Skip to contentSkip to conversation

mm/memory_hotplug: Shrink zones before removing memory

PATCH Superseded David Hildenbrand 4 participants 33 messages Aug 30, 2019 – Sep 27, 2019

Subsystems (11)
Memory management x86 PowerPC arm64 +7 more subsystems

v4 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v4 cover mm/memory_hotplug: Shrink zones before removing memory

This series fixes the access of uninitialized memmaps when shrinking zones/nodes and when removing memory.

We stop trying to shrink ZONE_DEVICE, as it's buggy, fixing it would be more involved (we don't have SECTION_IS_ONLINE as an indicator), and shrinking is only of limited use (set_zone_contiguous() cannot detect the ZONE_DEVICE as contiguous). As far as I can tell, this should be fine for ZONE_DEVICE.

We continue shrinking zones, but I reduced the amount of code to a minimum. Shrinking is especially necessary to keep zone->contiguous set where possible, especially on memory unplug of DIMMs at zone boundaries.

--------------------------------------------------------------------------

Zones are now properly shrunk when offlining memory blocks or when onlining failed. This allows to properly shrink zones on memory unplug even if the separate memory blocks of a DIMM were onlined to different zones or re-onlined to a different zone after offlining.

Example:

:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  0
        present  0
        managed  0
:/# echo "online_movable" > /sys/devices/system/memory/memory41/state
:/# echo "online_movable" > /sys/devices/system/memory/memory43/state
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  98304
        present  65536
        managed  65536
:/# echo 0 > /sys/devices/system/memory/memory43/online
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  32768
        present  32768
        managed  32768
:/# echo 0 > /sys/devices/system/memory/memory41/online
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  0
        present  0
        managed  0

--------------------------------------------------------------------------

I tested this with DIMMs on x86, but didn't test the ZONE_DEVICE part yet.

v3 -> v4:
- Drop "mm/memremap: Get rid of memmap_init_zone_device()"
-- As Alexander noticed, it was messy either way :)
- Drop "mm/memory_hotplug: Exit early in __remove_pages() on BUGs"
- Drop "mm: Exit early in set_zone_contiguous() if already contiguous"
- Drop "mm/memory_hotplug: Optimize zone shrinking code when checking for holes"
- Merged "mm/memory_hotplug: Remove pages from a zone before removing memory" and "mm/memory_hotplug: Remove zone parameter from __remove_pages()" into "mm/memory_hotplug: Shrink zones when offlining memory"
- Added "mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()"
- Stop shrinking ZONE_DEVICE
- Reshuffle patches, moving all fixes to the front. Add Fixes: tags.
- Change subject/description of various patches
- Minor changes (too many to mention)

Cc Aneesh Kumar K.V <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Dan Williams <[email protected]>
Cc Michal Hocko <[email protected]>

David Hildenbrand (8):
mm/memory_hotplug: Don't access uninitialized memmaps in
shrink_pgdat_span()
mm/memory_hotplug: Don't access uninitialized memmaps in
shrink_zone_span()
mm/memory_hotplug: Shrink zones when offlining memory mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone() mm/memory_hotplug: We always have a zone in
find_(smallest|biggest)_section_pfn
mm/memory_hotplug: Don't check for "all holes" in shrink_zone_span() mm/memory_hotplug: Drop local variables in shrink_zone_span() mm/memory_hotplug: Cleanup __remove_pages()

 arch/arm64/mm/mmu.c            |   4 +-
 arch/ia64/mm/init.c            |   4 +-
 arch/powerpc/mm/mem.c          |   3 +-
 arch/s390/mm/init.c            |   4 +-
 arch/sh/mm/init.c              |   4 +-
 arch/x86/mm/init_32.c          |   4 +-
 arch/x86/mm/init_64.c          |   4 +-
 include/linux/memory_hotplug.h |   7 +-
 mm/memory_hotplug.c            | 184 +++++++++++----------------------
 mm/memremap.c                  |  10 +-
 10 files changed, 80 insertions(+), 148 deletions(-)
Signature · 1 line
2.21.0
#1 lore raw

PATCH v4 1/8 mm/memory_hotplug: Don't access uninitialized memmaps in shrink_pgdat_span()

We might use the nid of memmaps that were never initialized. For example, if the memmap was poisoned, we will crash the kernel in pfn_to_nid() right now. Let's use the calculated boundaries of the separate zones instead. This now also avoids having to iterate over a whole bunch of subsections again, after shrinking one zone.

Before commit d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug"), the memmap was initialized to 0 and the node was set to the right value. After that commit, the node might be garbage.

We'll have to fix shrink_zone_span() next.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Cc Wei Yang <[email protected]>
Fixes d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug")
---
 mm/memory_hotplug.c | 72 ++++++++++-----------------------------------
 1 file changed, 15 insertions(+), 57 deletions(-)
mm/memory_hotplug.c +15 -57
index 49f7bf91c25a..ddba8d786e4a 100644
@@ -436,67 +436,25 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 	zone_span_writeunlock(zone);
 }
 
-static void shrink_pgdat_span(struct pglist_data *pgdat,
-			      unsigned long start_pfn, unsigned long end_pfn)
+static void update_pgdat_span(struct pglist_data *pgdat)
 {
-	unsigned long pgdat_start_pfn = pgdat->node_start_pfn;
-	unsigned long p = pgdat_end_pfn(pgdat); /* pgdat_end_pfn namespace clash */
-	unsigned long pgdat_end_pfn = p;
-	unsigned long pfn;
-	int nid = pgdat->node_id;
-
-	if (pgdat_start_pfn == start_pfn) {
-		/*
-		 * If the section is smallest section in the pgdat, it need
-		 * shrink pgdat->node_start_pfn and pgdat->node_spanned_pages.
-		 * In this case, we find second smallest valid mem_section
-		 * for shrinking zone.
-		 */
-		pfn = find_smallest_section_pfn(nid, NULL, end_pfn,
-						pgdat_end_pfn);
-		if (pfn) {
-			pgdat->node_start_pfn = pfn;
-			pgdat->node_spanned_pages = pgdat_end_pfn - pfn;
-		}
-	} else if (pgdat_end_pfn == end_pfn) {
-		/*
-		 * If the section is biggest section in the pgdat, it need
-		 * shrink pgdat->node_spanned_pages.
-		 * In this case, we find second biggest valid mem_section for
-		 * shrinking zone.
-		 */
-		pfn = find_biggest_section_pfn(nid, NULL, pgdat_start_pfn,
-					       start_pfn);
-		if (pfn)
-			pgdat->node_spanned_pages = pfn - pgdat_start_pfn + 1;
-	}
-
-	/*
-	 * If the section is not biggest or smallest mem_section in the pgdat,
-	 * it only creates a hole in the pgdat. So in this case, we need not
-	 * change the pgdat.
-	 * But perhaps, the pgdat has only hole data. Thus it check the pgdat
-	 * has only hole or not.
-	 */
-	pfn = pgdat_start_pfn;
-	for (; pfn < pgdat_end_pfn; pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
-			continue;
-
-		if (pfn_to_nid(pfn) != nid)
-			continue;
+	unsigned long node_start_pfn = 0, node_end_pfn = 0;
+	struct zone *zone;
 
-		/* Skip range to be removed */
-		if (pfn >= start_pfn && pfn < end_pfn)
-			continue;
+	for (zone = pgdat->node_zones;
+	     zone < pgdat->node_zones + MAX_NR_ZONES; zone++) {
+		unsigned long zone_end_pfn = zone->zone_start_pfn +
+					     zone->spanned_pages;
 
-		/* If we find valid section, we have nothing to do */
-		return;
+		/* No need to lock the zones, they can't change. */
+		if (zone_end_pfn > node_end_pfn)
+			node_end_pfn = zone_end_pfn;
+		if (zone->zone_start_pfn < node_start_pfn)
+			node_start_pfn = zone->zone_start_pfn;
 	}
 
-	/* The pgdat has no valid section */
-	pgdat->node_start_pfn = 0;
-	pgdat->node_spanned_pages = 0;
+	pgdat->node_start_pfn = node_start_pfn;
+	pgdat->node_spanned_pages = node_end_pfn - node_start_pfn;
 }
 
 static void __remove_zone(struct zone *zone, unsigned long start_pfn,
@@ -507,7 +465,7 @@ static void __remove_zone(struct zone *zone, unsigned long start_pfn,
 
 	pgdat_resize_lock(zone->zone_pgdat, &flags);
 	shrink_zone_span(zone, start_pfn, start_pfn + nr_pages);
-	shrink_pgdat_span(pgdat, start_pfn, start_pfn + nr_pages);
+	update_pgdat_span(pgdat);
 	pgdat_resize_unlock(zone->zone_pgdat, &flags);
 }
 
Signature · 1 line
2.21.0
#2 lore raw

PATCH v4 2/8 mm/memory_hotplug: Don't access uninitialized memmaps in shrink_zone_span()

In reply to David Hildenbrand

Let's limit shrinking to !ZONE_DEVICE so we can fix the current code. We should never try to touch the memmap of offline sections where we could have uninitialized memmaps and could trigger BUGs when calling page_to_nid() on poisoned pages.

Stopping to shrink the ZONE_DEVICE is fine as set_zone_contiguous() cannot deal with ZONE_DEVICE either way. The zones will always be !contiguous already and zone shrinking is therefore of limited use.

