Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
On Fri, Oct 09, 2009 at 21:43:26 +0100, Mal Gorman wrote: As you have tested this recently, would you be willing to post the results? While it's not a requirement of the patch, it would be nice to have an idea of how the effectiveness of memory hot-remove is improved when used with the powerpc balloon. This might convince others developers for balloons to register with the notifier. I did ten test runs without my patches and ten test runs with my patches on a 2.6.32-rc3 kernel. Without the patch: 6 out of 10 memory-remove operations without the patch removed 1 LMB (64Mb), the rest of the memory-remove attempts failed to remove any LMBs. With the patch: All of the memory-remove operations removed some LMBs. The average removed was just over 11 LMBs (704Mb) per attempt. Linux was given 2Gb of memory. During the test runs the average memory in use was 140Mb, not including cache and buffers, and the average amount consumed by the balloon was 1217Mb. The system was idle while the memory remove operation was performed. After each attempt the system was rebooted and allowed ~10 minutes to settle after boot. With a 2Gb configuration on POWER the LMB size is 64Mb. The drmgr command (part of powerpc-utils) was used to remove memory by LBM, just as an end-user would. Below is a list of the runs and the number of LMBs removed. Stock kernel (v2.6.32-rc3) -- LMBsUsed kb Loaned kb removed 0 135232 1257280 0 151168 1231744 1 152128 1234176 1 150976 1239232 1 151808 1232064 0 136064 1249152 0 137088 1246976 1 135296 1289984 1 136384 1263104 1 152960 1243904 === 0.60143910 1248762 Average 0.49 792916960 StdDev Patched kernel -- LMBsUsed kb Loaned kb removed 12 134336 1294336 10 152192 1250432 9 152832 1235520 15 153152 1237952 12 152320 1232704 13 135360 1252224 11 154176 1237056 10 153920 1243264 10 150720 1236416 13 151040 1230848 === 11.50 149005 1245075 Average 1.75 715817738 StdDev Regards, Robert Jennings ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
On Fri, Oct 02, 2009 at 01:44:58PM -0500, Robert Jennings wrote:
Memory balloon drivers can allocate a large amount of memory which
is not movable but could be freed to accomodate memory hotplug remove.
Prior to calling the memory hotplug notifier chain the memory in the
pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the
isolation will not proceed, causing the memory removal for that page
range to fail.
Rather than failing pageblock isolation if the the migrateteype is not
s/the the migrateteype/the migratetype/
MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock
are owned by a registered balloon driver (or other entity) using a
notifier chain. If all of the non-movable pages are owned by a balloon,
they can be freed later through the memory notifier chain and the range
can still be isolated in set_migratetype_isolate().
Signed-off-by: Robert Jennings r...@linux.vnet.ibm.com
---
drivers/base/memory.c | 19 +++
include/linux/memory.h | 26 ++
mm/page_alloc.c| 45 ++---
3 files changed, 83 insertions(+), 7 deletions(-)
Index: b/drivers/base/memory.c
===
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -63,6 +63,20 @@ void unregister_memory_notifier(struct n
}
EXPORT_SYMBOL(unregister_memory_notifier);
+static BLOCKING_NOTIFIER_HEAD(memory_isolate_chain);
+
+int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return blocking_notifier_chain_register(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(register_memory_isolate_notifier);
+
+void unregister_memory_isolate_notifier(struct notifier_block *nb)
+{
+ blocking_notifier_chain_unregister(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(unregister_memory_isolate_notifier);
+
/*
* register_memory - Setup a sysfs device for a memory block
*/
@@ -157,6 +171,11 @@ int memory_notify(unsigned long val, voi
return blocking_notifier_call_chain(memory_chain, val, v);
}
+int memory_isolate_notify(unsigned long val, void *v)
+{
+ return blocking_notifier_call_chain(memory_isolate_chain, val, v);
+}
+
/*
* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
* OK to have direct references to sparsemem variables in here.
Index: b/include/linux/memory.h
===
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -50,6 +50,18 @@ struct memory_notify {
int status_change_nid;
};
+/*
+ * During pageblock isolation, count the number of pages in the
+ * range [start_pfn, start_pfn + nr_pages)
+ */
The comment could have been slightly better. The count of pages in the
range is nr_pages - memory_holes but what you're counting is the number
of pages owned by the balloon driver in the notification chain.
