* [PATCH v4] x86,cpu-hotplug: assign same CPU number to readded CPU
@ 2014-07-23 3:22 Yasuaki Ishimatsu
2014-07-28 4:18 ` Yasuaki Ishimatsu
0 siblings, 1 reply; 9+ messages in thread
From: Yasuaki Ishimatsu @ 2014-07-23 3:22 UTC (permalink / raw)
To: tglx, mingo, hpa, bp, gong.chen, tony.luck, toshi.kani
Cc: x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, umgwanakikbuti, peterz
llc_shared_map is not cleared even if CPU is offline or hot removed.
So when hot-plugging CPU and assigning new CPU number to hot-added CPU,
the mask has wrong value. The mask is used by CSF schduler to create
sched_domain. So it breaks CFS scheduler.
Here is a example on my system.
My system has 4 sockets and each socket has 15 cores and HT is enabled.
In this case, each core of sockes is numbered as follows:
| CPU#
Socket#0 | 0-14 , 60-74
Socket#1 | 15-29, 75-89
Socket#2 | 30-44, 90-104
Socket#3 | 45-59, 105-119
Then llc_shared_mask of CPU#30 has 0x3fff80000001fffc0000000.
It means that last level cache of Socket#2 is shared with
CPU#30-44 and 90-104.
When hot-removing socket#2 and #3, each core of sockets is numbered
as follows:
| CPU#
Socket#0 | 0-14 , 60-74
Socket#1 | 15-29, 75-89
But llc_shared_mask is not cleared. So llc_shared_mask of CPU#30 remains
having 0x3fff80000001fffc0000000.
After that, when hot-adding socket#2 and #3, each core of sockets is
numbered as follows:
| CPU#
Socket#0 | 0-14 , 60-74
Socket#1 | 15-29, 75-89
Socket#2 | 30-59
Socket#3 | 90-119
Then llc_shared_mask of CPU#30 becomes 0x3fff8000fffffffc0000000.
It means that last level cache of Socket#2 is shared with CPU#30-59
and 90-104. So the mask has wrong value.
At first, I cleared hot-removed CPU number's bit from llc_shared_map
when hot removing CPU. But Borislav suggested that the problem will
disappear if readded CPU is assigned same CPU number. And llc_shared_map
must not be changed.
So the patch assigns same CPU number to readded CPU by linking CPU
number to APIC ID. And by the patch, the problem disappers.
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Suggested-by: Borislav Petkov <bp@alien8.de>
Reviewed-by: Toshi Kani <toshi.kani@hp.com>
---
v2: change cpuid to cpunum
v3: fix Borislav's email address of Suggested-by
fix typo (ACPI ID to APIC ID)
v4: change cpu_used_xxx to cpu_number_xxx
---
arch/x86/kernel/apic/apic.c | 33 ++++++++++++++++++++++++++++++++-
1 file changed, 32 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index ad28db7..5cecc3b 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -220,6 +220,23 @@ static void apic_pm_activate(void);
static unsigned long apic_phys;
/*
+ * Bind APIC ID to Logical CPU number
+ * Logical CPU number to APIC ID does not change by this array
+ * even if CPU is hotplugged. So don't clear the array even if
+ * CPU is hot-removed
+ */
+static int apicid_to_cpunum[MAX_LOCAL_APIC] = {
+ [0 ... MAX_LOCAL_APIC-1] = -1,
+};
+
+/*
+ * Represent Logical CPU number bound to APIC ID
+ * Don't clear a bit even if CPU is hot-removed
+ */
+static DECLARE_BITMAP(cpu_number_bits, CONFIG_NR_CPUS);
+static struct cpumask *const cpu_number_mask = to_cpumask(cpu_number_bits);
+
+/*
* Get the LAPIC version
*/
static inline int lapic_get_version(void)
@@ -2122,6 +2139,17 @@ void disconnect_bsp_APIC(int virt_wire_setup)
apic_write(APIC_LVT1, value);
}
+static int get_cpunum(int apicid)
+{
+ int cpu;
+
+ cpu = apicid_to_cpunum[apicid];
+ if (cpu < 0)
+ cpu = cpumask_next_zero(-1, cpu_number_mask);
+
+ return cpu;
+}
+
int generic_processor_info(int apicid, int version)
{
int cpu, max = nr_cpu_ids;
@@ -2199,7 +2227,9 @@ int generic_processor_info(int apicid, int version)
*/
cpu = 0;
} else
- cpu = cpumask_next_zero(-1, cpu_present_mask);
+ cpu = get_cpunum(apicid);
+
+ apicid_to_cpunum[apicid] = cpu;
/*
* Validate version
@@ -2228,6 +2258,7 @@ int generic_processor_info(int apicid, int version)
early_per_cpu(x86_cpu_to_logical_apicid, cpu) =
apic->x86_32_early_logical_apicid(cpu);
#endif
+ cpumask_set_cpu(cpu, cpu_number_mask);
set_cpu_possible(cpu, true);
set_cpu_present(cpu, true);
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH v4] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-07-23 3:22 [PATCH v4] x86,cpu-hotplug: assign same CPU number to readded CPU Yasuaki Ishimatsu
@ 2014-07-28 4:18 ` Yasuaki Ishimatsu
2014-09-04 2:46 ` [PATCH v5] " Yasuaki Ishimatsu
0 siblings, 1 reply; 9+ messages in thread
From: Yasuaki Ishimatsu @ 2014-07-28 4:18 UTC (permalink / raw)
To: tglx, mingo, hpa, bp, gong.chen, tony.luck, toshi.kani
Cc: x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, umgwanakikbuti, peterz
Ping.
