[PATCH v10 0/8] Allocate memmap from hotadded memory (per device)

[PATCH v10 0/8] Allocate memmap from hotadded memory (per device)

[Oscar Salvador]

Hi Andrew,

Please, consider pulling out the patch that sits currently in linux-mm and
put this one instead.
This patchset collects some Acked-by from Michal and rewords some changelogs/
comments.

Thanks

changes from v9 -> v10:
 - Reworded some changelogs and comments
 - Unify loop in online_pages_range()
 - Collect Michal's Acked-by

Changes from v8 -> v9:
 - Change order of kasan calls and offline_mem_sections
 - Collect Reviewed-by

Changes from v7 -> v8:
 - Addressed feedback from David for patch#4

Changes from v6 -> v7:
 - Fix check from "mm,memory_hotplug: Relax fully spanned sections check"
 - Add fixup from "mm,memory_hotplug: Allocate memmap from the added memory range"
 - Add Reviewed-by from David for patch#2
 - Fix changelog in "mm,memory_hotplug: Factor out adjusting present pages into
    adjust_present_page_count()"

Changes from v5 -> v6:
 - Create memory_block_{online,offline} functions
 - Create vmemmap_* functions to deal with vmemmap stuff, so
   {online,offline}_pages remain untouched
 - Add adjust_present_page_count's patch from David
 - Relax check in {offline,online}_pages
 - Rework changelogs

Changes from v4 -> v5:
 - Addressed feedback from David (patch#1)
 - Tested on x86_64 with different struct page sizes and on large/small memory
   blocks
 - Tested on arm64 with 4K, 64K (with and without THP) and with different struct
   page sizes

 NOTE: We might need to make this feature and hugetlb-vmemmap feature [1] mutually
       exclusive. I raised an issue I see in [2].
       Hugetlb-vmemmap feature has been withdrawn for the time being due to the need
       in further changes  wrt. locking/freeing context.
       I will keep an eye, and when the time comes again I will see how the two
       features play together and how one another can be disabled when needed.

Changes from v3 -> v4:
 - Addressed feedback from David
 - Wrap memmap_on_memory module thingy with #ifdef
   on MHP_MEMMAP_ON_MEMORY
 - Move "depend on MEMORY_HOTPLUG" to MHP_MEMMAP_ON_MEMORY
   in generic mm/Kconfig
 - Collect David's Reviewed-bys

Changes from v2 -> v3:
 - Addressed feedback from David
 - Squash former patch#4 and and patch#5 into patch#1
 - Fix config dependency CONFIR_SPARSE_VMEMMAP vs CONFIG_SPARSE_VMEMMAP_ENABLE
 - Simplify module parameter functions

Changes from v1 -> v2
 - Addressed feedback from David
 - Fence off the feature in case struct page size is not
   multiple of PMD size or pageblock alignment cannot be guaranted
 - Tested on x86_64 small and large memory_blocks
 - Tested on arm64 4KB and 64KB page sizes (for some reason I cannot boot
   my VM with 16KB page size).

 Arm64 with 4KB page size behaves like x86_64 after [1], which made section
 size smaller.
 With 64KB, the feature gets fenced off due to pageblock alignment.

Changes from RFCv3 -> v1:
 - Addressed feedback from David
 - Re-order patches

Changes from v2 -> v3 (RFC):
 - Re-order patches (Michal)
 - Fold "mm,memory_hotplug: Introduce MHP_MEMMAP_ON_MEMORY" in patch#1
 - Add kernel boot option to enable this feature (Michal)

Changes from v1 -> v2 (RFC):
 - Addressed feedback provided by David
 - Add a arch_support_memmap_on_memory to be called
   from mhp_supports_memmap_on_memory, as atm,
   only ARM, powerpc and x86_64 have altmat support.

[1] https://lore.kernel.org/lkml/cover.1611206601.git.sudaraj...

Original cover letter:

 The primary goal of this patchset is to reduce memory overhead of the
 hot-added memory (at least for SPARSEMEM_VMEMMAP memory model).
 The current way we use to populate memmap (struct page array) has two main drawbacks:

 a) it consumes an additional memory until the hotadded memory itself is
    onlined and
 b) memmap might end up on a different numa node which is especially true
    for movable_node configuration.
 c) due to fragmentation we might end up populating memmap with base
    pages

 One way to mitigate all these issues is to simply allocate memmap array
 (which is the largest memory footprint of the physical memory hotplug)
 from the hot-added memory itself. SPARSEMEM_VMEMMAP memory model allows
 us to map any pfn range so the memory doesn't need to be online to be
 usable for the array. See patch 4 for more details.
 This feature is only usable when CONFIG_SPARSEMEM_VMEMMAP is set.

 [Overall design]:

 Implementation wise we reuse vmem_altmap infrastructure to override
 the default allocator used by vmemap_populate.
 memory_block structure gains a new field called nr_vmemmap_pages,
 which accounts for the number of vmemmap pages used by that memory_block.
 E.g: On x86_64, that is 512 vmemmap pages on small memory bloks and 4096
      on large memory blocks (1GB)

 We also introduce new two functions: memory_block_{online,offline}.
 These functions take care of initializing/unitializing vmemmap pages
 prior to calling {online,offline}_pages, so the latter functions can
 remain totally untouched.

 More details can be found in the respective changelogs.


David Hildenbrand (1):
  mm,memory_hotplug: Factor out adjusting present pages into
    adjust_present_page_count()

Oscar Salvador (7):
  drivers/base/memory: Introduce memory_block_{online,offline}
  mm,memory_hotplug: Relax fully spanned sections check
  mm,memory_hotplug: Allocate memmap from the added memory range
  acpi,memhotplug: Enable MHP_MEMMAP_ON_MEMORY when supported
  mm,memory_hotplug: Add kernel boot option to enable memmap_on_memory
  x86/Kconfig: Introduce ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
  arm64/Kconfig: Introduce ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE

 Documentation/admin-guide/kernel-parameters.txt |  17 ++
 arch/arm64/Kconfig                              |   3 +
 arch/x86/Kconfig                                |   3 +
 drivers/acpi/acpi_memhotplug.c                  |   5 +-
 drivers/base/memory.c                           | 101 ++++++++++--
 include/linux/memory.h                          |   8 +-
 include/linux/memory_hotplug.h                  |  15 +-
 include/linux/memremap.h                        |   2 +-
 include/linux/mmzone.h                          |   7 +-
 mm/Kconfig                                      |   5 +
 mm/Makefile                                     |   5 +-
 mm/memory_hotplug.c                             | 211 +++++++++++++++++++++---
 mm/sparse.c                                     |   2 -
 13 files changed, 337 insertions(+), 47 deletions(-)

-- 
2.16.3

[PATCH v10 1/8] drivers/base/memory: Introduce memory_block_{online,offline}

[Oscar Salvador]

This is a preparatory patch that introduces two new functions:
memory_block_online() and memory_block_offline().

