Re: [PATCH 5/6] mm, memory_hotplug: do not associate hotadded memory to zones until online

[Michal Hocko <mhocko@kernel.org>]

On Thu 30-03-17 13:54:53, Michal Hocko wrote:
[...]
> -static int __meminit __add_section(int nid, struct zone *zone,
> -					unsigned long phys_start_pfn)
> +static int __meminit __add_section(int nid, unsigned long phys_start_pfn)
>  {
>  	int ret;
> +	int i;
>  
>  	if (pfn_valid(phys_start_pfn))
>  		return -EEXIST;
>  
> -	ret = sparse_add_one_section(zone, phys_start_pfn);
> -
> +	ret = sparse_add_one_section(NODE_DATA(nid), phys_start_pfn);
>  	if (ret < 0)
>  		return ret;
>  
> -	ret = __add_zone(zone, phys_start_pfn);
> +	/*
> +	 * Make all the pages reserved so that nobody will stumble over half
> +	 * initialized state.
> +	 */
> +	for (i = 0; i < PAGES_PER_SECTION; i++) {
> +		unsigned long pfn = phys_start_pfn + i;
> +		if (!pfn_valid(pfn))
> +			continue;
>  
> -	if (ret < 0)
> -		return ret;
> +		SetPageReserved(pfn_to_page(phys_start_pfn + i));
> +	}
>  
>  	return register_new_memory(nid, __pfn_to_section(phys_start_pfn));

I have just realized one more dependency on the zone initialization.
register_new_memory relies on is_zone_device_section to rule out
memblock specific operations including sysfs infrastructure. I have come
up with the following to handle this.
---
>From 377eb5691706de179fbf570915c1cf365d3253c0 Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.com>
Date: Thu, 6 Apr 2017 09:50:35 +0200
Subject: [PATCH] mm, memory_hotplug: get rid of is_zone_device_section

device memory hotplug hooks into regular memory hotplug only half way.
It needs memory sections to track struct pages but there is no
need/desire to associate those sections with memory blocks and export
them to the userspace via sysfs because they cannot be onlined anyway.

This is currently expressed by for_device argument to arch_add_memory
which then makes sure to associate the given memory range with
ZONE_DEVICE. register_new_memory then relies on is_zone_device_section
to distinguish special memory hotplug from the regular one. While this
works now, later patches in this series want to move __add_zone outside
of arch_add_memory path so we have to come up with something else.

Add want_memblock down the __add_pages path and use it to control
whether the section->memblock association should be done. arch_add_memory
the just trivially want memblock for everything but for_device hotplug.

remove_memory_section doesn't need is_zone_device_section either. We can
simply skip all the memblock specific cleanup if there is no memblock
for the given section.

Cc: Dan Williams <dan.j.williams@gmail.com>
Signed-off-by: Michal Hocko <mhocko@suse.com>
---
 arch/ia64/mm/init.c            |  2 +-
 arch/powerpc/mm/mem.c          |  2 +-
 arch/s390/mm/init.c            |  2 +-
 arch/sh/mm/init.c              |  2 +-
 arch/x86/mm/init_32.c          |  2 +-
 arch/x86/mm/init_64.c          |  2 +-
 drivers/base/memory.c          | 22 ++++++++--------------
 include/linux/memory_hotplug.h |  2 +-
 mm/memory_hotplug.c            | 11 +++++++----
 9 files changed, 22 insertions(+), 25 deletions(-)

diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index 06cdaef54b2e..62085fd902e6 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -657,7 +657,7 @@ int arch_add_memory(int nid, u64 start, u64 size, bool for_device)
 
 	zone = pgdat->node_zones +
 		zone_for_memory(nid, start, size, ZONE_NORMAL, for_device);
-	ret = __add_pages(nid, zone, start_pfn, nr_pages);
+	ret = __add_pages(nid, zone, start_pfn, nr_pages, !for_device);
 
 	if (ret)
 		printk("%s: Problem encountered in __add_pages() as ret=%d\n",
diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index 5f844337de21..ea3e09a62f38 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -149,7 +149,7 @@ int arch_add_memory(int nid, u64 start, u64 size, bool for_device)
 	zone = pgdata->node_zones +
 		zone_for_memory(nid, start, size, 0, for_device);
 
-	return __add_pages(nid, zone, start_pfn, nr_pages);
+	return __add_pages(nid, zone, start_pfn, nr_pages, !for_device);
 }
 
 #ifdef CONFIG_MEMORY_HOTREMOVE
diff --git a/arch/s390/mm/init.c b/arch/s390/mm/init.c
index bf5b8a0c4ff7..5c84346e5211 100644
--- a/arch/s390/mm/init.c
+++ b/arch/s390/mm/init.c
@@ -182,7 +182,7 @@ int arch_add_memory(int nid, u64 start, u64 size, bool for_device)
 			continue;
 		nr_pages = (start_pfn + size_pages > zone_end_pfn) ?
 			   zone_end_pfn - start_pfn : size_pages;
-		rc = __add_pages(nid, zone, start_pfn, nr_pages);
+		rc = __add_pages(nid, zone, start_pfn, nr_pages, !for_device);
 		if (rc)
 			break;
 		start_pfn += nr_pages;
diff --git a/arch/sh/mm/init.c b/arch/sh/mm/init.c
index 75491862d900..a9d57f75ae8c 100644
--- a/arch/sh/mm/init.c
+++ b/arch/sh/mm/init.c
@@ -498,7 +498,7 @@ int arch_add_memory(int nid, u64 start, u64 size, bool for_device)
 	ret = __add_pages(nid, pgdat->node_zones +
 			zone_for_memory(nid, start, size, ZONE_NORMAL,
 			for_device),
-			start_pfn, nr_pages);
+			start_pfn, nr_pages, !for_device);
 	if (unlikely(ret))
 		printk("%s: Failed, __add_pages() == %d\n", __func__, ret);
 
diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c
index 928cfde76232..bc5530cc0746 100644
--- a/arch/x86/mm/init_32.c
+++ b/arch/x86/mm/init_32.c
@@ -824,7 +824,7 @@ int arch_add_memory(int nid, u64 start, u64 size, bool for_device)
 	unsigned long start_pfn = start >> PAGE_SHIFT;
 	unsigned long nr_pages = size >> PAGE_SHIFT;
 
-	return __add_pages(nid, zone, start_pfn, nr_pages);
+	return __add_pages(nid, zone, start_pfn, nr_pages, !for_device);
 }
 
 #ifdef CONFIG_MEMORY_HOTREMOVE
diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 7eef17239378..39cfaee93975 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -652,7 +652,7 @@ int arch_add_memory(int nid, u64 start, u64 size, bool for_device)
 
 	init_memory_mapping(start, start + size);
 
-	ret = __add_pages(nid, zone, start_pfn, nr_pages);
+	ret = __add_pages(nid, zone, start_pfn, nr_pages, !for_device);
 	WARN_ON_ONCE(ret);
 
 	/* update max_pfn, max_low_pfn and high_memory */
diff --git a/drivers/base/memory.c b/drivers/base/memory.c
index cc4f1d0cbffe..89c15e942852 100644
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -685,14 +685,6 @@ static int add_memory_block(int base_section_nr)
 	return 0;
 }
 
-static bool is_zone_device_section(struct mem_section *ms)
-{
-	struct page *page;
-
-	page = sparse_decode_mem_map(ms->section_mem_map, __section_nr(ms));
-	return is_zone_device_page(page);
-}
-
 /*
  * need an interface for the VM to add new memory regions,
  * but without onlining it.
@@ -702,9 +694,6 @@ int register_new_memory(int nid, struct mem_section *section)
 	int ret = 0;
 	struct memory_block *mem;
 
-	if (is_zone_device_section(section))
-		return 0;
-
 	mutex_lock(&mem_sysfs_mutex);
 
 	mem = find_memory_block(section);
@@ -741,11 +730,16 @@ static int remove_memory_section(unsigned long node_id,
 {
 	struct memory_block *mem;
 
-	if (is_zone_device_section(section))
-		return 0;
-
 	mutex_lock(&mem_sysfs_mutex);
+
+	/*
+	 * Some users of the memory hotplug do not want/need memblock to
+	 * track all sections. Skip over those.
+	 */
 	mem = find_memory_block(section);
+	if (!mem)
+		return 0;
+
 	unregister_mem_sect_under_nodes(mem, __section_nr(section));
 
 	mem->section_count--;
diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
index 134a2f69c21a..3c8cf86201c3 100644
--- a/include/linux/memory_hotplug.h
+++ b/include/linux/memory_hotplug.h
@@ -111,7 +111,7 @@ extern int __remove_pages(struct zone *zone, unsigned long start_pfn,
 
 /* reasonably generic interface to expand the physical pages in a zone  */
 extern int __add_pages(int nid, struct zone *zone, unsigned long start_pfn,
-	unsigned long nr_pages);
+	unsigned long nr_pages, bool want_memblock);
 
 #ifdef CONFIG_NUMA
 extern int memory_add_physaddr_to_nid(u64 start);
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 28b16a831d40..5a8924a41a3b 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -489,7 +489,7 @@ static int __meminit __add_zone(struct zone *zone, unsigned long phys_start_pfn)
 }
 
 static int __meminit __add_section(int nid, struct zone *zone,
-					unsigned long phys_start_pfn)
+					unsigned long phys_start_pfn, bool want_memblock)
 {
 	int ret;
 
@@ -506,7 +506,10 @@ static int __meminit __add_section(int nid, struct zone *zone,
 	if (ret < 0)
 		return ret;
 
-	return register_new_memory(nid, __pfn_to_section(phys_start_pfn));
+	if (want_memblock)
+		ret = register_new_memory(nid, __pfn_to_section(phys_start_pfn));
+
+	return ret;
 }
 
 /*
@@ -516,7 +519,7 @@ static int __meminit __add_section(int nid, struct zone *zone,
  * add the new pages.
  */
 int __ref __add_pages(int nid, struct zone *zone, unsigned long phys_start_pfn,
-			unsigned long nr_pages)
+			unsigned long nr_pages, bool want_memblock)
 {
 	unsigned long i;
 	int err = 0;
@@ -544,7 +547,7 @@ int __ref __add_pages(int nid, struct zone *zone, unsigned long phys_start_pfn,
 	}
 
 	for (i = start_sec; i <= end_sec; i++) {
-		err = __add_section(nid, zone, section_nr_to_pfn(i));
+		err = __add_section(nid, zone, section_nr_to_pfn(i), want_memblock);
 
 		/*
 		 * EEXIST is finally dealt with by ioresource collision
-- 
2.11.0

-- 
Michal Hocko
SUSE Labs