(2014/07/23 12:22), Yasuaki Ishimatsu wrote:
> llc_shared_map is not cleared even if CPU is offline or hot removed.
> So when hot-plugging CPU and assigning new CPU number to hot-added CPU,
> the mask has wrong value. The mask is used by CSF schduler to create
> sched_domain. So it breaks CFS scheduler.
>
> Here is a example on my system.
> My system has 4 sockets and each socket has 15 cores and HT is enabled.
> In this case, each core of sockes is numbered as follows:
>
> | CPU#
> Socket#0 | 0-14 , 60-74
> Socket#1 | 15-29, 75-89
> Socket#2 | 30-44, 90-104
> Socket#3 | 45-59, 105-119
>
> Then llc_shared_mask of CPU#30 has 0x3fff80000001fffc0000000.
> It means that last level cache of Socket#2 is shared with
> CPU#30-44 and 90-104.
>
> When hot-removing socket#2 and #3, each core of sockets is numbered
> as follows:
>
> | CPU#
> Socket#0 | 0-14 , 60-74
> Socket#1 | 15-29, 75-89
>
> But llc_shared_mask is not cleared. So llc_shared_mask of CPU#30 remains
> having 0x3fff80000001fffc0000000.
>
> After that, when hot-adding socket#2 and #3, each core of sockets is
> numbered as follows:
>
> | CPU#
> Socket#0 | 0-14 , 60-74
> Socket#1 | 15-29, 75-89
> Socket#2 | 30-59
> Socket#3 | 90-119
>
> Then llc_shared_mask of CPU#30 becomes 0x3fff8000fffffffc0000000.
> It means that last level cache of Socket#2 is shared with CPU#30-59
> and 90-104. So the mask has wrong value.
>
> At first, I cleared hot-removed CPU number's bit from llc_shared_map
> when hot removing CPU. But Borislav suggested that the problem will
> disappear if readded CPU is assigned same CPU number. And llc_shared_map
> must not be changed.
>
> So the patch assigns same CPU number to readded CPU by linking CPU
> number to APIC ID. And by the patch, the problem disappers.
