[PATCH v9 0/4] Support SMT control on arm64

[PATCH v9 0/4] Support SMT control on arm64

[Yicong Yang]

From: Yicong Yang <yangyicong@hisilicon.com>

The core CPU control framework supports runtime SMT control which
is not yet supported on arm64. Besides the general vulnerabilities
concerns we want this runtime control on our arm64 server for:

- better single CPU performance in some cases
- saving overall power consumption

This patchset implements it in the following aspects:

- Provides a default topology_is_primary_thread()
- support retrieve SMT thread number on OF based system
- support retrieve SMT thread number on ACPI based system
- select HOTPLUG_SMT for arm64

Tests has been done on our real ACPI based arm64 server and on
ACPI/OF based QEMU VMs.

Change since v8:
- Fix WARN on ACPI based non-SMT platform noticed in v7, per Pierre.
Link: https://lore.kernel.org/all/20241105093237.63565-1-yangyicong@huawei.com/

Change since v7:
Address the comments from Thomas:
- Add a newline between the glue define and function of topology_is_primary_thread
- Explicitly mention the sibling mask won't be empty in the comment
Link: https://lore.kernel.org/lkml/20241030125415.18994-1-yangyicong@huawei.com/

Change since v6:
- Fix unused variable if !CONFIG_ARM64 || !CONFIG_RISV found by lkp-test
- Fix max_smt_thread_num updating in OF path pointed by Pierre
- Drop unused variable and refine the comments/commit per Pierre
Link: https://lore.kernel.org/linux-arm-kernel/20241015021841.35713-1-yangyicong@huawei.com/

Change since v5:
- Drop the dependency on CONFIG_SMP since it's always on on arm64, per Pierre
- Avoid potential multiple calls of cpu_smt_set_num_threads() on asymmetric system, per Dietmar
- Detect heterogenous SMT topology and issue a warning for partly support, per Pierre
- Thanks Dietmar for testing, didn't pickup the tag due to code changes. Thanks testing by Pierre
Link: https://lore.kernel.org/linux-arm-kernel/20240806085320.63514-1-yangyicong@huawei.com/

Change since v4:
- Provide a default topology_is_primary_thread() in the framework, Per Will
Link: https://lore.kernel.org/linux-arm-kernel/20231121092602.47792-1-yangyicong@huawei.com/

Change since v3:
- Fix some build and kconfig error reported by kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/linux-arm-kernel/20231114040110.54590-1-yangyicong@huawei.com/

Change since v2:
- Detect SMT thread number at topology build from ACPI/DT, avoid looping CPUs
- Split patches into ACPI/OF/arch_topology path and enable the kconfig for arm64
Link: https://lore.kernel.org/linux-arm-kernel/20231010115335.13862-1-yangyicong@huawei.com/

Yicong Yang (4):
  cpu/SMT: Provide a default topology_is_primary_thread()
  arch_topology: Support SMT control for OF based system
  arm64: topology: Support SMT control on ACPI based system
  arm64: Kconfig: Enable HOTPLUG_SMT

 arch/arm64/Kconfig                  |  1 +
 arch/arm64/kernel/topology.c        | 59 +++++++++++++++++++++++++++++
 arch/powerpc/include/asm/topology.h |  1 +
 arch/x86/include/asm/topology.h     |  2 +-
 drivers/base/arch_topology.c        | 24 ++++++++++++
 include/linux/topology.h            | 20 ++++++++++
 6 files changed, 106 insertions(+), 1 deletion(-)

-- 
2.24.0

[PATCH v9 1/4] cpu/SMT: Provide a default topology_is_primary_thread()

[Yicong Yang]

From: Yicong Yang <yangyicong@hisilicon.com>

Currently if architectures want to support HOTPLUG_SMT they need to
provide a topology_is_primary_thread() telling the framework which
thread in the SMT cannot offline. However arm64 doesn't have a
restriction on which thread in the SMT cannot offline, a simplest
choice is that just make 1st thread as the "primary" thread. So
just make this as the default implementation in the framework and
let architectures like x86 that have special primary thread to
override this function (which they've already done).

