[PATCH] x86/acpi: Don't add CPUs that are not online capable

[PATCH] x86/acpi: Don't add CPUs that are not online capable

[Mario Limonciello]

A number of systems are showing "hotplug capable" CPUs when they
are not really hotpluggable.  This is because the MADT has extra
CPU entries to support different CPUs that may be inserted into
the socket with different numbers of cores.

The ACPI spec is clear that the Online Capable bit in the
MADT should be used to determine whether or not a CPU is hotplug
capable when the enabled bit is not set.

Link: https://uefi.org/htmlspecs/ACPI_Spec_6_4_html/05_ACPI_Software_Programming_Model/ACPI_Software_Programming_Model.html?#local-apic-flags
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
---
 arch/x86/kernel/acpi/boot.c | 6 ++++++
 include/acpi/actbl2.h       | 1 +
 2 files changed, 7 insertions(+)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index e55e0c1fad8c..eeb10b27d6de 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -239,6 +239,12 @@ acpi_parse_lapic(union acpi_subtable_headers * header, const unsigned long end)
 	if (processor->id == 0xff)
 		return 0;
 
+	/* don't register processors that can not be onlined */
+	if (!(processor->lapic_flags & ACPI_MADT_ENABLED)) {
+		if (!(processor->lapic_flags & ACPI_MADT_ONLINE_CAPABLE))
+			return 0;
+	}
+
 	/*
 	 * We need to register disabled CPU as well to permit
 	 * counting disabled CPUs. This allows us to size
diff --git a/include/acpi/actbl2.h b/include/acpi/actbl2.h
index 2069ac38a4e2..fae45e383987 100644
--- a/include/acpi/actbl2.h
+++ b/include/acpi/actbl2.h
@@ -808,6 +808,7 @@ struct acpi_madt_multiproc_wakeup_mailbox {
 /* MADT Local APIC flags */
 
 #define ACPI_MADT_ENABLED           (1)	/* 00: Processor is usable if set */
+#define ACPI_MADT_ONLINE_CAPABLE    (2)	/* 01: System HW supports enabling processor at runtime */
 
 /* MADT MPS INTI flags (inti_flags) */
 
-- 
2.25.1

Re: [PATCH] x86/acpi: Don't add CPUs that are not online capable

[Hanjun Guo]

On 2021/8/12 13:16, Mario Limonciello wrote:
> A number of systems are showing "hotplug capable" CPUs when they
> are not really hotpluggable.  This is because the MADT has extra
> CPU entries to support different CPUs that may be inserted into
> the socket with different numbers of cores.
> 
> The ACPI spec is clear that the Online Capable bit in the
> MADT should be used to determine whether or not a CPU is hotplug
> capable when the enabled bit is not set.

This was introduced in ACPI 6.3 spec, which means ACPI 6.2 and
earlier versions don't include the "Online Capable bit".

> 
> Link: https://uefi.org/htmlspecs/ACPI_Spec_6_4_html/05_ACPI_Software_Programming_Model/ACPI_Software_Programming_Model.html?#local-apic-flags
> Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
> Reviewed-by: Huang Rui <ray.huang@amd.com>
> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> ---
>   arch/x86/kernel/acpi/boot.c | 6 ++++++
>   include/acpi/actbl2.h       | 1 +
>   2 files changed, 7 insertions(+)
> 
> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> index e55e0c1fad8c..eeb10b27d6de 100644
> --- a/arch/x86/kernel/acpi/boot.c
> +++ b/arch/x86/kernel/acpi/boot.c
> @@ -239,6 +239,12 @@ acpi_parse_lapic(union acpi_subtable_headers * header, const unsigned long end)
>   	if (processor->id == 0xff)
>   		return 0;
>   
> +	/* don't register processors that can not be onlined */
> +	if (!(processor->lapic_flags & ACPI_MADT_ENABLED)) {
> +		if (!(processor->lapic_flags & ACPI_MADT_ONLINE_CAPABLE))
> +			return 0;
> +	}

For firmware using ACPI 6.2 and early versions, the
ACPI_MADT_ONLINE_CAPABLE bit is reserved as zero, so if
we set CPU as disabled, the code here will always return
0 in those firmwares.

> +
>   	/*
>   	 * We need to register disabled CPU as well to permit
>   	 * counting disabled CPUs. This allows us to size

So we will not register the disabled CPU and will break
CPU hotplug features.

I think we need to consider the compatibility with old versions
of firmware.

Thanks
Hanjun

[Devel] Re: [PATCH] x86/acpi: Don't add CPUs that are not online capable

[Hanjun Guo]

[-- Attachment #1: Type: text/plain, Size: 2117 bytes --]

On 2021/8/12 13:16, Mario Limonciello wrote:
> A number of systems are showing "hotplug capable" CPUs when they
> are not really hotpluggable.  This is because the MADT has extra
> CPU entries to support different CPUs that may be inserted into
> the socket with different numbers of cores.
> 
> The ACPI spec is clear that the Online Capable bit in the
> MADT should be used to determine whether or not a CPU is hotplug
> capable when the enabled bit is not set.

This was introduced in ACPI 6.3 spec, which means ACPI 6.2 and
earlier versions don't include the "Online Capable bit".

> 
> Link: https://uefi.org/htmlspecs/ACPI_Spec_6_4_html/05_ACPI_Software_Programming_Model/ACPI_Software_Programming_Model.html?#local-apic-flags
> Reviewed-by: Alex Deucher <alexander.deucher(a)amd.com>
> Reviewed-by: Huang Rui <ray.huang(a)amd.com>
> Signed-off-by: Mario Limonciello <mario.limonciello(a)amd.com>
> ---
>   arch/x86/kernel/acpi/boot.c | 6 ++++++
>   include/acpi/actbl2.h       | 1 +
>   2 files changed, 7 insertions(+)
> 
> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> index e55e0c1fad8c..eeb10b27d6de 100644
> --- a/arch/x86/kernel/acpi/boot.c
> +++ b/arch/x86/kernel/acpi/boot.c
> @@ -239,6 +239,12 @@ acpi_parse_lapic(union acpi_subtable_headers * header, const unsigned long end)
>   	if (processor->id == 0xff)
>   		return 0;
>   
> +	/* don't register processors that can not be onlined */
> +	if (!(processor->lapic_flags & ACPI_MADT_ENABLED)) {
> +		if (!(processor->lapic_flags & ACPI_MADT_ONLINE_CAPABLE))
> +			return 0;
> +	}

For firmware using ACPI 6.2 and early versions, the
ACPI_MADT_ONLINE_CAPABLE bit is reserved as zero, so if
we set CPU as disabled, the code here will always return
0 in those firmwares.

> +
>   	/*
>   	 * We need to register disabled CPU as well to permit
>   	 * counting disabled CPUs. This allows us to size

So we will not register the disabled CPU and will break
CPU hotplug features.

I think we need to consider the compatibility with old versions
of firmware.

Thanks
Hanjun