>
> Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
> Suggested-by: Borislav Petkov <bp@alien8.de>
> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
> ---
> v2: change cpuid to cpunum
> v3: fix Borislav's email address of Suggested-by
> fix typo (ACPI ID to APIC ID)
> v4: change cpu_used_xxx to cpu_number_xxx
> ---
> arch/x86/kernel/apic/apic.c | 33 ++++++++++++++++++++++++++++++++-
> 1 file changed, 32 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
> index ad28db7..5cecc3b 100644
> --- a/arch/x86/kernel/apic/apic.c
> +++ b/arch/x86/kernel/apic/apic.c
> @@ -220,6 +220,23 @@ static void apic_pm_activate(void);
> static unsigned long apic_phys;
>
> /*
> + * Bind APIC ID to Logical CPU number
> + * Logical CPU number to APIC ID does not change by this array
> + * even if CPU is hotplugged. So don't clear the array even if
> + * CPU is hot-removed
> + */
> +static int apicid_to_cpunum[MAX_LOCAL_APIC] = {
> + [0 ... MAX_LOCAL_APIC-1] = -1,
> +};
> +
> +/*
> + * Represent Logical CPU number bound to APIC ID
> + * Don't clear a bit even if CPU is hot-removed
> + */
> +static DECLARE_BITMAP(cpu_number_bits, CONFIG_NR_CPUS);
> +static struct cpumask *const cpu_number_mask = to_cpumask(cpu_number_bits);
> +
> +/*
> * Get the LAPIC version
> */
> static inline int lapic_get_version(void)
> @@ -2122,6 +2139,17 @@ void disconnect_bsp_APIC(int virt_wire_setup)
> apic_write(APIC_LVT1, value);
> }
>
> +static int get_cpunum(int apicid)
> +{
> + int cpu;
> +
> + cpu = apicid_to_cpunum[apicid];
> + if (cpu < 0)
> + cpu = cpumask_next_zero(-1, cpu_number_mask);
> +
> + return cpu;
> +}
> +
> int generic_processor_info(int apicid, int version)
> {
> int cpu, max = nr_cpu_ids;
> @@ -2199,7 +2227,9 @@ int generic_processor_info(int apicid, int version)
> */
> cpu = 0;
> } else
> - cpu = cpumask_next_zero(-1, cpu_present_mask);
> + cpu = get_cpunum(apicid);
> +
> + apicid_to_cpunum[apicid] = cpu;
>
> /*
> * Validate version
> @@ -2228,6 +2258,7 @@ int generic_processor_info(int apicid, int version)
> early_per_cpu(x86_cpu_to_logical_apicid, cpu) =
> apic->x86_32_early_logical_apicid(cpu);
> #endif
> + cpumask_set_cpu(cpu, cpu_number_mask);
> set_cpu_possible(cpu, true);
> set_cpu_present(cpu, true);
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-07-28 4:18 ` Yasuaki Ishimatsu
@ 2014-09-04 2:46 ` Yasuaki Ishimatsu
2014-09-11 7:21 ` Yasuaki Ishimatsu
0 siblings, 1 reply; 9+ messages in thread
From: Yasuaki Ishimatsu @ 2014-09-04 2:46 UTC (permalink / raw)
To: tglx, mingo, hpa, bp, gong.chen, tony.luck, toshi.kani
Cc: x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, umgwanakikbuti, peterz
llc_shared_map is not cleared even if CPU is offline or hot removed.
So when hot-plugging CPU and assigning new CPU number to hot-added CPU,
the mask has wrong value. The mask is used by CSF schduler to create
sched_domain. So it breaks CFS scheduler.
Here is a example on my system.
My system has 4 sockets and each socket has 15 cores and HT is enabled.
In this case, each core of sockes is numbered as follows:
| CPU#
Socket#0 | 0-14 , 60-74
Socket#1 | 15-29, 75-89
Socket#2 | 30-44, 90-104
Socket#3 | 45-59, 105-119
Then llc_shared_mask of CPU#30 has 0x3fff80000001fffc0000000.
It means that last level cache of Socket#2 is shared with
CPU#30-44 and 90-104.
When hot-removing socket#2 and #3, each core of sockets is numbered
as follows:
| CPU#
Socket#0 | 0-14 , 60-74
Socket#1 | 15-29, 75-89
But llc_shared_mask is not cleared. So llc_shared_mask of CPU#30 remains
having 0x3fff80000001fffc0000000.
After that, when hot-adding socket#2 and #3, each core of sockets is
numbered as follows:
| CPU#
Socket#0 | 0-14 , 60-74
Socket#1 | 15-29, 75-89
Socket#2 | 30-59
Socket#3 | 90-119
Then llc_shared_mask of CPU#30 becomes 0x3fff8000fffffffc0000000.
It means that last level cache of Socket#2 is shared with CPU#30-59
and 90-104. So the mask has wrong value.
At first, I cleared hot-removed CPU number's bit from llc_shared_map
when hot removing CPU. But Borislav suggested that the problem will
disappear if readded CPU is assigned same CPU number. And llc_shared_map
must not be changed.
So the patch assigns same CPU number to readded CPU by linking CPU
number to APIC ID. And by the patch, the problem disappers.
Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Suggested-by: Borislav Petkov <bp@alien8.de>
Reviewed-by: Toshi Kani <toshi.kani@hp.com>
---
v2: change cpuid to cpunum
v3: fix Borislav's email address of Suggested-by
fix typo (ACPI ID to APIC ID)
v4: change cpu_used_xxx to cpu_number_xxx
v5: rebase to 3.17-rc3
---
arch/x86/kernel/apic/apic.c | 33 ++++++++++++++++++++++++++++++++-
1 file changed, 32 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 6776027..c476827 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -220,6 +220,23 @@ static void apic_pm_activate(void);
static unsigned long apic_phys;
/*
+ * Bind APIC ID to Logical CPU number
+ * Logical CPU number to APIC ID does not change by this array
+ * even if CPU is hotplugged. So don't clear the array even if
+ * CPU is hot-removed
+ */
+static int apicid_to_cpunum[MAX_LOCAL_APIC] = {
+ [0 ... MAX_LOCAL_APIC-1] = -1,
+};
+
+/*
+ * Represent Logical CPU number bound to APIC ID
+ * Don't clear a bit even if CPU is hot-removed
+ */
+static DECLARE_BITMAP(cpu_number_bits, CONFIG_NR_CPUS);
+static struct cpumask *const cpu_number_mask = to_cpumask(cpu_number_bits);
+
+/*
* Get the LAPIC version
*/
static inline int lapic_get_version(void)
@@ -2109,6 +2126,17 @@ void disconnect_bsp_APIC(int virt_wire_setup)
apic_write(APIC_LVT1, value);
}
+static int get_cpunum(int apicid)
+{
+ int cpu;
+
+ cpu = apicid_to_cpunum[apicid];
+ if (cpu < 0)
+ cpu = cpumask_next_zero(-1, cpu_number_mask);
+
+ return cpu;
+}
+
int generic_processor_info(int apicid, int version)
{
int cpu, max = nr_cpu_ids;
@@ -2186,7 +2214,9 @@ int generic_processor_info(int apicid, int version)
*/
cpu = 0;
} else
- cpu = cpumask_next_zero(-1, cpu_present_mask);
+ cpu = get_cpunum(apicid);
+
+ apicid_to_cpunum[apicid] = cpu;
/*
* Validate version
@@ -2215,6 +2245,7 @@ int generic_processor_info(int apicid, int version)
early_per_cpu(x86_cpu_to_logical_apicid, cpu) =
apic->x86_32_early_logical_apicid(cpu);
#endif
+ cpumask_set_cpu(cpu, cpu_number_mask);
set_cpu_possible(cpu, true);
set_cpu_present(cpu, true);
--
1.8.3.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-09-04 2:46 ` [PATCH v5] " Yasuaki Ishimatsu
@ 2014-09-11 7:21 ` Yasuaki Ishimatsu
2014-09-15 4:25 ` Mike Galbraith
0 siblings, 1 reply; 9+ messages in thread
From: Yasuaki Ishimatsu @ 2014-09-11 7:21 UTC (permalink / raw)
To: tglx, mingo, hpa, bp, gong.chen, tony.luck, toshi.kani
Cc: x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, umgwanakikbuti, peterz
There is no response for two months since posting v4.
What can I do for pushing the patch to upstream?
Thanks,
Yasuaki Ishimatsu
(2014/09/04 11:46), Yasuaki Ishimatsu wrote:
> llc_shared_map is not cleared even if CPU is offline or hot removed.
> So when hot-plugging CPU and assigning new CPU number to hot-added CPU,
> the mask has wrong value. The mask is used by CSF schduler to create
> sched_domain. So it breaks CFS scheduler.
>
> Here is a example on my system.
> My system has 4 sockets and each socket has 15 cores and HT is enabled.
> In this case, each core of sockes is numbered as follows:
>
> | CPU#
> Socket#0 | 0-14 , 60-74
> Socket#1 | 15-29, 75-89
> Socket#2 | 30-44, 90-104
> Socket#3 | 45-59, 105-119
>
> Then llc_shared_mask of CPU#30 has 0x3fff80000001fffc0000000.
> It means that last level cache of Socket#2 is shared with
> CPU#30-44 and 90-104.
>
> When hot-removing socket#2 and #3, each core of sockets is numbered
> as follows:
>
> | CPU#
> Socket#0 | 0-14 , 60-74
> Socket#1 | 15-29, 75-89
>
> But llc_shared_mask is not cleared. So llc_shared_mask of CPU#30 remains
> having 0x3fff80000001fffc0000000.
>
> After that, when hot-adding socket#2 and #3, each core of sockets is
> numbered as follows:
>
> | CPU#
> Socket#0 | 0-14 , 60-74
> Socket#1 | 15-29, 75-89
> Socket#2 | 30-59
> Socket#3 | 90-119
>
> Then llc_shared_mask of CPU#30 becomes 0x3fff8000fffffffc0000000.
> It means that last level cache of Socket#2 is shared with CPU#30-59
> and 90-104. So the mask has wrong value.
>
> At first, I cleared hot-removed CPU number's bit from llc_shared_map
> when hot removing CPU. But Borislav suggested that the problem will
> disappear if readded CPU is assigned same CPU number. And llc_shared_map
> must not be changed.