There's no need to provide a stub function if !CONFIG_SMP or
!CONFIG_HOTPLUG_SMP. In such case the testing CPU is already
the 1st CPU in the SMT so it's always the primary thread.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
---
 arch/powerpc/include/asm/topology.h |  1 +
 arch/x86/include/asm/topology.h     |  2 +-
 include/linux/topology.h            | 20 ++++++++++++++++++++
 3 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/include/asm/topology.h b/arch/powerpc/include/asm/topology.h
index 16bacfe8c7a2..da15b5efe807 100644
--- a/arch/powerpc/include/asm/topology.h
+++ b/arch/powerpc/include/asm/topology.h
@@ -152,6 +152,7 @@ static inline bool topology_is_primary_thread(unsigned int cpu)
 {
 	return cpu == cpu_first_thread_sibling(cpu);
 }
+#define topology_is_primary_thread topology_is_primary_thread
 
 static inline bool topology_smt_thread_allowed(unsigned int cpu)
 {
diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
index 92f3664dd933..d84d9b6d8678 100644
--- a/arch/x86/include/asm/topology.h
+++ b/arch/x86/include/asm/topology.h
@@ -219,11 +219,11 @@ static inline bool topology_is_primary_thread(unsigned int cpu)
 {
 	return cpumask_test_cpu(cpu, cpu_primary_thread_mask);
 }
+#define topology_is_primary_thread topology_is_primary_thread
 
 #else /* CONFIG_SMP */
 static inline int topology_phys_to_logical_pkg(unsigned int pkg) { return 0; }
 static inline int topology_max_smt_threads(void) { return 1; }
-static inline bool topology_is_primary_thread(unsigned int cpu) { return true; }
 static inline unsigned int topology_amd_nodes_per_pkg(void) { return 1; }
 #endif /* !CONFIG_SMP */
 
diff --git a/include/linux/topology.h b/include/linux/topology.h
index 52f5850730b3..b8e860276518 100644
--- a/include/linux/topology.h
+++ b/include/linux/topology.h
@@ -240,6 +240,26 @@ static inline const struct cpumask *cpu_smt_mask(int cpu)
 }
 #endif
 
+#ifndef topology_is_primary_thread
+
+#define topology_is_primary_thread topology_is_primary_thread
+
+static inline bool topology_is_primary_thread(unsigned int cpu)
+{
+	/*
+	 * On SMT hotplug the primary thread of the SMT won't be disabled.
+	 * Architectures do have a special primary thread (e.g. x86) need
+	 * to override this function. Otherwise just make the first thread
+	 * in the SMT as the primary thread.
+	 *
+	 * The sibling cpumask of an offline CPU contains always the CPU
+	 * itself.
+	 */
+	return cpu == cpumask_first(topology_sibling_cpumask(cpu));
+}
+
+#endif
+
 static inline const struct cpumask *cpu_cpu_mask(int cpu)
 {
 	return cpumask_of_node(cpu_to_node(cpu));
-- 
2.24.0

[PATCH v9 2/4] arch_topology: Support SMT control for OF based system

[Yicong Yang]

From: Yicong Yang <yangyicong@hisilicon.com>

On building the topology from the devicetree, we've already
gotten the SMT thread number of each core. Update the largest
SMT thread number and enable the SMT control by the end of
topology parsing.

The core's SMT control provides two interface to the users [1]:
1) enable/disable SMT by writing on/off
2) enable/disable SMT by writing thread number 1/max_thread_number

If a system have more than one SMT thread number the 2) may
not handle it well, since there're multiple thread numbers in the
system and 2) only accept 1/max_thread_number. So issue a warning
to notify the users if such system detected.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/ABI/testing/sysfs-devices-system-cpu#n542
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
---
 drivers/base/arch_topology.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
index 3ebe77566788..31e0291ee660 100644
--- a/drivers/base/arch_topology.c
+++ b/drivers/base/arch_topology.c
@@ -11,6 +11,7 @@
 #include <linux/cleanup.h>
 #include <linux/cpu.h>
 #include <linux/cpufreq.h>
+#include <linux/cpu_smt.h>
 #include <linux/device.h>
 #include <linux/of.h>
 #include <linux/slab.h>
@@ -506,6 +507,10 @@ core_initcall(free_raw_capacity);
 #endif
 
 #if defined(CONFIG_ARM64) || defined(CONFIG_RISCV)
+
+/* Maximum SMT thread number detected used to enable the SMT control */
+static unsigned int max_smt_thread_num;
+
 /*
  * This function returns the logic cpu number of the node.
  * There are basically three kinds of return values:
@@ -565,6 +570,17 @@ static int __init parse_core(struct device_node *core, int package_id,
 		i++;
 	} while (1);
 
