* [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
@ 2015-08-31 5:58 ` Changsheng Liu
2015-08-31 13:08 ` Vlastimil Babka
2015-09-01 15:37 ` Yasuaki Ishimatsu
0 siblings, 2 replies; 7+ messages in thread
From: Changsheng Liu @ 2015-08-31 5:58 UTC (permalink / raw)
To: akpm, isimatu.yasuaki, vbabka, yasu.isimatu
Cc: linux-mm, linux-kernel, wunan, yanxiaofeng, liuchangsheng, fandd,
Changsheng Liu
From: Changsheng Liu <liuchangcheng@inspur.com>
After the user config CONFIG_MOVABLE_NODE and movable_node kernel option,
When the memory is hot added, should_add_memory_movable() return 0
because all zones including movable zone are empty,
so the memory that was hot added will be added to the normal zone
and the normal zone will be created firstly.
But we want the whole node to be added to movable zone defaultly.
So we change should_add_memory_movable(): if the user config
CONFIG_MOVABLE_NODE and movable_node kernel option
it will always return 1 and all zones is empty at the same time,
so that the movable zone will be created firstly
and then the whole node will be added to movable zone defaultly.
If we want the node to be added to normal zone,
we can do it as follows:
"echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
If the memory is added to movable zone defaultly,
the user can offline it and add it to other zone again.
But if the memory is added to normal zone defaultly,
the user will not offline the memory used by kernel.
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
Tested-by: Dongdong Fan <fandd@inspur.com>
---
mm/memory_hotplug.c | 5 +++++
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 26fbba7..d1149ff 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
unsigned long start_pfn = start >> PAGE_SHIFT;
pg_data_t *pgdat = NODE_DATA(nid);
struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
+ struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
+
+ if (movable_node_is_enabled()
+ && (zone_end_pfn(normal_zone) <= start_pfn))
+ return 1;
if (zone_is_empty(movable_zone))
return 0;
--
1.7.1
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
2015-08-31 5:58 ` [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly Changsheng Liu
@ 2015-08-31 13:08 ` Vlastimil Babka
2015-09-01 6:05 ` Changsheng Liu
2015-09-01 15:37 ` Yasuaki Ishimatsu
1 sibling, 1 reply; 7+ messages in thread
From: Vlastimil Babka @ 2015-08-31 13:08 UTC (permalink / raw)
To: Changsheng Liu, akpm, isimatu.yasuaki, yasu.isimatu
Cc: linux-mm, linux-kernel, wunan, yanxiaofeng, fandd, Changsheng Liu
On 08/31/2015 07:58 AM, Changsheng Liu wrote:
> From: Changsheng Liu <liuchangcheng@inspur.com>
>
> After the user config CONFIG_MOVABLE_NODE and movable_node kernel option,
> When the memory is hot added, should_add_memory_movable() return 0
> because all zones including movable zone are empty,
> so the memory that was hot added will be added to the normal zone
> and the normal zone will be created firstly.
> But we want the whole node to be added to movable zone defaultly.
>
> So we change should_add_memory_movable(): if the user config
> CONFIG_MOVABLE_NODE and movable_node kernel option
> it will always return 1 and all zones is empty at the same time,
> so that the movable zone will be created firstly
> and then the whole node will be added to movable zone defaultly.
> If we want the node to be added to normal zone,
> we can do it as follows:
> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
>
> If the memory is added to movable zone defaultly,
> the user can offline it and add it to other zone again.
> But if the memory is added to normal zone defaultly,
> the user will not offline the memory used by kernel.
>
> Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
> Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
> Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
Thanks for the credit for commenting on the previous versions of the
patch. However, "Reviewed-by" currently means that the reviewer believes
the patch is OK, so you can add it only if the reviewer offers it
explicitly. See Documentation/SubmittingPatches section 13. There was a
discussion on ksummit-discuss about adding a new tag for this case, but
nothing was decided yet AFAIK.
> Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
> Tested-by: Dongdong Fan <fandd@inspur.com>
> ---
> mm/memory_hotplug.c | 5 +++++
> 1 files changed, 5 insertions(+), 0 deletions(-)
>
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 26fbba7..d1149ff 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
> unsigned long start_pfn = start >> PAGE_SHIFT;
> pg_data_t *pgdat = NODE_DATA(nid);
> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
> + struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
> +
> + if (movable_node_is_enabled()
> + && (zone_end_pfn(normal_zone) <= start_pfn))
> + return 1;
I wonder if the condition is true and ZONE_NORMAL exists (but it's
empty?) if you intend to only add movable memory to a node, so you can
still hot-remove it all with this patch?
>
> if (zone_is_empty(movable_zone))
> return 0;
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
2015-08-31 13:08 ` Vlastimil Babka
@ 2015-09-01 6:05 ` Changsheng Liu
0 siblings, 0 replies; 7+ messages in thread
From: Changsheng Liu @ 2015-09-01 6:05 UTC (permalink / raw)
To: Vlastimil Babka, akpm, isimatu.yasuaki, yasu.isimatu
Cc: linux-mm, linux-kernel, wunan, yanxiaofeng, fandd, Changsheng Liu
On 8/31/2015 21:08, Vlastimil Babka wrote:
> On 08/31/2015 07:58 AM, Changsheng Liu wrote:
>> From: Changsheng Liu <liuchangcheng@inspur.com>
>>
>> After the user config CONFIG_MOVABLE_NODE and movable_node kernel
>> option,
>> When the memory is hot added, should_add_memory_movable() return 0
>> because all zones including movable zone are empty,
>> so the memory that was hot added will be added to the normal zone
>> and the normal zone will be created firstly.
>> But we want the whole node to be added to movable zone defaultly.
>>
>> So we change should_add_memory_movable(): if the user config
>> CONFIG_MOVABLE_NODE and movable_node kernel option
>> it will always return 1 and all zones is empty at the same time,
>> so that the movable zone will be created firstly
>> and then the whole node will be added to movable zone defaultly.
>> If we want the node to be added to normal zone,
>> we can do it as follows:
>> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
>>
>> If the memory is added to movable zone defaultly,
>> the user can offline it and add it to other zone again.
>> But if the memory is added to normal zone defaultly,
>> the user will not offline the memory used by kernel.
>>
>> Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
>> Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
>> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
>> Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
>
> Thanks for the credit for commenting on the previous versions of the
> patch. However, "Reviewed-by" currently means that the reviewer
> believes the patch is OK, so you can add it only if the reviewer
> offers it explicitly. See Documentation/SubmittingPatches section 13.
> There was a discussion on ksummit-discuss about adding a new tag for
> this case, but nothing was decided yet AFAIK.
I'm sorry about it and thanks for your review,I will update the patch.
>
>> Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
>> Tested-by: Dongdong Fan <fandd@inspur.com>
>> ---
>> mm/memory_hotplug.c | 5 +++++
>> 1 files changed, 5 insertions(+), 0 deletions(-)
>>
>> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
>> index 26fbba7..d1149ff 100644
>> --- a/mm/memory_hotplug.c
>> +++ b/mm/memory_hotplug.c
>> @@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid,
>> u64 start, u64 size)
>> unsigned long start_pfn = start >> PAGE_SHIFT;
>> pg_data_t *pgdat = NODE_DATA(nid);
>> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
>> + struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
>> +
>> + if (movable_node_is_enabled()
>> + && (zone_end_pfn(normal_zone) <= start_pfn))
>> + return 1;
>
> I wonder if the condition is true and ZONE_NORMAL exists (but it's
> empty?) if you intend to only add movable memory to a node, so you can
> still hot-remove it all with this patch?
Yes
>
>
>>
>> if (zone_is_empty(movable_zone))
>> return 0;
>>
>
> .
