat.c        |  1 +
 drivers/virtio/virtio_mem.c     | 39 +++++++++++++++++++++++++++++++--
 include/uapi/linux/virtio_mem.h | 10 ++++++++-
 3 files changed, 47 insertions(+), 3 deletions(-)

diff --git a/drivers/acpi/numa/srat.c b/drivers/acpi/numa/srat.c
index 47b4969d9b93..5be5a977da1b 100644
--- a/drivers/acpi/numa/srat.c
+++ b/drivers/acpi/numa/srat.c
@@ -35,6 +35,7 @@ int pxm_to_node(int pxm)
 		return NUMA_NO_NODE;
 	return pxm_to_node_map[pxm];
 }
+EXPORT_SYMBOL(pxm_to_node);
 
 int node_to_pxm(int node)
 {
diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 5d1dcaa6fc42..270ddeaec059 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -21,6 +21,8 @@
 #include <linux/bitmap.h>
 #include <linux/lockdep.h>
 
+#include <acpi/acpi_numa.h>
+
 enum virtio_mem_mb_state {
 	/* Unplugged, not added to Linux. Can be reused later. */
 	VIRTIO_MEM_MB_STATE_UNUSED = 0,
@@ -72,6 +74,8 @@ struct virtio_mem {
 
 	/* The device block size (for communicating with the device). */
 	uint32_t device_block_size;
+	/* The translated node id. NUMA_NO_NODE in case not specified. */
+	int nid;
 	/* Physical start address of the memory region. */
 	uint64_t addr;
 	/* Maximum region size in bytes. */
@@ -389,7 +393,10 @@ static int virtio_mem_sb_bitmap_prepare_next_mb(struct virtio_mem *vm)
 static int virtio_mem_mb_add(struct virtio_mem *vm, unsigned long mb_id)
 {
 	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id);
-	int nid = memory_add_physaddr_to_nid(addr);
+	int nid = vm->nid;
+
+	if (nid == NUMA_NO_NODE)
+		nid = memory_add_physaddr_to_nid(addr);
 
 	dev_dbg(&vm->vdev->dev, "adding memory block: %lu\n", mb_id);
 	return add_memory(nid, addr, memory_block_size_bytes());
@@ -407,7 +414,10 @@ static int virtio_mem_mb_add(struct virtio_mem *vm, unsigned long mb_id)
 static int virtio_mem_mb_remove(struct virtio_mem *vm, unsigned long mb_id)
 {
 	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id);
-	int nid = memory_add_physaddr_to_nid(addr);
+	int nid = vm->nid;
+
+	if (nid == NUMA_NO_NODE)
+		nid = memory_add_physaddr_to_nid(addr);
 
 	dev_dbg(&vm->vdev->dev, "removing memory block: %lu\n", mb_id);
 	return remove_memory(nid, addr, memory_block_size_bytes());
@@ -426,6 +436,17 @@ static void virtio_mem_retry(struct virtio_mem *vm)
 	spin_unlock_irqrestore(&vm->removal_lock, flags);
 }
 
+static int virtio_mem_translate_node_id(struct virtio_mem *vm, uint16_t node_id)
+{
+	int node = NUMA_NO_NODE;
+
+#if defined(CONFIG_ACPI_NUMA)
+	if (virtio_has_feature(vm->vdev, VIRTIO_MEM_F_ACPI_PXM))
+		node = pxm_to_node(node_id);
+#endif
+	return node;
+}
+
 /*
  * Test if a virtio-mem device overlaps with the given range. Can be called
  * from (notifier) callbacks lockless.
@@ -1267,6 +1288,7 @@ static bool virtio_mem_any_memory_present(unsigned long start,
 static int virtio_mem_init(struct virtio_mem *vm)
 {
 	const uint64_t phys_limit = 1UL << MAX_PHYSMEM_BITS;
+	uint16_t node_id;
 
 	if (!vm->vdev->config->get) {
 		dev_err(&vm->vdev->dev, "config access disabled\n");
@@ -1287,6 +1309,9 @@ static int virtio_mem_init(struct virtio_mem *vm)
 		     &vm->plugged_size);
 	virtio_cread(vm->vdev, struct virtio_mem_config, block_size,
 		     &vm->device_block_size);
+	virtio_cread(vm->vdev, struct virtio_mem_config, node_id,
+		     &node_id);
+	vm->nid = virtio_mem_translate_node_id(vm, node_id);
 	virtio_cread(vm->vdev, struct virtio_mem_config, addr, &vm->addr);
 	virtio_cread(vm->vdev, struct virtio_mem_config, region_size,
 		     &vm->region_size);
@@ -1365,6 +1390,8 @@ static int virtio_mem_init(struct virtio_mem *vm)
 		 memory_block_size_bytes());
 	dev_info(&vm->vdev->dev, "subblock size: 0x%x",
 		 vm->subblock_size);
+	if (vm->nid != NUMA_NO_NODE)
+		dev_info(&vm->vdev->dev, "nid: %d", vm->nid);
 
 	return 0;
 }
@@ -1508,12 +1535,20 @@ static int virtio_mem_restore(struct virtio_device *vdev)
 }
 #endif
 
+static unsigned int virtio_mem_features[] = {
+#if defined(CONFIG_NUMA) && defined(CONFIG_ACPI_NUMA)
+	VIRTIO_MEM_F_ACPI_PXM,
+#endif
+};
+
 static struct virtio_device_id virtio_mem_id_table[] = {
 	{ VIRTIO_ID_MEM, VIRTIO_DEV_ANY_ID },
 	{ 0 },
 };
 
 static struct virtio_driver virtio_mem_driver = {
+	.feature_table = virtio_mem_features,
+	.feature_table_size = ARRAY_SIZE(virtio_mem_features),
 	.driver.name = KBUILD_MODNAME,
 	.driver.owner = THIS_MODULE,
 	.id_table = virtio_mem_id_table,
diff --git a/include/uapi/linux/virtio_mem.h b/include/uapi/linux/virtio_mem.h
index 1bfade78bdfd..e0a9dc7397c3 100644
--- a/include/uapi/linux/virtio_mem.h
+++ b/include/uapi/linux/virtio_mem.h
@@ -83,6 +83,12 @@
  * device is busy.
  */
 
+/* --- virtio-mem: feature bits --- */
+
+/* node_id is an ACPI PXM and is valid */
+#define VIRTIO_MEM_F_ACPI_PXM		0
+
+
 /* --- virtio-mem: guest -> host requests --- */
 
 /* request to plug memory blocks */
@@ -177,7 +183,9 @@ struct virtio_mem_resp {
 struct virtio_mem_config {
 	/* Block size and alignment. Cannot change. */
 	__u32 block_size;
-	__u32 padding;
+	/* Valid with VIRTIO_MEM_F_ACPI_PXM. Cannot change. */
+	__u16 node_id;
+	__u16 padding;
 	/* Start address of the memory region. Cannot change. */
 	__u64 addr;
 	/* Region size (maximum). Cannot change. */
-- 
2.25.3

[PATCH v4 04/15] virtio-mem: Paravirtualized memory hotunplug part 1

[David Hildenbrand]

Unplugging subblocks of memory blocks that are offline is easy. All we
have to do is watch out for concurrent onlining activity.

Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 114 insertions(+), 2 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 270ddeaec059..a3ec795be8be 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -123,7 +123,7 @@ struct virtio_mem {
 	 *
 	 * When this lock is held the pointers can't change, ONLINE and
 	 * OFFLINE blocks can't change the state and no subblocks will get
-	 * plugged.
+	 * plugged/unplugged.
 	 */
 	struct mutex hotplug_mutex;
 	bool hotplug_active;
@@ -280,6 +280,12 @@ static int virtio_mem_mb_state_prepare_next_mb(struct virtio_mem *vm)
 	     _mb_id++) \
 		if (virtio_mem_mb_get_state(_vm, _mb_id) == _state)
 
+#define virtio_mem_for_each_mb_state_rev(_vm, _mb_id, _state) \
+	for (_mb_id = _vm->next_mb_id - 1; \
+	     _mb_id >= _vm->first_mb_id && _vm->nb_mb_state[_state]; \
+	     _mb_id--) \
+		if (virtio_mem_mb_get_state(_vm, _mb_id) == _state)
+
 /*
  * Mark all selected subblocks plugged.
  *
@@ -325,6 +331,19 @@ static bool virtio_mem_mb_test_sb_plugged(struct virtio_mem *vm,
 	       bit + count;
 }
 
+/*
+ * Test if all selected subblocks are unplugged.
+ */
+static bool virtio_mem_mb_test_sb_unplugged(struct virtio_mem *vm,
+					    unsigned long mb_id, int sb_id,
+					    int count)
+{
+	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb + sb_id;
+
+	/* TODO: Helper similar to bitmap_set() */
+	return find_next_bit(vm->sb_bitmap, bit + count, bit) >= bit + count;
+}
+
 /*
  * Find the first plugged subblock. Returns vm->nb_sb_per_mb in case there is
  * none.
@@ -513,6 +532,9 @@ static void virtio_mem_notify_offline(struct virtio_mem *vm,
 		BUG();
 		break;
 	}
+
+	/* trigger the workqueue, maybe we can now unplug memory. */
+	virtio_mem_retry(vm);
 }
 
 static void virtio_mem_notify_online(struct virtio_mem *vm, unsigned long mb_id,
@@ -1122,6 +1144,94 @@ static int virtio_mem_plug_request(struct virtio_mem *vm, uint64_t diff)
 	return rc;
 }
 