+#define MEM_ISOLATE_COUNT(10)
+
+struct memory_isolate_notify {
+ unsigned long start_pfn;
+ unsigned int nr_pages;
+ unsigned int pages_found;
+};
+
struct notifier_block;
struct mem_section;
@@ -76,14 +88,28 @@ static inline int memory_notify(unsigned
{
return 0;
}
+static inline int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return 0;
+}
+static inline void unregister_memory_isolate_notifier(struct notifier_block
*nb)
+{
+}
+static inline int memory_isolate_notify(unsigned long val, void *v)
+{
+ return 0;
+}
#else
extern int register_memory_notifier(struct notifier_block *nb);
extern void unregister_memory_notifier(struct notifier_block *nb);
+extern int register_memory_isolate_notifier(struct notifier_block *nb);
+extern void unregister_memory_isolate_notifier(struct notifier_block *nb);
extern int register_new_memory(int, struct mem_section *);
extern int unregister_memory_section(struct mem_section *);
extern int memory_dev_init(void);
extern int remove_memory_block(unsigned long, struct mem_section *, int);
extern int memory_notify(unsigned long val, void *v);
+extern int memory_isolate_notify(unsigned long val, void *v);
extern struct memory_block *find_memory_block(struct mem_section *);
#define CONFIG_MEM_BLOCK_SIZE(PAGES_PER_SECTIONPAGE_SHIFT)
enum mem_add_context { BOOT, HOTPLUG };
Index: b/mm/page_alloc.c
===
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -48,6 +48,7 @@
#include linux/page_cgroup.h
#include linux/debugobjects.h
#include linux/kmemleak.h
+#include linux/memory.h
#include trace/events/kmem.h
#include asm/tlbflush.h
@@ -4985,23 +4986,53 @@ void set_pageblock_flags_group(struct pa
int set_migratetype_isolate(struct page *page)
{
struct zone *zone;
- unsigned long flags;
+ unsigned long flags, pfn, iter;
+ unsigned
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
* KAMEZAWA Hiroyuki (kamezawa.hir...@jp.fujitsu.com) wrote:
On Fri, 2 Oct 2009 13:44:58 -0500
Robert Jennings r...@linux.vnet.ibm.com wrote:
Memory balloon drivers can allocate a large amount of memory which
is not movable but could be freed to accomodate memory hotplug remove.
Prior to calling the memory hotplug notifier chain the memory in the
pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the
isolation will not proceed, causing the memory removal for that page
range to fail.
Rather than failing pageblock isolation if the the migrateteype is not
MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock
are owned by a registered balloon driver (or other entity) using a
notifier chain. If all of the non-movable pages are owned by a balloon,
they can be freed later through the memory notifier chain and the range
can still be isolated in set_migratetype_isolate().
Signed-off-by: Robert Jennings r...@linux.vnet.ibm.com
---
drivers/base/memory.c | 19 +++
include/linux/memory.h | 26 ++
mm/page_alloc.c| 45 ++---
3 files changed, 83 insertions(+), 7 deletions(-)
Index: b/drivers/base/memory.c
===
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -63,6 +63,20 @@ void unregister_memory_notifier(struct n
}
EXPORT_SYMBOL(unregister_memory_notifier);
+static BLOCKING_NOTIFIER_HEAD(memory_isolate_chain);
+
IIUC, this notifier is called under zone-lock.
please ATOMIC_NOTIFIER_HEAD().
I will correct this.
+int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return blocking_notifier_chain_register(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(register_memory_isolate_notifier);
+
+void unregister_memory_isolate_notifier(struct notifier_block *nb)
+{
+ blocking_notifier_chain_unregister(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(unregister_memory_isolate_notifier);
+
/*
* register_memory - Setup a sysfs device for a memory block
*/
@@ -157,6 +171,11 @@ int memory_notify(unsigned long val, voi
return blocking_notifier_call_chain(memory_chain, val, v);
}
+int memory_isolate_notify(unsigned long val, void *v)
+{
+ return blocking_notifier_call_chain(memory_isolate_chain, val, v);
+}
+
/*
* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
* OK to have direct references to sparsemem variables in here.