Before commit d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug"), the memmap was initialized with 0 and the node with the right value. So the zone might be wrong but not garbage. After that commit, both the zone and the node will be garbage when touching uninitialized memmaps.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Fixes d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug")
---
 mm/memory_hotplug.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +11 -3
index ddba8d786e4a..e0d1f6a9dfeb 100644
@@ -331,7 +331,7 @@ static unsigned long find_smallest_section_pfn(int nid, struct zone *zone,
 				     unsigned long end_pfn)
 {
 	for (; start_pfn < end_pfn; start_pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(start_pfn)))
+		if (unlikely(!pfn_to_online_page(start_pfn)))
 			continue;
 
 		if (unlikely(pfn_to_nid(start_pfn) != nid))
@@ -356,7 +356,7 @@ static unsigned long find_biggest_section_pfn(int nid, struct zone *zone,
 	/* pfn is the end pfn of a memory section. */
 	pfn = end_pfn - 1;
 	for (; pfn >= start_pfn; pfn -= PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
+		if (unlikely(!pfn_to_online_page(pfn)))
 			continue;
 
 		if (unlikely(pfn_to_nid(pfn) != nid))
@@ -415,7 +415,7 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 	 */
 	pfn = zone_start_pfn;
 	for (; pfn < zone_end_pfn; pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
+		if (unlikely(!pfn_to_online_page(pfn)))
 			continue;
 
 		if (page_zone(pfn_to_page(pfn)) != zone)
@@ -463,6 +463,14 @@ static void __remove_zone(struct zone *zone, unsigned long start_pfn,
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	unsigned long flags;
 
+	/*
+	 * Zone shrinking code cannot properly deal with ZONE_DEVICE. So
+	 * we will not try to shrink the zones - which is okay as
+	 * set_zone_contiguous() cannot deal with ZONE_DEVICE either way.
+	 */
+	if (zone_idx(zone) == ZONE_DEVICE)
+		return;
+
 	pgdat_resize_lock(zone->zone_pgdat, &flags);
 	shrink_zone_span(zone, start_pfn, start_pfn + nr_pages);
 	update_pgdat_span(pgdat);
Signature · 1 line
2.21.0
#3 lore raw

Re: [PATCH v4 2/8] mm/memory_hotplug: Don't access uninitialized memmaps in shrink_zone_span()

Quoted 8 lines · David Hildenbrand

Let's limit shrinking to !ZONE_DEVICE so we can fix the current code. We should never try to touch the memmap of offline sections where we could have uninitialized memmaps and could trigger BUGs when calling page_to_nid() on poisoned pages.

Stopping to shrink the ZONE_DEVICE is fine as set_zone_contiguous() cannot deal with ZONE_DEVICE either way. The zones will always be !contiguous already and zone shrinking is therefore of limited use.

Can you add more details w.r.t ZONE_DEVICE?

Quoted 52 lines

Before commit d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug"), the memmap was initialized with 0 and the node with the right value. So the zone might be wrong but not garbage. After that commit, both the zone and the node will be garbage when touching uninitialized memmaps.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Fixes d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug")
---
 mm/memory_hotplug.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +3 -3
index ddba8d786e4a..e0d1f6a9dfeb 100644
@@ -331,7 +331,7 @@ static unsigned long find_smallest_section_pfn(int nid, struct zone *zone,
 				     unsigned long end_pfn)
 {
 	for (; start_pfn < end_pfn; start_pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(start_pfn)))
+		if (unlikely(!pfn_to_online_page(start_pfn)))
 			continue;
 
 		if (unlikely(pfn_to_nid(start_pfn) != nid))
@@ -356,7 +356,7 @@ static unsigned long find_biggest_section_pfn(int nid, struct zone *zone,
 	/* pfn is the end pfn of a memory section. */
 	pfn = end_pfn - 1;
 	for (; pfn >= start_pfn; pfn -= PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
+		if (unlikely(!pfn_to_online_page(pfn)))
 			continue;
 
 		if (unlikely(pfn_to_nid(pfn) != nid))
@@ -415,7 +415,7 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 	 */
 	pfn = zone_start_pfn;
 	for (; pfn < zone_end_pfn; pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
+		if (unlikely(!pfn_to_online_page(pfn)))
 			continue;
 
 		if (page_zone(pfn_to_page(pfn)) != zone)

mm/memory_hotplug.c static void __remove_zone(struct zone *zone, unsigned long start_pfn, · line 472

Quoted 10 lines
mm/memory_hotplug.c +7 -0
@@ -463,6 +463,14 @@ static void __remove_zone(struct zone *zone, unsigned long start_pfn,
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	unsigned long flags;
 
+	/*
+	 * Zone shrinking code cannot properly deal with ZONE_DEVICE. So
+	 * we will not try to shrink the zones - which is okay as
+	 * set_zone_contiguous() cannot deal with ZONE_DEVICE either way.
+	 */
+	if (zone_idx(zone) == ZONE_DEVICE)
+		return;

Can you describe here what is special about ZONE_DEVICE?

Quoted 5 lines
mm/memory_hotplug.c +0 -0
 	pgdat_resize_lock(zone->zone_pgdat, &flags);
 	shrink_zone_span(zone, start_pfn, start_pfn + nr_pages);
 	update_pgdat_span(pgdat);
Signature · 1 line
2.21.0
#4 lore raw

Re: [PATCH v4 2/8] mm/memory_hotplug: Don't access uninitialized memmaps in shrink_zone_span()

Quoted 7 lines · Aneesh Kumar K.V
Quoted 5 lines · David Hildenbrand

Let's limit shrinking to !ZONE_DEVICE so we can fix the current code. We should never try to touch the memmap of offline sections where we could have uninitialized memmaps and could trigger BUGs when calling page_to_nid() on poisoned pages.

Quoted 5 lines
Quoted 3 lines

Stopping to shrink the ZONE_DEVICE is fine as set_zone_contiguous() cannot deal with ZONE_DEVICE either way. The zones will always be !contiguous already and zone shrinking is therefore of limited use.

Can you add more details w.r.t ZONE_DEVICE?

I can convert that to

"There is no reliable way to distinguish an uninitialized memmap from an initialized memmap that belongs to ZONE_DEVICE, as we don't have anything like SECTION_IS_ONLINE we can use similar to pfn_to_online_section() for !ZONE_DEVICE memory. E.g., set_zone_contiguous() similarly relies on pfn_to_online_section() and will therefore never set a ZONE_DEVICE zone consecutive. Stopping to shrink the ZONE_DEVICE therefore results in no observable changes, besides /proc/zoneinfo indicating different boundaries - something we can totally live with."

Quoted 56 lines
Quoted 56 lines

Before commit d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug"), the memmap was initialized with 0 and the node with the right value. So the zone might be wrong but not garbage. After that commit, both the zone and the node will be garbage when touching uninitialized memmaps.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Fixes d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug")
---
 mm/memory_hotplug.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +3 -3
index ddba8d786e4a..e0d1f6a9dfeb 100644
@@ -331,7 +331,7 @@ static unsigned long find_smallest_section_pfn(int nid, struct zone *zone,
 				     unsigned long end_pfn)
 {
 	for (; start_pfn < end_pfn; start_pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(start_pfn)))
+		if (unlikely(!pfn_to_online_page(start_pfn)))
 			continue;
 
 		if (unlikely(pfn_to_nid(start_pfn) != nid))
@@ -356,7 +356,7 @@ static unsigned long find_biggest_section_pfn(int nid, struct zone *zone,
 	/* pfn is the end pfn of a memory section. */
 	pfn = end_pfn - 1;
 	for (; pfn >= start_pfn; pfn -= PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
+		if (unlikely(!pfn_to_online_page(pfn)))
 			continue;
 
 		if (unlikely(pfn_to_nid(pfn) != nid))
@@ -415,7 +415,7 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 	 */
 	pfn = zone_start_pfn;
 	for (; pfn < zone_end_pfn; pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_valid(pfn)))
+		if (unlikely(!pfn_to_online_page(pfn)))
 			continue;
 
 		if (page_zone(pfn_to_page(pfn)) != zone)
@@ -463,6 +463,14 @@ static void __remove_zone(struct zone *zone, unsigned long start_pfn,
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	unsigned long flags;
 
Quoted 9 lines

mm/memory_hotplug.c static void __remove_zone(struct zone *zone, unsigned long start_pfn, · line 472

Quoted 7 lines
mm/memory_hotplug.c +7 -0
+	/*
+	 * Zone shrinking code cannot properly deal with ZONE_DEVICE. So
+	 * we will not try to shrink the zones - which is okay as
+	 * set_zone_contiguous() cannot deal with ZONE_DEVICE either way.
+	 */
+	if (zone_idx(zone) == ZONE_DEVICE)
+		return;

Can you describe here what is special about ZONE_DEVICE?

I think adding more details to the description is sufficient, especially as this also applies to set_zone_contiguous().

Thanks!

Signature · 3 lines
Thanks,

David / dhildenb
#5 lore raw

PATCH v4 3/8 mm/memory_hotplug: Shrink zones when offlining memory

In reply to David Hildenbrand

We currently try to shrink a single zone when removing memory. We use the zone of the first page of the memory we are removing. If that memmap was never initialized (e.g., memory was never onlined), we will read garbage and can trigger kernel BUGs (due to a stale pointer):