+/*
+ * Unplug the desired number of plugged subblocks of an offline memory block.
+ * Will fail if any subblock cannot get unplugged (instead of skipping it).
+ *
+ * Will modify the state of the memory block. Might temporarily drop the
+ * hotplug_mutex.
+ *
+ * Note: Can fail after some subblocks were successfully unplugged.
+ */
+static int virtio_mem_mb_unplug_any_sb_offline(struct virtio_mem *vm,
+					       unsigned long mb_id,
+					       uint64_t *nb_sb)
+{
+	int rc;
+
+	rc = virtio_mem_mb_unplug_any_sb(vm, mb_id, nb_sb);
+
+	/* some subblocks might have been unplugged even on failure */
+	if (!virtio_mem_mb_test_sb_plugged(vm, mb_id, 0, vm->nb_sb_per_mb))
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL);
+	if (rc)
+		return rc;
+
+	if (virtio_mem_mb_test_sb_unplugged(vm, mb_id, 0, vm->nb_sb_per_mb)) {
+		/*
+		 * Remove the block from Linux - this should never fail.
+		 * Hinder the block from getting onlined by marking it
+		 * unplugged. Temporarily drop the mutex, so
+		 * any pending GOING_ONLINE requests can be serviced/rejected.
+		 */
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_UNUSED);
+
+		mutex_unlock(&vm->hotplug_mutex);
+		rc = virtio_mem_mb_remove(vm, mb_id);
+		BUG_ON(rc);
+		mutex_lock(&vm->hotplug_mutex);
+	}
+	return 0;
+}
+
+/*
+ * Try to unplug the requested amount of memory.
+ */
+static int virtio_mem_unplug_request(struct virtio_mem *vm, uint64_t diff)
+{
+	uint64_t nb_sb = diff / vm->subblock_size;
+	unsigned long mb_id;
+	int rc;
+
+	if (!nb_sb)
+		return 0;
+
+	/*
+	 * We'll drop the mutex a couple of times when it is safe to do so.
+	 * This might result in some blocks switching the state (online/offline)
+	 * and we could miss them in this run - we will retry again later.
+	 */
+	mutex_lock(&vm->hotplug_mutex);
+
+	/* Try to unplug subblocks of partially plugged offline blocks. */
+	virtio_mem_for_each_mb_state_rev(vm, mb_id,
+					 VIRTIO_MEM_MB_STATE_OFFLINE_PARTIAL) {
+		rc = virtio_mem_mb_unplug_any_sb_offline(vm, mb_id,
+							 &nb_sb);
+		if (rc || !nb_sb)
+			goto out_unlock;
+		cond_resched();
+	}
+
+	/* Try to unplug subblocks of plugged offline blocks. */
+	virtio_mem_for_each_mb_state_rev(vm, mb_id,
+					 VIRTIO_MEM_MB_STATE_OFFLINE) {
+		rc = virtio_mem_mb_unplug_any_sb_offline(vm, mb_id,
+							 &nb_sb);
+		if (rc || !nb_sb)
+			goto out_unlock;
+		cond_resched();
+	}
+
+	mutex_unlock(&vm->hotplug_mutex);
+	return 0;
+out_unlock:
+	mutex_unlock(&vm->hotplug_mutex);
+	return rc;
+}
+
 /*
  * Try to unplug all blocks that couldn't be unplugged before, for example,
  * because the hypervisor was busy.
@@ -1204,8 +1314,10 @@ static void virtio_mem_run_wq(struct work_struct *work)
 		if (vm->requested_size > vm->plugged_size) {
 			diff = vm->requested_size - vm->plugged_size;
 			rc = virtio_mem_plug_request(vm, diff);
+		} else {
+			diff = vm->plugged_size - vm->requested_size;
+			rc = virtio_mem_unplug_request(vm, diff);
 		}
-		/* TODO: try to unplug memory */
 	}
 
 	switch (rc) {
-- 
2.25.3

[PATCH v4 05/15] virtio-mem: Paravirtualized memory hotunplug part 2

[David Hildenbrand]

We also want to unplug online memory (contained in online memory blocks
and, therefore, managed by the buddy), and eventually replug it later.

When requested to unplug memory, we use alloc_contig_range() to allocate
subblocks in online memory blocks (so we are the owner) and send them to
our hypervisor. When requested to plug memory, we can replug such memory
using free_contig_range() after asking our hypervisor.

We also want to mark all allocated pages PG_offline, so nobody will
touch them. To differentiate pages that were never onlined when
onlining the memory block from pages allocated via alloc_contig_range(), we
use PageDirty(). Based on this flag, virtio_mem_fake_online() can either
online the pages for the first time or use free_contig_range().

It is worth noting that there are no guarantees on how much memory can
actually get unplugged again. All device memory might completely be
fragmented with unmovable data, such that no subblock can get unplugged.

We are not touching the ZONE_MOVABLE. If memory is onlined to the
ZONE_MOVABLE, it can only get unplugged after that memory was offlined
manually by user space. In normal operation, virtio-mem memory is
suggested to be onlined to ZONE_NORMAL. In the future, we will try to
make unplug more likely to succeed.

Add a module parameter to control if online memory shall be touched.

As we want to access alloc_contig_range()/free_contig_range() from
kernel module context, export the symbols.

Note: Whenever virtio-mem uses alloc_contig_range(), all affected pages
are on the same node, in the same zone, and contain no holes.

Acked-by: Michal Hocko <mhocko@suse.com> # to export contig range allocator API
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Cc: Alexander Potapenko <glider@google.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/Kconfig      |   1 +
 drivers/virtio/virtio_mem.c | 157 ++++++++++++++++++++++++++++++++----
 mm/page_alloc.c             |   2 +
 3 files changed, 146 insertions(+), 14 deletions(-)

diff --git a/drivers/virtio/Kconfig b/drivers/virtio/Kconfig
index d6dde7d2cf76..4c1e14615001 100644
--- a/drivers/virtio/Kconfig
+++ b/drivers/virtio/Kconfig
@@ -85,6 +85,7 @@ config VIRTIO_MEM
 	depends on VIRTIO
 	depends on MEMORY_HOTPLUG_SPARSE
 	depends on MEMORY_HOTREMOVE
+	select CONTIG_ALLOC
 	help
 	 This driver provides access to virtio-mem paravirtualized memory
 	 devices, allowing to hotplug and hotunplug memory.
diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index a3ec795be8be..74f0d3cb1d22 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -23,6 +23,10 @@
 
 #include <acpi/acpi_numa.h>
 
+static bool unplug_online = true;
+module_param(unplug_online, bool, 0644);
+MODULE_PARM_DESC(unplug_online, "Try to unplug online memory");
+
 enum virtio_mem_mb_state {
 	/* Unplugged, not added to Linux. Can be reused later. */
 	VIRTIO_MEM_MB_STATE_UNUSED = 0,
@@ -654,23 +658,35 @@ static int virtio_mem_memory_notifier_cb(struct notifier_block *nb,
 }
 
 /*
- * Set a range of pages PG_offline.
+ * Set a range of pages PG_offline. Remember pages that were never onlined
+ * (via generic_online_page()) using PageDirty().
  */
 static void virtio_mem_set_fake_offline(unsigned long pfn,
-					unsigned int nr_pages)
+					unsigned int nr_pages, bool onlined)
 {
-	for (; nr_pages--; pfn++)
-		__SetPageOffline(pfn_to_page(pfn));
+	for (; nr_pages--; pfn++) {
+		struct page *page = pfn_to_page(pfn);
+
+		__SetPageOffline(page);
+		if (!onlined)
+			SetPageDirty(page);
+	}
 }
 
 /*
- * Clear PG_offline from a range of pages.
+ * Clear PG_offline from a range of pages. If the pages were never onlined,
+ * (via generic_online_page()), clear PageDirty().
  */
 static void virtio_mem_clear_fake_offline(unsigned long pfn,
-					  unsigned int nr_pages)
+					  unsigned int nr_pages, bool onlined)
 {
-	for (; nr_pages--; pfn++)
-		__ClearPageOffline(pfn_to_page(pfn));
+	for (; nr_pages--; pfn++) {
+		struct page *page = pfn_to_page(pfn);
+
+		__ClearPageOffline(page);
+		if (!onlined)
+			ClearPageDirty(page);
+	}
 }
 
 /*
@@ -686,10 +702,26 @@ static void virtio_mem_fake_online(unsigned long pfn, unsigned int nr_pages)
 	 * We are always called with subblock granularity, which is at least
 	 * aligned to MAX_ORDER - 1.
 	 */
-	virtio_mem_clear_fake_offline(pfn, nr_pages);
+	for (i = 0; i < nr_pages; i += 1 << order) {
+		struct page *page = pfn_to_page(pfn + i);
 
-	for (i = 0; i < nr_pages; i += 1 << order)
-		generic_online_page(pfn_to_page(pfn + i), order);
+		/*
+		 * If the page is PageDirty(), it was kept fake-offline when
+		 * onlining the memory block. Otherwise, it was allocated
+		 * using alloc_contig_range(). All pages in a subblock are
+		 * alike.
+		 */
+		if (PageDirty(page)) {
+			virtio_mem_clear_fake_offline(pfn + i, 1 << order,
+						      false);
+			generic_online_page(page, order);
+		} else {
+			virtio_mem_clear_fake_offline(pfn + i, 1 << order,
+						      true);
+			free_contig_range(pfn + i, 1 << order);
+			adjust_managed_page_count(page, 1 << order);
+		}
+	}
 }
 
 static void virtio_mem_online_page_cb(struct page *page, unsigned int order)
@@ -718,7 +750,8 @@ static void virtio_mem_online_page_cb(struct page *page, unsigned int order)
 		if (virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id, 1))
 			generic_online_page(page, order);
 		else
-			virtio_mem_set_fake_offline(PFN_DOWN(addr), 1 << order);
+			virtio_mem_set_fake_offline(PFN_DOWN(addr), 1 << order,
+						    false);
 		rcu_read_unlock();
 		return;
 	}
@@ -1186,6 +1219,72 @@ static int virtio_mem_mb_unplug_any_sb_offline(struct virtio_mem *vm,
 	return 0;
 }
 