Index: b/include/linux/memory.h
===
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -50,6 +50,18 @@ struct memory_notify {
int status_change_nid;
};
+/*
+ * During pageblock isolation, count the number of pages in the
+ * range [start_pfn, start_pfn + nr_pages)
+ */
+#define MEM_ISOLATE_COUNT (10)
+
+struct memory_isolate_notify {
+ unsigned long start_pfn;
+ unsigned int nr_pages;
+ unsigned int pages_found;
+};
Could you add commentary for each field ?
This will be documented in the next version of the patch.
+
struct notifier_block;
struct mem_section;
@@ -76,14 +88,28 @@ static inline int memory_notify(unsigned
{
return 0;
}
+static inline int register_memory_isolate_notifier(struct notifier_block
*nb)
+{
+ return 0;
+}
+static inline void unregister_memory_isolate_notifier(struct
notifier_block *nb)
+{
+}
+static inline int memory_isolate_notify(unsigned long val, void *v)
+{
+ return 0;
+}
#else
extern int register_memory_notifier(struct notifier_block *nb);
extern void unregister_memory_notifier(struct notifier_block *nb);
+extern int register_memory_isolate_notifier(struct notifier_block *nb);
+extern void unregister_memory_isolate_notifier(struct notifier_block *nb);
extern int register_new_memory(int, struct mem_section *);
extern int unregister_memory_section(struct mem_section *);
extern int memory_dev_init(void);
extern int remove_memory_block(unsigned long, struct mem_section *, int);
extern int memory_notify(unsigned long val, void *v);
+extern int memory_isolate_notify(unsigned long val, void *v);
extern struct memory_block *find_memory_block(struct mem_section *);
#define CONFIG_MEM_BLOCK_SIZE (PAGES_PER_SECTIONPAGE_SHIFT)
enum mem_add_context { BOOT, HOTPLUG };
Index: b/mm/page_alloc.c
===
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -48,6 +48,7 @@
#include linux/page_cgroup.h
#include linux/debugobjects.h
#include linux/kmemleak.h
+#include linux/memory.h
#include trace/events/kmem.h
#include asm/tlbflush.h
@@ -4985,23 +4986,53 @@ void set_pageblock_flags_group(struct pa
int
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
* Mel Gorman (m...@csn.ul.ie) wrote:
On Fri, Oct 02, 2009 at 01:44:58PM -0500, Robert Jennings wrote:
Memory balloon drivers can allocate a large amount of memory which
is not movable but could be freed to accomodate memory hotplug remove.
Prior to calling the memory hotplug notifier chain the memory in the
pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the
isolation will not proceed, causing the memory removal for that page
range to fail.
Rather than failing pageblock isolation if the the migrateteype is not
s/the the migrateteype/the migratetype/
MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock
are owned by a registered balloon driver (or other entity) using a
notifier chain. If all of the non-movable pages are owned by a balloon,
they can be freed later through the memory notifier chain and the range
can still be isolated in set_migratetype_isolate().
Signed-off-by: Robert Jennings r...@linux.vnet.ibm.com
---
drivers/base/memory.c | 19 +++
include/linux/memory.h | 26 ++
mm/page_alloc.c| 45 ++---
3 files changed, 83 insertions(+), 7 deletions(-)
Index: b/drivers/base/memory.c
===
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -63,6 +63,20 @@ void unregister_memory_notifier(struct n
}
EXPORT_SYMBOL(unregister_memory_notifier);
+static BLOCKING_NOTIFIER_HEAD(memory_isolate_chain);
+
+int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return blocking_notifier_chain_register(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(register_memory_isolate_notifier);
+
+void unregister_memory_isolate_notifier(struct notifier_block *nb)
+{
+ blocking_notifier_chain_unregister(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(unregister_memory_isolate_notifier);
+
/*
* register_memory - Setup a sysfs device for a memory block
*/
@@ -157,6 +171,11 @@ int memory_notify(unsigned long val, voi
return blocking_notifier_call_chain(memory_chain, val, v);
}
+int memory_isolate_notify(unsigned long val, void *v)
+{
+ return blocking_notifier_call_chain(memory_isolate_chain, val, v);
+}
+
/*
* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
* OK to have direct references to sparsemem variables in here.