+	/*
+	 * If max_smt_thread_num has been initialized and doesn't match
+	 * the thread number of this entry, then the system has
+	 * heterogeneous SMT topology.
+	 */
+	if (max_smt_thread_num && max_smt_thread_num != i)
+		pr_warn_once("Heterogeneous SMT topology is partly supported by SMT control\n");
+
+	if (max_smt_thread_num < i)
+		max_smt_thread_num = i;
+
 	cpu = get_cpu_for_node(core);
 	if (cpu >= 0) {
 		if (!leaf) {
@@ -677,6 +693,14 @@ static int __init parse_socket(struct device_node *socket)
 	if (!has_socket)
 		ret = parse_cluster(socket, 0, -1, 0);
 
+	/*
+	 * Notify the CPU framework of the SMT support. A thread number of 1
+	 * can be handled by the framework so we don't need to check
+	 * max_smt_thread_num to see we support SMT or not.
+	 */
+	if (max_smt_thread_num)
+		cpu_smt_set_num_threads(max_smt_thread_num, max_smt_thread_num);
+
 	return ret;
 }
 
-- 
2.24.0

[PATCH v9 3/4] arm64: topology: Support SMT control on ACPI based system

[Yicong Yang]

From: Yicong Yang <yangyicong@hisilicon.com>

For ACPI we'll build the topology from PPTT and we cannot directly
get the SMT number of each core. Instead using a temporary xarray
to record the heterogeneous information (from ACPI_PPTT_ACPI_IDENTICAL)
and SMT information of the first core in its heterogeneous CPU cluster
when building the topology. Then we can know the largest SMT number
in the system. If a homogeneous system's using ACPI 6.2 or later,
all the CPUs should be under the root node of PPTT. There'll be
only one entry in the xarray and all the CPUs in the system will
be assumed identical.

The core's SMT control provides two interface to the users [1]:
1) enable/disable SMT by writing on/off
2) enable/disable SMT by writing thread number 1/max_thread_number

If a system have more than one SMT thread number the 2) may
not handle it well, since there're multiple thread numbers in the
system and 2) only accept 1/max_thread_number. So issue a warning
to notify the users if such system detected.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/ABI/testing/sysfs-devices-system-cpu#n542
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
---
 arch/arm64/kernel/topology.c | 59 ++++++++++++++++++++++++++++++++++++
 1 file changed, 59 insertions(+)

diff --git a/arch/arm64/kernel/topology.c b/arch/arm64/kernel/topology.c
index 1a2c72f3e7f8..d10ba4b8efee 100644
--- a/arch/arm64/kernel/topology.c
+++ b/arch/arm64/kernel/topology.c
@@ -15,8 +15,10 @@
 #include <linux/arch_topology.h>
 #include <linux/cacheinfo.h>
 #include <linux/cpufreq.h>
+#include <linux/cpu_smt.h>
 #include <linux/init.h>
 #include <linux/percpu.h>
+#include <linux/xarray.h>
 
 #include <asm/cpu.h>
 #include <asm/cputype.h>
@@ -37,17 +39,28 @@ static bool __init acpi_cpu_is_threaded(int cpu)
 	return !!is_threaded;
 }
 
+struct cpu_smt_info {
+	int thread_num;
+	int core_id;
+};
+
 /*
  * Propagate the topology information of the processor_topology_node tree to the
  * cpu_topology array.
  */
 int __init parse_acpi_topology(void)
 {
+	int max_smt_thread_num = 0;
+	struct cpu_smt_info *entry;
+	struct xarray hetero_cpu;
+	unsigned long hetero_id;
 	int cpu, topology_id;
 
 	if (acpi_disabled)
 		return 0;
 