+/*
+ * Unplug the desired number of plugged subblocks of an online memory block.
+ * Will skip subblock that are busy.
+ *
+ * Will modify the state of the memory block.
+ *
+ * Note: Can fail after some subblocks were successfully unplugged. Can
+ *       return 0 even if subblocks were busy and could not get unplugged.
+ */
+static int virtio_mem_mb_unplug_any_sb_online(struct virtio_mem *vm,
+					      unsigned long mb_id,
+					      uint64_t *nb_sb)
+{
+	const unsigned long nr_pages = PFN_DOWN(vm->subblock_size);
+	unsigned long start_pfn;
+	int rc, sb_id;
+
+	/*
+	 * TODO: To increase the performance we want to try bigger, consecutive
+	 * subblocks first before falling back to single subblocks. Also,
+	 * we should sense via something like is_mem_section_removable()
+	 * first if it makes sense to go ahead any try to allocate.
+	 */
+	for (sb_id = 0; sb_id < vm->nb_sb_per_mb && *nb_sb; sb_id++) {
+		/* Find the next candidate subblock */
+		while (sb_id < vm->nb_sb_per_mb &&
+		       !virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id, 1))
+			sb_id++;
+		if (sb_id >= vm->nb_sb_per_mb)
+			break;
+
+		start_pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
+				     sb_id * vm->subblock_size);
+		rc = alloc_contig_range(start_pfn, start_pfn + nr_pages,
+					MIGRATE_MOVABLE, GFP_KERNEL);
+		if (rc == -ENOMEM)
+			/* whoops, out of memory */
+			return rc;
+		if (rc)
+			/* memory busy, we can't unplug this chunk */
+			continue;
+
+		/* Mark it as fake-offline before unplugging it */
+		virtio_mem_set_fake_offline(start_pfn, nr_pages, true);
+		adjust_managed_page_count(pfn_to_page(start_pfn), -nr_pages);
+
+		/* Try to unplug the allocated memory */
+		rc = virtio_mem_mb_unplug_sb(vm, mb_id, sb_id, 1);
+		if (rc) {
+			/* Return the memory to the buddy. */
+			virtio_mem_fake_online(start_pfn, nr_pages);
+			return rc;
+		}
+
+		virtio_mem_mb_set_state(vm, mb_id,
+					VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL);
+		*nb_sb -= 1;
+	}
+
+	/*
+	 * TODO: Once all subblocks of a memory block were unplugged, we want
+	 * to offline the memory block and remove it.
+	 */
+	return 0;
+}
+
 /*
  * Try to unplug the requested amount of memory.
  */
@@ -1225,8 +1324,37 @@ static int virtio_mem_unplug_request(struct virtio_mem *vm, uint64_t diff)
 		cond_resched();
 	}
 
+	if (!unplug_online) {
+		mutex_unlock(&vm->hotplug_mutex);
+		return 0;
+	}
+
+	/* Try to unplug subblocks of partially plugged online blocks. */
+	virtio_mem_for_each_mb_state_rev(vm, mb_id,
+					 VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL) {
+		rc = virtio_mem_mb_unplug_any_sb_online(vm, mb_id,
+							&nb_sb);
+		if (rc || !nb_sb)
+			goto out_unlock;
+		mutex_unlock(&vm->hotplug_mutex);
+		cond_resched();
+		mutex_lock(&vm->hotplug_mutex);
+	}
+
+	/* Try to unplug subblocks of plugged online blocks. */
+	virtio_mem_for_each_mb_state_rev(vm, mb_id,
+					 VIRTIO_MEM_MB_STATE_ONLINE) {
+		rc = virtio_mem_mb_unplug_any_sb_online(vm, mb_id,
+							&nb_sb);
+		if (rc || !nb_sb)
+			goto out_unlock;
+		mutex_unlock(&vm->hotplug_mutex);
+		cond_resched();
+		mutex_lock(&vm->hotplug_mutex);
+	}
+
 	mutex_unlock(&vm->hotplug_mutex);
-	return 0;
+	return nb_sb ? -EBUSY : 0;
 out_unlock:
 	mutex_unlock(&vm->hotplug_mutex);
 	return rc;
@@ -1332,7 +1460,8 @@ static void virtio_mem_run_wq(struct work_struct *work)
 	case -EBUSY:
 		/*
 		 * The hypervisor cannot process our request right now
-		 * (e.g., out of memory, migrating).
+		 * (e.g., out of memory, migrating) or we cannot free up
+		 * any memory to unplug it (all plugged memory is busy).
 		 */
 	case -ENOMEM:
 		/* Out of memory, try again later. */
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 69827d4fa052..43fbee4ee606 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -8594,6 +8594,7 @@ int alloc_contig_range(unsigned long start, unsigned long end,
 				pfn_max_align_up(end), migratetype);
 	return ret;
 }
+EXPORT_SYMBOL(alloc_contig_range);
 
 static int __alloc_contig_pages(unsigned long start_pfn,
 				unsigned long nr_pages, gfp_t gfp_mask)
@@ -8709,6 +8710,7 @@ void free_contig_range(unsigned long pfn, unsigned int nr_pages)
 	}
 	WARN(count != 0, "%d pages are still in use!\n", count);
 }
+EXPORT_SYMBOL(free_contig_range);
 
 /*
  * The zone indicated has a new number of managed_pages; batch sizes and percpu
-- 
2.25.3

[PATCH v4 06/15] mm: Allow to offline unmovable PageOffline() pages via MEM_GOING_OFFLINE

[David Hildenbrand]

virtio-mem wants to allow to offline memory blocks of which some parts
were unplugged (allocated via alloc_contig_range()), especially, to later
offline and remove completely unplugged memory blocks. The important part
is that PageOffline() has to remain set until the section is offline, so
these pages will never get accessed (e.g., when dumping). The pages should
not be handed back to the buddy (which would require clearing PageOffline()
and result in issues if offlining fails and the pages are suddenly in the
buddy).

Let's allow to do that by allowing to isolate any PageOffline() page
when offlining. This way, we can reach the memory hotplug notifier
MEM_GOING_OFFLINE, where the driver can signal that he is fine with
offlining this page by dropping its reference count. PageOffline() pages
with a reference count of 0 can then be skipped when offlining the
pages (like if they were free, however they are not in the buddy).

Anybody who uses PageOffline() pages and does not agree to offline them
(e.g., Hyper-V balloon, XEN balloon, VMWare balloon for 2MB pages) will not
decrement the reference count and make offlining fail when trying to
migrate such an unmovable page. So there should be no observable change.
Same applies to balloon compaction users (movable PageOffline() pages), the
pages will simply be migrated.

Note 1: If offlining fails, a driver has to increment the reference
	count again in MEM_CANCEL_OFFLINE.

Note 2: A driver that makes use of this has to be aware that re-onlining
	the memory block has to be handled by hooking into onlining code
	(online_page_callback_t), resetting the page PageOffline() and
	not giving them to the buddy.

Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Juergen Gross <jgross@suse.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Pavel Tatashin <pavel.tatashin@microsoft.com>
Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Anthony Yznaga <anthony.yznaga@oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Pingfan Liu <kernelfans@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/page-flags.h | 10 +++++++++
 mm/memory_hotplug.c        | 44 +++++++++++++++++++++++++++++---------
 mm/page_alloc.c            | 24 +++++++++++++++++++++
 mm/page_isolation.c        |  9 ++++++++
 4 files changed, 77 insertions(+), 10 deletions(-)

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 222f6f7b2bb3..6be1aa559b1e 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -777,6 +777,16 @@ PAGE_TYPE_OPS(Buddy, buddy)
  * not onlined when onlining the section).
  * The content of these pages is effectively stale. Such pages should not
  * be touched (read/write/dump/save) except by their owner.
+ *
+ * If a driver wants to allow to offline unmovable PageOffline() pages without
+ * putting them back to the buddy, it can do so via the memory notifier by
+ * decrementing the reference count in MEM_GOING_OFFLINE and incrementing the
+ * reference count in MEM_CANCEL_OFFLINE. When offlining, the PageOffline()
+ * pages (now with a reference count of zero) are treated like free pages,
+ * allowing the containing memory block to get offlined. A driver that
+ * relies on this feature is aware that re-onlining the memory block will
+ * require to re-set the pages PageOffline() and not giving them to the
+ * buddy via online_page_callback_t.
  */
 PAGE_TYPE_OPS(Offline, offline)
 
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index fc0aad0bc1f5..008e4a7ed8bc 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1224,11 +1224,17 @@ struct zone *test_pages_in_a_zone(unsigned long start_pfn,
 
 /*
  * Scan pfn range [start,end) to find movable/migratable pages (LRU pages,
- * non-lru movable pages and hugepages). We scan pfn because it's much
- * easier than scanning over linked list. This function returns the pfn
- * of the first found movable page if it's found, otherwise 0.
+ * non-lru movable pages and hugepages). Will skip over most unmovable
+ * pages (esp., pages that can be skipped when offlining), but bail out on
+ * definitely unmovable pages.
+ *
+ * Returns:
+ *	0 in case a movable page is found and movable_pfn was updated.
+ *	-ENOENT in case no movable page was found.
+ *	-EBUSY in case a definitely unmovable page was found.
  */
-static unsigned long scan_movable_pages(unsigned long start, unsigned long end)
+static int scan_movable_pages(unsigned long start, unsigned long end,
+			      unsigned long *movable_pfn)
 {
 	unsigned long pfn;
 
@@ -1240,18 +1246,30 @@ static unsigned long scan_movable_pages(unsigned long start, unsigned long end)
 			continue;
 		page = pfn_to_page(pfn);
 		if (PageLRU(page))
-			return pfn;
+			goto found;
 		if (__PageMovable(page))
-			return pfn;
+			goto found;
+
+		/*
+		 * PageOffline() pages that are not marked __PageMovable() and
+		 * have a reference count > 0 (after MEM_GOING_OFFLINE) are
+		 * definitely unmovable. If their reference count would be 0,
+		 * they could at least be skipped when offlining memory.
+		 */
+		if (PageOffline(page) && page_count(page))
+			return -EBUSY;
 
 		if (!PageHuge(page))
 			continue;
 		head = compound_head(page);
 		if (page_huge_active(head))
-			return pfn;
+			goto found;
 		skip = compound_nr(head) - (page - head);
 		pfn += skip - 1;
 	}
+	return -ENOENT;
+found:
+	*movable_pfn = pfn;
 	return 0;
 }
 