Index: b/include/linux/memory.h
===
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -50,6 +50,18 @@ struct memory_notify {
int status_change_nid;
};
+/*
+ * During pageblock isolation, count the number of pages in the
+ * range [start_pfn, start_pfn + nr_pages)
+ */
The comment could have been slightly better. The count of pages in the
range is nr_pages - memory_holes but what you're counting is the number
of pages owned by the balloon driver in the notification chain.
Right, it is misleading. I'll fix this.
+#define MEM_ISOLATE_COUNT (10)
+
+struct memory_isolate_notify {
+ unsigned long start_pfn;
+ unsigned int nr_pages;
+ unsigned int pages_found;
+};
+
struct notifier_block;
struct mem_section;
@@ -76,14 +88,28 @@ static inline int memory_notify(unsigned
{
return 0;
}
+static inline int register_memory_isolate_notifier(struct notifier_block
*nb)
+{
+ return 0;
+}
+static inline void unregister_memory_isolate_notifier(struct
notifier_block *nb)
+{
+}
+static inline int memory_isolate_notify(unsigned long val, void *v)
+{
+ return 0;
+}
#else
extern int register_memory_notifier(struct notifier_block *nb);
extern void unregister_memory_notifier(struct notifier_block *nb);
+extern int register_memory_isolate_notifier(struct notifier_block *nb);
+extern void unregister_memory_isolate_notifier(struct notifier_block *nb);
extern int register_new_memory(int, struct mem_section *);
extern int unregister_memory_section(struct mem_section *);
extern int memory_dev_init(void);
extern int remove_memory_block(unsigned long, struct mem_section *, int);
extern int memory_notify(unsigned long val, void *v);
+extern int memory_isolate_notify(unsigned long val, void *v);
extern struct memory_block *find_memory_block(struct mem_section *);
#define CONFIG_MEM_BLOCK_SIZE (PAGES_PER_SECTIONPAGE_SHIFT)
enum mem_add_context { BOOT, HOTPLUG };
Index: b/mm/page_alloc.c
===
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -48,6 +48,7 @@
#include linux/page_cgroup.h
#include linux/debugobjects.h
#include linux/kmemleak.h
+#include linux/memory.h
#include trace/events/kmem.h
#include asm/tlbflush.h
@@ -4985,23 +4986,53 @@ void
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
* Andrew Morton (a...@linux-foundation.org) wrote: On Fri, 2 Oct 2009 13:44:58 -0500 Robert Jennings r...@linux.vnet.ibm.com wrote: Memory balloon drivers can allocate a large amount of memory which is not movable but could be freed to accomodate memory hotplug remove. Prior to calling the memory hotplug notifier chain the memory in the pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the isolation will not proceed, causing the memory removal for that page range to fail. Rather than failing pageblock isolation if the the migrateteype is not MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock are owned by a registered balloon driver (or other entity) using a notifier chain. If all of the non-movable pages are owned by a balloon, they can be freed later through the memory notifier chain and the range can still be isolated in set_migratetype_isolate(). The patch looks sane enough to me. I expect that if the powerpc and s390 guys want to work on CMM over the next couple of months, they'd like this patch merged into 2.6.32. It's a bit larger and more involved than one would like, but I guess we can do that if suitable people (Mel? Kamezawa?) have had a close look and are OK with it. What do people think? I'd love to get it in 2.6.32 if that's possible. I have gone over the comments from Mel and Kamezawa I produced a new patchset. I just finished testing it (and I also tested with CONFIG_MEMORY_HOTPLUG_SPARSE=n) and it will be posted shortly. Has it been carefully compile- and run-time tested with CONFIG_MEMORY_HOTPLUG_SPARSE=n? Yes, I have compiled the kernel CONFIG_MEMORY_HOTPLUG_SPARSE=n and made sure that we didn't have any problems. --Robert Jennings ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