>
> So the patch assigns same CPU number to readded CPU by linking CPU
> number to APIC ID. And by the patch, the problem disappers.
>
> Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
> Suggested-by: Borislav Petkov <bp@alien8.de>
> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
> ---
> v2: change cpuid to cpunum
> v3: fix Borislav's email address of Suggested-by
> fix typo (ACPI ID to APIC ID)
> v4: change cpu_used_xxx to cpu_number_xxx
> v5: rebase to 3.17-rc3
> ---
> arch/x86/kernel/apic/apic.c | 33 ++++++++++++++++++++++++++++++++-
> 1 file changed, 32 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
> index 6776027..c476827 100644
> --- a/arch/x86/kernel/apic/apic.c
> +++ b/arch/x86/kernel/apic/apic.c
> @@ -220,6 +220,23 @@ static void apic_pm_activate(void);
> static unsigned long apic_phys;
>
> /*
> + * Bind APIC ID to Logical CPU number
> + * Logical CPU number to APIC ID does not change by this array
> + * even if CPU is hotplugged. So don't clear the array even if
> + * CPU is hot-removed
> + */
> +static int apicid_to_cpunum[MAX_LOCAL_APIC] = {
> + [0 ... MAX_LOCAL_APIC-1] = -1,
> +};
> +
> +/*
> + * Represent Logical CPU number bound to APIC ID
> + * Don't clear a bit even if CPU is hot-removed
> + */
> +static DECLARE_BITMAP(cpu_number_bits, CONFIG_NR_CPUS);
> +static struct cpumask *const cpu_number_mask = to_cpumask(cpu_number_bits);
> +
> +/*
> * Get the LAPIC version
> */
> static inline int lapic_get_version(void)
> @@ -2109,6 +2126,17 @@ void disconnect_bsp_APIC(int virt_wire_setup)
> apic_write(APIC_LVT1, value);
> }
>
> +static int get_cpunum(int apicid)
> +{
> + int cpu;
> +
> + cpu = apicid_to_cpunum[apicid];
> + if (cpu < 0)
> + cpu = cpumask_next_zero(-1, cpu_number_mask);
> +
> + return cpu;
> +}
> +
> int generic_processor_info(int apicid, int version)
> {
> int cpu, max = nr_cpu_ids;
> @@ -2186,7 +2214,9 @@ int generic_processor_info(int apicid, int version)
> */
> cpu = 0;
> } else
> - cpu = cpumask_next_zero(-1, cpu_present_mask);
> + cpu = get_cpunum(apicid);
> +
> + apicid_to_cpunum[apicid] = cpu;
>
> /*
> * Validate version
> @@ -2215,6 +2245,7 @@ int generic_processor_info(int apicid, int version)
> early_per_cpu(x86_cpu_to_logical_apicid, cpu) =
> apic->x86_32_early_logical_apicid(cpu);
> #endif
> + cpumask_set_cpu(cpu, cpu_number_mask);
> set_cpu_possible(cpu, true);
> set_cpu_present(cpu, true);
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-09-11 7:21 ` Yasuaki Ishimatsu
@ 2014-09-15 4:25 ` Mike Galbraith
2014-09-15 16:44 ` Toshi Kani
0 siblings, 1 reply; 9+ messages in thread
From: Mike Galbraith @ 2014-09-15 4:25 UTC (permalink / raw)
To: Yasuaki Ishimatsu
Cc: tglx, mingo, hpa, bp, gong.chen, tony.luck, toshi.kani, x86,
imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, peterz, Wanpeng Li, Linn Crosetto
On Thu, 2014-09-11 at 16:21 +0900, Yasuaki Ishimatsu wrote:
> There is no response for two months since posting v4.
> What can I do for pushing the patch to upstream?
Looks to me like we have two patches floating about for more or less the
same problem, this one, and...
https://lkml.org/lkml/2014/7/29/159
..this one, which you reviewed, and HP both reviewed and tested.
We seem to kinda stuck with Boris having said don't diddle the
cpu_llc_shared_map, but HP/Intel saying that this map diddling fixes
their explosions. If your alternative is preferred over diddling
cpu_llc_shared_map, perhaps HP/Intel can test/confirm that their
explosions stay gone?
> Thanks,
> Yasuaki Ishimatsu
>
> (2014/09/04 11:46), Yasuaki Ishimatsu wrote:
> > llc_shared_map is not cleared even if CPU is offline or hot removed.