@@ -1518,7 +1536,8 @@ static int __ref __offline_pages(unsigned long start_pfn,
 	}
 
 	do {
-		for (pfn = start_pfn; pfn;) {
+		pfn = start_pfn;
+		do {
 			if (signal_pending(current)) {
 				ret = -EINTR;
 				reason = "signal backoff";
@@ -1528,14 +1547,19 @@ static int __ref __offline_pages(unsigned long start_pfn,
 			cond_resched();
 			lru_add_drain_all();
 
-			pfn = scan_movable_pages(pfn, end_pfn);
-			if (pfn) {
+			ret = scan_movable_pages(pfn, end_pfn, &pfn);
+			if (!ret) {
 				/*
 				 * TODO: fatal migration failures should bail
 				 * out
 				 */
 				do_migrate_range(pfn, end_pfn);
 			}
+		} while (!ret);
+
+		if (ret != -ENOENT) {
+			reason = "unmovable page";
+			goto failed_removal_isolated;
 		}
 
 		/*
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 43fbee4ee606..00854e3d08a8 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -8363,6 +8363,19 @@ struct page *has_unmovable_pages(struct zone *zone, struct page *page,
 		if ((flags & MEMORY_OFFLINE) && PageHWPoison(page))
 			continue;
 
+		/*
+		 * We treat all PageOffline() pages as movable when offlining
+		 * to give drivers a chance to decrement their reference count
+		 * in MEM_GOING_OFFLINE in order to indicate that these pages
+		 * can be offlined as there are no direct references anymore.
+		 * For actually unmovable PageOffline() where the driver does
+		 * not support this, we will fail later when trying to actually
+		 * move these pages that still have a reference count > 0.
+		 * (false negatives in this function only)
+		 */
+		if ((flags & MEMORY_OFFLINE) && PageOffline(page))
+			continue;
+
 		if (__PageMovable(page) || PageLRU(page))
 			continue;
 
@@ -8783,6 +8796,17 @@ __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
 			offlined_pages++;
 			continue;
 		}
+		/*
+		 * At this point all remaining PageOffline() pages have a
+		 * reference count of 0 and can simply be skipped.
+		 */
+		if (PageOffline(page)) {
+			BUG_ON(page_count(page));
+			BUG_ON(PageBuddy(page));
+			pfn++;
+			offlined_pages++;
+			continue;
+		}
 
 		BUG_ON(page_count(page));
 		BUG_ON(!PageBuddy(page));
diff --git a/mm/page_isolation.c b/mm/page_isolation.c
index 2c11a38d6e87..f6d07c5f0d34 100644
--- a/mm/page_isolation.c
+++ b/mm/page_isolation.c
@@ -151,6 +151,7 @@ __first_valid_page(unsigned long pfn, unsigned long nr_pages)
  *			a bit mask)
  *			MEMORY_OFFLINE - isolate to offline (!allocate) memory
  *					 e.g., skip over PageHWPoison() pages
+ *					 and PageOffline() pages.
  *			REPORT_FAILURE - report details about the failure to
  *			isolate the range
  *
@@ -259,6 +260,14 @@ __test_page_isolated_in_pageblock(unsigned long pfn, unsigned long end_pfn,
 		else if ((flags & MEMORY_OFFLINE) && PageHWPoison(page))
 			/* A HWPoisoned page cannot be also PageBuddy */
 			pfn++;
+		else if ((flags & MEMORY_OFFLINE) && PageOffline(page) &&
+			 !page_count(page))
+			/*
+			 * The responsible driver agreed to skip PageOffline()
+			 * pages when offlining memory by dropping its
+			 * reference in MEM_GOING_OFFLINE.
+			 */
+			pfn++;
 		else
 			break;
 	}
-- 
2.25.3

[PATCH v4 07/15] virtio-mem: Allow to offline partially unplugged memory blocks

[David Hildenbrand]

Dropping the reference count of PageOffline() pages during MEM_GOING_ONLINE
allows offlining code to skip them. However, we also have to clear
PG_reserved, because PG_reserved pages get detected as unmovable right
away. Take care of restoring the reference count when offlining is
canceled.

Clarify why we don't have to perform any action when unloading the
driver. Also, let's add a warning if anybody is still holding a
reference to unplugged pages when offlining.

Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 68 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 67 insertions(+), 1 deletion(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 74f0d3cb1d22..b0b41c73ce89 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -572,6 +572,57 @@ static void virtio_mem_notify_online(struct virtio_mem *vm, unsigned long mb_id,
 		virtio_mem_retry(vm);
 }
 
+static void virtio_mem_notify_going_offline(struct virtio_mem *vm,
+					    unsigned long mb_id)
+{
+	const unsigned long nr_pages = PFN_DOWN(vm->subblock_size);
+	struct page *page;
+	unsigned long pfn;
+	int sb_id, i;
+
+	for (sb_id = 0; sb_id < vm->nb_sb_per_mb; sb_id++) {
+		if (virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id, 1))
+			continue;
+		/*
+		 * Drop our reference to the pages so the memory can get
+		 * offlined and add the unplugged pages to the managed
+		 * page counters (so offlining code can correctly subtract
+		 * them again).
+		 */
+		pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
+			       sb_id * vm->subblock_size);
+		adjust_managed_page_count(pfn_to_page(pfn), nr_pages);
+		for (i = 0; i < nr_pages; i++) {
+			page = pfn_to_page(pfn + i);
+			if (WARN_ON(!page_ref_dec_and_test(page)))
+				dump_page(page, "unplugged page referenced");
+		}
+	}
+}
+
+static void virtio_mem_notify_cancel_offline(struct virtio_mem *vm,
+					     unsigned long mb_id)
+{
+	const unsigned long nr_pages = PFN_DOWN(vm->subblock_size);
+	unsigned long pfn;
+	int sb_id, i;
+
+	for (sb_id = 0; sb_id < vm->nb_sb_per_mb; sb_id++) {
+		if (virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id, 1))
+			continue;
+		/*
+		 * Get the reference we dropped when going offline and
+		 * subtract the unplugged pages from the managed page
+		 * counters.
+		 */
+		pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
+			       sb_id * vm->subblock_size);
+		adjust_managed_page_count(pfn_to_page(pfn), -nr_pages);
+		for (i = 0; i < nr_pages; i++)
+			page_ref_inc(pfn_to_page(pfn + i));
+	}
+}
+
 /*
  * This callback will either be called synchronously from add_memory() or
  * asynchronously (e.g., triggered via user space). We have to be careful
@@ -618,6 +669,7 @@ static int virtio_mem_memory_notifier_cb(struct notifier_block *nb,
 			break;
 		}
 		vm->hotplug_active = true;
+		virtio_mem_notify_going_offline(vm, mb_id);
 		break;
 	case MEM_GOING_ONLINE:
 		mutex_lock(&vm->hotplug_mutex);
@@ -642,6 +694,12 @@ static int virtio_mem_memory_notifier_cb(struct notifier_block *nb,
 		mutex_unlock(&vm->hotplug_mutex);
 		break;
 	case MEM_CANCEL_OFFLINE:
+		if (!vm->hotplug_active)
+			break;
+		virtio_mem_notify_cancel_offline(vm, mb_id);
+		vm->hotplug_active = false;
+		mutex_unlock(&vm->hotplug_mutex);
+		break;
 	case MEM_CANCEL_ONLINE:
 		if (!vm->hotplug_active)
 			break;
@@ -668,8 +726,11 @@ static void virtio_mem_set_fake_offline(unsigned long pfn,
 		struct page *page = pfn_to_page(pfn);
 
 		__SetPageOffline(page);
-		if (!onlined)
+		if (!onlined) {
 			SetPageDirty(page);
+			/* FIXME: remove after cleanups */
+			ClearPageReserved(page);
+		}
 	}
 }
 
@@ -1722,6 +1783,11 @@ static void virtio_mem_remove(struct virtio_device *vdev)
 		BUG_ON(rc);
 		virtio_mem_mb_set_state(vm, mb_id, VIRTIO_MEM_MB_STATE_UNUSED);
 	}
+	/*
+	 * After we unregistered our callbacks, user space can no longer
+	 * offline partially plugged online memory blocks. No need to worry
+	 * about them.
+	 */
 
 	/* unregister callbacks */
 	unregister_virtio_mem_device(vm);
-- 
2.25.3

[PATCH v4 08/15] mm/memory_hotplug: Introduce offline_and_remove_memory()

[David Hildenbrand]

virtio-mem wants to offline and remove a memory block once it unplugged
all subblocks (e.g., using alloc_contig_range()). Let's provide
an interface to do that from a driver. virtio-mem already supports to
offline partially unplugged memory blocks. Offlining a fully unplugged
memory block will not require to migrate any pages. All unplugged
subblocks are PageOffline() and have a reference count of 0 - so
offlining code will simply skip them.

All we need is an interface to offline and remove the memory from kernel
module context, where we don't have access to the memory block devices
(esp. find_memory_block() and device_offline()) and the device hotplug
lock.

To keep things simple, allow to only work on a single memory block.