On Fri, Oct 09, 2009 at 03:23:04PM -0500, Robert Jennings wrote: * Andrew Morton (a...@linux-foundation.org) wrote: On Fri, 2 Oct 2009 13:44:58 -0500 Robert Jennings r...@linux.vnet.ibm.com wrote: Memory balloon drivers can allocate a large amount of memory which is not movable but could be freed to accomodate memory hotplug remove. Prior to calling the memory hotplug notifier chain the memory in the pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the isolation will not proceed, causing the memory removal for that page range to fail. Rather than failing pageblock isolation if the the migrateteype is not MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock are owned by a registered balloon driver (or other entity) using a notifier chain. If all of the non-movable pages are owned by a balloon, they can be freed later through the memory notifier chain and the range can still be isolated in set_migratetype_isolate(). The patch looks sane enough to me. I expect that if the powerpc and s390 guys want to work on CMM over the next couple of months, they'd like this patch merged into 2.6.32. It's a bit larger and more involved than one would like, but I guess we can do that if suitable people (Mel? Kamezawa?) have had a close look and are OK with it. What do people think? I'd love to get it in 2.6.32 if that's possible. I have gone over the comments from Mel and Kamezawa I produced a new patchset. I just finished testing it (and I also tested with CONFIG_MEMORY_HOTPLUG_SPARSE=n) and it will be posted shortly. As you have tested this recently, would you be willing to post the results? While it's not a requirement of the patch, it would be nice to have an idea of how the effectiveness of memory hot-remove is improved when used with the powerpc balloon. This might convince others developers for balloons to register with the notifier. Total aside, I'm not overly clear why so much of balloon driver logic is in drivers and not in the core. At a casual glance, it would appear that balloon logic could be improved by combining it with similar logic as is used for lumpy reclaim. This comment is not intended to hurt the patch, but for the people working on CMM to consider if it hasn't been considered already. Has it been carefully compile- and run-time tested with CONFIG_MEMORY_HOTPLUG_SPARSE=n? Yes, I have compiled the kernel CONFIG_MEMORY_HOTPLUG_SPARSE=n and made sure that we didn't have any problems. The pfn_valid_within() was the biggie as far as the core is concerned. That sort of mistake causes fairly mad-looking oops. To be perfectly honest, I didn't review the powerpc-specific portion assuming that people are testing that and that there are developers more familiar with the area. -- Mel Gorman Part-time Phd Student Linux Technology Center University of Limerick IBM Dublin Software Lab ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
On Fri, 2 Oct 2009 13:44:58 -0500 Robert Jennings r...@linux.vnet.ibm.com wrote: Memory balloon drivers can allocate a large amount of memory which is not movable but could be freed to accomodate memory hotplug remove. Prior to calling the memory hotplug notifier chain the memory in the pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the isolation will not proceed, causing the memory removal for that page range to fail. Rather than failing pageblock isolation if the the migrateteype is not MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock are owned by a registered balloon driver (or other entity) using a notifier chain. If all of the non-movable pages are owned by a balloon, they can be freed later through the memory notifier chain and the range can still be isolated in set_migratetype_isolate(). The patch looks sane enough to me. I expect that if the powerpc and s390 guys want to work on CMM over the next couple of months, they'd like this patch merged into 2.6.32. It's a bit larger and more involved than one would like, but I guess we can do that if suitable people (Mel? Kamezawa?) have had a close look and are OK with it. What do people think? Has it been carefully compile- and run-time tested with CONFIG_MEMORY_HOTPLUG_SPARSE=n? ___ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
Re: [PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
On Fri, 2 Oct 2009 13:44:58 -0500
Robert Jennings r...@linux.vnet.ibm.com wrote:
Memory balloon drivers can allocate a large amount of memory which
is not movable but could be freed to accomodate memory hotplug remove.
Prior to calling the memory hotplug notifier chain the memory in the
pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the
isolation will not proceed, causing the memory removal for that page
range to fail.
Rather than failing pageblock isolation if the the migrateteype is not
MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock
are owned by a registered balloon driver (or other entity) using a
notifier chain. If all of the non-movable pages are owned by a balloon,
they can be freed later through the memory notifier chain and the range
can still be isolated in set_migratetype_isolate().