+	xa_init(&hetero_cpu);
+
 	for_each_possible_cpu(cpu) {
 		topology_id = find_acpi_cpu_topology(cpu, 0);
 		if (topology_id < 0)
@@ -57,6 +70,32 @@ int __init parse_acpi_topology(void)
 			cpu_topology[cpu].thread_id = topology_id;
 			topology_id = find_acpi_cpu_topology(cpu, 1);
 			cpu_topology[cpu].core_id   = topology_id;
+
+			/*
+			 * In the PPTT, CPUs below a node with the 'identical
+			 * implementation' flag have the same number of threads.
+			 * Count the number of threads for only one CPU (i.e.
+			 * one core_id) among those with the same hetero_id.
+			 * See the comment of find_acpi_cpu_topology_hetero_id()
+			 * for more details.
+			 *
+			 * One entry is created for each node having:
+			 * - the 'identical implementation' flag
+			 * - its parent not having the flag
+			 */
+			hetero_id = find_acpi_cpu_topology_hetero_id(cpu);
+			entry = (struct cpu_smt_info *)xa_load(&hetero_cpu, hetero_id);
+			if (!entry) {
+				entry = kzalloc(sizeof(*entry), GFP_KERNEL);
+				WARN_ON(!entry);
+
+				entry->core_id = topology_id;
+				entry->thread_num = 1;
+				xa_store(&hetero_cpu, hetero_id,
+					 entry, GFP_KERNEL);
+			} else if (entry->core_id == topology_id) {
+				entry->thread_num++;
+			}
 		} else {
 			cpu_topology[cpu].thread_id  = -1;
 			cpu_topology[cpu].core_id    = topology_id;
@@ -67,6 +106,26 @@ int __init parse_acpi_topology(void)
 		cpu_topology[cpu].package_id = topology_id;
 	}
 
+	/*
+	 * This should be a short loop depending on the number of heterogeneous
+	 * CPU clusters. Typically on a homogeneous system there's only one
+	 * entry in the XArray.
+	 */
+	xa_for_each(&hetero_cpu, hetero_id, entry) {
+		if (entry->thread_num != max_smt_thread_num && max_smt_thread_num)
+			pr_warn_once("Heterogeneous SMT topology is partly supported by SMT control\n");
+
+		if (entry->thread_num > max_smt_thread_num)
+			max_smt_thread_num = entry->thread_num;
+
+		xa_erase(&hetero_cpu, hetero_id);
+		kfree(entry);
+	}
+
+	if (max_smt_thread_num)
+		cpu_smt_set_num_threads(max_smt_thread_num, max_smt_thread_num);
+
+	xa_destroy(&hetero_cpu);
 	return 0;
 }
 #endif
-- 
2.24.0

[PATCH v9 4/4] arm64: Kconfig: Enable HOTPLUG_SMT

[Yicong Yang]

From: Yicong Yang <yangyicong@hisilicon.com>

Enable HOTPLUG_SMT for SMT control.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
---
 arch/arm64/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 70d7f4f20225..5eecd2e2c093 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -243,6 +243,7 @@ config ARM64
 	select HAVE_KRETPROBES
 	select HAVE_GENERIC_VDSO
 	select HOTPLUG_CORE_SYNC_DEAD if HOTPLUG_CPU
+	select HOTPLUG_SMT if HOTPLUG_CPU
 	select IRQ_DOMAIN
 	select IRQ_FORCED_THREADING
 	select KASAN_VMALLOC if KASAN
-- 
2.24.0

Re: [PATCH v9 1/4] cpu/SMT: Provide a default topology_is_primary_thread()

[Pierre Gondois]