Acked-by: Michal Hocko <mhocko@suse.com>
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Oscar Salvador <osalvador@suse.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Wei Yang <richard.weiyang@gmail.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Qian Cai <cai@lca.pw>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 include/linux/memory_hotplug.h |  1 +
 mm/memory_hotplug.c            | 37 ++++++++++++++++++++++++++++++++++
 2 files changed, 38 insertions(+)

diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
index 93d9ada74ddd..cb7499843f5c 100644
--- a/include/linux/memory_hotplug.h
+++ b/include/linux/memory_hotplug.h
@@ -319,6 +319,7 @@ extern void try_offline_node(int nid);
 extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages);
 extern int remove_memory(int nid, u64 start, u64 size);
 extern void __remove_memory(int nid, u64 start, u64 size);
+extern int offline_and_remove_memory(int nid, u64 start, u64 size);
 
 #else
 static inline bool is_mem_section_removable(unsigned long pfn,
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 008e4a7ed8bc..4acb99aa9bf4 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1821,4 +1821,41 @@ int remove_memory(int nid, u64 start, u64 size)
 	return rc;
 }
 EXPORT_SYMBOL_GPL(remove_memory);
+
+/*
+ * Try to offline and remove a memory block. Might take a long time to
+ * finish in case memory is still in use. Primarily useful for memory devices
+ * that logically unplugged all memory (so it's no longer in use) and want to
+ * offline + remove the memory block.
+ */
+int offline_and_remove_memory(int nid, u64 start, u64 size)
+{
+	struct memory_block *mem;
+	int rc = -EINVAL;
+
+	if (!IS_ALIGNED(start, memory_block_size_bytes()) ||
+	    size != memory_block_size_bytes())
+		return rc;
+
+	lock_device_hotplug();
+	mem = find_memory_block(__pfn_to_section(PFN_DOWN(start)));
+	if (mem)
+		rc = device_offline(&mem->dev);
+	/* Ignore if the device is already offline. */
+	if (rc > 0)
+		rc = 0;
+
+	/*
+	 * In case we succeeded to offline the memory block, remove it.
+	 * This cannot fail as it cannot get onlined in the meantime.
+	 */
+	if (!rc) {
+		rc = try_remove_memory(nid, start, size);
+		WARN_ON_ONCE(rc);
+	}
+	unlock_device_hotplug();
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(offline_and_remove_memory);
 #endif /* CONFIG_MEMORY_HOTREMOVE */
-- 
2.25.3

[PATCH v4 09/15] virtio-mem: Offline and remove completely unplugged memory blocks

[David Hildenbrand]

Let's offline+remove memory blocks once all subblocks are unplugged. We
can use the new Linux MM interface for that. As no memory is in use
anymore, this shouldn't take a long time and shouldn't fail. There might
be corner cases where the offlining could still fail (especially, if
another notifier NACKs the offlining request).

Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 47 +++++++++++++++++++++++++++++++++----
 1 file changed, 43 insertions(+), 4 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index b0b41c73ce89..a2edb87e5ed8 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -446,6 +446,28 @@ static int virtio_mem_mb_remove(struct virtio_mem *vm, unsigned long mb_id)
 	return remove_memory(nid, addr, memory_block_size_bytes());
 }
 
+/*
+ * Try to offline and remove a memory block from Linux.
+ *
+ * Must not be called with the vm->hotplug_mutex held (possible deadlock with
+ * onlining code).
+ *
+ * Will not modify the state of the memory block.
+ */
+static int virtio_mem_mb_offline_and_remove(struct virtio_mem *vm,
+					    unsigned long mb_id)
+{
+	const uint64_t addr = virtio_mem_mb_id_to_phys(mb_id);
+	int nid = vm->nid;
+
+	if (nid == NUMA_NO_NODE)
+		nid = memory_add_physaddr_to_nid(addr);
+
+	dev_dbg(&vm->vdev->dev, "offlining and removing memory block: %lu\n",
+		mb_id);
+	return offline_and_remove_memory(nid, addr, memory_block_size_bytes());
+}
+
 /*
  * Trigger the workqueue so the device can perform its magic.
  */
@@ -537,7 +559,13 @@ static void virtio_mem_notify_offline(struct virtio_mem *vm,
 		break;
 	}
 
-	/* trigger the workqueue, maybe we can now unplug memory. */
+	/*
+	 * Trigger the workqueue, maybe we can now unplug memory. Also,
+	 * when we offline and remove a memory block, this will re-trigger
+	 * us immediately - which is often nice because the removal of
+	 * the memory block (e.g., memmap) might have freed up memory
+	 * on other memory blocks we manage.
+	 */
 	virtio_mem_retry(vm);
 }
 
@@ -1284,7 +1312,8 @@ static int virtio_mem_mb_unplug_any_sb_offline(struct virtio_mem *vm,
  * Unplug the desired number of plugged subblocks of an online memory block.
  * Will skip subblock that are busy.
  *
- * Will modify the state of the memory block.
+ * Will modify the state of the memory block. Might temporarily drop the
+ * hotplug_mutex.
  *
  * Note: Can fail after some subblocks were successfully unplugged. Can
  *       return 0 even if subblocks were busy and could not get unplugged.
@@ -1340,9 +1369,19 @@ static int virtio_mem_mb_unplug_any_sb_online(struct virtio_mem *vm,
 	}
 
 	/*
-	 * TODO: Once all subblocks of a memory block were unplugged, we want
-	 * to offline the memory block and remove it.
+	 * Once all subblocks of a memory block were unplugged, offline and
+	 * remove it. This will usually not fail, as no memory is in use
+	 * anymore - however some other notifiers might NACK the request.
 	 */
+	if (virtio_mem_mb_test_sb_unplugged(vm, mb_id, 0, vm->nb_sb_per_mb)) {
+		mutex_unlock(&vm->hotplug_mutex);
+		rc = virtio_mem_mb_offline_and_remove(vm, mb_id);
+		mutex_lock(&vm->hotplug_mutex);
+		if (!rc)
+			virtio_mem_mb_set_state(vm, mb_id,
+						VIRTIO_MEM_MB_STATE_UNUSED);
+	}
+
 	return 0;
 }
 
-- 
2.25.3

[PATCH v4 10/15] virtio-mem: Better retry handling

[David Hildenbrand]

Let's start with a retry interval of 5 seconds and double the time until
we reach 5 minutes, in case we keep getting errors. Reset the retry
interval in case we succeeded.

The two main reasons for having to retry are
- The hypervisor is busy and cannot process our request
- We cannot reach the desired requested_size (esp., not enough memory can
  get unplugged because we can't allocate any subblocks).

Tested-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index a2edb87e5ed8..eb4c16d634e0 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -141,7 +141,9 @@ struct virtio_mem {
 
 	/* Timer for retrying to plug/unplug memory. */
 	struct hrtimer retry_timer;
-#define VIRTIO_MEM_RETRY_TIMER_MS		30000
+	unsigned int retry_timer_ms;
+#define VIRTIO_MEM_RETRY_TIMER_MIN_MS		50000
+#define VIRTIO_MEM_RETRY_TIMER_MAX_MS		300000
 
 	/* Memory notifier (online/offline events). */
 	struct notifier_block memory_notifier;
@@ -1550,6 +1552,7 @@ static void virtio_mem_run_wq(struct work_struct *work)
 
 	switch (rc) {
 	case 0:
+		vm->retry_timer_ms = VIRTIO_MEM_RETRY_TIMER_MIN_MS;
 		break;
 	case -ENOSPC:
 		/*
@@ -1565,8 +1568,7 @@ static void virtio_mem_run_wq(struct work_struct *work)
 		 */
 	case -ENOMEM:
 		/* Out of memory, try again later. */
-		hrtimer_start(&vm->retry_timer,
-			      ms_to_ktime(VIRTIO_MEM_RETRY_TIMER_MS),
+		hrtimer_start(&vm->retry_timer, ms_to_ktime(vm->retry_timer_ms),
 			      HRTIMER_MODE_REL);
 		break;
 	case -EAGAIN:
@@ -1586,6 +1588,8 @@ static enum hrtimer_restart virtio_mem_timer_expired(struct hrtimer *timer)
 					     retry_timer);
 
 	virtio_mem_retry(vm);
+	vm->retry_timer_ms = min_t(unsigned int, vm->retry_timer_ms * 2,
+				   VIRTIO_MEM_RETRY_TIMER_MAX_MS);
 	return HRTIMER_NORESTART;
 }
 
@@ -1754,6 +1758,7 @@ static int virtio_mem_probe(struct virtio_device *vdev)
 	spin_lock_init(&vm->removal_lock);
 	hrtimer_init(&vm->retry_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	vm->retry_timer.function = virtio_mem_timer_expired;
+	vm->retry_timer_ms = VIRTIO_MEM_RETRY_TIMER_MIN_MS;
 
 	/* register the virtqueue */
 	rc = virtio_mem_init_vq(vm);
-- 
2.25.3

[PATCH v4 11/15] virtio-mem: Add parent resource for all added "System RAM"

[David Hildenbrand]

Let's add a parent resource, named after the virtio device (inspired by
drivers/dax/kmem.c). This allows user space to identify which memory
belongs to which virtio-mem device.

With this change and two virtio-mem devices:
	:/# cat /proc/iomem
	00000000-00000fff : Reserved
	00001000-0009fbff : System RAM
	[...]
	140000000-333ffffff : virtio0
	  140000000-147ffffff : System RAM
	  148000000-14fffffff : System RAM
	  150000000-157ffffff : System RAM
	[...]
	334000000-3033ffffff : virtio1
	  338000000-33fffffff : System RAM
	  340000000-347ffffff : System RAM
	  348000000-34fffffff : System RAM
	[...]