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
2015-08-31 5:58 ` [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly Changsheng Liu
2015-08-31 13:08 ` Vlastimil Babka
@ 2015-09-01 15:37 ` Yasuaki Ishimatsu
2015-09-02 9:06 ` Changsheng Liu
2015-09-06 8:33 ` Changsheng Liu
1 sibling, 2 replies; 7+ messages in thread
From: Yasuaki Ishimatsu @ 2015-09-01 15:37 UTC (permalink / raw)
To: Changsheng Liu
Cc: akpm, isimatu.yasuaki, vbabka, linux-mm, linux-kernel, wunan,
yanxiaofeng, fandd, Changsheng Liu
On Mon, 31 Aug 2015 01:58:40 -0400
Changsheng Liu <liuchangsheng@inspur.com> wrote:
> From: Changsheng Liu <liuchangcheng@inspur.com>
>
> After the user config CONFIG_MOVABLE_NODE and movable_node kernel option,
> When the memory is hot added, should_add_memory_movable() return 0
> because all zones including movable zone are empty,
> so the memory that was hot added will be added to the normal zone
> and the normal zone will be created firstly.
> But we want the whole node to be added to movable zone defaultly.
>
> So we change should_add_memory_movable(): if the user config
> CONFIG_MOVABLE_NODE and movable_node kernel option
> it will always return 1 and all zones is empty at the same time,
> so that the movable zone will be created firstly
> and then the whole node will be added to movable zone defaultly.
> If we want the node to be added to normal zone,
> we can do it as follows:
> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
>
> If the memory is added to movable zone defaultly,
> the user can offline it and add it to other zone again.
> But if the memory is added to normal zone defaultly,
> the user will not offline the memory used by kernel.
>
> Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
> Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
> Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
> Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
> Tested-by: Dongdong Fan <fandd@inspur.com>
> ---
> mm/memory_hotplug.c | 5 +++++
> 1 files changed, 5 insertions(+), 0 deletions(-)
>
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 26fbba7..d1149ff 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
> unsigned long start_pfn = start >> PAGE_SHIFT;
> pg_data_t *pgdat = NODE_DATA(nid);
> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
> + struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
> +
> + if (movable_node_is_enabled()
> + && (zone_end_pfn(normal_zone) <= start_pfn))
> + return 1;
If system boots up without movable_node, kernel behavior is changed by the patch.
And you syould consider other zone.
How about it. The patch is no build and test.
---
mm/memory_hotplug.c | 36 ++++++++++++++++++++++++++++++++----
1 files changed, 32 insertions(+), 4 deletions(-)
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 6da82bc..321595d 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1198,6 +1198,8 @@ static int check_hotplug_memory_range(u64 start, u64 size)
/*
* If movable zone has already been setup, newly added memory should be check.
* If its address is higher than movable zone, it should be added as movable.
+ * And if system boots up with movable_zone and added memory does not overlap
+ * other zone except for movable zone, the memory is added as movable.
* Without this check, movable zone may overlap with other zone.
*/
static int should_add_memory_movable(int nid, u64 start, u64 size)
@@ -1205,14 +1207,40 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
unsigned long start_pfn = start >> PAGE_SHIFT;
pg_data_t *pgdat = NODE_DATA(nid);
struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
+ struct zone *zone;
+ enum zone_type zt = ZONE_MOVABLE - 1;
+
+ /*
+ * If memory is added after ZONE_MOVALBE, the memory is managed as
+ * movable.
+ */
+ if (!zone_is_empty(movable_zone) &&
+ (movable_zone->zone_start_pfn <= start_pfn))
+ return 1;
- if (zone_is_empty(movable_zone))
+ if (!movable_node_is_enabled())
return 0;
- if (movable_zone->zone_start_pfn <= start_pfn)
- return 1;
+ /*
+ * Find enabled zone and check the added memory.
+ * If the memory is added after the enabled zone, the memory is
+ * managed as movable.
+ *
+ * If all zones are empty, the memory is also managed as movable.
+ */
+ for (; zt >= ZONE_DMA; zt--) {
+ zone = pgdat->node_zones + zt;
- return 0;
+ if (zone_is_empty(zone))
+ continue;
+
+ if (zone_end_pfn(zone) <= start_pfn)
+ return 1;
+ else
+ return 0;
+ }
+
+ return 1;
}
int zone_for_memory(int nid, u64 start, u64 size, int zone_default)
--
1.7.1
>
> if (zone_is_empty(movable_zone))
> return 0;
> --
> 1.7.1
>
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
2015-09-01 15:37 ` Yasuaki Ishimatsu
@ 2015-09-02 9:06 ` Changsheng Liu
2015-09-06 8:33 ` Changsheng Liu
1 sibling, 0 replies; 7+ messages in thread
From: Changsheng Liu @ 2015-09-02 9:06 UTC (permalink / raw)
To: Yasuaki Ishimatsu
Cc: akpm, isimatu.yasuaki, vbabka, linux-mm, linux-kernel, wunan,
yanxiaofeng, fandd, Changsheng Liu
On 9/1/2015 23:37, Yasuaki Ishimatsu wrote:
> On Mon, 31 Aug 2015 01:58:40 -0400
> Changsheng Liu <liuchangsheng@inspur.com> wrote:
>
>> From: Changsheng Liu <liuchangcheng@inspur.com>
>>
>> After the user config CONFIG_MOVABLE_NODE and movable_node kernel option,
>> When the memory is hot added, should_add_memory_movable() return 0
>> because all zones including movable zone are empty,
>> so the memory that was hot added will be added to the normal zone
>> and the normal zone will be created firstly.
>> But we want the whole node to be added to movable zone defaultly.
>>
>> So we change should_add_memory_movable(): if the user config
>> CONFIG_MOVABLE_NODE and movable_node kernel option
>> it will always return 1 and all zones is empty at the same time,
>> so that the movable zone will be created firstly
>> and then the whole node will be added to movable zone defaultly.
>> If we want the node to be added to normal zone,
>> we can do it as follows:
>> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
>>
>> If the memory is added to movable zone defaultly,
>> the user can offline it and add it to other zone again.
>> But if the memory is added to normal zone defaultly,
>> the user will not offline the memory used by kernel.
>>
>> Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
>> Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
>> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
>> Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
>> Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
>> Tested-by: Dongdong Fan <fandd@inspur.com>
>> ---
>> mm/memory_hotplug.c | 5 +++++
>> 1 files changed, 5 insertions(+), 0 deletions(-)
>>
>> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
>> index 26fbba7..d1149ff 100644
>> --- a/mm/memory_hotplug.c
>> +++ b/mm/memory_hotplug.c
>> @@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
>> unsigned long start_pfn = start >> PAGE_SHIFT;
>> pg_data_t *pgdat = NODE_DATA(nid);
>> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
>> + struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
>> +
>> + if (movable_node_is_enabled()
>> + && (zone_end_pfn(normal_zone) <= start_pfn))
>> + return 1;
> If system boots up without movable_node, kernel behavior is changed by the patch.
> And you syould consider other zone.
>
> How about it. The patch is no build and test.
Thanks for your suggestion. We will test the patch on our platform
>
>
> ---
> mm/memory_hotplug.c | 36 ++++++++++++++++++++++++++++++++----
> 1 files changed, 32 insertions(+), 4 deletions(-)
>
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 6da82bc..321595d 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1198,6 +1198,8 @@ static int check_hotplug_memory_range(u64 start, u64 size)
> /*
> * If movable zone has already been setup, newly added memory should be check.
> * If its address is higher than movable zone, it should be added as movable.
> + * And if system boots up with movable_zone and added memory does not overlap
> + * other zone except for movable zone, the memory is added as movable.
> * Without this check, movable zone may overlap with other zone.
> */
> static int should_add_memory_movable(int nid, u64 start, u64 size)
> @@ -1205,14 +1207,40 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
> unsigned long start_pfn = start >> PAGE_SHIFT;
> pg_data_t *pgdat = NODE_DATA(nid);
> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
> + struct zone *zone;
> + enum zone_type zt = ZONE_MOVABLE - 1;
> +
> + /*
> + * If memory is added after ZONE_MOVALBE, the memory is managed as
> + * movable.
> + */
> + if (!zone_is_empty(movable_zone) &&
> + (movable_zone->zone_start_pfn <= start_pfn))
> + return 1;
>
> - if (zone_is_empty(movable_zone))
> + if (!movable_node_is_enabled())
> return 0;
>
> - if (movable_zone->zone_start_pfn <= start_pfn)
> - return 1;
> + /*
> + * Find enabled zone and check the added memory.
> + * If the memory is added after the enabled zone, the memory is
> + * managed as movable.
> + *
> + * If all zones are empty, the memory is also managed as movable.
> + */
> + for (; zt >= ZONE_DMA; zt--) {
> + zone = pgdat->node_zones + zt;
>
> - return 0;
> + if (zone_is_empty(zone))
> + continue;
> +
> + if (zone_end_pfn(zone) <= start_pfn)
> + return 1;
> + else
> + return 0;
> + }
> +
> + return 1;
> }
>
> int zone_for_memory(int nid, u64 start, u64 size, int zone_default)
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
2015-09-01 15:37 ` Yasuaki Ishimatsu
2015-09-02 9:06 ` Changsheng Liu
@ 2015-09-06 8:33 ` Changsheng Liu
2015-09-11 19:42 ` Yasuaki Ishimatsu
1 sibling, 1 reply; 7+ messages in thread
From: Changsheng Liu @ 2015-09-06 8:33 UTC (permalink / raw)
To: Yasuaki Ishimatsu
Cc: akpm, isimatu.yasuaki, vbabka, linux-mm, linux-kernel, wunan,
yanxiaofeng, fandd, Changsheng Liu
On 9/1/2015 23:37, Yasuaki Ishimatsu wrote:
> On Mon, 31 Aug 2015 01:58:40 -0400
> Changsheng Liu <liuchangsheng@inspur.com> wrote:
>
>> From: Changsheng Liu <liuchangcheng@inspur.com>
>>
>> After the user config CONFIG_MOVABLE_NODE and movable_node kernel option,
>> When the memory is hot added, should_add_memory_movable() return 0
>> because all zones including movable zone are empty,
>> so the memory that was hot added will be added to the normal zone
>> and the normal zone will be created firstly.
>> But we want the whole node to be added to movable zone defaultly.
>>
>> So we change should_add_memory_movable(): if the user config
>> CONFIG_MOVABLE_NODE and movable_node kernel option
>> it will always return 1 and all zones is empty at the same time,
>> so that the movable zone will be created firstly
>> and then the whole node will be added to movable zone defaultly.
>> If we want the node to be added to normal zone,
>> we can do it as follows:
>> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
>>
>> If the memory is added to movable zone defaultly,
>> the user can offline it and add it to other zone again.
>> But if the memory is added to normal zone defaultly,
>> the user will not offline the memory used by kernel.
>>
>> Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
>> Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
>> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
>> Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
>> Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
>> Tested-by: Dongdong Fan <fandd@inspur.com>
>> ---
>> mm/memory_hotplug.c | 5 +++++
>> 1 files changed, 5 insertions(+), 0 deletions(-)
>>
>> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
>> index 26fbba7..d1149ff 100644
>> --- a/mm/memory_hotplug.c
>> +++ b/mm/memory_hotplug.c
>> @@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
>> unsigned long start_pfn = start >> PAGE_SHIFT;
>> pg_data_t *pgdat = NODE_DATA(nid);
>> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
>> + struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
>> +
>> + if (movable_node_is_enabled()
>> + && (zone_end_pfn(normal_zone) <= start_pfn))
>> + return 1;
> If system boots up without movable_node, kernel behavior is changed by the patch.
> And you syould consider other zone.
>
> How about it. The patch is no build and test.
>
>
> ---
> mm/memory_hotplug.c | 36 ++++++++++++++++++++++++++++++++----
> 1 files changed, 32 insertions(+), 4 deletions(-)
>
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 6da82bc..321595d 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1198,6 +1198,8 @@ static int check_hotplug_memory_range(u64 start, u64 size)
> /*
> * If movable zone has already been setup, newly added memory should be check.
> * If its address is higher than movable zone, it should be added as movable.
> + * And if system boots up with movable_zone and added memory does not overlap
> + * other zone except for movable zone, the memory is added as movable.
> * Without this check, movable zone may overlap with other zone.
> */
> static int should_add_memory_movable(int nid, u64 start, u64 size)
> @@ -1205,14 +1207,40 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
> unsigned long start_pfn = start >> PAGE_SHIFT;
> pg_data_t *pgdat = NODE_DATA(nid);
> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
> + struct zone *zone;
> + enum zone_type zt = ZONE_MOVABLE - 1;
> +
> + /*
> + * If memory is added after ZONE_MOVALBE, the memory is managed as
> + * movable.
> + */
> + if (!zone_is_empty(movable_zone) &&
> + (movable_zone->zone_start_pfn <= start_pfn))
> + return 1;
>
> - if (zone_is_empty(movable_zone))
> + if (!movable_node_is_enabled())
> return 0;
>
> - if (movable_zone->zone_start_pfn <= start_pfn)
> - return 1;
> + /*
> + * Find enabled zone and check the added memory.
> + * If the memory is added after the enabled zone, the memory is
> + * managed as movable.
> + *
> + * If all zones are empty, the memory is also managed as movable.
> + */
> + for (; zt >= ZONE_DMA; zt--) {
> + zone = pgdat->node_zones + zt;
>
> - return 0;
> + if (zone_is_empty(zone))
> + continue;
> +
> + if (zone_end_pfn(zone) <= start_pfn)
> + return 1;
> + else
> + return 0;
> + }
> +
> + return 1;
> }
>
The function zone_for_memory() adds the memory to
ZONE_NORMAL(x86_64)/ZONE_HIGH(x86_32) defaultly, So I think the system
just need check the added-memory is whether after the ZONE_NORMAL/ZONE_HIGH
> int zone_for_memory(int nid, u64 start, u64 size, int zone_default)
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly
2015-09-06 8:33 ` Changsheng Liu
@ 2015-09-11 19:42 ` Yasuaki Ishimatsu
0 siblings, 0 replies; 7+ messages in thread
From: Yasuaki Ishimatsu @ 2015-09-11 19:42 UTC (permalink / raw)
To: Changsheng Liu
Cc: akpm, isimatu.yasuaki, vbabka, linux-mm, linux-kernel, wunan,
yanxiaofeng, fandd, Changsheng Liu
Hi Changsheng,
Thank you for your comments.
I'm on vacation. So After that, I'll repost the patch.
Thanks,
Yasuaki Ishimatsu
On Sun, 6 Sep 2015 16:33:42 +0800
Changsheng Liu <liuchangsheng@inspur.com> wrote:
>
>
> On 9/1/2015 23:37, Yasuaki Ishimatsu wrote:
> > On Mon, 31 Aug 2015 01:58:40 -0400
> > Changsheng Liu <liuchangsheng@inspur.com> wrote:
> >
> >> From: Changsheng Liu <liuchangcheng@inspur.com>
> >>
> >> After the user config CONFIG_MOVABLE_NODE and movable_node kernel option,
> >> When the memory is hot added, should_add_memory_movable() return 0
> >> because all zones including movable zone are empty,
> >> so the memory that was hot added will be added to the normal zone
> >> and the normal zone will be created firstly.
> >> But we want the whole node to be added to movable zone defaultly.
> >>
> >> So we change should_add_memory_movable(): if the user config
> >> CONFIG_MOVABLE_NODE and movable_node kernel option
> >> it will always return 1 and all zones is empty at the same time,
> >> so that the movable zone will be created firstly
> >> and then the whole node will be added to movable zone defaultly.
> >> If we want the node to be added to normal zone,
> >> we can do it as follows:
> >> "echo online_kernel > /sys/devices/system/memory/memoryXXX/state"
> >>
> >> If the memory is added to movable zone defaultly,
> >> the user can offline it and add it to other zone again.
> >> But if the memory is added to normal zone defaultly,
> >> the user will not offline the memory used by kernel.
> >>
> >> Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
> >> Reviewed-by: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
> >> Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
> >> Reviewed-by: Xiaofeng Yan <yanxiaofeng@inspur.com>
> >> Signed-off-by: Changsheng Liu <liuchangcheng@inspur.com>
> >> Tested-by: Dongdong Fan <fandd@inspur.com>
> >> ---
> >> mm/memory_hotplug.c | 5 +++++
> >> 1 files changed, 5 insertions(+), 0 deletions(-)
> >>
> >> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> >> index 26fbba7..d1149ff 100644
> >> --- a/mm/memory_hotplug.c
> >> +++ b/mm/memory_hotplug.c
> >> @@ -1197,6 +1197,11 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
> >> unsigned long start_pfn = start >> PAGE_SHIFT;
> >> pg_data_t *pgdat = NODE_DATA(nid);
> >> struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
> >> + struct zone *normal_zone = pgdat->node_zones + ZONE_NORMAL;
> >> +
> >> + if (movable_node_is_enabled()
> >> + && (zone_end_pfn(normal_zone) <= start_pfn))
> >> + return 1;
> > If system boots up without movable_node, kernel behavior is changed by the patch.
> > And you syould consider other zone.
> >
> > How about it. The patch is no build and test.
> >
> >
> > ---
> > mm/memory_hotplug.c | 36 ++++++++++++++++++++++++++++++++----
> > 1 files changed, 32 insertions(+), 4 deletions(-)
> >
> > diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> > index 6da82bc..321595d 100644
> > --- a/mm/memory_hotplug.c
> > +++ b/mm/memory_hotplug.c
> > @@ -1198,6 +1198,8 @@ static int check_hotplug_memory_range(u64 start, u64 size)
> > /*
> > * If movable zone has already been setup, newly added memory should be check.
> > * If its address is higher than movable zone, it should be added as movable.
> > + * And if system boots up with movable_zone and added memory does not overlap
> > + * other zone except for movable zone, the memory is added as movable.
> > * Without this check, movable zone may overlap with other zone.
> > */
> > static int should_add_memory_movable(int nid, u64 start, u64 size)
> > @@ -1205,14 +1207,40 @@ static int should_add_memory_movable(int nid, u64 start, u64 size)
> > unsigned long start_pfn = start >> PAGE_SHIFT;
> > pg_data_t *pgdat = NODE_DATA(nid);
> > struct zone *movable_zone = pgdat->node_zones + ZONE_MOVABLE;
> > + struct zone *zone;
> > + enum zone_type zt = ZONE_MOVABLE - 1;
> > +
> > + /*
> > + * If memory is added after ZONE_MOVALBE, the memory is managed as
> > + * movable.
> > + */
> > + if (!zone_is_empty(movable_zone) &&
> > + (movable_zone->zone_start_pfn <= start_pfn))
> > + return 1;
> >
> > - if (zone_is_empty(movable_zone))
> > + if (!movable_node_is_enabled())
> > return 0;
> >
> > - if (movable_zone->zone_start_pfn <= start_pfn)
> > - return 1;
> > + /*
> > + * Find enabled zone and check the added memory.
> > + * If the memory is added after the enabled zone, the memory is
> > + * managed as movable.
> > + *
> > + * If all zones are empty, the memory is also managed as movable.
> > + */
> > + for (; zt >= ZONE_DMA; zt--) {
> > + zone = pgdat->node_zones + zt;
> >
> > - return 0;
> > + if (zone_is_empty(zone))
> > + continue;
> > +
> > + if (zone_end_pfn(zone) <= start_pfn)
> > + return 1;
> > + else
> > + return 0;
> > + }
> > +
> > + return 1;
> > }
> >
> The function zone_for_memory() adds the memory to
> ZONE_NORMAL(x86_64)/ZONE_HIGH(x86_32) defaultly, So I think the system
> just need check the added-memory is whether after the ZONE_NORMAL/ZONE_HIGH
> > int zone_for_memory(int nid, u64 start, u64 size, int zone_default)
>
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2015-09-11 19:42 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <0bc3aaab6cea54112f1c444880f9b832@s.corp-email.com>
2015-08-31 5:58 ` [PATCH V4] mm: memory hot-add: memory can not be added to movable zone defaultly Changsheng Liu
2015-08-31 13:08 ` Vlastimil Babka
2015-09-01 6:05 ` Changsheng Liu
2015-09-01 15:37 ` Yasuaki Ishimatsu
2015-09-02 9:06 ` Changsheng Liu
2015-09-06 8:33 ` Changsheng Liu
2015-09-11 19:42 ` Yasuaki Ishimatsu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).