:/# [   23.912993] BUG: unable to handle page fault for address: 000000000000353d
[   23.914219] #PF: supervisor write access in kernel mode
[   23.915199] #PF: error_code(0x0002) - not-present page
[   23.916160] PGD 0 P4D 0
[   23.916627] Oops: 0002 [#1] SMP PTI
[   23.917256] CPU: 1 PID: 7 Comm: kworker/u8:0 Not tainted 5.3.0-rc5-next-20190820+ #317
[   23.918900] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.1-0-ga5cab58e9a3f-prebuilt.qemu.4
[   23.921194] Workqueue: kacpi_hotplug acpi_hotplug_work_fn
[   23.922249] RIP: 0010:clear_zone_contiguous+0x5/0x10
[   23.923173] Code: 48 89 c6 48 89 c3 e8 2a fe ff ff 48 85 c0 75 cf 5b 5d c3 c6 85 fd 05 00 00 01 5b 5d c3 0f 1f 840
[   23.926876] RSP: 0018:ffffad2400043c98 EFLAGS: 00010246
[   23.927928] RAX: 0000000000000000 RBX: 0000000200000000 RCX: 0000000000000000
[   23.929458] RDX: 0000000000200000 RSI: 0000000000140000 RDI: 0000000000002f40
[   23.930899] RBP: 0000000140000000 R08: 0000000000000000 R09: 0000000000000001
[   23.932362] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000140000
[   23.933603] R13: 0000000000140000 R14: 0000000000002f40 R15: ffff9e3e7aff3680
[   23.934913] FS:  0000000000000000(0000) GS:ffff9e3e7bb00000(0000) knlGS:0000000000000000
[   23.936294] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   23.937481] CR2: 000000000000353d CR3: 0000000058610000 CR4: 00000000000006e0
[   23.938687] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   23.939889] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[   23.941168] Call Trace:
[   23.941580]  __remove_pages+0x4b/0x640
[   23.942303]  ? mark_held_locks+0x49/0x70
[   23.943149]  arch_remove_memory+0x63/0x8d
[   23.943921]  try_remove_memory+0xdb/0x130
[   23.944766]  ? walk_memory_blocks+0x7f/0x9e
[   23.945616]  __remove_memory+0xa/0x11
[   23.946274]  acpi_memory_device_remove+0x70/0x100
[   23.947308]  acpi_bus_trim+0x55/0x90
[   23.947914]  acpi_device_hotplug+0x227/0x3a0
[   23.948714]  acpi_hotplug_work_fn+0x1a/0x30
[   23.949433]  process_one_work+0x221/0x550
[   23.950190]  worker_thread+0x50/0x3b0
[   23.950993]  kthread+0x105/0x140
[   23.951644]  ? process_one_work+0x550/0x550
[   23.952508]  ? kthread_park+0x80/0x80
[   23.953367]  ret_from_fork+0x3a/0x50
[   23.954025] Modules linked in:
[   23.954613] CR2: 000000000000353d
[   23.955248] ---[ end trace 93d982b1fb3e1a69 ]---

Instead, shrink the zones when offlining memory or when onlining failed. Introduce and use remove_pfn_range_from_zone(() for that. We now properly shrink the zones, even if we have DIMMs whereby
- Some memory blocks fall into no zone (never onlined)
- Some memory blocks fall into multiple zones (offlined+re-onlined)
- Multiple memory blocks that fall into different zones

Drop the zone parameter (with a potential dubious value) from __remove_pages() and __remove_section().

Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>
Cc Tony Luck <[email protected]>
Cc Fenghua Yu <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc Vasily Gorbik <[email protected]>
Cc Christian Borntraeger <[email protected]>
Cc Yoshinori Sato <[email protected]>
Cc Rich Felker <[email protected]>
Cc Dave Hansen <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Steve Capper <[email protected]>
Cc Mike Rapoport <[email protected]>
Cc Anshuman Khandual <[email protected]>
Cc Yu Zhao <[email protected]>
Cc Jun Yao <[email protected]>
Cc Robin Murphy <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc "Matthew Wilcox (Oracle)" <[email protected]>
Cc Christophe Leroy <[email protected]>
Cc "Aneesh Kumar K.V" <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Gerald Schaefer <[email protected]>
Cc Halil Pasic <[email protected]>
Cc Tom Lendacky <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Masahiro Yamada <[email protected]>
Cc Dan Williams <[email protected]>
Cc Wei Yang <[email protected]>
Cc Qian Cai <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Logan Gunthorpe <[email protected]>
Cc Ira Weiny <[email protected]>
Fixes d0dc12e86b31 ("mm/memory_hotplug: optimize memory hotplug")
---
 arch/arm64/mm/mmu.c            |  4 +---
 arch/ia64/mm/init.c            |  4 +---
 arch/powerpc/mm/mem.c          |  3 +--
 arch/s390/mm/init.c            |  4 +---
 arch/sh/mm/init.c              |  4 +---
 arch/x86/mm/init_32.c          |  4 +---
 arch/x86/mm/init_64.c          |  4 +---
 include/linux/memory_hotplug.h |  7 +++++--
 mm/memory_hotplug.c            | 31 ++++++++++++++++---------------
 mm/memremap.c                  |  3 +--
 10 files changed, 29 insertions(+), 39 deletions(-)
arch/arm64/mm/mmu.c +1 -3
index 60c929f3683b..d10247fab0fd 100644
@@ -1069,7 +1069,6 @@ void arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct zone *zone;
 
 	/*
 	 * FIXME: Cleanup page tables (also in arch_add_memory() in case
@@ -1078,7 +1077,6 @@ void arch_remove_memory(int nid, u64 start, u64 size,
 	 * unplug. ARCH_ENABLE_MEMORY_HOTREMOVE must not be
 	 * unlocked yet.
 	 */
-	zone = page_zone(pfn_to_page(start_pfn));
-	__remove_pages(zone, start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 }
 #endif
arch/ia64/mm/init.c +1 -3
index bf9df2625bc8..a6dd80a2c939 100644
@@ -689,9 +689,7 @@ void arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct zone *zone;
 
-	zone = page_zone(pfn_to_page(start_pfn));
-	__remove_pages(zone, start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 }
 #endif
arch/powerpc/mm/mem.c +1 -2
index 69f99128a8d6..f3a5e397b911 100644
@@ -130,10 +130,9 @@ void __ref arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct page *page = pfn_to_page(start_pfn) + vmem_altmap_offset(altmap);
 	int ret;
 
-	__remove_pages(page_zone(page), start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 
 	/* Remove htab bolted mappings for this section of memory */
 	start = (unsigned long)__va(start);
arch/s390/mm/init.c +1 -3
index 20340a03ad90..6f13eb66e375 100644
@@ -296,10 +296,8 @@ void arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct zone *zone;
 
-	zone = page_zone(pfn_to_page(start_pfn));
-	__remove_pages(zone, start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 	vmem_remove_mapping(start, size);
 }
 #endif /* CONFIG_MEMORY_HOTPLUG */
arch/sh/mm/init.c +1 -3
index dfdbaa50946e..d1b1ff2be17a 100644
@@ -434,9 +434,7 @@ void arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = PFN_DOWN(start);
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct zone *zone;
 
-	zone = page_zone(pfn_to_page(start_pfn));
-	__remove_pages(zone, start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 }
 #endif /* CONFIG_MEMORY_HOTPLUG */
arch/x86/mm/init_32.c +1 -3
index 4068abb9427f..9d036be27aaa 100644
@@ -865,10 +865,8 @@ void arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct zone *zone;
 
-	zone = page_zone(pfn_to_page(start_pfn));
-	__remove_pages(zone, start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 }
 #endif
 
arch/x86/mm/init_64.c +1 -3
index a6b5c653727b..b8541d77452c 100644
@@ -1212,10 +1212,8 @@ void __ref arch_remove_memory(int nid, u64 start, u64 size,
 {
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
-	struct page *page = pfn_to_page(start_pfn) + vmem_altmap_offset(altmap);
-	struct zone *zone = page_zone(page);
 
-	__remove_pages(zone, start_pfn, nr_pages, altmap);
+	__remove_pages(start_pfn, nr_pages, altmap);
 	kernel_physical_mapping_remove(start, start + size);
 }
 #endif /* CONFIG_MEMORY_HOTPLUG */
include/linux/memory_hotplug.h +5 -2
index f46ea71b4ffd..451efd4499cc 100644
@@ -125,8 +125,8 @@ static inline bool movable_node_is_enabled(void)
 
 extern void arch_remove_memory(int nid, u64 start, u64 size,
 			       struct vmem_altmap *altmap);
-extern void __remove_pages(struct zone *zone, unsigned long start_pfn,
-			   unsigned long nr_pages, struct vmem_altmap *altmap);
+extern void __remove_pages(unsigned long start_pfn, unsigned long nr_pages,
+			   struct vmem_altmap *altmap);
 
 /* reasonably generic interface to expand the physical pages */
 extern int __add_pages(int nid, unsigned long start_pfn, unsigned long nr_pages,
@@ -345,6 +345,9 @@ extern int add_memory(int nid, u64 start, u64 size);
 extern int add_memory_resource(int nid, struct resource *resource);
 extern void move_pfn_range_to_zone(struct zone *zone, unsigned long start_pfn,
 		unsigned long nr_pages, struct vmem_altmap *altmap);
+extern void remove_pfn_range_from_zone(struct zone *zone,
+				       unsigned long start_pfn,
+				       unsigned long nr_pages);
 extern bool is_memblock_offlined(struct memory_block *mem);
 extern int sparse_add_section(int nid, unsigned long pfn,
 		unsigned long nr_pages, struct vmem_altmap *altmap);
mm/memory_hotplug.c +16 -15
index e0d1f6a9dfeb..4da59ec14dbb 100644
@@ -457,8 +457,9 @@ static void update_pgdat_span(struct pglist_data *pgdat)
 	pgdat->node_spanned_pages = node_end_pfn - node_start_pfn;
 }
 
-static void __remove_zone(struct zone *zone, unsigned long start_pfn,
-		unsigned long nr_pages)
+void __ref remove_pfn_range_from_zone(struct zone *zone,
+				      unsigned long start_pfn,
+				      unsigned long nr_pages)
 {
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	unsigned long flags;
@@ -471,28 +472,30 @@ static void __remove_zone(struct zone *zone, unsigned long start_pfn,
 	if (zone_idx(zone) == ZONE_DEVICE)
 		return;
 
+	clear_zone_contiguous(zone);
+
 	pgdat_resize_lock(zone->zone_pgdat, &flags);
 	shrink_zone_span(zone, start_pfn, start_pfn + nr_pages);
 	update_pgdat_span(pgdat);
 	pgdat_resize_unlock(zone->zone_pgdat, &flags);
+
+	set_zone_contiguous(zone);
 }
 