Hello Yicong,


On 11/14/24 15:11, Yicong Yang wrote:
> From: Yicong Yang <yangyicong@hisilicon.com>
> 
> Currently if architectures want to support HOTPLUG_SMT they need to
> provide a topology_is_primary_thread() telling the framework which
> thread in the SMT cannot offline. However arm64 doesn't have a
> restriction on which thread in the SMT cannot offline, a simplest
> choice is that just make 1st thread as the "primary" thread. So
> just make this as the default implementation in the framework and
> let architectures like x86 that have special primary thread to
> override this function (which they've already done).
> 
> There's no need to provide a stub function if !CONFIG_SMP or
> !CONFIG_HOTPLUG_SMP. In such case the testing CPU is already
> the 1st CPU in the SMT so it's always the primary thread.
> 
> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
> ---
>   arch/powerpc/include/asm/topology.h |  1 +
>   arch/x86/include/asm/topology.h     |  2 +-
>   include/linux/topology.h            | 20 ++++++++++++++++++++
>   3 files changed, 22 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/powerpc/include/asm/topology.h b/arch/powerpc/include/asm/topology.h
> index 16bacfe8c7a2..da15b5efe807 100644
> --- a/arch/powerpc/include/asm/topology.h
> +++ b/arch/powerpc/include/asm/topology.h
> @@ -152,6 +152,7 @@ static inline bool topology_is_primary_thread(unsigned int cpu)
>   {
>   	return cpu == cpu_first_thread_sibling(cpu);
>   }
> +#define topology_is_primary_thread topology_is_primary_thread
>   
>   static inline bool topology_smt_thread_allowed(unsigned int cpu)
>   {
> diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
> index 92f3664dd933..d84d9b6d8678 100644
> --- a/arch/x86/include/asm/topology.h
> +++ b/arch/x86/include/asm/topology.h
> @@ -219,11 +219,11 @@ static inline bool topology_is_primary_thread(unsigned int cpu)
>   {
>   	return cpumask_test_cpu(cpu, cpu_primary_thread_mask);
>   }
> +#define topology_is_primary_thread topology_is_primary_thread
>   
>   #else /* CONFIG_SMP */
>   static inline int topology_phys_to_logical_pkg(unsigned int pkg) { return 0; }
>   static inline int topology_max_smt_threads(void) { return 1; }
> -static inline bool topology_is_primary_thread(unsigned int cpu) { return true; }
>   static inline unsigned int topology_amd_nodes_per_pkg(void) { return 1; }
>   #endif /* !CONFIG_SMP */
>   
> diff --git a/include/linux/topology.h b/include/linux/topology.h
> index 52f5850730b3..b8e860276518 100644
> --- a/include/linux/topology.h
> +++ b/include/linux/topology.h
> @@ -240,6 +240,26 @@ static inline const struct cpumask *cpu_smt_mask(int cpu)
>   }
>   #endif
>   
> +#ifndef topology_is_primary_thread
> +
> +#define topology_is_primary_thread topology_is_primary_thread
> +
> +static inline bool topology_is_primary_thread(unsigned int cpu)
> +{
> +	/*
> +	 * On SMT hotplug the primary thread of the SMT won't be disabled.
> +	 * Architectures do have a special primary thread (e.g. x86) need
> +	 * to override this function. Otherwise just make the first thread
> +	 * in the SMT as the primary thread.
> +	 *
> +	 * The sibling cpumask of an offline CPU contains always the CPU
> +	 * itself.

As Thomas suggested, would it be possible to check it for other
architectures ?
For instance for loongarch at arch/loongarch/kernel/smp.c,
clear_cpu_sibling_map() seems to completely clear &cpu_sibling_map[cpu]
when a CPU is put offline. This would make topology_sibling_cpumask(cpu)
to be empty and cpu_bootable() return false if the CPU never booted before.

Personal note:
cpu_bootable() is called from an already online CPU:
cpu_bootable (kernel/cpu.c:678)
cpu_up (kernel/cpu.c:1722 kernel/cpu.c:1702)
bringup_nonboot_cpus (kernel/cpu.c:1793 kernel/cpu.c:1901)
smp_init (./include/linux/bitmap.h:445 ./include/linux/nodemask.h:241 ./include/linux/nodemask.h:438 kernel/smp.c:1011)
kernel_init_freeable (init/main.c:1573)
kernel_init (init/main.c:1473)
ret_from_fork (arch/arm64/kernel/entry.S:861)

store_cpu_topology() for arm64 is called from the booting CPU:
store_cpu_topology (drivers/base/arch_topology.c:921)
secondary_start_kernel (arch/arm64/kernel/smp.c:251)
__secondary_switched (arch/arm64/kernel/head.S:418)

> +	 */
> +	return cpu == cpumask_first(topology_sibling_cpumask(cpu));
> +}
> +
> +#endif
> +
>   static inline const struct cpumask *cpu_cpu_mask(int cpu)
>   {
>   	return cpumask_of_node(cpu_to_node(cpu));

Re: [PATCH v9 1/4] cpu/SMT: Provide a default topology_is_primary_thread()

[Yicong Yang]