For now, these functions will only call online_pages() and offline_pages()
respectively, but they will be later in charge of preparing the vmemmap
pages, carrying out the initialization and proper accounting of such
pages.

Since memory_block struct contains all the information, pass this struct
down the chain till the end functions.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
---
 drivers/base/memory.c | 33 +++++++++++++++++++++------------
 1 file changed, 21 insertions(+), 12 deletions(-)

diff --git a/drivers/base/memory.c b/drivers/base/memory.c
index f35298425575..f209925a5d4e 100644
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -169,30 +169,41 @@ int memory_notify(unsigned long val, void *v)
 	return blocking_notifier_call_chain(&memory_chain, val, v);
 }
 
+static int memory_block_online(struct memory_block *mem)
+{
+	unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
+	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+
+	return online_pages(start_pfn, nr_pages, mem->online_type, mem->nid);
+}
+
+static int memory_block_offline(struct memory_block *mem)
+{
+	unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
+	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+
+	return offline_pages(start_pfn, nr_pages);
+}
+
 /*
  * MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
  * OK to have direct references to sparsemem variables in here.
  */
 static int
-memory_block_action(unsigned long start_section_nr, unsigned long action,
-		    int online_type, int nid)
+memory_block_action(struct memory_block *mem, unsigned long action)
 {
-	unsigned long start_pfn;
-	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
 	int ret;
 
-	start_pfn = section_nr_to_pfn(start_section_nr);
-
 	switch (action) {
 	case MEM_ONLINE:
-		ret = online_pages(start_pfn, nr_pages, online_type, nid);
+		ret = memory_block_online(mem);
 		break;
 	case MEM_OFFLINE:
-		ret = offline_pages(start_pfn, nr_pages);
+		ret = memory_block_offline(mem);
 		break;
 	default:
 		WARN(1, KERN_WARNING "%s(%ld, %ld) unknown action: "
-		     "%ld\n", __func__, start_section_nr, action, action);
+		     "%ld\n", __func__, mem->start_section_nr, action, action);
 		ret = -EINVAL;
 	}
 
@@ -210,9 +221,7 @@ static int memory_block_change_state(struct memory_block *mem,
 	if (to_state == MEM_OFFLINE)
 		mem->state = MEM_GOING_OFFLINE;
 
-	ret = memory_block_action(mem->start_section_nr, to_state,
-				  mem->online_type, mem->nid);
-
+	ret = memory_block_action(mem, to_state);
 	mem->state = ret ? from_state_req : to_state;
 
 	return ret;
-- 
2.16.3

[PATCH v10 2/8] mm,memory_hotplug: Relax fully spanned sections check

[Oscar Salvador]

We want {online,offline}_pages to operate on whole memblocks, but
memmap_on_memory will poke pageblock_nr_pages aligned holes in the
beginning, which is a special case we want to allow.
Relax the check to account for that case.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
---
 mm/memory_hotplug.c | 22 ++++++++++++++++++----
 1 file changed, 18 insertions(+), 4 deletions(-)

diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 0cdbbfbc5757..e6aafd17a01a 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -838,9 +838,16 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	int ret;
 	struct memory_notify arg;
 
-	/* We can only online full sections (e.g., SECTION_IS_ONLINE) */
+	/*
+	 * {on,off}lining is constrained to full memory sections (or more
+	 * precisly to memory blocks from the user space POV).
+	 * memmap_on_memory is an exception because it reserves initial part
+	 * of the physical memory space for vmemmaps. That space is pageblock
+	 * aligned.
+	 */
 	if (WARN_ON_ONCE(!nr_pages ||
-			 !IS_ALIGNED(pfn | nr_pages, PAGES_PER_SECTION)))
+			 !IS_ALIGNED(pfn, pageblock_nr_pages) ||
+			 !IS_ALIGNED(pfn + nr_pages, PAGES_PER_SECTION)))
 		return -EINVAL;
 
 	mem_hotplug_begin();
@@ -1573,9 +1580,16 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages)
 	int ret, node;
 	char *reason;
 
-	/* We can only offline full sections (e.g., SECTION_IS_ONLINE) */
+	/*
+	 * {on,off}lining is constrained to full memory sections (or more
+	 * precisly to memory blocks from the user space POV).
+	 * memmap_on_memory is an exception because it reserves initial part
+	 * of the physical memory space for vmemmaps. That space is pageblock
+	 * aligned.
+	 */
 	if (WARN_ON_ONCE(!nr_pages ||
-			 !IS_ALIGNED(start_pfn | nr_pages, PAGES_PER_SECTION)))
+			 !IS_ALIGNED(start_pfn, pageblock_nr_pages) ||
+			 !IS_ALIGNED(start_pfn + nr_pages, PAGES_PER_SECTION)))
 		return -EINVAL;
 
 	mem_hotplug_begin();
-- 
2.16.3

[PATCH v10 3/8] mm,memory_hotplug: Factor out adjusting present pages into adjust_present_page_count()

[Oscar Salvador]

From: David Hildenbrand <david@redhat.com>

Let's have a single place (inspired by adjust_managed_page_count()) where
we adjust present pages.
In contrast to adjust_managed_page_count(), only memory onlining/offlining
is allowed to modify the number of present pages.

Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
---
 mm/memory_hotplug.c | 22 ++++++++++++----------
 1 file changed, 12 insertions(+), 10 deletions(-)

diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index e6aafd17a01a..b954fd10474e 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -829,6 +829,16 @@ struct zone * zone_for_pfn_range(int online_type, int nid, unsigned start_pfn,
 	return default_zone_for_pfn(nid, start_pfn, nr_pages);
 }
 
+static void adjust_present_page_count(struct zone *zone, long nr_pages)
+{
+	unsigned long flags;
+
+	zone->present_pages += nr_pages;
+	pgdat_resize_lock(zone->zone_pgdat, &flags);
+	zone->zone_pgdat->node_present_pages += nr_pages;
+	pgdat_resize_unlock(zone->zone_pgdat, &flags);
+}
+
 int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 		       int online_type, int nid)
 {
@@ -884,11 +894,7 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	}
 
 	online_pages_range(pfn, nr_pages);
-	zone->present_pages += nr_pages;
-
-	pgdat_resize_lock(zone->zone_pgdat, &flags);
-	zone->zone_pgdat->node_present_pages += nr_pages;
-	pgdat_resize_unlock(zone->zone_pgdat, &flags);
+	adjust_present_page_count(zone, nr_pages);
 
 	node_states_set_node(nid, &arg);
 	if (need_zonelists_rebuild)
@@ -1705,11 +1711,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages)
 