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 52 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 51 insertions(+), 1 deletion(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index eb4c16d634e0..80cdb9e6b3c4 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -99,6 +99,9 @@ struct virtio_mem {
 	/* Id of the next memory bock to prepare when needed. */
 	unsigned long next_mb_id;
 
+	/* The parent resource for all memory added via this device. */
+	struct resource *parent_resource;
+
 	/* Summary of all memory block states. */
 	unsigned long nb_mb_state[VIRTIO_MEM_MB_STATE_COUNT];
 #define VIRTIO_MEM_NB_OFFLINE_THRESHOLD		10
@@ -1741,6 +1744,44 @@ static int virtio_mem_init(struct virtio_mem *vm)
 	return 0;
 }
 
+static int virtio_mem_create_resource(struct virtio_mem *vm)
+{
+	/*
+	 * When force-unloading the driver and removing the device, we
+	 * could have a garbage pointer. Duplicate the string.
+	 */
+	const char *name = kstrdup(dev_name(&vm->vdev->dev), GFP_KERNEL);
+
+	if (!name)
+		return -ENOMEM;
+
+	vm->parent_resource = __request_mem_region(vm->addr, vm->region_size,
+						   name, IORESOURCE_SYSTEM_RAM);
+	if (!vm->parent_resource) {
+		kfree(name);
+		dev_warn(&vm->vdev->dev, "could not reserve device region\n");
+		return -EBUSY;
+	}
+
+	/* The memory is not actually busy - make add_memory() work. */
+	vm->parent_resource->flags &= ~IORESOURCE_BUSY;
+	return 0;
+}
+
+static void virtio_mem_delete_resource(struct virtio_mem *vm)
+{
+	const char *name;
+
+	if (!vm->parent_resource)
+		return;
+
+	name = vm->parent_resource->name;
+	release_resource(vm->parent_resource);
+	kfree(vm->parent_resource);
+	kfree(name);
+	vm->parent_resource = NULL;
+}
+
 static int virtio_mem_probe(struct virtio_device *vdev)
 {
 	struct virtio_mem *vm;
@@ -1770,11 +1811,16 @@ static int virtio_mem_probe(struct virtio_device *vdev)
 	if (rc)
 		goto out_del_vq;
 
+	/* create the parent resource for all memory */
+	rc = virtio_mem_create_resource(vm);
+	if (rc)
+		goto out_del_vq;
+
 	/* register callbacks */
 	vm->memory_notifier.notifier_call = virtio_mem_memory_notifier_cb;
 	rc = register_memory_notifier(&vm->memory_notifier);
 	if (rc)
-		goto out_del_vq;
+		goto out_del_resource;
 	rc = register_virtio_mem_device(vm);
 	if (rc)
 		goto out_unreg_mem;
@@ -1788,6 +1834,8 @@ static int virtio_mem_probe(struct virtio_device *vdev)
 	return 0;
 out_unreg_mem:
 	unregister_memory_notifier(&vm->memory_notifier);
+out_del_resource:
+	virtio_mem_delete_resource(vm);
 out_del_vq:
 	vdev->config->del_vqs(vdev);
 out_free_vm:
@@ -1848,6 +1896,8 @@ static void virtio_mem_remove(struct virtio_device *vdev)
 	    vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL] ||
 	    vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE_MOVABLE])
 		dev_warn(&vdev->dev, "device still has system memory added\n");
+	else
+		virtio_mem_delete_resource(vm);
 
 	/* remove all tracking data - no locking needed */
 	vfree(vm->mb_state);
-- 
2.25.3

[PATCH v4 12/15] virtio-mem: Drop manual check for already present memory

[David Hildenbrand]

Registering our parent resource will fail if any memory is still present
(e.g., because somebody unloaded the driver and tries to reload it). No
need for the manual check.

Move our "unplug all" handling to after registering the resource.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 55 ++++++++-----------------------------
 1 file changed, 12 insertions(+), 43 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 80cdb9e6b3c4..8dd57b61b09b 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -1616,23 +1616,6 @@ static int virtio_mem_init_vq(struct virtio_mem *vm)
 	return 0;
 }
 
-/*
- * Test if any memory in the range is present in Linux.
- */
-static bool virtio_mem_any_memory_present(unsigned long start,
-					  unsigned long size)
-{
-	const unsigned long start_pfn = PFN_DOWN(start);
-	const unsigned long end_pfn = PFN_UP(start + size);
-	unsigned long pfn;
-
-	for (pfn = start_pfn; pfn != end_pfn; pfn++)
-		if (present_section_nr(pfn_to_section_nr(pfn)))
-			return true;
-
-	return false;
-}
-
 static int virtio_mem_init(struct virtio_mem *vm)
 {
 	const uint64_t phys_limit = 1UL << MAX_PHYSMEM_BITS;
@@ -1664,32 +1647,6 @@ static int virtio_mem_init(struct virtio_mem *vm)
 	virtio_cread(vm->vdev, struct virtio_mem_config, region_size,
 		     &vm->region_size);
 
-	/*
-	 * If we still have memory plugged, we might have to unplug all
-	 * memory first. However, if somebody simply unloaded the driver
-	 * we would have to reinitialize the old state - something we don't
-	 * support yet. Detect if we have any memory in the area present.
-	 */
-	if (vm->plugged_size) {
-		uint64_t usable_region_size;
-
-		virtio_cread(vm->vdev, struct virtio_mem_config,
-			     usable_region_size, &usable_region_size);
-
-		if (virtio_mem_any_memory_present(vm->addr,
-						  usable_region_size)) {
-			dev_err(&vm->vdev->dev,
-				"reloading the driver is not supported\n");
-			return -EINVAL;
-		}
-		/*
-		 * Note: it might happen that the device is busy and
-		 * unplugging all memory might take some time.
-		 */
-		dev_info(&vm->vdev->dev, "unplugging all memory required\n");
-		vm->unplug_all_required = 1;
-	}
-
 	/*
 	 * We always hotplug memory in memory block granularity. This way,
 	 * we have to wait for exactly one memory block to online.
@@ -1760,6 +1717,8 @@ static int virtio_mem_create_resource(struct virtio_mem *vm)
 	if (!vm->parent_resource) {
 		kfree(name);
 		dev_warn(&vm->vdev->dev, "could not reserve device region\n");
+		dev_info(&vm->vdev->dev,
+			 "reloading the driver is not supported\n");
 		return -EBUSY;
 	}
 
@@ -1816,6 +1775,16 @@ static int virtio_mem_probe(struct virtio_device *vdev)
 	if (rc)
 		goto out_del_vq;
 
+	/*
+	 * If we still have memory plugged, we have to unplug all memory first.
+	 * Registering our parent resource makes sure that this memory isn't
+	 * actually in use (e.g., trying to reload the driver).
+	 */
+	if (vm->plugged_size) {
+		vm->unplug_all_required = 1;
+		dev_info(&vm->vdev->dev, "unplugging all memory is required\n");
+	}
+
 	/* register callbacks */
 	vm->memory_notifier.notifier_call = virtio_mem_memory_notifier_cb;
 	rc = register_memory_notifier(&vm->memory_notifier);
-- 
2.25.3

[PATCH v4 13/15] virtio-mem: Unplug subblocks right-to-left

[David Hildenbrand]

We unplug blocks right-to-left, let's also unplug subblocks within a block
right-to-left.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 38 ++++++++++++++++---------------------
 1 file changed, 16 insertions(+), 22 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 8dd57b61b09b..a719e1a04ac7 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -353,18 +353,6 @@ static bool virtio_mem_mb_test_sb_unplugged(struct virtio_mem *vm,
 	return find_next_bit(vm->sb_bitmap, bit + count, bit) >= bit + count;
 }
 
-/*
- * Find the first plugged subblock. Returns vm->nb_sb_per_mb in case there is
- * none.
- */
-static int virtio_mem_mb_first_plugged_sb(struct virtio_mem *vm,
-					  unsigned long mb_id)
-{
-	const int bit = (mb_id - vm->first_mb_id) * vm->nb_sb_per_mb;
-
-	return find_next_bit(vm->sb_bitmap, bit + vm->nb_sb_per_mb, bit) - bit;
-}
-
 /*
  * Find the first unplugged subblock. Returns vm->nb_sb_per_mb in case there is
  * none.
@@ -1016,21 +1004,27 @@ static int virtio_mem_mb_unplug_any_sb(struct virtio_mem *vm,
 	int sb_id, count;
 	int rc;
 
+	sb_id = vm->nb_sb_per_mb - 1;
 	while (*nb_sb) {
-		sb_id = virtio_mem_mb_first_plugged_sb(vm, mb_id);
-		if (sb_id >= vm->nb_sb_per_mb)
+		/* Find the next candidate subblock */
+		while (sb_id >= 0 &&
+		       virtio_mem_mb_test_sb_unplugged(vm, mb_id, sb_id, 1))
+			sb_id--;
+		if (sb_id < 0)
 			break;
+		/* Try to unplug multiple subblocks at a time */
 		count = 1;
-		while (count < *nb_sb &&
-		       sb_id + count  < vm->nb_sb_per_mb &&
-		       virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id + count,
-						     1))
+		while (count < *nb_sb && sb_id > 0 &&
+		       virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id - 1, 1)) {
 			count++;
+			sb_id--;
+		}
 
 		rc = virtio_mem_mb_unplug_sb(vm, mb_id, sb_id, count);
 		if (rc)
 			return rc;
 		*nb_sb -= count;
+		sb_id--;
 	}
 
 	return 0;
@@ -1337,12 +1331,12 @@ static int virtio_mem_mb_unplug_any_sb_online(struct virtio_mem *vm,
 	 * we should sense via something like is_mem_section_removable()
 	 * first if it makes sense to go ahead any try to allocate.
 	 */
-	for (sb_id = 0; sb_id < vm->nb_sb_per_mb && *nb_sb; sb_id++) {
+	for (sb_id = vm->nb_sb_per_mb - 1; sb_id >= 0 && *nb_sb; sb_id--) {
 		/* Find the next candidate subblock */
-		while (sb_id < vm->nb_sb_per_mb &&
+		while (sb_id >= 0 &&
 		       !virtio_mem_mb_test_sb_plugged(vm, mb_id, sb_id, 1))
-			sb_id++;
-		if (sb_id >= vm->nb_sb_per_mb)
+			sb_id--;
+		if (sb_id < 0)
 			break;
 
 		start_pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
-- 
2.25.3

[PATCH v4 14/15] virtio-mem: Use -ETXTBSY as error code if the device is busy

[David Hildenbrand]