On 2024/11/15 17:42, Pierre Gondois wrote:
> Hello Yicong,
> 
> 
> On 11/14/24 15:11, Yicong Yang wrote:
>> From: Yicong Yang <yangyicong@hisilicon.com>
>>
>> Currently if architectures want to support HOTPLUG_SMT they need to
>> provide a topology_is_primary_thread() telling the framework which
>> thread in the SMT cannot offline. However arm64 doesn't have a
>> restriction on which thread in the SMT cannot offline, a simplest
>> choice is that just make 1st thread as the "primary" thread. So
>> just make this as the default implementation in the framework and
>> let architectures like x86 that have special primary thread to
>> override this function (which they've already done).
>>
>> There's no need to provide a stub function if !CONFIG_SMP or
>> !CONFIG_HOTPLUG_SMP. In such case the testing CPU is already
>> the 1st CPU in the SMT so it's always the primary thread.
>>
>> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
>> ---
>>   arch/powerpc/include/asm/topology.h |  1 +
>>   arch/x86/include/asm/topology.h     |  2 +-
>>   include/linux/topology.h            | 20 ++++++++++++++++++++
>>   3 files changed, 22 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/powerpc/include/asm/topology.h b/arch/powerpc/include/asm/topology.h
>> index 16bacfe8c7a2..da15b5efe807 100644
>> --- a/arch/powerpc/include/asm/topology.h
>> +++ b/arch/powerpc/include/asm/topology.h
>> @@ -152,6 +152,7 @@ static inline bool topology_is_primary_thread(unsigned int cpu)
>>   {
>>       return cpu == cpu_first_thread_sibling(cpu);
>>   }
>> +#define topology_is_primary_thread topology_is_primary_thread
>>     static inline bool topology_smt_thread_allowed(unsigned int cpu)
>>   {
>> diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h
>> index 92f3664dd933..d84d9b6d8678 100644
>> --- a/arch/x86/include/asm/topology.h
>> +++ b/arch/x86/include/asm/topology.h
>> @@ -219,11 +219,11 @@ static inline bool topology_is_primary_thread(unsigned int cpu)
>>   {
>>       return cpumask_test_cpu(cpu, cpu_primary_thread_mask);
>>   }
>> +#define topology_is_primary_thread topology_is_primary_thread
>>     #else /* CONFIG_SMP */
>>   static inline int topology_phys_to_logical_pkg(unsigned int pkg) { return 0; }
>>   static inline int topology_max_smt_threads(void) { return 1; }
>> -static inline bool topology_is_primary_thread(unsigned int cpu) { return true; }
>>   static inline unsigned int topology_amd_nodes_per_pkg(void) { return 1; }
>>   #endif /* !CONFIG_SMP */
>>   diff --git a/include/linux/topology.h b/include/linux/topology.h
>> index 52f5850730b3..b8e860276518 100644
>> --- a/include/linux/topology.h
>> +++ b/include/linux/topology.h
>> @@ -240,6 +240,26 @@ static inline const struct cpumask *cpu_smt_mask(int cpu)
>>   }
>>   #endif
>>   +#ifndef topology_is_primary_thread
>> +
>> +#define topology_is_primary_thread topology_is_primary_thread
>> +
>> +static inline bool topology_is_primary_thread(unsigned int cpu)
>> +{
>> +    /*
>> +     * On SMT hotplug the primary thread of the SMT won't be disabled.
>> +     * Architectures do have a special primary thread (e.g. x86) need
>> +     * to override this function. Otherwise just make the first thread
>> +     * in the SMT as the primary thread.
>> +     *
>> +     * The sibling cpumask of an offline CPU contains always the CPU
>> +     * itself.
> 
> As Thomas suggested, would it be possible to check it for other
> architectures ?
> For instance for loongarch at arch/loongarch/kernel/smp.c,
> clear_cpu_sibling_map() seems to completely clear &cpu_sibling_map[cpu]
> when a CPU is put offline. This would make topology_sibling_cpumask(cpu)
> to be empty and cpu_bootable() return false if the CPU never booted before.
> 

cpu_bootable() only affects architectures select HOTPLUG_SMT, otherwise it'll always
return true. Since x86 and powerpc have their own illustration of primary thread and
have an override version of this funciton, arm64 is the only user now by this patchset.
We have this guarantee for arm64 and also for other architectures using arch_topology.c
(see clear_cpu_topology()). So if loogarch has a different implementation, they
should implement a topology_is_primary_thread() variant to support HOTPLUG_SMT.