 	/* removal success */
 	adjust_managed_page_count(pfn_to_page(start_pfn), -nr_pages);
-	zone->present_pages -= nr_pages;
-
-	pgdat_resize_lock(zone->zone_pgdat, &flags);
-	zone->zone_pgdat->node_present_pages -= nr_pages;
-	pgdat_resize_unlock(zone->zone_pgdat, &flags);
+	adjust_present_page_count(zone, -nr_pages);
 
 	init_per_zone_wmark_min();
 
-- 
2.16.3

[PATCH v10 4/8] mm,memory_hotplug: Allocate memmap from the added memory range

[Oscar Salvador]

Physical memory hotadd has to allocate a memmap (struct page array) for
the newly added memory section. Currently, alloc_pages_node() is used
for those allocations.

This has some disadvantages:
 a) an existing memory is consumed for that purpose
    (eg: ~2MB per 128MB memory section on x86_64)
    This can even lead to extreme cases where system goes OOM because
    the physically hotplugged memory depletes the available memory before
    it is onlined.
 b) if the whole node is movable then we have off-node struct pages
    which has performance drawbacks.
 c) It might be there are no PMD_ALIGNED chunks so memmap array gets
    populated with base pages.

This can be improved when CONFIG_SPARSEMEM_VMEMMAP is enabled.

Vmemap page tables can map arbitrary memory. That means that we can
reserve a part of the physically hotadded memory to back vmemmap page
tables. This implementation uses the beginning of the hotplugged memory
for that purpose.

There are some non-obviously things to consider though.  Vmemmap
pages are allocated/freed during the memory hotplug events
(add_memory_resource(), try_remove_memory()) when the memory is
added/removed. This means that the reserved physical range is not online
although it is used. The most obvious side effect is that pfn_to_online_page()
returns NULL for those pfns. The current design expects that this
should be OK as the hotplugged memory is considered a garbage until it
is onlined. For example hibernation wouldn't save the content of those
vmmemmaps into the image so it wouldn't be restored on resume but this
should be OK as there no real content to recover anyway while metadata
is reachable from other data structures (e.g. vmemmap page tables).

The reserved space is therefore (de)initialized during the {on,off}line
events (mhp_{de}init_memmap_on_memory). That is done by extracting page
allocator independent initialization from the regular onlining path.
The primary reason to handle the reserved space outside of {on,off}line_pages
is to make each initialization specific to the purpose rather than
special case them in a single function.

As per above, the functions that are introduced are:

 - mhp_init_memmap_on_memory:
		       Initializes vmemmap pages by calling move_pfn_range_to_zone(),
		       calls kasan_add_zero_shadow(), and onlines as many sections
		       as vmemmap pages fully span.
 - mhp_deinit_memmap_on_memory:
		       Offlines as many sections as vmemmap pages fully span,
		       removes the range from zhe zone by remove_pfn_range_from_zone(),
		       and calls kasan_remove_zero_shadow() for the range.

The new function memory_block_online() calls mhp_init_memmap_on_memory() before
doing the actual online_pages(). Should online_pages() fail, we clean up
by calling mhp_deinit_memmap_on_memory().
Adjusting of present_pages is done at the end once we know that online_pages()
succedeed.

On offline, memory_block_offline() needs to unaccount vmemmap pages from
present_pages() before calling offline_pages().
This is necessary because offline_pages() tears down some structures based
on the fact whether the node or the zone become empty.
If offline_pages() fails, we account back vmemmap pages.
If it succeeds, we call mhp_deinit_memmap_on_memory().

Hot-remove:

 We need to be careful when removing memory, as adding and
 removing memory needs to be done with the same granularity.
 To check that this assumption is not violated, we check the
 memory range we want to remove and if a) any memory block has
 vmemmap pages and b) the range spans more than a single memory
 block, we scream out loud and refuse to proceed.

 If all is good and the range was using memmap on memory (aka vmemmap pages),
 we construct an altmap structure so free_hugepage_table does the right
 thing and calls vmem_altmap_free instead of free_pagetable.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
---
 drivers/base/memory.c          |  72 ++++++++++++++++--
 include/linux/memory.h         |   8 +-
 include/linux/memory_hotplug.h |  15 +++-
 include/linux/memremap.h       |   2 +-
 include/linux/mmzone.h         |   7 +-
 mm/Kconfig                     |   5 ++
 mm/memory_hotplug.c            | 161 ++++++++++++++++++++++++++++++++++++++---
 mm/sparse.c                    |   2 -
 8 files changed, 250 insertions(+), 22 deletions(-)

diff --git a/drivers/base/memory.c b/drivers/base/memory.c
index f209925a5d4e..b31b3af5c490 100644
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -173,16 +173,73 @@ static int memory_block_online(struct memory_block *mem)
 {
 	unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
 	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+	unsigned long nr_vmemmap_pages = mem->nr_vmemmap_pages;
+	struct zone *zone;
+	int ret;
+
+	zone = zone_for_pfn_range(mem->online_type, mem->nid, start_pfn, nr_pages);
+
+	/*
+	 * Although vmemmap pages have a different lifecycle than the pages
+	 * they describe (they remain until the memory is unplugged), doing
+	 * their initialization and accounting at memory onlining/offlining
+	 * stage helps to keep accounting easier to follow - e.g vmemmaps
+	 * belong to the same zone as the memory they backed.
+	 */
+	if (nr_vmemmap_pages) {
+		ret = mhp_init_memmap_on_memory(start_pfn, nr_vmemmap_pages, zone);
+		if (ret)
+			return ret;
+	}
+
+	ret = online_pages(start_pfn + nr_vmemmap_pages,
+			   nr_pages - nr_vmemmap_pages, zone);
+	if (ret) {
+		if (nr_vmemmap_pages)
+			mhp_deinit_memmap_on_memory(start_pfn, nr_vmemmap_pages);
+		return ret;
+	}
+
+	/*
+	 * Account once onlining succeeded. If the zone was unpopulated, it is
+	 * now already properly populated.
+	 */
+	if (nr_vmemmap_pages)
+		adjust_present_page_count(zone, nr_vmemmap_pages);
 
