Skip to contentSkip to conversation

mm/memory_hotplug: build zonelist for managed_zone

PATCH In review Wei Yang 3 participants 9 messages Jan 27, 2022 – Feb 1, 2022

Subsystems (2)
lore
Share & tools
lore

Wei Yang

author
#0 lore raw

PATCH 1/1 mm/memory_hotplug: build zonelist for managed_zone

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Fixes 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")
CC Mel Gorman <[email protected]>
---
 mm/memory_hotplug.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +3 -3
index 2a9627dc784c..8f1906b33937 100644
@@ -1102,11 +1102,11 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	spin_unlock_irqrestore(&zone->lock, flags);
 
 	/*
-	 * If this zone is not populated, then it is not in zonelist.
+	 * If this zone is not managed, then it is not in zonelist.
 	 * This means the page allocator ignores this zone.
 	 * So, zonelist must be updated after online.
 	 */
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		need_zonelists_rebuild = 1;
 		setup_zone_pageset(zone);
 	}
@@ -1985,7 +1985,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
 	/* reinitialise watermarks and update pcp limits */
 	init_per_zone_wmark_min();
 
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		zone_pcp_reset(zone);
 		build_all_zonelists(NULL);
 	}
Signature · 1 line
2.33.1
#1 lore raw
Quoted 16 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

Quoted 7 lines

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Why?

Quoted 36 lines
Fixes 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")
CC Mel Gorman <[email protected]>
---
 mm/memory_hotplug.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +3 -3
index 2a9627dc784c..8f1906b33937 100644
@@ -1102,11 +1102,11 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	spin_unlock_irqrestore(&zone->lock, flags);
 
 	/*
-	 * If this zone is not populated, then it is not in zonelist.
+	 * If this zone is not managed, then it is not in zonelist.
 	 * This means the page allocator ignores this zone.
 	 * So, zonelist must be updated after online.
 	 */
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		need_zonelists_rebuild = 1;
 		setup_zone_pageset(zone);
 	}
@@ -1985,7 +1985,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
 	/* reinitialise watermarks and update pcp limits */
 	init_per_zone_wmark_min();
 
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		zone_pcp_reset(zone);
 		build_all_zonelists(NULL);
 	}
Signature · 1 line
2.33.1
Signature · 2 lines
Michal Hocko
SUSE Labs
author
#2 lore raw
Quoted 22 lines · Michal Hocko
Quoted 21 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

Quoted 5 lines
Quoted 2 lines

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Why?

In case we online a populated zone, but not managed. Then this zone will not be in zonelist. Right?

Signature · 2 lines
Wei Yang
Help you, Help me
#3 lore raw
Quoted 23 lines · Wei Yang
Quoted 22 lines · Michal Hocko
Quoted 21 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Why?

In case we online a populated zone, but not managed. Then this zone will not be in zonelist. Right?

yeah. We can theoretically end up with a zone without any managed menory on the zonelists. But my primary question is why do we need this change? Does it fix any existing problem? Does it make the code easier to read/understand? Does it improve performance? Every patch should have a justification. Your changelog merely lists the history and then states the new behavior without any explanation of why that is needed or desired. See?

Signature · 2 lines
Michal Hocko
SUSE Labs
author
#4 lore raw
Quoted 11 lines · Michal Hocko
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Why?

In case we online a populated zone, but not managed. Then this zone will not be in zonelist. Right?

yeah. We can theoretically end up with a zone without any managed menory on the zonelists. But my primary question is why do we need this change?

Currently I don't see a real case for this.

Quoted 5 lines

Does it fix any existing problem? Does it make the code easier to read/understand? Does it improve performance? Every patch should have a justification. Your changelog merely lists the history and then states the new behavior without any explanation of why that is needed or desired. See?

I don't have such reason now.