-static void __remove_section(struct zone *zone, unsigned long pfn,
-		unsigned long nr_pages, unsigned long map_offset,
-		struct vmem_altmap *altmap)
+static void __remove_section(unsigned long pfn, unsigned long nr_pages,
+			     unsigned long map_offset,
+			     struct vmem_altmap *altmap)
 {
 	struct mem_section *ms = __nr_to_section(pfn_to_section_nr(pfn));
 
 	if (WARN_ON_ONCE(!valid_section(ms)))
 		return;
 
-	__remove_zone(zone, pfn, nr_pages);
 	sparse_remove_section(ms, pfn, nr_pages, map_offset, altmap);
 }
 
 /**
- * __remove_pages() - remove sections of pages from a zone
- * @zone: zone from which pages need to be removed
+ * __remove_pages() - remove sections of pages
  * @pfn: starting pageframe (must be aligned to start of a section)
  * @nr_pages: number of pages to remove (must be multiple of section size)
  * @altmap: alternative device page map or %NULL if default memmap is used
@@ -502,16 +505,14 @@ static void __remove_section(struct zone *zone, unsigned long pfn,
  * sure that pages are marked reserved and zones are adjust properly by
  * calling offline_pages().
  */
-void __remove_pages(struct zone *zone, unsigned long pfn,
-		    unsigned long nr_pages, struct vmem_altmap *altmap)
+void __remove_pages(unsigned long pfn, unsigned long nr_pages,
+		    struct vmem_altmap *altmap)
 {
 	unsigned long map_offset = 0;
 	unsigned long nr, start_sec, end_sec;
 
 	map_offset = vmem_altmap_offset(altmap);
 
-	clear_zone_contiguous(zone);
-
 	if (check_pfn_span(pfn, nr_pages, "remove"))
 		return;
 
@@ -523,13 +524,11 @@ void __remove_pages(struct zone *zone, unsigned long pfn,
 		cond_resched();
 		pfns = min(nr_pages, PAGES_PER_SECTION
 				- (pfn & ~PAGE_SECTION_MASK));
-		__remove_section(zone, pfn, pfns, map_offset, altmap);
+		__remove_section(pfn, pfns, map_offset, altmap);
 		pfn += pfns;
 		nr_pages -= pfns;
 		map_offset = 0;
 	}
-
-	set_zone_contiguous(zone);
 }
 
 int set_online_page_callback(online_page_callback_t callback)
@@ -857,6 +856,7 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages, int online_typ
 		 (unsigned long long) pfn << PAGE_SHIFT,
 		 (((unsigned long long) pfn + nr_pages) << PAGE_SHIFT) - 1);
 	memory_notify(MEM_CANCEL_ONLINE, &arg);
+	remove_pfn_range_from_zone(zone, pfn, nr_pages);
 	mem_hotplug_done();
 	return ret;
 }
@@ -1592,6 +1592,7 @@ static int __ref __offline_pages(unsigned long start_pfn,
 	writeback_set_ratelimit();
 
 	memory_notify(MEM_OFFLINE, &arg);
+	remove_pfn_range_from_zone(zone, start_pfn, nr_pages);
 	mem_hotplug_done();
 	return 0;
 
mm/memremap.c +1 -2
index f6c17339cd0d..cb90c3e8804a 100644
@@ -139,8 +139,7 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 	mem_hotplug_begin();
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
 		pfn = PHYS_PFN(res->start);
-		__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
-				 PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
 				pgmap_altmap(pgmap));
Signature · 1 line
2.21.0
#6 lore raw

PATCH v4 4/8 mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()

In reply to David Hildenbrand

Let's poison the pages similar to when adding new memory in sparse_add_section(). Also call remove_pfn_range_from_zone() from memunmap_pages(), so we can poison the memmap from there as well.

While at it, calculate the pfn in memunmap_pages() only once.

Cc Andrew Morton <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
---
 mm/memory_hotplug.c | 3 +++
 mm/memremap.c       | 7 ++++---
 2 files changed, 7 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +3 -0
index 4da59ec14dbb..5bfca690a922 100644
@@ -464,6 +464,9 @@ void __ref remove_pfn_range_from_zone(struct zone *zone,
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	unsigned long flags;
 
+	/* Poison struct pages because they are now uninitialized again. */
+	page_init_poison(pfn_to_page(start_pfn), sizeof(struct page) * nr_pages);
+
 	/*
 	 * Zone shrinking code cannot properly deal with ZONE_DEVICE. So
 	 * we will not try to shrink the zones - which is okay as
mm/memremap.c +4 -3
index cb90c3e8804a..48f573502f88 100644
@@ -125,7 +125,7 @@ static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
 void memunmap_pages(struct dev_pagemap *pgmap)
 {
 	struct resource *res = &pgmap->res;
-	unsigned long pfn;
+	unsigned long pfn = PHYS_PFN(res->start);
 	int nid;
 
 	dev_pagemap_kill(pgmap);
@@ -134,11 +134,12 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 	dev_pagemap_cleanup(pgmap);
 
 	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(PHYS_PFN(res->start)));
+	nid = page_to_nid(pfn_to_page(pfn));
 
 	mem_hotplug_begin();
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
+				   PHYS_PFN(resource_size(res)));
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		pfn = PHYS_PFN(res->start);
 		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
Signature · 1 line
2.21.0
#7 lore raw

Re: [PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()

David Hildenbrand <[email protected]> writes:
@@ -134,11 +134,12 @@ void memunmap_pages(struct dev_pagemap *pgmap)

mm/memremap.c void memunmap_pages(struct dev_pagemap *pgmap) · line 141

Quoted 3 lines
mm/memremap.c +2 -0
 	mem_hotplug_begin();
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
+				   PHYS_PFN(resource_size(res)));

That should be part of PATCH 3?

Quoted 7 lines
mm/memremap.c +0 -1
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		pfn = PHYS_PFN(res->start);
 		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
Signature · 1 line
2.21.0

-aneesh

#8 lore raw

Re: [PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()

Quoted 8 lines · Aneesh Kumar K.V

David Hildenbrand <[email protected]> writes:
@@ -134,11 +134,12 @@ void memunmap_pages(struct dev_pagemap *pgmap)

mm/memremap.c void memunmap_pages(struct dev_pagemap *pgmap) · line 141

Quoted 3 lines
mm/memremap.c +2 -0
 	mem_hotplug_begin();
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
+				   PHYS_PFN(resource_size(res)));

That should be part of PATCH 3?

I thought about that but moved it to #4, because it easily gets lost in the already-big-enough-patch and is a NOP for ZONE_DEVICE before this change.

Thanks!

Signature · 3 lines
Thanks,

David / dhildenb
#9 lore raw

PATCH v4 5/8 mm/memory_hotplug: We always have a zone in find_(smallest|biggest)_section_pfn

In reply to David Hildenbrand

With shrink_pgdat_span() out of the way, we now always have a valid zone.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Cc Wei Yang <[email protected]>
---
 mm/memory_hotplug.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
mm/memory_hotplug.c +2 -2
index 5bfca690a922..d6807934bb30 100644
@@ -337,7 +337,7 @@ static unsigned long find_smallest_section_pfn(int nid, struct zone *zone,
 		if (unlikely(pfn_to_nid(start_pfn) != nid))
 			continue;
 
-		if (zone && zone != page_zone(pfn_to_page(start_pfn)))
+		if (zone != page_zone(pfn_to_page(start_pfn)))
 			continue;
 
 		return start_pfn;
@@ -362,7 +362,7 @@ static unsigned long find_biggest_section_pfn(int nid, struct zone *zone,
 		if (unlikely(pfn_to_nid(pfn) != nid))
 			continue;
 
-		if (zone && zone != page_zone(pfn_to_page(pfn)))
+		if (zone != page_zone(pfn_to_page(pfn)))
 			continue;
 
 		return pfn;
Signature · 1 line
2.21.0
#10 lore raw

PATCH v4 6/8 mm/memory_hotplug: Don't check for "all holes" in shrink_zone_span()

In reply to David Hildenbrand

If we have holes, the holes will automatically get detected and removed once we remove the next bigger/smaller section. The extra checks can go.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Michal Hocko <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Cc Wei Yang <[email protected]>
---
 mm/memory_hotplug.c | 34 +++++++---------------------------
 1 file changed, 7 insertions(+), 27 deletions(-)
mm/memory_hotplug.c +7 -27
index d6807934bb30..82f5012cea3c 100644
@@ -393,6 +393,9 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 		if (pfn) {
 			zone->zone_start_pfn = pfn;
 			zone->spanned_pages = zone_end_pfn - pfn;
+		} else {
+			zone->zone_start_pfn = 0;
+			zone->spanned_pages = 0;
 		}
 	} else if (zone_end_pfn == end_pfn) {
 		/*
@@ -405,34 +408,11 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 					       start_pfn);
 		if (pfn)
 			zone->spanned_pages = pfn - zone_start_pfn + 1;
+		else {
+			zone->zone_start_pfn = 0;
+			zone->spanned_pages = 0;
+		}
 	}
-
-	/*
-	 * The section is not biggest or smallest mem_section in the zone, it
-	 * only creates a hole in the zone. So in this case, we need not
-	 * change the zone. But perhaps, the zone has only hole data. Thus
-	 * it check the zone has only hole or not.
-	 */
-	pfn = zone_start_pfn;
-	for (; pfn < zone_end_pfn; pfn += PAGES_PER_SUBSECTION) {
-		if (unlikely(!pfn_to_online_page(pfn)))
-			continue;
-
-		if (page_zone(pfn_to_page(pfn)) != zone)
-			continue;
-
-		/* Skip range to be removed */
-		if (pfn >= start_pfn && pfn < end_pfn)
-			continue;
-
-		/* If we find valid section, we have nothing to do */
-		zone_span_writeunlock(zone);
-		return;
-	}
-
-	/* The zone has no valid section */
-	zone->zone_start_pfn = 0;
-	zone->spanned_pages = 0;
 	zone_span_writeunlock(zone);
 }
 
Signature · 1 line
2.21.0
#11 lore raw

PATCH v4 7/8 mm/memory_hotplug: Drop local variables in shrink_zone_span()