-	return online_pages(start_pfn, nr_pages, mem->online_type, mem->nid);
+	return ret;
 }
 
 static int memory_block_offline(struct memory_block *mem)
 {
 	unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
 	unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+	unsigned long nr_vmemmap_pages = mem->nr_vmemmap_pages;
+	struct zone *zone;
+	int ret;
+
+	zone = page_zone(pfn_to_page(start_pfn));
+
+	/*
+	 * Unaccount before offlining, such that unpopulated zone and kthreads
+	 * can properly be torn down in offline_pages().
+	 */
+	if (nr_vmemmap_pages)
+		adjust_present_page_count(zone, -nr_vmemmap_pages);
 
-	return offline_pages(start_pfn, nr_pages);
+	ret = offline_pages(start_pfn + nr_vmemmap_pages,
+			    nr_pages - nr_vmemmap_pages);
+	if (ret) {
+		/* offline_pages() failed. Account back. */
+		if (nr_vmemmap_pages)
+			adjust_present_page_count(zone, nr_vmemmap_pages);
+		return ret;
+	}
+
+	if (nr_vmemmap_pages)
+		mhp_deinit_memmap_on_memory(start_pfn, nr_vmemmap_pages);
+
+	return ret;
 }
 
 /*
@@ -576,7 +633,8 @@ int register_memory(struct memory_block *memory)
 	return ret;
 }
 
-static int init_memory_block(unsigned long block_id, unsigned long state)
+static int init_memory_block(unsigned long block_id, unsigned long state,
+			     unsigned long nr_vmemmap_pages)
 {
 	struct memory_block *mem;
 	int ret = 0;
@@ -593,6 +651,7 @@ static int init_memory_block(unsigned long block_id, unsigned long state)
 	mem->start_section_nr = block_id * sections_per_block;
 	mem->state = state;
 	mem->nid = NUMA_NO_NODE;
+	mem->nr_vmemmap_pages = nr_vmemmap_pages;
 
 	ret = register_memory(mem);
 
@@ -612,7 +671,7 @@ static int add_memory_block(unsigned long base_section_nr)
 	if (section_count == 0)
 		return 0;
 	return init_memory_block(memory_block_id(base_section_nr),
-				 MEM_ONLINE);
+				 MEM_ONLINE, 0);
 }
 
 static void unregister_memory(struct memory_block *memory)
@@ -634,7 +693,8 @@ static void unregister_memory(struct memory_block *memory)
  *
  * Called under device_hotplug_lock.
  */
-int create_memory_block_devices(unsigned long start, unsigned long size)
+int create_memory_block_devices(unsigned long start, unsigned long size,
+				unsigned long vmemmap_pages)
 {
 	const unsigned long start_block_id = pfn_to_block_id(PFN_DOWN(start));
 	unsigned long end_block_id = pfn_to_block_id(PFN_DOWN(start + size));
@@ -647,7 +707,7 @@ int create_memory_block_devices(unsigned long start, unsigned long size)
 		return -EINVAL;
 
 	for (block_id = start_block_id; block_id != end_block_id; block_id++) {
-		ret = init_memory_block(block_id, MEM_OFFLINE);
+		ret = init_memory_block(block_id, MEM_OFFLINE, vmemmap_pages);
 		if (ret)
 			break;
 	}
diff --git a/include/linux/memory.h b/include/linux/memory.h
index 4da95e684e20..97e92e8b556a 100644
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -29,6 +29,11 @@ struct memory_block {
 	int online_type;		/* for passing data to online routine */
 	int nid;			/* NID for this memory block */
 	struct device dev;
+	/*
+	 * Number of vmemmap pages. These pages
+	 * lay at the beginning of the memory block.
+	 */
+	unsigned long nr_vmemmap_pages;
 };
 
 int arch_get_memory_phys_device(unsigned long start_pfn);
@@ -80,7 +85,8 @@ static inline int memory_notify(unsigned long val, void *v)
 #else
 extern int register_memory_notifier(struct notifier_block *nb);
 extern void unregister_memory_notifier(struct notifier_block *nb);
-int create_memory_block_devices(unsigned long start, unsigned long size);
+int create_memory_block_devices(unsigned long start, unsigned long size,
+				unsigned long vmemmap_pages);
 void remove_memory_block_devices(unsigned long start, unsigned long size);
 extern void memory_dev_init(void);
 extern int memory_notify(unsigned long val, void *v);
diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
index 7288aa5ef73b..28f32fd00fe9 100644
--- a/include/linux/memory_hotplug.h
+++ b/include/linux/memory_hotplug.h
@@ -55,6 +55,14 @@ typedef int __bitwise mhp_t;
  */
 #define MHP_MERGE_RESOURCE	((__force mhp_t)BIT(0))
 
+/*
+ * We want memmap (struct page array) to be self contained.
+ * To do so, we will use the beginning of the hot-added range to build
+ * the page tables for the memmap array that describes the entire range.
+ * Only selected architectures support it with SPARSE_VMEMMAP.
+ */
+#define MHP_MEMMAP_ON_MEMORY   ((__force mhp_t)BIT(1))
+
 /*
  * Extended parameters for memory hotplug:
  * altmap: alternative allocator for memmap array (optional)
@@ -99,9 +107,13 @@ static inline void zone_seqlock_init(struct zone *zone)
 extern int zone_grow_free_lists(struct zone *zone, unsigned long new_nr_pages);
 extern int zone_grow_waitqueues(struct zone *zone, unsigned long nr_pages);
 extern int add_one_highpage(struct page *page, int pfn, int bad_ppro);
+extern void adjust_present_page_count(struct zone *zone, long nr_pages);
 /* VM interface that may be used by firmware interface */
+extern int mhp_init_memmap_on_memory(unsigned long pfn, unsigned long nr_pages,
+				     struct zone *zone);
+extern void mhp_deinit_memmap_on_memory(unsigned long pfn, unsigned long nr_pages);
 extern int online_pages(unsigned long pfn, unsigned long nr_pages,
-			int online_type, int nid);
+			struct zone *zone);
 extern struct zone *test_pages_in_a_zone(unsigned long start_pfn,
 					 unsigned long end_pfn);
 extern void __offline_isolated_pages(unsigned long start_pfn,
@@ -359,6 +371,7 @@ extern struct zone *zone_for_pfn_range(int online_type, int nid, unsigned start_
 extern int arch_create_linear_mapping(int nid, u64 start, u64 size,
 				      struct mhp_params *params);
 void arch_remove_linear_mapping(u64 start, u64 size);
+extern bool mhp_supports_memmap_on_memory(unsigned long size);
 #endif /* CONFIG_MEMORY_HOTPLUG */
 