Quoted 3 lines
Signature · 2 lines
Michal Hocko
SUSE Labs
Signature · 2 lines
Wei Yang
Help you, Help me
#5 lore raw
In reply to Wei Yang
Quoted 50 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Fixes 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")
CC Mel Gorman <[email protected]>
---
 mm/memory_hotplug.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +2 -2
index 2a9627dc784c..8f1906b33937 100644
@@ -1102,11 +1102,11 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	spin_unlock_irqrestore(&zone->lock, flags);
 
 	/*
-	 * If this zone is not populated, then it is not in zonelist.
+	 * If this zone is not managed, then it is not in zonelist.
 	 * This means the page allocator ignores this zone.
 	 * So, zonelist must be updated after online.
 	 */
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		need_zonelists_rebuild = 1;
 		setup_zone_pageset(zone);
 	}

mm/memory_hotplug.c int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages, · line 1991

Quoted 8 lines
mm/memory_hotplug.c +1 -1
@@ -1985,7 +1985,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
 	/* reinitialise watermarks and update pcp limits */
 	init_per_zone_wmark_min();
 
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		zone_pcp_reset(zone);
 		build_all_zonelists(NULL);
 	}

A note that managed_zone() is a moving target w.r.t. memory ballooning. In extreme cases, we can have whole zones (temporarily) be completely !managed for that reason.

IMHO memory hot(un)plug is usually the wrong place to check for managed_zone(), it cares about populated_zone().

Signature · 3 lines
Thanks,

David / dhildenb
author
#6 lore raw
Quoted 64 lines · David Hildenbrand
Quoted 50 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Fixes 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")
CC Mel Gorman <[email protected]>
---
 mm/memory_hotplug.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +2 -2
index 2a9627dc784c..8f1906b33937 100644
@@ -1102,11 +1102,11 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	spin_unlock_irqrestore(&zone->lock, flags);
 
 	/*
-	 * If this zone is not populated, then it is not in zonelist.
+	 * If this zone is not managed, then it is not in zonelist.
 	 * This means the page allocator ignores this zone.
 	 * So, zonelist must be updated after online.
 	 */
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		need_zonelists_rebuild = 1;
 		setup_zone_pageset(zone);
 	}

mm/memory_hotplug.c int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages, · line 1991

Quoted 8 lines
mm/memory_hotplug.c +1 -1
@@ -1985,7 +1985,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
 	/* reinitialise watermarks and update pcp limits */
 	init_per_zone_wmark_min();
 
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		zone_pcp_reset(zone);
 		build_all_zonelists(NULL);
 	}

A note that managed_zone() is a moving target w.r.t. memory ballooning. In extreme cases, we can have whole zones (temporarily) be completely !managed for that reason.

Quoted 3 lines

IMHO memory hot(un)plug is usually the wrong place to check for managed_zone(), it cares about populated_zone().

So we need to check populated_zone when building zonelist?

Quoted 4 lines
Signature · 3 lines
Thanks,

David / dhildenb
Signature · 2 lines
Wei Yang
Help you, Help me
#7 lore raw
Quoted 65 lines · Wei Yang
Quoted 64 lines · David Hildenbrand
Quoted 50 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Fixes 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")
CC Mel Gorman <[email protected]>
---
 mm/memory_hotplug.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +2 -2
index 2a9627dc784c..8f1906b33937 100644
@@ -1102,11 +1102,11 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	spin_unlock_irqrestore(&zone->lock, flags);
 
 	/*
-	 * If this zone is not populated, then it is not in zonelist.
+	 * If this zone is not managed, then it is not in zonelist.
 	 * This means the page allocator ignores this zone.
 	 * So, zonelist must be updated after online.
 	 */
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		need_zonelists_rebuild = 1;
 		setup_zone_pageset(zone);
 	}

mm/memory_hotplug.c int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages, · line 1991

Quoted 8 lines
mm/memory_hotplug.c +1 -1
@@ -1985,7 +1985,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
 	/* reinitialise watermarks and update pcp limits */
 	init_per_zone_wmark_min();
 
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		zone_pcp_reset(zone);
 		build_all_zonelists(NULL);
 	}

A note that managed_zone() is a moving target w.r.t. memory ballooning. In extreme cases, we can have whole zones (temporarily) be completely !managed for that reason.