> Personal note:
> cpu_bootable() is called from an already online CPU:
> cpu_bootable (kernel/cpu.c:678)
> cpu_up (kernel/cpu.c:1722 kernel/cpu.c:1702)
> bringup_nonboot_cpus (kernel/cpu.c:1793 kernel/cpu.c:1901)
> smp_init (./include/linux/bitmap.h:445 ./include/linux/nodemask.h:241 ./include/linux/nodemask.h:438 kernel/smp.c:1011)
> kernel_init_freeable (init/main.c:1573)
> kernel_init (init/main.c:1473)
> ret_from_fork (arch/arm64/kernel/entry.S:861)
> 
> store_cpu_topology() for arm64 is called from the booting CPU:
> store_cpu_topology (drivers/base/arch_topology.c:921)
> secondary_start_kernel (arch/arm64/kernel/smp.c:251)
> __secondary_switched (arch/arm64/kernel/head.S:418)
> 
>> +     */
>> +    return cpu == cpumask_first(topology_sibling_cpumask(cpu));
>> +}
>> +
>> +#endif
>> +
>>   static inline const struct cpumask *cpu_cpu_mask(int cpu)
>>   {
>>       return cpumask_of_node(cpu_to_node(cpu));
> 
> .

Re: [PATCH v9 1/4] cpu/SMT: Provide a default topology_is_primary_thread()

[Dietmar Eggemann]

On 18/11/2024 11:50, Yicong Yang wrote:
> On 2024/11/15 17:42, Pierre Gondois wrote:
>> Hello Yicong,
>>
>>
>> On 11/14/24 15:11, Yicong Yang wrote:
>>> From: Yicong Yang <yangyicong@hisilicon.com>

[...]

>>>   diff --git a/include/linux/topology.h b/include/linux/topology.h
>>> index 52f5850730b3..b8e860276518 100644
>>> --- a/include/linux/topology.h
>>> +++ b/include/linux/topology.h
>>> @@ -240,6 +240,26 @@ static inline const struct cpumask *cpu_smt_mask(int cpu)
>>>   }
>>>   #endif
>>>   +#ifndef topology_is_primary_thread
>>> +
>>> +#define topology_is_primary_thread topology_is_primary_thread
>>> +
>>> +static inline bool topology_is_primary_thread(unsigned int cpu)
>>> +{
>>> +    /*
>>> +     * On SMT hotplug the primary thread of the SMT won't be disabled.
>>> +     * Architectures do have a special primary thread (e.g. x86) need
>>> +     * to override this function. Otherwise just make the first thread
>>> +     * in the SMT as the primary thread.
>>> +     *
>>> +     * The sibling cpumask of an offline CPU contains always the CPU
>>> +     * itself.
>>
>> As Thomas suggested, would it be possible to check it for other
>> architectures ?
>> For instance for loongarch at arch/loongarch/kernel/smp.c,
>> clear_cpu_sibling_map() seems to completely clear &cpu_sibling_map[cpu]
>> when a CPU is put offline. This would make topology_sibling_cpumask(cpu)
>> to be empty and cpu_bootable() return false if the CPU never booted before.
>>
> 
> cpu_bootable() only affects architectures select HOTPLUG_SMT, otherwise it'll always
> return true. Since x86 and powerpc have their own illustration of primary thread and
> have an override version of this funciton, arm64 is the only user now by this patchset.
> We have this guarantee for arm64 and also for other architectures using arch_topology.c
> (see clear_cpu_topology()). So if loogarch has a different implementation, they
> should implement a topology_is_primary_thread() variant to support HOTPLUG_SMT.

I also stumbled over this sentence.

drivers/base/arch_topology.c:

void clear_cpu_topology(int cpu)                   (2)

  ...
  cpumask_set_cpu(cpu, &cpu_topo->thread_sibling)  (4)

void __init reset_cpu_topology(void)               (1)

  for_each_possible_cpu(cpu)

    ...
    clear_cpu_topology(cpu)                        (2)

#if defined(CONFIG_ARM64) || defined(CONFIG_RISCV) (3)
void __init init_cpu_topology(void)

  reset_cpu_topology()                             (1)
  ...

Does this mean the default implementation relies on (4), i.e. is only
valid for arm64 and riscv? (3)
Do all the other archs then have to overwrite the default implementation
 (like x86 and powerpc) if they want to implement CONFIG_HOTPLUG_SMT?

[...]