In reply to David Hildenbrand

Get rid of the unnecessary local variables.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Cc Wei Yang <[email protected]>
---
 mm/memory_hotplug.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
mm/memory_hotplug.c +6 -9
index 82f5012cea3c..80cb32cd105e 100644
@@ -374,14 +374,11 @@ static unsigned long find_biggest_section_pfn(int nid, struct zone *zone,
 static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 			     unsigned long end_pfn)
 {
-	unsigned long zone_start_pfn = zone->zone_start_pfn;
-	unsigned long z = zone_end_pfn(zone); /* zone_end_pfn namespace clash */
-	unsigned long zone_end_pfn = z;
 	unsigned long pfn;
 	int nid = zone_to_nid(zone);
 
 	zone_span_writelock(zone);
-	if (zone_start_pfn == start_pfn) {
+	if (zone->zone_start_pfn == start_pfn) {
 		/*
 		 * If the section is smallest section in the zone, it need
 		 * shrink zone->zone_start_pfn and zone->zone_spanned_pages.
@@ -389,25 +386,25 @@ static void shrink_zone_span(struct zone *zone, unsigned long start_pfn,
 		 * for shrinking zone.
 		 */
 		pfn = find_smallest_section_pfn(nid, zone, end_pfn,
-						zone_end_pfn);
+						zone_end_pfn(zone));
 		if (pfn) {
+			zone->spanned_pages = zone_end_pfn(zone) - pfn;
 			zone->zone_start_pfn = pfn;
-			zone->spanned_pages = zone_end_pfn - pfn;
 		} else {
 			zone->zone_start_pfn = 0;
 			zone->spanned_pages = 0;
 		}
-	} else if (zone_end_pfn == end_pfn) {
+	} else if (zone_end_pfn(zone) == end_pfn) {
 		/*
 		 * If the section is biggest section in the zone, it need
 		 * shrink zone->spanned_pages.
 		 * In this case, we find second biggest valid mem_section for
 		 * shrinking zone.
 		 */
-		pfn = find_biggest_section_pfn(nid, zone, zone_start_pfn,
+		pfn = find_biggest_section_pfn(nid, zone, zone->zone_start_pfn,
 					       start_pfn);
 		if (pfn)
-			zone->spanned_pages = pfn - zone_start_pfn + 1;
+			zone->spanned_pages = pfn - zone->zone_start_pfn + 1;
 		else {
 			zone->zone_start_pfn = 0;
 			zone->spanned_pages = 0;
Signature · 1 line
2.21.0
#12 lore raw

PATCH v4 8/8 mm/memory_hotplug: Cleanup __remove_pages()

In reply to David Hildenbrand

Let's drop the basically unused section stuff and simplify.

Also, let's use a shorter variant to calculate the number of pages to the next section boundary.

Cc Andrew Morton <[email protected]>
Cc Oscar Salvador <[email protected]>
Cc Michal Hocko <[email protected]>
Cc Pavel Tatashin <[email protected]>
Cc Dan Williams <[email protected]>
Cc Wei Yang <[email protected]>
---
 mm/memory_hotplug.c | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)
mm/memory_hotplug.c +6 -11
index 80cb32cd105e..2b9dad7c5821 100644
@@ -488,25 +488,20 @@ static void __remove_section(unsigned long pfn, unsigned long nr_pages,
 void __remove_pages(unsigned long pfn, unsigned long nr_pages,
 		    struct vmem_altmap *altmap)
 {
+	const unsigned long end_pfn = pfn + nr_pages;
+	unsigned long cur_nr_pages;
 	unsigned long map_offset = 0;
-	unsigned long nr, start_sec, end_sec;
 
 	map_offset = vmem_altmap_offset(altmap);
 
 	if (check_pfn_span(pfn, nr_pages, "remove"))
 		return;
 
-	start_sec = pfn_to_section_nr(pfn);
-	end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
-	for (nr = start_sec; nr <= end_sec; nr++) {
-		unsigned long pfns;
-
+	for (; pfn < end_pfn; pfn += cur_nr_pages) {
 		cond_resched();
-		pfns = min(nr_pages, PAGES_PER_SECTION
-				- (pfn & ~PAGE_SECTION_MASK));
-		__remove_section(pfn, pfns, map_offset, altmap);
-		pfn += pfns;
-		nr_pages -= pfns;
+		/* Select all remaining pages up to the next section boundary */
+		cur_nr_pages = min(end_pfn - pfn, -(pfn | PAGE_SECTION_MASK));
+		__remove_section(pfn, cur_nr_pages, map_offset, altmap);
 		map_offset = 0;
 	}
 }
Signature · 1 line
2.21.0
#13 lore raw
In reply to David Hildenbrand
Quoted 70 lines · David Hildenbrand

This series fixes the access of uninitialized memmaps when shrinking zones/nodes and when removing memory.

We stop trying to shrink ZONE_DEVICE, as it's buggy, fixing it would be more involved (we don't have SECTION_IS_ONLINE as an indicator), and shrinking is only of limited use (set_zone_contiguous() cannot detect the ZONE_DEVICE as contiguous). As far as I can tell, this should be fine for ZONE_DEVICE.

We continue shrinking zones, but I reduced the amount of code to a minimum. Shrinking is especially necessary to keep zone->contiguous set where possible, especially on memory unplug of DIMMs at zone boundaries.

--------------------------------------------------------------------------

Zones are now properly shrunk when offlining memory blocks or when onlining failed. This allows to properly shrink zones on memory unplug even if the separate memory blocks of a DIMM were onlined to different zones or re-onlined to a different zone after offlining.

Example:

:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  0
        present  0
        managed  0
:/# echo "online_movable" > /sys/devices/system/memory/memory41/state
:/# echo "online_movable" > /sys/devices/system/memory/memory43/state
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  98304
        present  65536
        managed  65536
:/# echo 0 > /sys/devices/system/memory/memory43/online
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  32768
        present  32768
        managed  32768
:/# echo 0 > /sys/devices/system/memory/memory41/online
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  0
        present  0
        managed  0

--------------------------------------------------------------------------

I tested this with DIMMs on x86, but didn't test the ZONE_DEVICE part yet.

v3 -> v4:
- Drop "mm/memremap: Get rid of memmap_init_zone_device()"
-- As Alexander noticed, it was messy either way :)
- Drop "mm/memory_hotplug: Exit early in __remove_pages() on BUGs"
- Drop "mm: Exit early in set_zone_contiguous() if already contiguous"
- Drop "mm/memory_hotplug: Optimize zone shrinking code when checking for holes"
- Merged "mm/memory_hotplug: Remove pages from a zone before removing memory" and "mm/memory_hotplug: Remove zone parameter from __remove_pages()" into "mm/memory_hotplug: Shrink zones when offlining memory"
- Added "mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()"
- Stop shrinking ZONE_DEVICE
- Reshuffle patches, moving all fixes to the front. Add Fixes: tags.
- Change subject/description of various patches
- Minor changes (too many to mention)

Quoted 4 lines
Cc Aneesh Kumar K.V <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Dan Williams <[email protected]>
Cc Michal Hocko <[email protected]>

Friendly ping. Can any of the devmem folks verify that this fixes the devmem issues (and not breaks it :) )?

Signature · 3 lines
Thanks,

David / dhildenb
#14 lore raw
In reply to David Hildenbrand
Quoted 89 lines · David Hildenbrand

This series fixes the access of uninitialized memmaps when shrinking zones/nodes and when removing memory.

We stop trying to shrink ZONE_DEVICE, as it's buggy, fixing it would be more involved (we don't have SECTION_IS_ONLINE as an indicator), and shrinking is only of limited use (set_zone_contiguous() cannot detect the ZONE_DEVICE as contiguous). As far as I can tell, this should be fine for ZONE_DEVICE.

We continue shrinking zones, but I reduced the amount of code to a minimum. Shrinking is especially necessary to keep zone->contiguous set where possible, especially on memory unplug of DIMMs at zone boundaries.

--------------------------------------------------------------------------

Zones are now properly shrunk when offlining memory blocks or when onlining failed. This allows to properly shrink zones on memory unplug even if the separate memory blocks of a DIMM were onlined to different zones or re-onlined to a different zone after offlining.

Example:

:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  0
        present  0
        managed  0
:/# echo "online_movable" > /sys/devices/system/memory/memory41/state
:/# echo "online_movable" > /sys/devices/system/memory/memory43/state
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  98304
        present  65536
        managed  65536
:/# echo 0 > /sys/devices/system/memory/memory43/online
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  32768
        present  32768
        managed  32768
:/# echo 0 > /sys/devices/system/memory/memory41/online
:/# cat /proc/zoneinfo
Node 1, zone  Movable
        spanned  0
        present  0
        managed  0

--------------------------------------------------------------------------

I tested this with DIMMs on x86, but didn't test the ZONE_DEVICE part yet.

v3 -> v4:
- Drop "mm/memremap: Get rid of memmap_init_zone_device()"
-- As Alexander noticed, it was messy either way :)
- Drop "mm/memory_hotplug: Exit early in __remove_pages() on BUGs"
- Drop "mm: Exit early in set_zone_contiguous() if already contiguous"
- Drop "mm/memory_hotplug: Optimize zone shrinking code when checking for holes"
- Merged "mm/memory_hotplug: Remove pages from a zone before removing memory" and "mm/memory_hotplug: Remove zone parameter from __remove_pages()" into "mm/memory_hotplug: Shrink zones when offlining memory"
- Added "mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()"
- Stop shrinking ZONE_DEVICE
- Reshuffle patches, moving all fixes to the front. Add Fixes: tags.
- Change subject/description of various patches
- Minor changes (too many to mention)

Cc Aneesh Kumar K.V <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Dan Williams <[email protected]>
Cc Michal Hocko <[email protected]>

David Hildenbrand (8):
mm/memory_hotplug: Don't access uninitialized memmaps in
shrink_pgdat_span()
mm/memory_hotplug: Don't access uninitialized memmaps in
shrink_zone_span()
mm/memory_hotplug: Shrink zones when offlining memory mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone() mm/memory_hotplug: We always have a zone in
find_(smallest|biggest)_section_pfn
mm/memory_hotplug: Don't check for "all holes" in shrink_zone_span() mm/memory_hotplug: Drop local variables in shrink_zone_span() mm/memory_hotplug: Cleanup __remove_pages()

Quoted 12 lines
 arch/arm64/mm/mmu.c            |   4 +-
 arch/ia64/mm/init.c            |   4 +-
 arch/powerpc/mm/mem.c          |   3 +-
 arch/s390/mm/init.c            |   4 +-
 arch/sh/mm/init.c              |   4 +-
 arch/x86/mm/init_32.c          |   4 +-
 arch/x86/mm/init_64.c          |   4 +-
 include/linux/memory_hotplug.h |   7 +-
 mm/memory_hotplug.c            | 184 +++++++++++----------------------
 mm/memremap.c                  |  10 +-
 10 files changed, 80 insertions(+), 148 deletions(-)

@Andrew, if there are no further comments, can we give this some testing?