Signed-off-by: Robert Jennings r...@linux.vnet.ibm.com
---
drivers/base/memory.c | 19 +++
include/linux/memory.h | 26 ++
mm/page_alloc.c| 45 ++---
3 files changed, 83 insertions(+), 7 deletions(-)
Index: b/drivers/base/memory.c
===
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -63,6 +63,20 @@ void unregister_memory_notifier(struct n
}
EXPORT_SYMBOL(unregister_memory_notifier);
+static BLOCKING_NOTIFIER_HEAD(memory_isolate_chain);
+
IIUC, this notifier is called under zone-lock.
please ATOMIC_NOTIFIER_HEAD().
+int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return blocking_notifier_chain_register(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(register_memory_isolate_notifier);
+
+void unregister_memory_isolate_notifier(struct notifier_block *nb)
+{
+ blocking_notifier_chain_unregister(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(unregister_memory_isolate_notifier);
+
/*
* register_memory - Setup a sysfs device for a memory block
*/
@@ -157,6 +171,11 @@ int memory_notify(unsigned long val, voi
return blocking_notifier_call_chain(memory_chain, val, v);
}
+int memory_isolate_notify(unsigned long val, void *v)
+{
+ return blocking_notifier_call_chain(memory_isolate_chain, val, v);
+}
+
/*
* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
* OK to have direct references to sparsemem variables in here.
Index: b/include/linux/memory.h
===
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -50,6 +50,18 @@ struct memory_notify {
int status_change_nid;
};
+/*
+ * During pageblock isolation, count the number of pages in the
+ * range [start_pfn, start_pfn + nr_pages)
+ */
+#define MEM_ISOLATE_COUNT(10)
+
+struct memory_isolate_notify {
+ unsigned long start_pfn;
+ unsigned int nr_pages;
+ unsigned int pages_found;
+};
Could you add commentary for each field ?
+
struct notifier_block;
struct mem_section;
@@ -76,14 +88,28 @@ static inline int memory_notify(unsigned
{
return 0;
}
+static inline int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return 0;
+}
+static inline void unregister_memory_isolate_notifier(struct notifier_block
*nb)
+{
+}
+static inline int memory_isolate_notify(unsigned long val, void *v)
+{
+ return 0;
+}
#else
extern int register_memory_notifier(struct notifier_block *nb);
extern void unregister_memory_notifier(struct notifier_block *nb);
+extern int register_memory_isolate_notifier(struct notifier_block *nb);
+extern void unregister_memory_isolate_notifier(struct notifier_block *nb);
extern int register_new_memory(int, struct mem_section *);
extern int unregister_memory_section(struct mem_section *);
extern int memory_dev_init(void);
extern int remove_memory_block(unsigned long, struct mem_section *, int);
extern int memory_notify(unsigned long val, void *v);
+extern int memory_isolate_notify(unsigned long val, void *v);
extern struct memory_block *find_memory_block(struct mem_section *);
#define CONFIG_MEM_BLOCK_SIZE(PAGES_PER_SECTIONPAGE_SHIFT)
enum mem_add_context { BOOT, HOTPLUG };
Index: b/mm/page_alloc.c
===
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -48,6 +48,7 @@
#include linux/page_cgroup.h
#include linux/debugobjects.h
#include linux/kmemleak.h
+#include linux/memory.h
#include trace/events/kmem.h
#include asm/tlbflush.h
@@ -4985,23 +4986,53 @@ void set_pageblock_flags_group(struct pa
int set_migratetype_isolate(struct page *page)
{
struct zone *zone;
- unsigned long flags;
+ unsigned long flags, pfn, iter;
+ unsigned long immobile = 0;
+ struct memory_isolate_notify arg;
+ int notifier_ret;
int ret = -EBUSY;
[PATCH 1/2][v2] mm: add notifier in pageblock isolation for balloon drivers
Memory balloon drivers can allocate a large amount of memory which
is not movable but could be freed to accomodate memory hotplug remove.
Prior to calling the memory hotplug notifier chain the memory in the
pageblock is isolated. If the migrate type is not MIGRATE_MOVABLE the
isolation will not proceed, causing the memory removal for that page
range to fail.
Rather than failing pageblock isolation if the the migrateteype is not
MIGRATE_MOVABLE, this patch checks if all of the pages in the pageblock
are owned by a registered balloon driver (or other entity) using a
notifier chain. If all of the non-movable pages are owned by a balloon,
they can be freed later through the memory notifier chain and the range
can still be isolated in set_migratetype_isolate().