 #endif /* __LINUX_MEMORY_HOTPLUG_H */
diff --git a/include/linux/memremap.h b/include/linux/memremap.h
index f5b464daeeca..45a79da89c5f 100644
--- a/include/linux/memremap.h
+++ b/include/linux/memremap.h
@@ -17,7 +17,7 @@ struct device;
  * @alloc: track pages consumed, private to vmemmap_populate()
  */
 struct vmem_altmap {
-	const unsigned long base_pfn;
+	unsigned long base_pfn;
 	const unsigned long end_pfn;
 	const unsigned long reserve;
 	unsigned long free;
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 47946cec7584..76f4ca5ed230 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -427,6 +427,11 @@ enum zone_type {
 	 *    techniques might use alloc_contig_range() to hide previously
 	 *    exposed pages from the buddy again (e.g., to implement some sort
 	 *    of memory unplug in virtio-mem).
+	 * 6. Memory-hotplug: when using memmap_on_memory and onlining the memory
+	 *    to the MOVABLE zone, the vmemmap pages are also placed in such
+	 *    zone. Such pages cannot be really moved around as they are
+	 *    self-stored in the range, but they are treated as movable when
+	 *    the range they describe is about to be offlined.
 	 *
 	 * In general, no unmovable allocations that degrade memory offlining
 	 * should end up in ZONE_MOVABLE. Allocators (like alloc_contig_range())
@@ -1378,10 +1383,8 @@ static inline int online_section_nr(unsigned long nr)
 
 #ifdef CONFIG_MEMORY_HOTPLUG
 void online_mem_sections(unsigned long start_pfn, unsigned long end_pfn);
-#ifdef CONFIG_MEMORY_HOTREMOVE
 void offline_mem_sections(unsigned long start_pfn, unsigned long end_pfn);
 #endif
-#endif
 
 static inline struct mem_section *__pfn_to_section(unsigned long pfn)
 {
diff --git a/mm/Kconfig b/mm/Kconfig
index 24c045b24b95..febf805000f8 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -183,6 +183,11 @@ config MEMORY_HOTREMOVE
 	depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
 	depends on MIGRATION
 
+config MHP_MEMMAP_ON_MEMORY
+	def_bool y
+	depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP
+	depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
+
 # Heavily threaded applications may benefit from splitting the mm-wide
 # page_table_lock, so that faults on different parts of the user address
 # space can be handled with less contention: split it at this NR_CPUS.
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index b954fd10474e..d2de1fab75e2 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -42,6 +42,8 @@
 #include "internal.h"
 #include "shuffle.h"
 
+static bool memmap_on_memory;
+
 /*
  * online_page_callback contains pointer to current page onlining function.
  * Initially it is generic_online_page(). If it is required it could be
@@ -648,9 +650,16 @@ static void online_pages_range(unsigned long start_pfn, unsigned long nr_pages)
 	 * decide to not expose all pages to the buddy (e.g., expose them
 	 * later). We account all pages as being online and belonging to this
 	 * zone ("present").
+	 * When using memmap_on_memory, the range might not be aligned to
+	 * MAX_ORDER_NR_PAGES - 1, but pageblock aligned. __ffs() will detect
+	 * this and the first chunk to online will be pageblock_nr_pages.
 	 */
-	for (pfn = start_pfn; pfn < end_pfn; pfn += MAX_ORDER_NR_PAGES)
-		(*online_page_callback)(pfn_to_page(pfn), MAX_ORDER - 1);
+	for (pfn = start_pfn; pfn < end_pfn;) {
+		int order = min(MAX_ORDER - 1UL, __ffs(pfn));
+
+		(*online_page_callback)(pfn_to_page(pfn), order);
+		pfn += (1UL << order);
+	}
 
 	/* mark all involved sections as online */
 	online_mem_sections(start_pfn, end_pfn);
@@ -829,7 +838,11 @@ struct zone * zone_for_pfn_range(int online_type, int nid, unsigned start_pfn,
 	return default_zone_for_pfn(nid, start_pfn, nr_pages);
 }
 
-static void adjust_present_page_count(struct zone *zone, long nr_pages)
+/*
+ * This function should only be called by memory_block_{online,offline},
+ * and {online,offline}_pages.
+ */
+void adjust_present_page_count(struct zone *zone, long nr_pages)
 {
 	unsigned long flags;
 
@@ -839,12 +852,54 @@ static void adjust_present_page_count(struct zone *zone, long nr_pages)
 	pgdat_resize_unlock(zone->zone_pgdat, &flags);
 }
 
-int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
-		       int online_type, int nid)
+int mhp_init_memmap_on_memory(unsigned long pfn, unsigned long nr_pages,
+			      struct zone *zone)
+{
+	unsigned long end_pfn = pfn + nr_pages;
+	int ret;
+
+	ret = kasan_add_zero_shadow(__va(PFN_PHYS(pfn)), PFN_PHYS(nr_pages));
+	if (ret)
+		return ret;
+
+	move_pfn_range_to_zone(zone, pfn, nr_pages, NULL, MIGRATE_UNMOVABLE);
+
+	/*
+	 * It might be that the vmemmap_pages fully span sections. If that is
+	 * the case, mark those sections online here as otherwise they will be
+	 * left offline.
+	 */
+	if (nr_pages >= PAGES_PER_SECTION)
+	        online_mem_sections(pfn, ALIGN_DOWN(end_pfn, PAGES_PER_SECTION));
+
+	return ret;
+}
+
+void mhp_deinit_memmap_on_memory(unsigned long pfn, unsigned long nr_pages)
+{
+	unsigned long end_pfn = pfn + nr_pages;
+
+	/*
+	 * It might be that the vmemmap_pages fully span sections. If that is
+	 * the case, mark those sections offline here as otherwise they will be
+	 * left online.
+	 */
+	if (nr_pages >= PAGES_PER_SECTION)
+		offline_mem_sections(pfn, ALIGN_DOWN(end_pfn, PAGES_PER_SECTION));
+
+        /*
+	 * The pages associated with this vmemmap have been offlined, so
+	 * we can reset its state here.
+	 */
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn, nr_pages);
+	kasan_remove_zero_shadow(__va(PFN_PHYS(pfn)), PFN_PHYS(nr_pages));
+}
+
+int __ref online_pages(unsigned long pfn, unsigned long nr_pages, struct zone *zone)
 {
 	unsigned long flags;
-	struct zone *zone;
 	int need_zonelists_rebuild = 0;
+	const int nid = zone_to_nid(zone);
 	int ret;
 	struct memory_notify arg;
 
@@ -863,7 +918,6 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	mem_hotplug_begin();
 