Signature · 3 lines
Thanks,

David / dhildenb
maintainer
#15 lore raw
Quoted 1 line · On Thu, 19 Sep 2019 15:58:43 +0200 David Hildenbrand <[email protected]> wrote:

@Andrew, if there are no further comments, can we give this some testing?

Yup, it's in my queue for after the current material is merged up.

#16 lore raw
Quoted 6 lines · Andrew Morton
Quoted 1 line · On Thu, 19 Sep 2019 15:58:43 +0200 David Hildenbrand <[email protected]> wrote:

@Andrew, if there are no further comments, can we give this some testing?

Yup, it's in my queue for after the current material is merged up.

Awesome, thanks Andrew!

Signature · 3 lines
Thanks,

David / dhildenb
#17 lore raw

PATCH 1/2 mm/memunmap: Use the correct start and end pfn when removing pages from zone

In reply to David Hildenbrand

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
    pc: c0000000000b58b0: memset+0x68/0x104
    lr: c0000000003eb008: page_init_poison+0x38/0x50
    ...
  current = 0xc000000271c67d80
  paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
    pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50
...

---
 mm/memremap.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)
mm/memremap.c +10 -5
index 390bb3544589..76b98110031e 100644
@@ -113,7 +113,8 @@ static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
 void memunmap_pages(struct dev_pagemap *pgmap)
 {
 	struct resource *res = &pgmap->res;
-	unsigned long pfn = PHYS_PFN(res->start);
+	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, nr_pages;
 	int nid;
 
 	dev_pagemap_kill(pgmap);
@@ -121,14 +122,18 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 		put_page(pfn_to_page(pfn));
 	dev_pagemap_cleanup(pgmap);
 
+	start_pfn = pfn_first(pgmap);
+	end_pfn = pfn_end(pgmap);
+	nr_pages = end_pfn - start_pfn;
+
 	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(pfn));
+	nid = page_to_nid(pfn_to_page(start_pfn));
 
 	mem_hotplug_begin();
-	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
-				   PHYS_PFN(resource_size(res)));
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(start_pfn)),
+				   start_pfn, nr_pages);
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(start_pfn, nr_pages, NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
 				pgmap_altmap(pgmap));
Signature · 1 line
2.21.0
#18 lore raw

PATCH 2/2 mm/memmap_init: Update variable name in memmap_init_zone

The third argument is actually number of pages. Changes the variable name from size to nr_pages to indicate this better.

No functional change in this patch.

---
 mm/page_alloc.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
mm/page_alloc.c +4 -4
index 3334a769eb91..df9e09a5359f 100644
@@ -5914,10 +5914,10 @@ void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
 #ifdef CONFIG_ZONE_DEVICE
 void __ref memmap_init_zone_device(struct zone *zone,
 				   unsigned long start_pfn,
-				   unsigned long size,
+				   unsigned long nr_pages,
 				   struct dev_pagemap *pgmap)
 {
-	unsigned long pfn, end_pfn = start_pfn + size;
+	unsigned long pfn, end_pfn = start_pfn + nr_pages;
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	struct vmem_altmap *altmap = pgmap_altmap(pgmap);
 	unsigned long zone_idx = zone_idx(zone);
@@ -5934,7 +5934,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
 	 */
 	if (altmap) {
 		start_pfn = altmap->base_pfn + vmem_altmap_offset(altmap);
-		size = end_pfn - start_pfn;
+		nr_pages = end_pfn - start_pfn;
 	}
 
 	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
@@ -5981,7 +5981,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
 	}
 
 	pr_info("%s initialised %lu pages in %ums\n", __func__,
-		size, jiffies_to_msecs(jiffies - start));
+		nr_pages, jiffies_to_msecs(jiffies - start));
 }
 
 #endif
Signature · 1 line
2.21.0
author Reviewed-by on v1 2/2
#19 lore raw

Re: [PATCH 2/2] mm/memmap_init: Update variable name in memmap_init_zone

Quoted 38 lines · Aneesh Kumar K.V

The third argument is actually number of pages. Changes the variable name from size to nr_pages to indicate this better.

No functional change in this patch.

---
 mm/page_alloc.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
mm/page_alloc.c +3 -3
index 3334a769eb91..df9e09a5359f 100644
@@ -5914,10 +5914,10 @@ void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
 #ifdef CONFIG_ZONE_DEVICE
 void __ref memmap_init_zone_device(struct zone *zone,
 				   unsigned long start_pfn,
-				   unsigned long size,
+				   unsigned long nr_pages,
 				   struct dev_pagemap *pgmap)
 {
-	unsigned long pfn, end_pfn = start_pfn + size;
+	unsigned long pfn, end_pfn = start_pfn + nr_pages;
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	struct vmem_altmap *altmap = pgmap_altmap(pgmap);
 	unsigned long zone_idx = zone_idx(zone);
@@ -5934,7 +5934,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
 	 */
 	if (altmap) {
 		start_pfn = altmap->base_pfn + vmem_altmap_offset(altmap);
-		size = end_pfn - start_pfn;
+		nr_pages = end_pfn - start_pfn;
 	}
 
 	for (pfn = start_pfn; pfn < end_pfn; pfn++) {

mm/page_alloc.c void __ref memmap_init_zone_device(struct zone *zone, · line 5987

Quoted 8 lines
mm/page_alloc.c +1 -1
@@ -5981,7 +5981,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
 	}
 
 	pr_info("%s initialised %lu pages in %ums\n", __func__,
-		size, jiffies_to_msecs(jiffies - start));
+		nr_pages, jiffies_to_msecs(jiffies - start));
 }
 
 #endif
Signature · 7 lines
Thanks,

David / dhildenb
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm
Reviewed-by on v1 2/2
#20 lore raw

Re: [PATCH 2/2] mm/memmap_init: Update variable name in memmap_init_zone

In reply to Aneesh Kumar K.V
Quoted 50 lines

The third argument is actually number of pages. Changes the variable name from size to nr_pages to indicate this better.

No functional change in this patch.

---
 mm/page_alloc.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
mm/page_alloc.c +2 -2
index 3334a769eb91..df9e09a5359f 100644
@@ -5914,10 +5914,10 @@ void __meminit memmap_init_zone(unsigned long size,
 int nid, unsigned long zone,
 #ifdef CONFIG_ZONE_DEVICE
 void __ref memmap_init_zone_device(struct zone *zone,
 				   unsigned long start_pfn,
-				   unsigned long size,
+				   unsigned long nr_pages,
 				   struct dev_pagemap *pgmap)
 {
-	unsigned long pfn, end_pfn = start_pfn + size;
+	unsigned long pfn, end_pfn = start_pfn + nr_pages;
 	struct pglist_data *pgdat = zone->zone_pgdat;
 	struct vmem_altmap *altmap = pgmap_altmap(pgmap);
mm/page_alloc.c +2 -3
 	unsigned long zone_idx = zone_idx(zone);
@@ -5934,7 +5934,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
 	 */
 	if (altmap) {
 		start_pfn = altmap->base_pfn + vmem_altmap_offset(altmap);
-		size = end_pfn - start_pfn;
+		nr_pages = end_pfn - start_pfn;
 	}
 
 	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
@@ -5981,7 +5981,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
 	}
 
 	pr_info("%s initialised %lu pages in %ums\n", __func__,
-		size, jiffies_to_msecs(jiffies - start));
+		nr_pages, jiffies_to_msecs(jiffies - start));
 }
 
 #endif
--
Quoted 1 line

2.21.0

Quoted 3 lines

_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm

#21 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

In reply to Aneesh Kumar K.V
Quoted 55 lines · Aneesh Kumar K.V

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
    pc: c0000000000b58b0: memset+0x68/0x104
    lr: c0000000003eb008: page_init_poison+0x38/0x50
    ...
  current = 0xc000000271c67d80
  paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
    pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50
...

---
 mm/memremap.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)
mm/memremap.c +7 -2
index 390bb3544589..76b98110031e 100644
@@ -113,7 +113,8 @@ static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
 void memunmap_pages(struct dev_pagemap *pgmap)
 {
 	struct resource *res = &pgmap->res;
-	unsigned long pfn = PHYS_PFN(res->start);
+	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, nr_pages;
 	int nid;
 
 	dev_pagemap_kill(pgmap);
@@ -121,14 +122,18 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 		put_page(pfn_to_page(pfn));
 	dev_pagemap_cleanup(pgmap);
 
+	start_pfn = pfn_first(pgmap);
+	end_pfn = pfn_end(pgmap);
+	nr_pages = end_pfn - start_pfn;
+
 	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(pfn));
+	nid = page_to_nid(pfn_to_page(start_pfn));
 
 	mem_hotplug_begin();

mm/memremap.c void memunmap_pages(struct dev_pagemap *pgmap) · line 139

Quoted 10 lines
mm/memremap.c +3 -3
-	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
-				   PHYS_PFN(resource_size(res)));
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(start_pfn)),
+				   start_pfn, nr_pages);
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(start_pfn, nr_pages, NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
 				pgmap_altmap(pgmap));

Just to make sure, my patches did not break that, right (IOW, broken upstream)?