Let's be able to distinguish if the device or if memory is busy.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index a719e1a04ac7..abd93b778a26 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -893,7 +893,7 @@ static int virtio_mem_send_plug_request(struct virtio_mem *vm, uint64_t addr,
 	case VIRTIO_MEM_RESP_NACK:
 		return -EAGAIN;
 	case VIRTIO_MEM_RESP_BUSY:
-		return -EBUSY;
+		return -ETXTBSY;
 	case VIRTIO_MEM_RESP_ERROR:
 		return -EINVAL;
 	default:
@@ -919,7 +919,7 @@ static int virtio_mem_send_unplug_request(struct virtio_mem *vm, uint64_t addr,
 		vm->plugged_size -= size;
 		return 0;
 	case VIRTIO_MEM_RESP_BUSY:
-		return -EBUSY;
+		return -ETXTBSY;
 	case VIRTIO_MEM_RESP_ERROR:
 		return -EINVAL;
 	default:
@@ -941,7 +941,7 @@ static int virtio_mem_send_unplug_all_request(struct virtio_mem *vm)
 		atomic_set(&vm->config_changed, 1);
 		return 0;
 	case VIRTIO_MEM_RESP_BUSY:
-		return -EBUSY;
+		return -ETXTBSY;
 	default:
 		return -ENOMEM;
 	}
@@ -1557,11 +1557,15 @@ static void virtio_mem_run_wq(struct work_struct *work)
 		 * or we have too many offline memory blocks.
 		 */
 		break;
-	case -EBUSY:
+	case -ETXTBSY:
 		/*
 		 * The hypervisor cannot process our request right now
-		 * (e.g., out of memory, migrating) or we cannot free up
-		 * any memory to unplug it (all plugged memory is busy).
+		 * (e.g., out of memory, migrating);
+		 */
+	case -EBUSY:
+		/*
+		 * We cannot free up any memory to unplug it (all plugged memory
+		 * is busy).
 		 */
 	case -ENOMEM:
 		/* Out of memory, try again later. */
-- 
2.25.3

[PATCH v4 15/15] virtio-mem: Try to unplug the complete online memory block first

[David Hildenbrand]

Right now, we always try to unplug single subblocks when processing an
online memory block. Let's try to unplug the complete online memory block
first, in case it is fully plugged and the unplug request is large
enough. Fallback to single subblocks in case the memory block cannot get
unplugged as a whole.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---
 drivers/virtio/virtio_mem.c | 88 ++++++++++++++++++++++++-------------
 1 file changed, 57 insertions(+), 31 deletions(-)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index abd93b778a26..9e523db3bee1 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -1307,6 +1307,46 @@ static int virtio_mem_mb_unplug_any_sb_offline(struct virtio_mem *vm,
 	return 0;
 }
 
+/*
+ * Unplug the given plugged subblocks of an online memory block.
+ *
+ * Will modify the state of the memory block.
+ */
+static int virtio_mem_mb_unplug_sb_online(struct virtio_mem *vm,
+					  unsigned long mb_id, int sb_id,
+					  int count)
+{
+	const unsigned long nr_pages = PFN_DOWN(vm->subblock_size) * count;
+	unsigned long start_pfn;
+	int rc;
+
+	start_pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
+			     sb_id * vm->subblock_size);
+	rc = alloc_contig_range(start_pfn, start_pfn + nr_pages,
+				MIGRATE_MOVABLE, GFP_KERNEL);
+	if (rc == -ENOMEM)
+		/* whoops, out of memory */
+		return rc;
+	if (rc)
+		return -EBUSY;
+
+	/* Mark it as fake-offline before unplugging it */
+	virtio_mem_set_fake_offline(start_pfn, nr_pages, true);
+	adjust_managed_page_count(pfn_to_page(start_pfn), -nr_pages);
+
+	/* Try to unplug the allocated memory */
+	rc = virtio_mem_mb_unplug_sb(vm, mb_id, sb_id, count);
+	if (rc) {
+		/* Return the memory to the buddy. */
+		virtio_mem_fake_online(start_pfn, nr_pages);
+		return rc;
+	}
+
+	virtio_mem_mb_set_state(vm, mb_id,
+				VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL);
+	return 0;
+}
+
 /*
  * Unplug the desired number of plugged subblocks of an online memory block.
  * Will skip subblock that are busy.
@@ -1321,16 +1361,21 @@ static int virtio_mem_mb_unplug_any_sb_online(struct virtio_mem *vm,
 					      unsigned long mb_id,
 					      uint64_t *nb_sb)
 {
-	const unsigned long nr_pages = PFN_DOWN(vm->subblock_size);
-	unsigned long start_pfn;
 	int rc, sb_id;
 
-	/*
-	 * TODO: To increase the performance we want to try bigger, consecutive
-	 * subblocks first before falling back to single subblocks. Also,
-	 * we should sense via something like is_mem_section_removable()
-	 * first if it makes sense to go ahead any try to allocate.
-	 */
+	/* If possible, try to unplug the complete block in one shot. */
+	if (*nb_sb >= vm->nb_sb_per_mb &&
+	    virtio_mem_mb_test_sb_plugged(vm, mb_id, 0, vm->nb_sb_per_mb)) {
+		rc = virtio_mem_mb_unplug_sb_online(vm, mb_id, 0,
+						    vm->nb_sb_per_mb);
+		if (!rc) {
+			*nb_sb -= vm->nb_sb_per_mb;
+			goto unplugged;
+		} else if (rc != -EBUSY)
+			return rc;
+	}
+
+	/* Fallback to single subblocks. */
 	for (sb_id = vm->nb_sb_per_mb - 1; sb_id >= 0 && *nb_sb; sb_id--) {
 		/* Find the next candidate subblock */
 		while (sb_id >= 0 &&
@@ -1339,34 +1384,15 @@ static int virtio_mem_mb_unplug_any_sb_online(struct virtio_mem *vm,
 		if (sb_id < 0)
 			break;
 
-		start_pfn = PFN_DOWN(virtio_mem_mb_id_to_phys(mb_id) +
-				     sb_id * vm->subblock_size);
-		rc = alloc_contig_range(start_pfn, start_pfn + nr_pages,
-					MIGRATE_MOVABLE, GFP_KERNEL);
-		if (rc == -ENOMEM)
-			/* whoops, out of memory */
-			return rc;
-		if (rc)
-			/* memory busy, we can't unplug this chunk */
+		rc = virtio_mem_mb_unplug_sb_online(vm, mb_id, sb_id, 1);
+		if (rc == -EBUSY)
 			continue;
-
-		/* Mark it as fake-offline before unplugging it */
-		virtio_mem_set_fake_offline(start_pfn, nr_pages, true);
-		adjust_managed_page_count(pfn_to_page(start_pfn), -nr_pages);
-
-		/* Try to unplug the allocated memory */
-		rc = virtio_mem_mb_unplug_sb(vm, mb_id, sb_id, 1);
-		if (rc) {
-			/* Return the memory to the buddy. */
-			virtio_mem_fake_online(start_pfn, nr_pages);
+		else if (rc)
 			return rc;
-		}
-
-		virtio_mem_mb_set_state(vm, mb_id,
-					VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL);
 		*nb_sb -= 1;
 	}
 
+unplugged:
 	/*
 	 * Once all subblocks of a memory block were unplugged, offline and
 	 * remove it. This will usually not fail, as no memory is in use
-- 
2.25.3

Re: [PATCH v4 11/15] virtio-mem: Add parent resource for all added "System RAM"

[Pankaj Gupta]

> Let's add a parent resource, named after the virtio device (inspired by
> drivers/dax/kmem.c). This allows user space to identify which memory
> belongs to which virtio-mem device.
>
> With this change and two virtio-mem devices:
>         :/# cat /proc/iomem
>         00000000-00000fff : Reserved
>         00001000-0009fbff : System RAM
>         [...]
>         140000000-333ffffff : virtio0
>           140000000-147ffffff : System RAM
>           148000000-14fffffff : System RAM
>           150000000-157ffffff : System RAM
>         [...]
>         334000000-3033ffffff : virtio1
>           338000000-33fffffff : System RAM
>           340000000-347ffffff : System RAM
>           348000000-34fffffff : System RAM
>         [...]
>
> Cc: "Michael S. Tsirkin" <mst@redhat.com>
> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
> Signed-off-by: David Hildenbrand <david@redhat.com>
> ---
>  drivers/virtio/virtio_mem.c | 52 ++++++++++++++++++++++++++++++++++++-
>  1 file changed, 51 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
> index eb4c16d634e0..80cdb9e6b3c4 100644
> --- a/drivers/virtio/virtio_mem.c
> +++ b/drivers/virtio/virtio_mem.c
> @@ -99,6 +99,9 @@ struct virtio_mem {
>         /* Id of the next memory bock to prepare when needed. */
>         unsigned long next_mb_id;
>
> +       /* The parent resource for all memory added via this device. */
> +       struct resource *parent_resource;
> +
>         /* Summary of all memory block states. */
>         unsigned long nb_mb_state[VIRTIO_MEM_MB_STATE_COUNT];
>  #define VIRTIO_MEM_NB_OFFLINE_THRESHOLD                10
> @@ -1741,6 +1744,44 @@ static int virtio_mem_init(struct virtio_mem *vm)
>         return 0;
>  }
>
> +static int virtio_mem_create_resource(struct virtio_mem *vm)
> +{
> +       /*
> +        * When force-unloading the driver and removing the device, we
> +        * could have a garbage pointer. Duplicate the string.
> +        */
> +       const char *name = kstrdup(dev_name(&vm->vdev->dev), GFP_KERNEL);
> +
> +       if (!name)
> +               return -ENOMEM;
> +
> +       vm->parent_resource = __request_mem_region(vm->addr, vm->region_size,
> +                                                  name, IORESOURCE_SYSTEM_RAM);
> +       if (!vm->parent_resource) {
> +               kfree(name);
> +               dev_warn(&vm->vdev->dev, "could not reserve device region\n");
> +               return -EBUSY;
> +       }
> +
> +       /* The memory is not actually busy - make add_memory() work. */
> +       vm->parent_resource->flags &= ~IORESOURCE_BUSY;
> +       return 0;
> +}
> +
> +static void virtio_mem_delete_resource(struct virtio_mem *vm)
> +{
> +       const char *name;
> +
> +       if (!vm->parent_resource)
> +               return;
> +
> +       name = vm->parent_resource->name;
> +       release_resource(vm->parent_resource);
> +       kfree(vm->parent_resource);
> +       kfree(name);
> +       vm->parent_resource = NULL;
> +}
> +
>  static int virtio_mem_probe(struct virtio_device *vdev)
>  {
>         struct virtio_mem *vm;
> @@ -1770,11 +1811,16 @@ static int virtio_mem_probe(struct virtio_device *vdev)
>         if (rc)
>                 goto out_del_vq;
>
> +       /* create the parent resource for all memory */
> +       rc = virtio_mem_create_resource(vm);
> +       if (rc)
> +               goto out_del_vq;
> +
>         /* register callbacks */
>         vm->memory_notifier.notifier_call = virtio_mem_memory_notifier_cb;
>         rc = register_memory_notifier(&vm->memory_notifier);
>         if (rc)
> -               goto out_del_vq;
> +               goto out_del_resource;
>         rc = register_virtio_mem_device(vm);
>         if (rc)
>                 goto out_unreg_mem;
> @@ -1788,6 +1834,8 @@ static int virtio_mem_probe(struct virtio_device *vdev)
>         return 0;
>  out_unreg_mem:
>         unregister_memory_notifier(&vm->memory_notifier);
> +out_del_resource:
> +       virtio_mem_delete_resource(vm);
>  out_del_vq:
>         vdev->config->del_vqs(vdev);
>  out_free_vm:
> @@ -1848,6 +1896,8 @@ static void virtio_mem_remove(struct virtio_device *vdev)
>             vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE_PARTIAL] ||
>             vm->nb_mb_state[VIRTIO_MEM_MB_STATE_ONLINE_MOVABLE])
>                 dev_warn(&vdev->dev, "device still has system memory added\n");
> +       else
> +               virtio_mem_delete_resource(vm);
>
>         /* remove all tracking data - no locking needed */
>         vfree(vm->mb_state);
> --