> > So when hot-plugging CPU and assigning new CPU number to hot-added CPU,
> > the mask has wrong value. The mask is used by CSF schduler to create
> > sched_domain. So it breaks CFS scheduler.
> >
> > Here is a example on my system.
> > My system has 4 sockets and each socket has 15 cores and HT is enabled.
> > In this case, each core of sockes is numbered as follows:
> >
> > | CPU#
> > Socket#0 | 0-14 , 60-74
> > Socket#1 | 15-29, 75-89
> > Socket#2 | 30-44, 90-104
> > Socket#3 | 45-59, 105-119
> >
> > Then llc_shared_mask of CPU#30 has 0x3fff80000001fffc0000000.
> > It means that last level cache of Socket#2 is shared with
> > CPU#30-44 and 90-104.
> >
> > When hot-removing socket#2 and #3, each core of sockets is numbered
> > as follows:
> >
> > | CPU#
> > Socket#0 | 0-14 , 60-74
> > Socket#1 | 15-29, 75-89
> >
> > But llc_shared_mask is not cleared. So llc_shared_mask of CPU#30 remains
> > having 0x3fff80000001fffc0000000.
> >
> > After that, when hot-adding socket#2 and #3, each core of sockets is
> > numbered as follows:
> >
> > | CPU#
> > Socket#0 | 0-14 , 60-74
> > Socket#1 | 15-29, 75-89
> > Socket#2 | 30-59
> > Socket#3 | 90-119
> >
> > Then llc_shared_mask of CPU#30 becomes 0x3fff8000fffffffc0000000.
> > It means that last level cache of Socket#2 is shared with CPU#30-59
> > and 90-104. So the mask has wrong value.
> >
> > At first, I cleared hot-removed CPU number's bit from llc_shared_map
> > when hot removing CPU. But Borislav suggested that the problem will
> > disappear if readded CPU is assigned same CPU number. And llc_shared_map
> > must not be changed.
> >
> > So the patch assigns same CPU number to readded CPU by linking CPU
> > number to APIC ID. And by the patch, the problem disappers.
> >
> > Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
> > Suggested-by: Borislav Petkov <bp@alien8.de>
> > Reviewed-by: Toshi Kani <toshi.kani@hp.com>
> > ---
> > v2: change cpuid to cpunum
> > v3: fix Borislav's email address of Suggested-by
> > fix typo (ACPI ID to APIC ID)
> > v4: change cpu_used_xxx to cpu_number_xxx
> > v5: rebase to 3.17-rc3
> > ---
> > arch/x86/kernel/apic/apic.c | 33 ++++++++++++++++++++++++++++++++-
> > 1 file changed, 32 insertions(+), 1 deletion(-)
> >
> > diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
> > index 6776027..c476827 100644
> > --- a/arch/x86/kernel/apic/apic.c
> > +++ b/arch/x86/kernel/apic/apic.c
> > @@ -220,6 +220,23 @@ static void apic_pm_activate(void);
> > static unsigned long apic_phys;
> >
> > /*
> > + * Bind APIC ID to Logical CPU number
> > + * Logical CPU number to APIC ID does not change by this array
> > + * even if CPU is hotplugged. So don't clear the array even if
> > + * CPU is hot-removed
> > + */
> > +static int apicid_to_cpunum[MAX_LOCAL_APIC] = {
> > + [0 ... MAX_LOCAL_APIC-1] = -1,
> > +};
> > +
> > +/*
> > + * Represent Logical CPU number bound to APIC ID
> > + * Don't clear a bit even if CPU is hot-removed
> > + */
> > +static DECLARE_BITMAP(cpu_number_bits, CONFIG_NR_CPUS);
> > +static struct cpumask *const cpu_number_mask = to_cpumask(cpu_number_bits);
> > +
> > +/*
> > * Get the LAPIC version
> > */
> > static inline int lapic_get_version(void)
> > @@ -2109,6 +2126,17 @@ void disconnect_bsp_APIC(int virt_wire_setup)
> > apic_write(APIC_LVT1, value);
> > }
> >
> > +static int get_cpunum(int apicid)
> > +{
> > + int cpu;
> > +
> > + cpu = apicid_to_cpunum[apicid];
> > + if (cpu < 0)
> > + cpu = cpumask_next_zero(-1, cpu_number_mask);
> > +
> > + return cpu;
> > +}
> > +
> > int generic_processor_info(int apicid, int version)
> > {
> > int cpu, max = nr_cpu_ids;
> > @@ -2186,7 +2214,9 @@ int generic_processor_info(int apicid, int version)
> > */
> > cpu = 0;
> > } else
> > - cpu = cpumask_next_zero(-1, cpu_present_mask);
> > + cpu = get_cpunum(apicid);
> > +
> > + apicid_to_cpunum[apicid] = cpu;
> >
> > /*
> > * Validate version
> > @@ -2215,6 +2245,7 @@ int generic_processor_info(int apicid, int version)