 	/* associate pfn range with the zone */
-	zone = zone_for_pfn_range(online_type, nid, pfn, nr_pages);
 	move_pfn_range_to_zone(zone, pfn, nr_pages, NULL, MIGRATE_ISOLATE);
 
 	arg.start_pfn = pfn;
@@ -1077,6 +1131,45 @@ static int online_memory_block(struct memory_block *mem, void *arg)
 	return device_online(&mem->dev);
 }
 
+bool mhp_supports_memmap_on_memory(unsigned long size)
+{
+	unsigned long nr_vmemmap_pages = size / PAGE_SIZE;
+	unsigned long vmemmap_size = nr_vmemmap_pages * sizeof(struct page);
+	unsigned long remaining_size = size - vmemmap_size;
+
+	/*
+	 * Besides having arch support and the feature enabled at runtime, we
+	 * need a few more assumptions to hold true:
+	 *
+	 * a) We span a single memory block: memory onlining/offlinin;g happens
+	 *    in memory block granularity. We don't want the vmemmap of online
+	 *    memory blocks to reside on offline memory blocks. In the future,
+	 *    we might want to support variable-sized memory blocks to make the
+	 *    feature more versatile.
+	 *
+	 * b) The vmemmap pages span complete PMDs: We don't want vmemmap code
+	 *    to populate memory from the altmap for unrelated parts (i.e.,
+	 *    other memory blocks)
+	 *
+	 * c) The vmemmap pages (and thereby the pages that will be exposed to
+	 *    the buddy) have to cover full pageblocks: memory onlining/offlining
+	 *    code requires applicable ranges to be page-aligned, for example, to
+	 *    set the migratetypes properly.
+	 *
+	 * TODO: Although we have a check here to make sure that vmemmap pages
+	 *       fully populate a PMD, it is not the right place to check for
+	 *       this. A much better solution involves improving vmemmap code
+	 *       to fallback to base pages when trying to populate vmemmap using
+	 *       altmap as an alternative source of memory, and we do not exactly
+	 *       populate a single PMD.
+	 */
+	return memmap_on_memory &&
+	       IS_ENABLED(CONFIG_MHP_MEMMAP_ON_MEMORY) &&
+	       size == memory_block_size_bytes() &&
+	       IS_ALIGNED(vmemmap_size, PMD_SIZE) &&
+	       IS_ALIGNED(remaining_size, (pageblock_nr_pages << PAGE_SHIFT));
+}
+
 /*
  * NOTE: The caller must call lock_device_hotplug() to serialize hotplug
  * and online/offline operations (triggered e.g. by sysfs).
@@ -1086,6 +1179,7 @@ static int online_memory_block(struct memory_block *mem, void *arg)
 int __ref add_memory_resource(int nid, struct resource *res, mhp_t mhp_flags)
 {
 	struct mhp_params params = { .pgprot = pgprot_mhp(PAGE_KERNEL) };
+	struct vmem_altmap mhp_altmap = {};
 	u64 start, size;
 	bool new_node = false;
 	int ret;
@@ -1112,13 +1206,26 @@ int __ref add_memory_resource(int nid, struct resource *res, mhp_t mhp_flags)
 		goto error;
 	new_node = ret;
 
+	/*
+	 * Self hosted memmap array
+	 */
+	if (mhp_flags & MHP_MEMMAP_ON_MEMORY) {
+		if (!mhp_supports_memmap_on_memory(size)) {
+			ret = -EINVAL;
+			goto error;
+		}
+		mhp_altmap.free = PHYS_PFN(size);
+		mhp_altmap.base_pfn = PHYS_PFN(start);
+		params.altmap = &mhp_altmap;
+	}
+
 	/* call arch's memory hotadd */
 	ret = arch_add_memory(nid, start, size, &params);
 	if (ret < 0)
 		goto error;
 
 	/* create memory block devices after memory was added */
-	ret = create_memory_block_devices(start, size);
+	ret = create_memory_block_devices(start, size, mhp_altmap.alloc);
 	if (ret) {
 		arch_remove_memory(nid, start, size, NULL);
 		goto error;
@@ -1766,6 +1873,14 @@ static int check_memblock_offlined_cb(struct memory_block *mem, void *arg)
 	return 0;
 }
 
+static int get_nr_vmemmap_pages_cb(struct memory_block *mem, void *arg)
+{
+	/*
+	 * If not set, continue with the next block.
+	 */
+	return mem->nr_vmemmap_pages;
+}
+
 static int check_cpu_on_node(pg_data_t *pgdat)
 {
 	int cpu;
@@ -1840,6 +1955,9 @@ EXPORT_SYMBOL(try_offline_node);
 static int __ref try_remove_memory(int nid, u64 start, u64 size)
 {
 	int rc = 0;
+	struct vmem_altmap mhp_altmap = {};
+	struct vmem_altmap *altmap = NULL;
+	unsigned long nr_vmemmap_pages;
 
 	BUG_ON(check_hotplug_memory_range(start, size));
 
@@ -1852,6 +1970,31 @@ static int __ref try_remove_memory(int nid, u64 start, u64 size)
 	if (rc)
 		return rc;
 
+	/*
+	 * We only support removing memory added with MHP_MEMMAP_ON_MEMORY in
+	 * the same granularity it was added - a single memory block.
+	 */
+	if (memmap_on_memory) {
+		nr_vmemmap_pages = walk_memory_blocks(start, size, NULL,
+						      get_nr_vmemmap_pages_cb);
+		if (nr_vmemmap_pages) {
+			if (size != memory_block_size_bytes()) {
+				pr_warn("Refuse to remove %#llx - %#llx,"
+					"wrong granularity\n",
+					start, start + size);
+				return -EINVAL;
+			}
+
+			/*
+			 * Let remove_pmd_table->free_hugepage_table do the
+			 * right thing if we used vmem_altmap when hot-adding
+			 * the range.
+			 */
+			mhp_altmap.alloc = nr_vmemmap_pages;
+			altmap = &mhp_altmap;
+		}
+	}
+
 	/* remove memmap entry */
 	firmware_map_remove(start, start + size, "System RAM");
 
@@ -1863,7 +2006,7 @@ static int __ref try_remove_memory(int nid, u64 start, u64 size)
 
 	mem_hotplug_begin();
 
-	arch_remove_memory(nid, start, size, NULL);
+	arch_remove_memory(nid, start, size, altmap);
 
 	if (IS_ENABLED(CONFIG_ARCH_KEEP_MEMBLOCK)) {
 		memblock_free(start, size);
diff --git a/mm/sparse.c b/mm/sparse.c
index 7bd23f9d6cef..8e96cf00536b 100644
--- a/mm/sparse.c
+++ b/mm/sparse.c
@@ -623,7 +623,6 @@ void online_mem_sections(unsigned long start_pfn, unsigned long end_pfn)
 	}
 }
 