Signed-off-by: Robert Jennings r...@linux.vnet.ibm.com
---
drivers/base/memory.c | 19 +++
include/linux/memory.h | 26 ++
mm/page_alloc.c| 45 ++---
3 files changed, 83 insertions(+), 7 deletions(-)
Index: b/drivers/base/memory.c
===
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -63,6 +63,20 @@ void unregister_memory_notifier(struct n
}
EXPORT_SYMBOL(unregister_memory_notifier);
+static BLOCKING_NOTIFIER_HEAD(memory_isolate_chain);
+
+int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return blocking_notifier_chain_register(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(register_memory_isolate_notifier);
+
+void unregister_memory_isolate_notifier(struct notifier_block *nb)
+{
+ blocking_notifier_chain_unregister(memory_isolate_chain, nb);
+}
+EXPORT_SYMBOL(unregister_memory_isolate_notifier);
+
/*
* register_memory - Setup a sysfs device for a memory block
*/
@@ -157,6 +171,11 @@ int memory_notify(unsigned long val, voi
return blocking_notifier_call_chain(memory_chain, val, v);
}
+int memory_isolate_notify(unsigned long val, void *v)
+{
+ return blocking_notifier_call_chain(memory_isolate_chain, val, v);
+}
+
/*
* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
* OK to have direct references to sparsemem variables in here.
Index: b/include/linux/memory.h
===
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -50,6 +50,18 @@ struct memory_notify {
int status_change_nid;
};
+/*
+ * During pageblock isolation, count the number of pages in the
+ * range [start_pfn, start_pfn + nr_pages)
+ */
+#define MEM_ISOLATE_COUNT (10)
+
+struct memory_isolate_notify {
+ unsigned long start_pfn;
+ unsigned int nr_pages;
+ unsigned int pages_found;
+};
+
struct notifier_block;
struct mem_section;
@@ -76,14 +88,28 @@ static inline int memory_notify(unsigned
{
return 0;
}
+static inline int register_memory_isolate_notifier(struct notifier_block *nb)
+{
+ return 0;
+}
+static inline void unregister_memory_isolate_notifier(struct notifier_block
*nb)
+{
+}
+static inline int memory_isolate_notify(unsigned long val, void *v)
+{
+ return 0;
+}
#else
extern int register_memory_notifier(struct notifier_block *nb);
extern void unregister_memory_notifier(struct notifier_block *nb);
+extern int register_memory_isolate_notifier(struct notifier_block *nb);
+extern void unregister_memory_isolate_notifier(struct notifier_block *nb);
extern int register_new_memory(int, struct mem_section *);
extern int unregister_memory_section(struct mem_section *);
extern int memory_dev_init(void);
extern int remove_memory_block(unsigned long, struct mem_section *, int);
extern int memory_notify(unsigned long val, void *v);
+extern int memory_isolate_notify(unsigned long val, void *v);
extern struct memory_block *find_memory_block(struct mem_section *);
#define CONFIG_MEM_BLOCK_SIZE (PAGES_PER_SECTIONPAGE_SHIFT)
enum mem_add_context { BOOT, HOTPLUG };
Index: b/mm/page_alloc.c
===
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -48,6 +48,7 @@
#include linux/page_cgroup.h
#include linux/debugobjects.h
#include linux/kmemleak.h
+#include linux/memory.h
#include trace/events/kmem.h
#include asm/tlbflush.h
@@ -4985,23 +4986,53 @@ void set_pageblock_flags_group(struct pa
int set_migratetype_isolate(struct page *page)
{
struct zone *zone;
- unsigned long flags;
+ unsigned long flags, pfn, iter;
+ unsigned long immobile = 0;
+ struct memory_isolate_notify arg;
+ int notifier_ret;
int ret = -EBUSY;
int zone_idx;
zone = page_zone(page);
zone_idx = zone_idx(zone);
+
spin_lock_irqsave(zone-lock, flags);
+ if (get_pageblock_migratetype(page) == MIGRATE_MOVABLE ||
+ zone_idx == ZONE_MOVABLE) {
+ ret = 0;
+ goto out;
+ }
+
+