Reviewed-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>

[PATCH v4 16/15] virtio-mem: Don't rely on implicit compiler padding for requests

[David Hildenbrand]

The compiler will add padding after the last member, make that explicit.
The size of a request is always 24 bytes. The size of a response always
10 bytes. Add compile-time checks.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
---

Something I noticed while working on the spec (which proves that writing a
virtio-spec makes sense :) ).

---
 drivers/virtio/virtio_mem.c     | 3 +++
 include/uapi/linux/virtio_mem.h | 3 +++
 2 files changed, 6 insertions(+)

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index 9e523db3bee1..f658fe9149be 100644
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -1770,6 +1770,9 @@ static int virtio_mem_probe(struct virtio_device *vdev)
 	struct virtio_mem *vm;
 	int rc = -EINVAL;
 
+	BUILD_BUG_ON(sizeof(struct virtio_mem_req) != 24);
+	BUILD_BUG_ON(sizeof(struct virtio_mem_resp) != 10);
+
 	vdev->priv = vm = kzalloc(sizeof(*vm), GFP_KERNEL);
 	if (!vm)
 		return -ENOMEM;
diff --git a/include/uapi/linux/virtio_mem.h b/include/uapi/linux/virtio_mem.h
index e0a9dc7397c3..a455c488a995 100644
--- a/include/uapi/linux/virtio_mem.h
+++ b/include/uapi/linux/virtio_mem.h
@@ -103,16 +103,19 @@
 struct virtio_mem_req_plug {
 	__virtio64 addr;
 	__virtio16 nb_blocks;
+	__virtio16 padding[3];
 };
 
 struct virtio_mem_req_unplug {
 	__virtio64 addr;
 	__virtio16 nb_blocks;
+	__virtio16 padding[3];
 };
 
 struct virtio_mem_req_state {
 	__virtio64 addr;
 	__virtio16 nb_blocks;
+	__virtio16 padding[3];
 };
 
 struct virtio_mem_req {
-- 
2.25.4

Re: [PATCH v4 00/15] virtio-mem: paravirtualized memory

[teawater]

Hi David,

Thanks for your work.
I tried this version with cloud-hypervisor master.  It worked very well.

Best,
Hui

> 2020年5月7日 22:01，David Hildenbrand <david@redhat.com> 写道：
> 
> This series is based on v5.7-rc4. The patches are located at:
>    https://github.com/davidhildenbrand/linux.git virtio-mem-v4
> 
> This is basically a resend of v3 [1], now based on v5.7-rc4 and restested.
> One patch was reshuffled and two ACKs I missed to add were added. The
> rebase did not require any modifications to patches.
> 
> Details about virtio-mem can be found in the cover letter of v2 [2]. A
> basic QEMU implementation was posted yesterday [3].
> 
> [1] https://lkml.kernel.org/r/20200507103119.11219-1-david@redhat.com
> [2] https://lkml.kernel.org/r/20200311171422.10484-1-david@redhat.com
> [3] https://lkml.kernel.org/r/20200506094948.76388-1-david@redhat.com
> 
> v3 -> v4:
> - Move "MAINTAINERS: Add myself as virtio-mem maintainer" to #2
> - Add two ACKs from Andrew (in reply to v2)
> -- "mm: Allow to offline unmovable PageOffline() pages via ..."
> -- "mm/memory_hotplug: Introduce offline_and_remove_memory()"
> 
> v2 -> v3:
> - "virtio-mem: Paravirtualized memory hotplug"
> -- Include "linux/slab.h" to fix build issues
> -- Remember the "region_size", helpful for patch #11
> -- Minor simplifaction in virtio_mem_overlaps_range()
> -- Use notifier_from_errno() instead of notifier_to_errno() in notifier
> -- More reliable check for added memory when unloading the driver
> - "virtio-mem: Allow to specify an ACPI PXM as nid"
> -- Also print the nid
> - Added patch #11-#15
> 
> David Hildenbrand (15):
>  virtio-mem: Paravirtualized memory hotplug
>  MAINTAINERS: Add myself as virtio-mem maintainer
>  virtio-mem: Allow to specify an ACPI PXM as nid
>  virtio-mem: Paravirtualized memory hotunplug part 1
>  virtio-mem: Paravirtualized memory hotunplug part 2
>  mm: Allow to offline unmovable PageOffline() pages via
>    MEM_GOING_OFFLINE
>  virtio-mem: Allow to offline partially unplugged memory blocks
>  mm/memory_hotplug: Introduce offline_and_remove_memory()
>  virtio-mem: Offline and remove completely unplugged memory blocks
>  virtio-mem: Better retry handling
>  virtio-mem: Add parent resource for all added "System RAM"
>  virtio-mem: Drop manual check for already present memory
>  virtio-mem: Unplug subblocks right-to-left
>  virtio-mem: Use -ETXTBSY as error code if the device is busy
>  virtio-mem: Try to unplug the complete online memory block first
> 
> MAINTAINERS                     |    7 +
> drivers/acpi/numa/srat.c        |    1 +
> drivers/virtio/Kconfig          |   17 +
> drivers/virtio/Makefile         |    1 +
> drivers/virtio/virtio_mem.c     | 1962 +++++++++++++++++++++++++++++++
> include/linux/memory_hotplug.h  |    1 +
> include/linux/page-flags.h      |   10 +
> include/uapi/linux/virtio_ids.h |    1 +
> include/uapi/linux/virtio_mem.h |  208 ++++
> mm/memory_hotplug.c             |   81 +-
> mm/page_alloc.c                 |   26 +
> mm/page_isolation.c             |    9 +
> 12 files changed, 2314 insertions(+), 10 deletions(-)
> create mode 100644 drivers/virtio/virtio_mem.c
> create mode 100644 include/uapi/linux/virtio_mem.h
> 
> -- 
> 2.25.3

Re: [PATCH v4 00/15] virtio-mem: paravirtualized memory

[David Hildenbrand]

On 20.05.20 07:25, teawater wrote:
> Hi David,
> 
> Thanks for your work.
> I tried this version with cloud-hypervisor master.  It worked very well.
> 
> Best,
> Hui

Hi Hui,

thanks for testing!

Cheers!

-- 
Thanks,

David / dhildenb

Re: [PATCH v4 00/15] virtio-mem: paravirtualized memory

[David Hildenbrand]

On 07.05.20 16:01, David Hildenbrand wrote:
> This series is based on v5.7-rc4. The patches are located at:
>     https://github.com/davidhildenbrand/linux.git virtio-mem-v4
> 
> This is basically a resend of v3 [1], now based on v5.7-rc4 and restested.
> One patch was reshuffled and two ACKs I missed to add were added. The
> rebase did not require any modifications to patches.
> 
> Details about virtio-mem can be found in the cover letter of v2 [2]. A
> basic QEMU implementation was posted yesterday [3].
> 
> [1] https://lkml.kernel.org/r/20200507103119.11219-1-david@redhat.com
> [2] https://lkml.kernel.org/r/20200311171422.10484-1-david@redhat.com
> [3] https://lkml.kernel.org/r/20200506094948.76388-1-david@redhat.com
> 
> v3 -> v4:
> - Move "MAINTAINERS: Add myself as virtio-mem maintainer" to #2
> - Add two ACKs from Andrew (in reply to v2)
> -- "mm: Allow to offline unmovable PageOffline() pages via ..."
> -- "mm/memory_hotplug: Introduce offline_and_remove_memory()"
> 
> v2 -> v3:
> - "virtio-mem: Paravirtualized memory hotplug"
> -- Include "linux/slab.h" to fix build issues
> -- Remember the "region_size", helpful for patch #11
> -- Minor simplifaction in virtio_mem_overlaps_range()
> -- Use notifier_from_errno() instead of notifier_to_errno() in notifier
> -- More reliable check for added memory when unloading the driver
> - "virtio-mem: Allow to specify an ACPI PXM as nid"
> -- Also print the nid
> - Added patch #11-#15

@MST ping, v5.7 has been released

-- 
Thanks,

David / dhildenb