Signature · 7 lines
Thanks,

David / dhildenb
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm
#22 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 54 lines · David Hildenbrand
Quoted 53 lines · Aneesh Kumar K.V

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
     pc: c0000000000b58b0: memset+0x68/0x104
     lr: c0000000003eb008: page_init_poison+0x38/0x50
     ...
   current = 0xc000000271c67d80
   paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
     pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50
...

---
  mm/memremap.c | 15 ++++++++++-----
  1 file changed, 10 insertions(+), 5 deletions(-)
mm/memremap.c +7 -2
index 390bb3544589..76b98110031e 100644
@@ -113,7 +113,8 @@ static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
  void memunmap_pages(struct dev_pagemap *pgmap)
  {
  	struct resource *res = &pgmap->res;
-	unsigned long pfn = PHYS_PFN(res->start);
+	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, nr_pages;
  	int nid;
 
  	dev_pagemap_kill(pgmap);
@@ -121,14 +122,18 @@ void memunmap_pages(struct dev_pagemap *pgmap)
  		put_page(pfn_to_page(pfn));
  	dev_pagemap_cleanup(pgmap);
 
+	start_pfn = pfn_first(pgmap);
+	end_pfn = pfn_end(pgmap);
+	nr_pages = end_pfn - start_pfn;
+
  	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(pfn));
+	nid = page_to_nid(pfn_to_page(start_pfn));
 
Quoted 16 lines

mm/memremap.c void memunmap_pages(struct dev_pagemap *pgmap) · line 139

Quoted 12 lines
mm/memremap.c +3 -3
  	mem_hotplug_begin();
-	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
-				   PHYS_PFN(resource_size(res)));
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(start_pfn)),
+				   start_pfn, nr_pages);
  	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(start_pfn, nr_pages, NULL);
  	} else {
  		arch_remove_memory(nid, res->start, resource_size(res),
  				pgmap_altmap(pgmap));

Just to make sure, my patches did not break that, right (IOW, broken upstream)?

That is correct. Your patches helped to remove other usages of wrong pfns. The last few left got fixed in this patch.

-aneesh

Reviewed-by on v1 1/2
#23 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

In reply to Aneesh Kumar K.V
Quoted 68 lines

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
    pc: c0000000000b58b0: memset+0x68/0x104
    lr: c0000000003eb008: page_init_poison+0x38/0x50
    ...
  current = 0xc000000271c67d80
  paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
    pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50
...

---
 mm/memremap.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)
mm/memremap.c +2 -1
index 390bb3544589..76b98110031e 100644
@@ -113,7 +113,8 @@ static void dev_pagemap_cleanup(struct dev_pagemap
 *pgmap)
 void memunmap_pages(struct dev_pagemap *pgmap)
 {
 	struct resource *res = &pgmap->res;
-	unsigned long pfn = PHYS_PFN(res->start);
+	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, nr_pages;
 	int nid;
 
mm/memremap.c +8 -5
 	dev_pagemap_kill(pgmap);
@@ -121,14 +122,18 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 		put_page(pfn_to_page(pfn));
 	dev_pagemap_cleanup(pgmap);
 
+	start_pfn = pfn_first(pgmap);
+	end_pfn = pfn_end(pgmap);
+	nr_pages = end_pfn - start_pfn;
+
 	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(pfn));
+	nid = page_to_nid(pfn_to_page(start_pfn));
 
 	mem_hotplug_begin();
-	remove_pfn_range_from_zone(page_zone(pfn_to_page(pfn)), pfn,
-				   PHYS_PFN(resource_size(res)));
+	remove_pfn_range_from_zone(page_zone(pfn_to_page(start_pfn)),
+				   start_pfn, nr_pages);
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		__remove_pages(pfn, PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(start_pfn, nr_pages, NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
 				pgmap_altmap(pgmap));
--
Quoted 1 line

2.21.0

Quoted 3 lines

_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm

maintainer
#24 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

In reply to Aneesh Kumar K.V
Quoted 16 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
    pc: c0000000000b58b0: memset+0x68/0x104
    lr: c0000000003eb008: page_init_poison+0x38/0x50
    ...
  current = 0xc000000271c67d80
  paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
    pid   = 3665, comm = ndctl
Quoted 4 lines

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Doesn't apply to mainline or -next. Which tree is this against?
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm

#25 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 23 lines · Andrew Morton
Quoted 16 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
     pc: c0000000000b58b0: memset+0x68/0x104
     lr: c0000000003eb008: page_init_poison+0x38/0x50
     ...
   current = 0xc000000271c67d80
   paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
     pid   = 3665, comm = ndctl
Quoted 4 lines

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Quoted 2 lines

Doesn't apply to mainline or -next. Which tree is this against?

After applying the patches from David on mainline. That is the reason I replied to this thread. I should have mentioned in the email that it is based on patch series "[PATCH v4 0/8] mm/memory_hotplug: Shrink zones before removing memory"

-aneesh

#26 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 24 lines · Aneesh Kumar K.V
Quoted 23 lines · Andrew Morton
Quoted 20 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
     pc: c0000000000b58b0: memset+0x68/0x104
     lr: c0000000003eb008: page_init_poison+0x38/0x50
     ...
   current = 0xc000000271c67d80
   paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
     pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Quoted 7 lines
Quoted 2 lines

Doesn't apply to mainline or -next. Which tree is this against?

After applying the patches from David on mainline. That is the reason I replied to this thread. I should have mentioned in the email that it is based on patch series "[PATCH v4 0/8] mm/memory_hotplug: Shrink zones before removing memory"

So if I am not wrong, my patch "[PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()" makes it show up that we actually call _remove_pages() with wrong parameters, right?

If so, I guess it would be better for you to fix it before my series and I will rebase my series on top of that.

Signature · 3 lines
Thanks,

David / dhildenb
#27 lore raw

PATCH 1/1 mm/memunmap: Use the correct start and end pfn when removing pages from zone

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   81.356173] kernel BUG at include/linux/mm.h:1107!
cpu 0x1: Vector: 700 (Program Check) at [c000000274087890]
    pc: c0000000004b9728: memunmap_pages+0x238/0x340
    lr: c0000000004b9724: memunmap_pages+0x234/0x340
...
    pid   = 3669, comm = ndctl

kernel BUG at include/linux/mm.h:1107!
[c000000274087ba0] c0000000009e3500 devm_action_release+0x30/0x50 [c000000274087bc0] c0000000009e4758 release_nodes+0x268/0x2d0 [c000000274087c30] c0000000009dd144 device_release_driver_internal+0x174/0x240 [c000000274087c70] c0000000009d9dfc unbind_store+0x13c/0x190 [c000000274087cb0] c0000000009d8a24 drv_attr_store+0x44/0x60 [c000000274087cd0] c0000000005a7470 sysfs_kf_write+0x70/0xa0 [c000000274087d10] c0000000005a5cac kernfs_fop_write+0x1ac/0x290 [c000000274087d60] c0000000004be45c __vfs_write+0x3c/0x70 [c000000274087d80] c0000000004c26e4 vfs_write+0xe4/0x200 [c000000274087dd0] c0000000004c2a6c ksys_write+0x7c/0x140 [c000000274087e20] c00000000000bbd0 system_call+0x5c/0x68

---
Note:
This patch alone won't fix all the kernel crashes related wrong usage of pfn. For ndctl destroy-namespace to work correctly we need rest of patches from the series posted at

https://lore.kernel.org/linux-mm/[email protected]

 mm/memremap.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
mm/memremap.c +9 -5
index 32c79b51af86..4b31f0b7c42d 100644
@@ -105,7 +105,8 @@ static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
 void memunmap_pages(struct dev_pagemap *pgmap)
 {
 	struct resource *res = &pgmap->res;
-	unsigned long pfn;
+	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, nr_pages;
 	int nid;
 
 	dev_pagemap_kill(pgmap);
@@ -113,14 +114,17 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 		put_page(pfn_to_page(pfn));
 	dev_pagemap_cleanup(pgmap);
 
+	start_pfn = pfn_first(pgmap);
+	end_pfn = pfn_end(pgmap);
+	nr_pages = end_pfn - start_pfn;
+
 	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(PHYS_PFN(res->start)));
+	nid = page_to_nid(pfn_to_page(start_pfn));
 
 	mem_hotplug_begin();
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		pfn = PHYS_PFN(res->start);
-		__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
-				 PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(page_zone(pfn_to_page(start_pfn)), start_pfn,
+			       nr_pages, NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
 				pgmap_altmap(pgmap));
Signature · 1 line
2.21.0
#28 lore raw

Re: [PATCH] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 43 lines · Aneesh Kumar K.V

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   81.356173] kernel BUG at include/linux/mm.h:1107!
cpu 0x1: Vector: 700 (Program Check) at [c000000274087890]
    pc: c0000000004b9728: memunmap_pages+0x238/0x340
    lr: c0000000004b9724: memunmap_pages+0x234/0x340
...
    pid   = 3669, comm = ndctl

kernel BUG at include/linux/mm.h:1107!
[c000000274087ba0] c0000000009e3500 devm_action_release+0x30/0x50 [c000000274087bc0] c0000000009e4758 release_nodes+0x268/0x2d0 [c000000274087c30] c0000000009dd144 device_release_driver_internal+0x174/0x240 [c000000274087c70] c0000000009d9dfc unbind_store+0x13c/0x190 [c000000274087cb0] c0000000009d8a24 drv_attr_store+0x44/0x60 [c000000274087cd0] c0000000005a7470 sysfs_kf_write+0x70/0xa0 [c000000274087d10] c0000000005a5cac kernfs_fop_write+0x1ac/0x290 [c000000274087d60] c0000000004be45c __vfs_write+0x3c/0x70 [c000000274087d80] c0000000004c26e4 vfs_write+0xe4/0x200 [c000000274087dd0] c0000000004c2a6c ksys_write+0x7c/0x140 [c000000274087e20] c00000000000bbd0 system_call+0x5c/0x68

---
Note:
This patch alone won't fix all the kernel crashes related wrong usage of pfn. For ndctl destroy-namespace to work correctly we need rest of patches from the series posted at

https://lore.kernel.org/linux-mm/[email protected]

 mm/memremap.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
mm/memremap.c +0 -0
index 32c79b51af86..4b31f0b7c42d 100644

mm/memremap.c static void dev_pagemap_cleanup(struct dev_pagemap *pgmap) · line 109

Quoted 6 lines
mm/memremap.c +2 -1
@@ -105,7 +105,8 @@ static void dev_pagemap_cleanup(struct dev_pagemap *pgmap)
 void memunmap_pages(struct dev_pagemap *pgmap)
 {
 	struct resource *res = &pgmap->res;
-	unsigned long pfn;
+	unsigned long start_pfn, end_pfn;
+	unsigned long pfn, nr_pages;

pack all into a single line?