> > early_per_cpu(x86_cpu_to_logical_apicid, cpu) =
> > apic->x86_32_early_logical_apicid(cpu);
> > #endif
> > + cpumask_set_cpu(cpu, cpu_number_mask);
> > set_cpu_possible(cpu, true);
> > set_cpu_present(cpu, true);
> >
>
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-09-15 4:25 ` Mike Galbraith
@ 2014-09-15 16:44 ` Toshi Kani
2014-09-16 3:56 ` Mike Galbraith
0 siblings, 1 reply; 9+ messages in thread
From: Toshi Kani @ 2014-09-15 16:44 UTC (permalink / raw)
To: Mike Galbraith
Cc: Yasuaki Ishimatsu, tglx, mingo, hpa, bp, gong.chen, tony.luck,
x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, peterz, Wanpeng Li, Linn Crosetto
On Mon, 2014-09-15 at 06:25 +0200, Mike Galbraith wrote:
> On Thu, 2014-09-11 at 16:21 +0900, Yasuaki Ishimatsu wrote:
> > There is no response for two months since posting v4.
> > What can I do for pushing the patch to upstream?
>
> Looks to me like we have two patches floating about for more or less the
> same problem, this one, and...
>
> https://lkml.org/lkml/2014/7/29/159
>
> ..this one, which you reviewed, and HP both reviewed and tested.
>
> We seem to kinda stuck with Boris having said don't diddle the
> cpu_llc_shared_map, but HP/Intel saying that this map diddling fixes
> their explosions. If your alternative is preferred over diddling
> cpu_llc_shared_map, perhaps HP/Intel can test/confirm that their
> explosions stay gone?
Well, Boris mentioned later in his email:
https://lkml.org/lkml/2014/7/22/201
And I agree with his assessment that both patches make sense.
Thanks,
-Toshi
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-09-15 16:44 ` Toshi Kani
@ 2014-09-16 3:56 ` Mike Galbraith
2014-09-16 9:53 ` Wanpeng Li
0 siblings, 1 reply; 9+ messages in thread
From: Mike Galbraith @ 2014-09-16 3:56 UTC (permalink / raw)
To: Toshi Kani
Cc: Yasuaki Ishimatsu, tglx, mingo, hpa, bp, gong.chen, tony.luck,
x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, peterz, Wanpeng Li, Linn Crosetto
On Mon, 2014-09-15 at 10:44 -0600, Toshi Kani wrote:
> On Mon, 2014-09-15 at 06:25 +0200, Mike Galbraith wrote:
> > On Thu, 2014-09-11 at 16:21 +0900, Yasuaki Ishimatsu wrote:
> > > There is no response for two months since posting v4.
> > > What can I do for pushing the patch to upstream?
> >
> > Looks to me like we have two patches floating about for more or less the
> > same problem, this one, and...
> >
> > https://lkml.org/lkml/2014/7/29/159
> >
> > ..this one, which you reviewed, and HP both reviewed and tested.
> >
> > We seem to kinda stuck with Boris having said don't diddle the
> > cpu_llc_shared_map, but HP/Intel saying that this map diddling fixes
> > their explosions. If your alternative is preferred over diddling
> > cpu_llc_shared_map, perhaps HP/Intel can test/confirm that their
> > explosions stay gone?
>
> Well, Boris mentioned later in his email:
> https://lkml.org/lkml/2014/7/22/201
>
> And I agree with his assessment that both patches make sense.
Nonetheless, this just reeks of "department of redundancy department".
I have nothing against doing both really, but it does leave me wondering
if we would not then be merging the mask clearing "just because".
-Mike
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-09-16 3:56 ` Mike Galbraith
@ 2014-09-16 9:53 ` Wanpeng Li
2014-09-16 10:00 ` Mike Galbraith
0 siblings, 1 reply; 9+ messages in thread
From: Wanpeng Li @ 2014-09-16 9:53 UTC (permalink / raw)
To: Mike Galbraith, Toshi Kani
Cc: Yasuaki Ishimatsu, tglx, mingo, hpa, bp, gong.chen, tony.luck,
x86, imammedo, huawei.libin, paul.gortmaker, linux-kernel,
srivatsa.bhat, peterz, Wanpeng Li, Linn Crosetto
Hi Mike,
于 14-9-16 上午11:56, Mike Galbraith 写道:
> On Mon, 2014-09-15 at 10:44 -0600, Toshi Kani wrote:
>> On Mon, 2014-09-15 at 06:25 +0200, Mike Galbraith wrote:
>>> On Thu, 2014-09-11 at 16:21 +0900, Yasuaki Ishimatsu wrote:
>>>> There is no response for two months since posting v4.