Re: [PATCH v9 1/4] cpu/SMT: Provide a default topology_is_primary_thread()

[Yicong Yang]

On 2024/11/18 23:04, Dietmar Eggemann wrote:
> On 18/11/2024 11:50, Yicong Yang wrote:
>> On 2024/11/15 17:42, Pierre Gondois wrote:
>>> Hello Yicong,
>>>
>>>
>>> On 11/14/24 15:11, Yicong Yang wrote:
>>>> From: Yicong Yang <yangyicong@hisilicon.com>
> 
> [...]
> 
>>>>   diff --git a/include/linux/topology.h b/include/linux/topology.h
>>>> index 52f5850730b3..b8e860276518 100644
>>>> --- a/include/linux/topology.h
>>>> +++ b/include/linux/topology.h
>>>> @@ -240,6 +240,26 @@ static inline const struct cpumask *cpu_smt_mask(int cpu)
>>>>   }
>>>>   #endif
>>>>   +#ifndef topology_is_primary_thread
>>>> +
>>>> +#define topology_is_primary_thread topology_is_primary_thread
>>>> +
>>>> +static inline bool topology_is_primary_thread(unsigned int cpu)
>>>> +{
>>>> +    /*
>>>> +     * On SMT hotplug the primary thread of the SMT won't be disabled.
>>>> +     * Architectures do have a special primary thread (e.g. x86) need
>>>> +     * to override this function. Otherwise just make the first thread
>>>> +     * in the SMT as the primary thread.
>>>> +     *
>>>> +     * The sibling cpumask of an offline CPU contains always the CPU
>>>> +     * itself.
>>>
>>> As Thomas suggested, would it be possible to check it for other
>>> architectures ?
>>> For instance for loongarch at arch/loongarch/kernel/smp.c,
>>> clear_cpu_sibling_map() seems to completely clear &cpu_sibling_map[cpu]
>>> when a CPU is put offline. This would make topology_sibling_cpumask(cpu)
>>> to be empty and cpu_bootable() return false if the CPU never booted before.
>>>
>>
>> cpu_bootable() only affects architectures select HOTPLUG_SMT, otherwise it'll always
>> return true. Since x86 and powerpc have their own illustration of primary thread and
>> have an override version of this funciton, arm64 is the only user now by this patchset.
>> We have this guarantee for arm64 and also for other architectures using arch_topology.c
>> (see clear_cpu_topology()). So if loogarch has a different implementation, they
>> should implement a topology_is_primary_thread() variant to support HOTPLUG_SMT.
> 
> I also stumbled over this sentence.
> 
> drivers/base/arch_topology.c:
> 
> void clear_cpu_topology(int cpu)                   (2)
> 
>   ...
>   cpumask_set_cpu(cpu, &cpu_topo->thread_sibling)  (4)
> 
> void __init reset_cpu_topology(void)               (1)
> 
>   for_each_possible_cpu(cpu)
> 
>     ...
>     clear_cpu_topology(cpu)                        (2)
> 
> #if defined(CONFIG_ARM64) || defined(CONFIG_RISCV) (3)
> void __init init_cpu_topology(void)
> 
>   reset_cpu_topology()                             (1)
>   ...
> 
> Does this mean the default implementation relies on (4), i.e. is only
> valid for arm64 and riscv? (3)
> Do all the other archs then have to overwrite the default implementation
>  (like x86 and powerpc) if they want to implement CONFIG_HOTPLUG_SMT?
> 

I think yes if they have problems with the default implementation. That's
what used to be to support HOTPLUG_SMT before this, each arthitecture
provides their own variant of topology_is_primary_thread.

The current approach may also work since cpu_bootable() will make all the
CPUs boot at least once:
static inline bool cpu_bootable(unsigned int cpu) {
[...]
	if (topology_is_primary_thread(cpu))
		return true;

	return !cpumask_test_cpu(cpu, &cpus_booted_once_mask); // True if not booted
}

The boot process will be like below. cpu_bootable() is checked twice:
-> cpu_up()
     cpu_bootable() (1)
     [...]
     cpuhp_bringup_ap()
       [ archs usually update the sibling_mask in start_secondary() here ]
       bringup_wait_for_ap_online()
         if (!cpu_bootable(cpu)) (2)
           return -ECANCELED // roll back and offline this CPU

So an empty mask in (1) won't block the CPU going online. And the default
topology_is_primary_thread() should work if sibling mask updated before
the checking in (2). I hacked x86 to use the default topology_is_primary_thread
and tested on a qemu vm and it seems also work (just for test since the
primary thread should not always be the 1st thread in the core on x86,
not quite sure).

Thanks.