Quoted 16 lines
mm/memremap.c +5 -1
 	int nid;
 
 	dev_pagemap_kill(pgmap);
@@ -113,14 +114,17 @@ void memunmap_pages(struct dev_pagemap *pgmap)
 		put_page(pfn_to_page(pfn));
 	dev_pagemap_cleanup(pgmap);
 
+	start_pfn = pfn_first(pgmap);
+	end_pfn = pfn_end(pgmap);
+	nr_pages = end_pfn - start_pfn;
+
 	/* pages are dead and unused, undo the arch mapping */
-	nid = page_to_nid(pfn_to_page(PHYS_PFN(res->start)));
+	nid = page_to_nid(pfn_to_page(start_pfn));
 

mm/memremap.c void memunmap_pages(struct dev_pagemap *pgmap) · line 130

Quoted 10 lines
mm/memremap.c +2 -3
 	mem_hotplug_begin();
 	if (pgmap->type == MEMORY_DEVICE_PRIVATE) {
-		pfn = PHYS_PFN(res->start);
-		__remove_pages(page_zone(pfn_to_page(pfn)), pfn,
-				 PHYS_PFN(resource_size(res)), NULL);
+		__remove_pages(page_zone(pfn_to_page(start_pfn)), start_pfn,
+			       nr_pages, NULL);
 	} else {
 		arch_remove_memory(nid, res->start, resource_size(res),
 				pgmap_altmap(pgmap));

I am no expert on the details of the whole pgmap, what-to-add-what-to-initialize-whatsoever devmem specific stuff, but this change looks sane to me as far as I can tell.

Signature · 3 lines
Thanks,

David / dhildenb
#29 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

In reply to David Hildenbrand
Quoted 33 lines · David Hildenbrand
Quoted 24 lines · Aneesh Kumar K.V
Quoted 23 lines · Andrew Morton
Quoted 20 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
      pc: c0000000000b58b0: memset+0x68/0x104
      lr: c0000000003eb008: page_init_poison+0x38/0x50
      ...
    current = 0xc000000271c67d80
    paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
      pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Quoted 7 lines
Quoted 2 lines

Doesn't apply to mainline or -next. Which tree is this against?

After applying the patches from David on mainline. That is the reason I replied to this thread. I should have mentioned in the email that it is based on patch series "[PATCH v4 0/8] mm/memory_hotplug: Shrink zones before removing memory"

Quoted 7 lines

So if I am not wrong, my patch "[PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()" makes it show up that we actually call _remove_pages() with wrong parameters, right?

If so, I guess it would be better for you to fix it before my series and I will rebase my series on top of that.

I posted a patch that can be applied to mainline. I sent that as a reply to this email. Can you include that and PATCH 2 as first two patches in your series? That should help to locate the full patch series needed for fixing the kernel crash.

-aneesh

#30 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 38 lines · Aneesh Kumar K.V
Quoted 37 lines · David Hildenbrand
Quoted 24 lines · Aneesh Kumar K.V
Quoted 23 lines · Andrew Morton
Quoted 20 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
      pc: c0000000000b58b0: memset+0x68/0x104
      lr: c0000000003eb008: page_init_poison+0x38/0x50
      ...
    current = 0xc000000271c67d80
    paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
      pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Quoted 7 lines
Quoted 2 lines

Doesn't apply to mainline or -next. Which tree is this against?

After applying the patches from David on mainline. That is the reason I replied to this thread. I should have mentioned in the email that it is based on patch series "[PATCH v4 0/8] mm/memory_hotplug: Shrink zones before removing memory"

So if I am not wrong, my patch "[PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()" makes it show up that we actually call _remove_pages() with wrong parameters, right?

Quoted 8 lines
Quoted 3 lines

If so, I guess it would be better for you to fix it before my series and I will rebase my series on top of that.

I posted a patch that can be applied to mainline. I sent that as a reply to this email. Can you include that and PATCH 2 as first two patches in your series? That should help to locate the full patch series needed for fixing the kernel crash.

I can drag these along, unless Andrew wants to pick them up right away (or we're waiting for more feedback).

Is there a Fixes: Tag we can add to the first patch?

Signature · 7 lines
Thanks,

David / dhildenb
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm
#31 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 43 lines · David Hildenbrand
Quoted 42 lines · Aneesh Kumar K.V
Quoted 40 lines · David Hildenbrand
Quoted 24 lines · Aneesh Kumar K.V
Quoted 23 lines · Andrew Morton
Quoted 20 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
       pc: c0000000000b58b0: memset+0x68/0x104
       lr: c0000000003eb008: page_init_poison+0x38/0x50
       ...
     current = 0xc000000271c67d80
     paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
       pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Quoted 7 lines
Quoted 2 lines

Doesn't apply to mainline or -next. Which tree is this against?

After applying the patches from David on mainline. That is the reason I replied to this thread. I should have mentioned in the email that it is based on patch series "[PATCH v4 0/8] mm/memory_hotplug: Shrink zones before removing memory"

So if I am not wrong, my patch "[PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()" makes it show up that we actually call _remove_pages() with wrong parameters, right?

If so, I guess it would be better for you to fix it before my series and I will rebase my series on top of that.

Quoted 7 lines
Quoted 4 lines

I posted a patch that can be applied to mainline. I sent that as a reply to this email. Can you include that and PATCH 2 as first two patches in your series? That should help to locate the full patch series needed for fixing the kernel crash.

I can drag these along, unless Andrew wants to pick them up right away (or we're waiting for more feedback).

Considering this patch alone won't fix the issue, It would be nice if we could club them with rest of the changes.

Quoted 3 lines

Is there a Fixes: Tag we can add to the first patch?

IIUC this was always broken.

-aneesh

#32 lore raw

Re: [PATCH 1/2] mm/memunmap: Use the correct start and end pfn when removing pages from zone

Quoted 49 lines · Aneesh Kumar K.V
Quoted 48 lines · David Hildenbrand
Quoted 46 lines · Aneesh Kumar K.V
Quoted 33 lines · David Hildenbrand
Quoted 24 lines · Aneesh Kumar K.V
Quoted 23 lines · Andrew Morton
Quoted 20 lines · On Thu, 26 Sep 2019 17:55:51 +0530 "Aneesh Kumar K.V"…

With altmap, all the resource pfns are not initialized. While initializing pfn, altmap reserve space is skipped. Hence when removing pfn from zone skip pfns that were never initialized.

Update memunmap_pages to calculate start and end pfn based on altmap values. This fixes a kernel crash that is observed when destroying namespace.

[   74.745056] BUG: Unable to handle kernel data access at 0xc00c000001400000
[   74.745256] Faulting instruction address: 0xc0000000000b58b0
cpu 0x2: Vector: 300 (Data Access) at [c00000026ea93580]
       pc: c0000000000b58b0: memset+0x68/0x104
       lr: c0000000003eb008: page_init_poison+0x38/0x50
       ...
     current = 0xc000000271c67d80
     paca    = 0xc00000003fffd680   irqmask: 0x03   irq_happened: 0x01
       pid   = 3665, comm = ndctl

[link register ] c0000000003eb008 page_init_poison+0x38/0x50
[c00000026ea93830] c0000000004754d4 remove_pfn_range_from_zone+0x64/0x3e0 [c00000026ea938a0] c0000000004b8a60 memunmap_pages+0x300/0x400 [c00000026ea93930] c0000000009e32a0 devm_action_release+0x30/0x50

Quoted 7 lines
Quoted 2 lines

Doesn't apply to mainline or -next. Which tree is this against?

After applying the patches from David on mainline. That is the reason I replied to this thread. I should have mentioned in the email that it is based on patch series "[PATCH v4 0/8] mm/memory_hotplug: Shrink zones before removing memory"

Quoted 7 lines

So if I am not wrong, my patch "[PATCH v4 4/8] mm/memory_hotplug: Poison memmap in remove_pfn_range_from_zone()" makes it show up that we actually call _remove_pages() with wrong parameters, right?

If so, I guess it would be better for you to fix it before my series and I will rebase my series on top of that.

I posted a patch that can be applied to mainline. I sent that as a reply to this email. Can you include that and PATCH 2 as first two patches in your series? That should help to locate the full patch series needed for fixing the kernel crash.

Quoted 6 lines
Quoted 2 lines

I can drag these along, unless Andrew wants to pick them up right away (or we're waiting for more feedback).

Considering this patch alone won't fix the issue, It would be nice if we could club them with rest of the changes.

I'll drag them along, adding Pankaj's RB's. If they get picked up independently, fine :)

Signature · 7 lines
Thanks,

David / dhildenb
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm