Current ACPI cpu hotplug driver fails to associate hot-added CPUs with
corresponding NUMA node when doing socket online. The code path to
associate CPU with NUMA node is as below:
acpi_processor_add()
->acpi_processor_get_info()
->acpi_processor_hotadd_init()
->acpi_map_lsapic()
->_acpi_map_lsapic()
->acpi_map_cpu2node()
cpu_subsys_online()
->try_online_node()
->node_set_online()
When doing socket online, a new NUMA node is introduced in addition to
hot-added CPU and memory device. And the new NUMA node is marked as
online when onlining hot-added CPUs through sysfs interface
/sys/devices/system/cpu/cpuxx/online.
On the other hand, acpi_map_cpu2node() will only build the CPU to node
map if corresponding NUMA node is already online, so it always fails
to associate hot-added CPUs with corresponding NUMA node because the
NUMA node is still in offline state.
For the fix, we could safely remove the "node_online(node)" check in
function acpi_map_cpu2node() because it's only called for hot-added CPUs
by acpi_processor_hotadd_init().
Signed-off-by: Jiang Liu <[email protected]>
---
arch/x86/kernel/acpi/boot.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 6c0b43b..7625de9 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -614,10 +614,10 @@ static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
int nid;
nid = acpi_get_node(handle);
- if (nid == -1 || !node_online(nid))
- return;
- set_apicid_to_node(physid, nid);
- numa_set_node(cpu, nid);
+ if (nid != -1) {
+ set_apicid_to_node(physid, nid);
+ numa_set_node(cpu, nid);
+ }
#endif
}
--
1.7.10.4
On Monday, January 20, 2014 10:31:54 AM Jiang Liu wrote:
> Current ACPI cpu hotplug driver fails to associate hot-added CPUs with
> corresponding NUMA node when doing socket online. The code path to
> associate CPU with NUMA node is as below:
> acpi_processor_add()
> ->acpi_processor_get_info()
> ->acpi_processor_hotadd_init()
> ->acpi_map_lsapic()
> ->_acpi_map_lsapic()
> ->acpi_map_cpu2node()
> cpu_subsys_online()
> ->try_online_node()
> ->node_set_online()
>
> When doing socket online, a new NUMA node is introduced in addition to
> hot-added CPU and memory device. And the new NUMA node is marked as
> online when onlining hot-added CPUs through sysfs interface
> /sys/devices/system/cpu/cpuxx/online.
>
> On the other hand, acpi_map_cpu2node() will only build the CPU to node
> map if corresponding NUMA node is already online, so it always fails
> to associate hot-added CPUs with corresponding NUMA node because the
> NUMA node is still in offline state.
>
> For the fix, we could safely remove the "node_online(node)" check in
> function acpi_map_cpu2node() because it's only called for hot-added CPUs
> by acpi_processor_hotadd_init().
>
> Signed-off-by: Jiang Liu <[email protected]>
I wonder what the status here is? Did this patch go anywhere?
> ---
> arch/x86/kernel/acpi/boot.c | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> index 6c0b43b..7625de9 100644
> --- a/arch/x86/kernel/acpi/boot.c
> +++ b/arch/x86/kernel/acpi/boot.c
> @@ -614,10 +614,10 @@ static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
> int nid;
>
> nid = acpi_get_node(handle);
> - if (nid == -1 || !node_online(nid))
> - return;
> - set_apicid_to_node(physid, nid);
> - numa_set_node(cpu, nid);
> + if (nid != -1) {
> + set_apicid_to_node(physid, nid);
> + numa_set_node(cpu, nid);
> + }
> #endif
> }
>
>
--
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.
On 2014/2/5 8:14, Rafael J. Wysocki wrote:
> On Monday, January 20, 2014 10:31:54 AM Jiang Liu wrote:
>> Current ACPI cpu hotplug driver fails to associate hot-added CPUs with
>> corresponding NUMA node when doing socket online. The code path to
>> associate CPU with NUMA node is as below:
>> acpi_processor_add()
>> ->acpi_processor_get_info()
>> ->acpi_processor_hotadd_init()
>> ->acpi_map_lsapic()
>> ->_acpi_map_lsapic()
>> ->acpi_map_cpu2node()
>> cpu_subsys_online()
>> ->try_online_node()
>> ->node_set_online()
>>
>> When doing socket online, a new NUMA node is introduced in addition to
>> hot-added CPU and memory device. And the new NUMA node is marked as
>> online when onlining hot-added CPUs through sysfs interface
>> /sys/devices/system/cpu/cpuxx/online.
>>
>> On the other hand, acpi_map_cpu2node() will only build the CPU to node
>> map if corresponding NUMA node is already online, so it always fails
>> to associate hot-added CPUs with corresponding NUMA node because the
>> NUMA node is still in offline state.
>>
>> For the fix, we could safely remove the "node_online(node)" check in
>> function acpi_map_cpu2node() because it's only called for hot-added CPUs
>> by acpi_processor_hotadd_init().
>>
>> Signed-off-by: Jiang Liu <[email protected]>
>
> I wonder what the status here is? Did this patch go anywhere?
Hi Rafael,
It's still in review stage, hasn't been accepted by any
maintainer yet.
Thanks!
Gerry
>
>> ---
>> arch/x86/kernel/acpi/boot.c | 8 ++++----
>> 1 file changed, 4 insertions(+), 4 deletions(-)
>>
>> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
>> index 6c0b43b..7625de9 100644
>> --- a/arch/x86/kernel/acpi/boot.c
>> +++ b/arch/x86/kernel/acpi/boot.c
>> @@ -614,10 +614,10 @@ static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
>> int nid;
>>
>> nid = acpi_get_node(handle);
>> - if (nid == -1 || !node_online(nid))
>> - return;
>> - set_apicid_to_node(physid, nid);
>> - numa_set_node(cpu, nid);
>> + if (nid != -1) {
>> + set_apicid_to_node(physid, nid);
>> + numa_set_node(cpu, nid);
>> + }
>> #endif
>> }
>>
>>
>
On Friday, February 07, 2014 05:17:45 PM Jiang Liu wrote:
>
> On 2014/2/5 8:14, Rafael J. Wysocki wrote:
> > On Monday, January 20, 2014 10:31:54 AM Jiang Liu wrote:
> >> Current ACPI cpu hotplug driver fails to associate hot-added CPUs with
> >> corresponding NUMA node when doing socket online. The code path to
> >> associate CPU with NUMA node is as below:
> >> acpi_processor_add()
> >> ->acpi_processor_get_info()
> >> ->acpi_processor_hotadd_init()
> >> ->acpi_map_lsapic()
> >> ->_acpi_map_lsapic()
> >> ->acpi_map_cpu2node()
> >> cpu_subsys_online()
> >> ->try_online_node()
> >> ->node_set_online()
> >>
> >> When doing socket online, a new NUMA node is introduced in addition to
> >> hot-added CPU and memory device. And the new NUMA node is marked as
> >> online when onlining hot-added CPUs through sysfs interface
> >> /sys/devices/system/cpu/cpuxx/online.
> >>
> >> On the other hand, acpi_map_cpu2node() will only build the CPU to node
> >> map if corresponding NUMA node is already online, so it always fails
> >> to associate hot-added CPUs with corresponding NUMA node because the
> >> NUMA node is still in offline state.
> >>
> >> For the fix, we could safely remove the "node_online(node)" check in
> >> function acpi_map_cpu2node() because it's only called for hot-added CPUs
> >> by acpi_processor_hotadd_init().
> >>
> >> Signed-off-by: Jiang Liu <[email protected]>
> >
> > I wonder what the status here is? Did this patch go anywhere?
> Hi Rafael,
> It's still in review stage, hasn't been accepted by any
> maintainer yet.
OK
Peter, are you fine with the patch below?
Rafael
> >
> >> ---
> >> arch/x86/kernel/acpi/boot.c | 8 ++++----
> >> 1 file changed, 4 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> >> index 6c0b43b..7625de9 100644
> >> --- a/arch/x86/kernel/acpi/boot.c
> >> +++ b/arch/x86/kernel/acpi/boot.c
> >> @@ -614,10 +614,10 @@ static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
> >> int nid;
> >>
> >> nid = acpi_get_node(handle);
> >> - if (nid == -1 || !node_online(nid))
> >> - return;
> >> - set_apicid_to_node(physid, nid);
> >> - numa_set_node(cpu, nid);
> >> + if (nid != -1) {
> >> + set_apicid_to_node(physid, nid);
> >> + numa_set_node(cpu, nid);
> >> + }
> >> #endif
> >> }
> >>
> >>
> >
--
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.
Ping...
On 2014/2/7 20:03, Rafael J. Wysocki wrote:
> On Friday, February 07, 2014 05:17:45 PM Jiang Liu wrote:
>>
>> On 2014/2/5 8:14, Rafael J. Wysocki wrote:
>>> On Monday, January 20, 2014 10:31:54 AM Jiang Liu wrote:
>>>> Current ACPI cpu hotplug driver fails to associate hot-added CPUs with
>>>> corresponding NUMA node when doing socket online. The code path to
>>>> associate CPU with NUMA node is as below:
>>>> acpi_processor_add()
>>>> ->acpi_processor_get_info()
>>>> ->acpi_processor_hotadd_init()
>>>> ->acpi_map_lsapic()
>>>> ->_acpi_map_lsapic()
>>>> ->acpi_map_cpu2node()
>>>> cpu_subsys_online()
>>>> ->try_online_node()
>>>> ->node_set_online()
>>>>
>>>> When doing socket online, a new NUMA node is introduced in addition to
>>>> hot-added CPU and memory device. And the new NUMA node is marked as
>>>> online when onlining hot-added CPUs through sysfs interface
>>>> /sys/devices/system/cpu/cpuxx/online.
>>>>
>>>> On the other hand, acpi_map_cpu2node() will only build the CPU to node
>>>> map if corresponding NUMA node is already online, so it always fails
>>>> to associate hot-added CPUs with corresponding NUMA node because the
>>>> NUMA node is still in offline state.
>>>>
>>>> For the fix, we could safely remove the "node_online(node)" check in
>>>> function acpi_map_cpu2node() because it's only called for hot-added CPUs
>>>> by acpi_processor_hotadd_init().
>>>>
>>>> Signed-off-by: Jiang Liu <[email protected]>
>>>
>>> I wonder what the status here is? Did this patch go anywhere?
>> Hi Rafael,
>> It's still in review stage, hasn't been accepted by any
>> maintainer yet.
>
> OK
>
> Peter, are you fine with the patch below?
>
> Rafael
>
>
>>>
>>>> ---
>>>> arch/x86/kernel/acpi/boot.c | 8 ++++----
>>>> 1 file changed, 4 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
>>>> index 6c0b43b..7625de9 100644
>>>> --- a/arch/x86/kernel/acpi/boot.c
>>>> +++ b/arch/x86/kernel/acpi/boot.c
>>>> @@ -614,10 +614,10 @@ static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
>>>> int nid;
>>>>
>>>> nid = acpi_get_node(handle);
>>>> - if (nid == -1 || !node_online(nid))
>>>> - return;
>>>> - set_apicid_to_node(physid, nid);
>>>> - numa_set_node(cpu, nid);
>>>> + if (nid != -1) {
>>>> + set_apicid_to_node(physid, nid);
>>>> + numa_set_node(cpu, nid);
>>>> + }
>>>> #endif
>>>> }
>>>>
>>>>
>>>
>
On 02/12/2014 06:32 PM, Jiang Liu wrote:
> Ping...
Sorry, will look at it tomorrow.
-hpa
Commit-ID: 896dc5064083063981954f142e43548765199792
Gitweb: http://git.kernel.org/tip/896dc5064083063981954f142e43548765199792
Author: Jiang Liu <[email protected]>
AuthorDate: Mon, 20 Jan 2014 10:31:54 +0800
Committer: H. Peter Anvin <[email protected]>
CommitDate: Thu, 20 Feb 2014 19:01:22 -0800
x86, acpi: Fix bug in associating hot-added CPUs with corresponding NUMA node
Current ACPI cpu hotplug driver fails to associate hot-added CPUs with
corresponding NUMA node when doing socket online. The code path to
associate CPU with NUMA node is as below:
acpi_processor_add()
->acpi_processor_get_info()
->acpi_processor_hotadd_init()
->acpi_map_lsapic()
->_acpi_map_lsapic()
->acpi_map_cpu2node()
cpu_subsys_online()
->try_online_node()
->node_set_online()
When doing socket online, a new NUMA node is introduced in addition to
hot-added CPU and memory device. And the new NUMA node is marked as
online when onlining hot-added CPUs through sysfs interface
/sys/devices/system/cpu/cpuxx/online.
On the other hand, acpi_map_cpu2node() will only build the CPU to node
map if corresponding NUMA node is already online, so it always fails
to associate hot-added CPUs with corresponding NUMA node because the
NUMA node is still in offline state.
For the fix, we could safely remove the "node_online(node)" check in
function acpi_map_cpu2node() because it's only called for hot-added CPUs
by acpi_processor_hotadd_init().
Signed-off-by: Jiang Liu <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Acked-by: Rafael J. Wysocki <[email protected]>
Signed-off-by: H. Peter Anvin <[email protected]>
---
arch/x86/kernel/acpi/boot.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 1dac942..9f46f2b 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -613,10 +613,10 @@ static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid)
int nid;
nid = acpi_get_node(handle);
- if (nid == -1 || !node_online(nid))
- return;
- set_apicid_to_node(physid, nid);
- numa_set_node(cpu, nid);
+ if (nid != -1) {
+ set_apicid_to_node(physid, nid);
+ numa_set_node(cpu, nid);
+ }
#endif
}