Quoted 5 lines
Quoted 3 lines

IMHO memory hot(un)plug is usually the wrong place to check for managed_zone(), it cares about populated_zone().

So we need to check populated_zone when building zonelist?

I think the issue is that

a) We can have zones without any managed pages put present page during boot, for example, if all pages in the zone were allocated via memblock.

b) We can have zones without any managed pages temporarily -- extreme cases of memory ballooning / virtio-mem.

I tend to think that populated_zone() might actually be the right check whenever building a zonelist. Because even in case of a) we might end up freeing a memblock allocation later, suddenly having free pages in such a zone, but the zone not in any zonelist.

Signature · 3 lines
Thanks,

David / dhildenb
author
#8 lore raw
Quoted 77 lines · David Hildenbrand
Quoted 65 lines · Wei Yang
Quoted 64 lines · David Hildenbrand
Quoted 50 lines · Wei Yang

During memory hotplug, when online/offline a zone, we need to rebuild the zonelist for all node. There are two checks to decide whether a zone would be added to zonelist:

* one in online_pages/offline_pages to decide necessity
* one in build_zonerefs_node to do real add

Currently we use different criteria at these two places, which is different from the original behavior.

Originally during memory hotplug, zonelist is re-built when zone hasn't been populated. This in introduced in 'commit 6811378e7d8b ("[PATCH] wait_table and zonelist initializing for memory hotadd: update zonelists")'. And at that moment, build_zonelists_node() also use populated_zone() to decide whether the zone should be added to zonelist.

While in 'commit 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")', build_zonelists_node() changed to use managed_zone() to add zonelist. But we still use populated_zone() to decide the necessity.

This patch restore the original behavior by using the same criteria to add a zone in zonelist during memory hotplug.

Fixes 6aa303defb74 ("mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator")
CC Mel Gorman <[email protected]>
---
 mm/memory_hotplug.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
mm/memory_hotplug.c +2 -2
index 2a9627dc784c..8f1906b33937 100644
@@ -1102,11 +1102,11 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
 	spin_unlock_irqrestore(&zone->lock, flags);
 
 	/*
-	 * If this zone is not populated, then it is not in zonelist.
+	 * If this zone is not managed, then it is not in zonelist.
 	 * This means the page allocator ignores this zone.
 	 * So, zonelist must be updated after online.
 	 */
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		need_zonelists_rebuild = 1;
 		setup_zone_pageset(zone);
 	}

mm/memory_hotplug.c int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages, · line 1991

Quoted 8 lines
mm/memory_hotplug.c +1 -1
@@ -1985,7 +1985,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
 	/* reinitialise watermarks and update pcp limits */
 	init_per_zone_wmark_min();
 
-	if (!populated_zone(zone)) {
+	if (!managed_zone(zone)) {
 		zone_pcp_reset(zone);
 		build_all_zonelists(NULL);
 	}

A note that managed_zone() is a moving target w.r.t. memory ballooning. In extreme cases, we can have whole zones (temporarily) be completely !managed for that reason.

Quoted 5 lines
Quoted 3 lines

IMHO memory hot(un)plug is usually the wrong place to check for managed_zone(), it cares about populated_zone().

So we need to check populated_zone when building zonelist?

I think the issue is that

a) We can have zones without any managed pages put present page during boot, for example, if all pages in the zone were allocated via memblock.

Quoted 8 lines

b) We can have zones without any managed pages temporarily -- extreme cases of memory ballooning / virtio-mem.

I tend to think that populated_zone() might actually be the right check whenever building a zonelist. Because even in case of a) we might end up freeing a memblock allocation later, suddenly having free pages in such a zone, but the zone not in any zonelist.

I agree with you for this point.

Quoted 5 lines
Signature · 4 lines
-- 
Thanks,

David / dhildenb
Signature · 2 lines
Wei Yang
Help you, Help me