-#ifdef CONFIG_MEMORY_HOTREMOVE
 /* Mark all memory sections within the pfn range as offline */
 void offline_mem_sections(unsigned long start_pfn, unsigned long end_pfn)
 {
@@ -644,7 +643,6 @@ void offline_mem_sections(unsigned long start_pfn, unsigned long end_pfn)
 		ms->section_mem_map &= ~SECTION_IS_ONLINE;
 	}
 }
-#endif
 
 #ifdef CONFIG_SPARSEMEM_VMEMMAP
 static struct page * __meminit populate_section_memmap(unsigned long pfn,
-- 
2.16.3

[PATCH v10 5/8] acpi,memhotplug: Enable MHP_MEMMAP_ON_MEMORY when supported

[Oscar Salvador]

Let the caller check whether it can pass MHP_MEMMAP_ON_MEMORY by
checking mhp_supports_memmap_on_memory().
MHP_MEMMAP_ON_MEMORY can only be set in case
ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE is enabled, the architecture supports
altmap, and the range to be added spans a single memory block.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
---
 drivers/acpi/acpi_memhotplug.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/acpi/acpi_memhotplug.c b/drivers/acpi/acpi_memhotplug.c
index b02fd51e5589..8cc195c4c861 100644
--- a/drivers/acpi/acpi_memhotplug.c
+++ b/drivers/acpi/acpi_memhotplug.c
@@ -171,6 +171,7 @@ static int acpi_memory_enable_device(struct acpi_memory_device *mem_device)
 	acpi_handle handle = mem_device->device->handle;
 	int result, num_enabled = 0;
 	struct acpi_memory_info *info;
+	mhp_t mhp_flags = MHP_NONE;
 	int node;
 
 	node = acpi_get_node(handle);
@@ -194,8 +195,10 @@ static int acpi_memory_enable_device(struct acpi_memory_device *mem_device)
 		if (node < 0)
 			node = memory_add_physaddr_to_nid(info->start_addr);
 
+		if (mhp_supports_memmap_on_memory(info->length))
+			mhp_flags |= MHP_MEMMAP_ON_MEMORY;
 		result = __add_memory(node, info->start_addr, info->length,
-				      MHP_NONE);
+				      mhp_flags);
 
 		/*
 		 * If the memory block has been used by the kernel, add_memory()
-- 
2.16.3

[PATCH v10 6/8] mm,memory_hotplug: Add kernel boot option to enable memmap_on_memory

[Oscar Salvador]

Self stored memmap leads to a sparse memory situation which is unsuitable
for workloads that requires large contiguous memory chunks, so make this
an opt-in which needs to be explicitly enabled.

To control this, let memory_hotplug have its own memory space, as suggested
by David, so we can add memory_hotplug.memmap_on_memory parameter.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
---
 Documentation/admin-guide/kernel-parameters.txt | 17 +++++++++++++++++
 mm/Makefile                                     |  5 ++++-
 mm/memory_hotplug.c                             | 10 +++++++++-
 3 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 04545725f187..af32c17cd4eb 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -2794,6 +2794,23 @@
 			seconds.  Use this parameter to check at some
 			other rate.  0 disables periodic checking.
 
+	memory_hotplug.memmap_on_memory
+			[KNL,X86,ARM] Boolean flag to enable this feature.
+			Format: {on | off (default)}
+			When enabled, runtime hotplugged memory will
+			allocate its internal metadata (struct pages)
+			from the hotadded memory which will allow to
+			hotadd a lot of memory without requiring
+			additional memory to do so.
+			This feature is disabled by default because it
+			has some implication on large (e.g. GB)
+			allocations in some configurations (e.g. small
+			memory blocks).
+			The state of the flag can be read in
+			/sys/module/memory_hotplug/parameters/memmap_on_memory.
+			Note that even when enabled, there are a few cases where
+			the feature is not effective.
+
 	memtest=	[KNL,X86,ARM,PPC] Enable memtest
 			Format: <integer>
 			default : 0 <disable>
diff --git a/mm/Makefile b/mm/Makefile
index 72227b24a616..82ae9482f5e3 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -58,9 +58,13 @@ obj-y			:= filemap.o mempool.o oom_kill.o fadvise.o \
 page-alloc-y := page_alloc.o
 page-alloc-$(CONFIG_SHUFFLE_PAGE_ALLOCATOR) += shuffle.o
 
+# Give 'memory_hotplug' its own module-parameter namespace
+memory-hotplug-$(CONFIG_MEMORY_HOTPLUG) += memory_hotplug.o
+
 obj-y += page-alloc.o
 obj-y += init-mm.o
 obj-y += memblock.o
+obj-y += $(memory-hotplug-y)
 
 ifdef CONFIG_MMU
 	obj-$(CONFIG_ADVISE_SYSCALLS)	+= madvise.o
@@ -83,7 +87,6 @@ obj-$(CONFIG_SLUB) += slub.o
 obj-$(CONFIG_KASAN)	+= kasan/
 obj-$(CONFIG_KFENCE) += kfence/
 obj-$(CONFIG_FAILSLAB) += failslab.o
-obj-$(CONFIG_MEMORY_HOTPLUG) += memory_hotplug.o
 obj-$(CONFIG_MEMTEST)		+= memtest.o
 obj-$(CONFIG_MIGRATION) += migrate.o
 obj-$(CONFIG_TRANSPARENT_HUGEPAGE) += huge_memory.o khugepaged.o
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index d2de1fab75e2..e39db4321f71 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -42,7 +42,15 @@
 #include "internal.h"
 #include "shuffle.h"
 
-static bool memmap_on_memory;
+
+/*
+ * memory_hotplug.memmap_on_memory parameter
+ */
+static bool memmap_on_memory __ro_after_init;
+#ifdef CONFIG_MHP_MEMMAP_ON_MEMORY
+module_param(memmap_on_memory, bool, 0444);
+MODULE_PARM_DESC(memmap_on_memory, "Enable memmap on memory for memory hotplug");
+#endif
 
 /*
  * online_page_callback contains pointer to current page onlining function.
-- 
2.16.3

[PATCH v10 7/8] x86/Kconfig: Introduce ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE

[Oscar Salvador]

Enable x86_64 platform to use the MHP_MEMMAP_ON_MEMORY feature.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
---
 arch/x86/Kconfig | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 2792879d398e..9f0211df1746 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -2433,6 +2433,9 @@ config ARCH_ENABLE_MEMORY_HOTREMOVE
 	def_bool y
 	depends on MEMORY_HOTPLUG
 
+config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
+	def_bool y
+
 config USE_PERCPU_NUMA_NODE_ID
 	def_bool y
 	depends on NUMA
-- 
2.16.3

[PATCH v10 8/8] arm64/Kconfig: Introduce ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE

[Oscar Salvador]

Enable arm64 platform to use the MHP_MEMMAP_ON_MEMORY feature.

Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
---
 arch/arm64/Kconfig | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index e4e1b6550115..68735831b236 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -309,6 +309,9 @@ config ARCH_ENABLE_MEMORY_HOTPLUG
 config ARCH_ENABLE_MEMORY_HOTREMOVE
 	def_bool y
 
+config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
+	def_bool y
+
 config SMP
 	def_bool y
 
-- 
2.16.3

Re: [PATCH v10 4/8] mm,memory_hotplug: Allocate memmap from the added memory range

[Michal Hocko]

On Wed 21-04-21 12:26:57, Oscar Salvador wrote:
> Physical memory hotadd has to allocate a memmap (struct page array) for
> the newly added memory section. Currently, alloc_pages_node() is used
> for those allocations.
> 
> This has some disadvantages:
>  a) an existing memory is consumed for that purpose
>     (eg: ~2MB per 128MB memory section on x86_64)
>     This can even lead to extreme cases where system goes OOM because
>     the physically hotplugged memory depletes the available memory before
>     it is onlined.
>  b) if the whole node is movable then we have off-node struct pages
>     which has performance drawbacks.
>  c) It might be there are no PMD_ALIGNED chunks so memmap array gets
>     populated with base pages.
> 
> This can be improved when CONFIG_SPARSEMEM_VMEMMAP is enabled.
> 
> Vmemap page tables can map arbitrary memory. That means that we can
> reserve a part of the physically hotadded memory to back vmemmap page
> tables. This implementation uses the beginning of the hotplugged memory
> for that purpose.
> 
> There are some non-obviously things to consider though.  Vmemmap
> pages are allocated/freed during the memory hotplug events
> (add_memory_resource(), try_remove_memory()) when the memory is
> added/removed. This means that the reserved physical range is not online
> although it is used. The most obvious side effect is that pfn_to_online_page()
> returns NULL for those pfns. The current design expects that this
> should be OK as the hotplugged memory is considered a garbage until it
> is onlined. For example hibernation wouldn't save the content of those
> vmmemmaps into the image so it wouldn't be restored on resume but this
> should be OK as there no real content to recover anyway while metadata
> is reachable from other data structures (e.g. vmemmap page tables).
> 
> The reserved space is therefore (de)initialized during the {on,off}line
> events (mhp_{de}init_memmap_on_memory). That is done by extracting page
> allocator independent initialization from the regular onlining path.
> The primary reason to handle the reserved space outside of {on,off}line_pages
> is to make each initialization specific to the purpose rather than
> special case them in a single function.
> 
> As per above, the functions that are introduced are:
> 
>  - mhp_init_memmap_on_memory:
> 		       Initializes vmemmap pages by calling move_pfn_range_to_zone(),
> 		       calls kasan_add_zero_shadow(), and onlines as many sections
> 		       as vmemmap pages fully span.
>  - mhp_deinit_memmap_on_memory:
> 		       Offlines as many sections as vmemmap pages fully span,
> 		       removes the range from zhe zone by remove_pfn_range_from_zone(),
> 		       and calls kasan_remove_zero_shadow() for the range.
> 
> The new function memory_block_online() calls mhp_init_memmap_on_memory() before
> doing the actual online_pages(). Should online_pages() fail, we clean up
> by calling mhp_deinit_memmap_on_memory().
> Adjusting of present_pages is done at the end once we know that online_pages()
> succedeed.
> 
> On offline, memory_block_offline() needs to unaccount vmemmap pages from
> present_pages() before calling offline_pages().
> This is necessary because offline_pages() tears down some structures based
> on the fact whether the node or the zone become empty.
> If offline_pages() fails, we account back vmemmap pages.
> If it succeeds, we call mhp_deinit_memmap_on_memory().
> 
> Hot-remove:
> 
>  We need to be careful when removing memory, as adding and
>  removing memory needs to be done with the same granularity.
>  To check that this assumption is not violated, we check the
>  memory range we want to remove and if a) any memory block has
>  vmemmap pages and b) the range spans more than a single memory
>  block, we scream out loud and refuse to proceed.
> 
>  If all is good and the range was using memmap on memory (aka vmemmap pages),
>  we construct an altmap structure so free_hugepage_table does the right
>  thing and calls vmem_altmap_free instead of free_pagetable.
> 
> Signed-off-by: Oscar Salvador <osalvador@suse.de>
> Reviewed-by: David Hildenbrand <david@redhat.com>

Thanks for updating the changelog.

Acked-by: Michal Hocko <mhocko@suse.com>

[...]
> @@ -648,9 +650,16 @@ static void online_pages_range(unsigned long start_pfn, unsigned long nr_pages)
>  	 * decide to not expose all pages to the buddy (e.g., expose them
>  	 * later). We account all pages as being online and belonging to this
>  	 * zone ("present").
> +	 * When using memmap_on_memory, the range might not be aligned to
> +	 * MAX_ORDER_NR_PAGES - 1, but pageblock aligned. __ffs() will detect
> +	 * this and the first chunk to online will be pageblock_nr_pages.
>  	 */
> -	for (pfn = start_pfn; pfn < end_pfn; pfn += MAX_ORDER_NR_PAGES)
> -		(*online_page_callback)(pfn_to_page(pfn), MAX_ORDER - 1);
> +	for (pfn = start_pfn; pfn < end_pfn;) {
> +		int order = min(MAX_ORDER - 1UL, __ffs(pfn));
> +
> +		(*online_page_callback)(pfn_to_page(pfn), order);
> +		pfn += (1UL << order);
> +	}
>  
>  	/* mark all involved sections as online */
>  	online_mem_sections(start_pfn, end_pfn);

You have dropped the check for the overflow beyond end_pfn and this made
me think whether that is safe in general. It took me a while to realize
that end_pfn is always going to be within MAX_ORDER - 1 due to section
constrains (hopefully no surprises on some arches). Early init code is
in a much more complicated situation because the early memory maps can
have many oddities.

... just thinking out loud in case I need too to look that up again in
future...
-- 
Michal Hocko
SUSE Labs