>>>> What can I do for pushing the patch to upstream?
>>> Looks to me like we have two patches floating about for more or less the
>>> same problem, this one, and...
>>>
>>> https://lkml.org/lkml/2014/7/29/159
>>>
>>> ..this one, which you reviewed, and HP both reviewed and tested.
>>>
>>> We seem to kinda stuck with Boris having said don't diddle the
>>> cpu_llc_shared_map, but HP/Intel saying that this map diddling fixes
>>> their explosions. If your alternative is preferred over diddling
>>> cpu_llc_shared_map, perhaps HP/Intel can test/confirm that their
>>> explosions stay gone?
>> Well, Boris mentioned later in his email:
>> https://lkml.org/lkml/2014/7/22/201
>>
>> And I agree with his assessment that both patches make sense.
> Nonetheless, this just reeks of "department of redundancy department".
> I have nothing against doing both really, but it does leave me wondering
> if we would not then be merging the mask clearing "just because".
Maybe you miss my reply.
https://lkml.org/lkml/2014/7/29/40
Regards,
Wanpeng Li
>
> -Mike
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v5] x86,cpu-hotplug: assign same CPU number to readded CPU
2014-09-16 9:53 ` Wanpeng Li
@ 2014-09-16 10:00 ` Mike Galbraith
0 siblings, 0 replies; 9+ messages in thread
From: Mike Galbraith @ 2014-09-16 10:00 UTC (permalink / raw)
To: Wanpeng Li
Cc: Toshi Kani, Yasuaki Ishimatsu, tglx, mingo, hpa, bp, gong.chen,
tony.luck, x86, imammedo, huawei.libin, paul.gortmaker,
linux-kernel, srivatsa.bhat, peterz, Wanpeng Li, Linn Crosetto
On Tue, 2014-09-16 at 17:53 +0800, Wanpeng Li wrote:
> Hi Mike,
>
> 于 14-9-16 上午11:56, Mike Galbraith 写道:
> > On Mon, 2014-09-15 at 10:44 -0600, Toshi Kani wrote:
> >> On Mon, 2014-09-15 at 06:25 +0200, Mike Galbraith wrote:
> >>> On Thu, 2014-09-11 at 16:21 +0900, Yasuaki Ishimatsu wrote:
> >>>> There is no response for two months since posting v4.
> >>>> What can I do for pushing the patch to upstream?
> >>> Looks to me like we have two patches floating about for more or less the
> >>> same problem, this one, and...
> >>>
> >>> https://lkml.org/lkml/2014/7/29/159
> >>>
> >>> ..this one, which you reviewed, and HP both reviewed and tested.
> >>>
> >>> We seem to kinda stuck with Boris having said don't diddle the
> >>> cpu_llc_shared_map, but HP/Intel saying that this map diddling fixes
> >>> their explosions. If your alternative is preferred over diddling
> >>> cpu_llc_shared_map, perhaps HP/Intel can test/confirm that their
> >>> explosions stay gone?
> >> Well, Boris mentioned later in his email:
> >> https://lkml.org/lkml/2014/7/22/201
> >>
> >> And I agree with his assessment that both patches make sense.
> > Nonetheless, this just reeks of "department of redundancy department".
> > I have nothing against doing both really, but it does leave me wondering
> > if we would not then be merging the mask clearing "just because".
>
> Maybe you miss my reply.
>
> https://lkml.org/lkml/2014/7/29/40
Yes, I did. Thanks.
-Mike
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2014-09-16 10:08 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-07-23 3:22 [PATCH v4] x86,cpu-hotplug: assign same CPU number to readded CPU Yasuaki Ishimatsu
2014-07-28 4:18 ` Yasuaki Ishimatsu
2014-09-04 2:46 ` [PATCH v5] " Yasuaki Ishimatsu
2014-09-11 7:21 ` Yasuaki Ishimatsu
2014-09-15 4:25 ` Mike Galbraith
2014-09-15 16:44 ` Toshi Kani
2014-09-16 3:56 ` Mike Galbraith
2014-09-16 9:53 ` Wanpeng Li
2014-09-16 10:00 ` Mike Galbraith
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).